* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 0:54 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-04-28 0:54 UTC (permalink / raw)
To: lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas, linux-pci, jcm
Cc: patches, linux-kernel, linux-arm-kernel, rjw, Khuong Dinh, Duc Dang
From: Khuong Dinh <kdinh@apm.com>
This patch makes pci-xgene-msi driver ACPI-aware and provides
MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
Signed-off-by: Khuong Dinh <kdinh@apm.com>
Signed-off-by: Duc Dang <dhdang@apm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
---
v2:
- Verify with BIOS version 3.06.25 and 3.07.09
v1:
- Initial version
---
drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
1 files changed, 32 insertions(+), 3 deletions(-)
diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
index f1b633b..00aaa3d 100644
--- a/drivers/pci/host/pci-xgene-msi.c
+++ b/drivers/pci/host/pci-xgene-msi.c
@@ -24,6 +24,7 @@
#include <linux/pci.h>
#include <linux/platform_device.h>
#include <linux/of_pci.h>
+#include <linux/acpi.h>
#define MSI_IR0 0x000000
#define MSI_INT0 0x800000
@@ -39,7 +40,7 @@ struct xgene_msi_group {
};
struct xgene_msi {
- struct device_node *node;
+ struct fwnode_handle *fwnode;
struct irq_domain *inner_domain;
struct irq_domain *msi_domain;
u64 msi_addr;
@@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
.free = xgene_irq_domain_free,
};
+#ifdef CONFIG_ACPI
+static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
+{
+ return xgene_msi_ctrl.fwnode;
+}
+#endif
+
static int xgene_allocate_domains(struct xgene_msi *msi)
{
msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
@@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
if (!msi->inner_domain)
return -ENOMEM;
- msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
+ msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
&xgene_msi_domain_info,
msi->inner_domain);
@@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
return -ENOMEM;
}
+#ifdef CONFIG_ACPI
+ pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
+#endif
return 0;
}
@@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
{},
};
+#ifdef CONFIG_ACPI
+static const struct acpi_device_id xgene_msi_acpi_ids[] = {
+ {"APMC0D0E", 0},
+ { },
+};
+#endif
+
static int xgene_msi_probe(struct platform_device *pdev)
{
struct resource *res;
@@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
goto error;
}
xgene_msi->msi_addr = res->start;
- xgene_msi->node = pdev->dev.of_node;
+
+ xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
+ if (!xgene_msi->fwnode) {
+ xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
+ if (!xgene_msi->fwnode) {
+ dev_err(&pdev->dev, "Failed to create fwnode\n");
+ rc = ENOMEM;
+ goto error;
+ }
+ }
+
xgene_msi->num_cpus = num_possible_cpus();
rc = xgene_msi_init_allocator(xgene_msi);
@@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
.driver = {
.name = "xgene-msi",
.of_match_table = xgene_msi_match_table,
+ .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
},
.probe = xgene_msi_probe,
.remove = xgene_msi_remove,
--
1.7.1
^ permalink raw reply related [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 0:54 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-04-28 0:54 UTC (permalink / raw)
To: lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas, linux-pci, jcm
Cc: patches, Duc Dang, rjw, linux-kernel, Khuong Dinh, linux-arm-kernel
From: Khuong Dinh <kdinh@apm.com>
This patch makes pci-xgene-msi driver ACPI-aware and provides
MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
Signed-off-by: Khuong Dinh <kdinh@apm.com>
Signed-off-by: Duc Dang <dhdang@apm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
---
v2:
- Verify with BIOS version 3.06.25 and 3.07.09
v1:
- Initial version
---
drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
1 files changed, 32 insertions(+), 3 deletions(-)
diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
index f1b633b..00aaa3d 100644
--- a/drivers/pci/host/pci-xgene-msi.c
+++ b/drivers/pci/host/pci-xgene-msi.c
@@ -24,6 +24,7 @@
#include <linux/pci.h>
#include <linux/platform_device.h>
#include <linux/of_pci.h>
+#include <linux/acpi.h>
#define MSI_IR0 0x000000
#define MSI_INT0 0x800000
@@ -39,7 +40,7 @@ struct xgene_msi_group {
};
struct xgene_msi {
- struct device_node *node;
+ struct fwnode_handle *fwnode;
struct irq_domain *inner_domain;
struct irq_domain *msi_domain;
u64 msi_addr;
@@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
.free = xgene_irq_domain_free,
};
+#ifdef CONFIG_ACPI
+static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
+{
+ return xgene_msi_ctrl.fwnode;
+}
+#endif
+
static int xgene_allocate_domains(struct xgene_msi *msi)
{
msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
@@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
if (!msi->inner_domain)
return -ENOMEM;
- msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
+ msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
&xgene_msi_domain_info,
msi->inner_domain);
@@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
return -ENOMEM;
}
+#ifdef CONFIG_ACPI
+ pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
+#endif
return 0;
}
@@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
{},
};
+#ifdef CONFIG_ACPI
+static const struct acpi_device_id xgene_msi_acpi_ids[] = {
+ {"APMC0D0E", 0},
+ { },
+};
+#endif
+
static int xgene_msi_probe(struct platform_device *pdev)
{
struct resource *res;
@@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
goto error;
}
xgene_msi->msi_addr = res->start;
- xgene_msi->node = pdev->dev.of_node;
+
+ xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
+ if (!xgene_msi->fwnode) {
+ xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
+ if (!xgene_msi->fwnode) {
+ dev_err(&pdev->dev, "Failed to create fwnode\n");
+ rc = ENOMEM;
+ goto error;
+ }
+ }
+
xgene_msi->num_cpus = num_possible_cpus();
rc = xgene_msi_init_allocator(xgene_msi);
@@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
.driver = {
.name = "xgene-msi",
.of_match_table = xgene_msi_match_table,
+ .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
},
.probe = xgene_msi_probe,
.remove = xgene_msi_remove,
--
1.7.1
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply related [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 0:54 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-04-28 0:54 UTC (permalink / raw)
To: linux-arm-kernel
From: Khuong Dinh <kdinh@apm.com>
This patch makes pci-xgene-msi driver ACPI-aware and provides
MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
Signed-off-by: Khuong Dinh <kdinh@apm.com>
Signed-off-by: Duc Dang <dhdang@apm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
---
v2:
- Verify with BIOS version 3.06.25 and 3.07.09
v1:
- Initial version
---
drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
1 files changed, 32 insertions(+), 3 deletions(-)
diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
index f1b633b..00aaa3d 100644
--- a/drivers/pci/host/pci-xgene-msi.c
+++ b/drivers/pci/host/pci-xgene-msi.c
@@ -24,6 +24,7 @@
#include <linux/pci.h>
#include <linux/platform_device.h>
#include <linux/of_pci.h>
+#include <linux/acpi.h>
#define MSI_IR0 0x000000
#define MSI_INT0 0x800000
@@ -39,7 +40,7 @@ struct xgene_msi_group {
};
struct xgene_msi {
- struct device_node *node;
+ struct fwnode_handle *fwnode;
struct irq_domain *inner_domain;
struct irq_domain *msi_domain;
u64 msi_addr;
@@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
.free = xgene_irq_domain_free,
};
+#ifdef CONFIG_ACPI
+static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
+{
+ return xgene_msi_ctrl.fwnode;
+}
+#endif
+
static int xgene_allocate_domains(struct xgene_msi *msi)
{
msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
@@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
if (!msi->inner_domain)
return -ENOMEM;
- msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
+ msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
&xgene_msi_domain_info,
msi->inner_domain);
@@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
return -ENOMEM;
}
+#ifdef CONFIG_ACPI
+ pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
+#endif
return 0;
}
@@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
{},
};
+#ifdef CONFIG_ACPI
+static const struct acpi_device_id xgene_msi_acpi_ids[] = {
+ {"APMC0D0E", 0},
+ { },
+};
+#endif
+
static int xgene_msi_probe(struct platform_device *pdev)
{
struct resource *res;
@@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
goto error;
}
xgene_msi->msi_addr = res->start;
- xgene_msi->node = pdev->dev.of_node;
+
+ xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
+ if (!xgene_msi->fwnode) {
+ xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
+ if (!xgene_msi->fwnode) {
+ dev_err(&pdev->dev, "Failed to create fwnode\n");
+ rc = ENOMEM;
+ goto error;
+ }
+ }
+
xgene_msi->num_cpus = num_possible_cpus();
rc = xgene_msi_init_allocator(xgene_msi);
@@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
.driver = {
.name = "xgene-msi",
.of_match_table = xgene_msi_match_table,
+ .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
},
.probe = xgene_msi_probe,
.remove = xgene_msi_remove,
--
1.7.1
^ permalink raw reply related [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-04-28 0:54 ` Khuong Dinh
(?)
@ 2017-04-28 5:11 ` Jon Masters
-1 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-04-28 5:11 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas,
linux-pci
Cc: patches, linux-kernel, linux-arm-kernel, rjw, Duc Dang
On 04/27/2017 08:54 PM, Khuong Dinh wrote:
> From: Khuong Dinh <kdinh@apm.com>
>
> This patch makes pci-xgene-msi driver ACPI-aware and provides
> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>
> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Thanks. Currently testing. Will review and followup shortly.
Jon.
--
Computer Architect | Sent from my Fedora powered laptop
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 5:11 ` Jon Masters
0 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-04-28 5:11 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas,
linux-pci
Cc: rjw, Duc Dang, patches, linux-kernel, linux-arm-kernel
On 04/27/2017 08:54 PM, Khuong Dinh wrote:
> From: Khuong Dinh <kdinh@apm.com>
>
> This patch makes pci-xgene-msi driver ACPI-aware and provides
> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>
> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Thanks. Currently testing. Will review and followup shortly.
Jon.
--
Computer Architect | Sent from my Fedora powered laptop
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 5:11 ` Jon Masters
0 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-04-28 5:11 UTC (permalink / raw)
To: linux-arm-kernel
On 04/27/2017 08:54 PM, Khuong Dinh wrote:
> From: Khuong Dinh <kdinh@apm.com>
>
> This patch makes pci-xgene-msi driver ACPI-aware and provides
> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>
> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Thanks. Currently testing. Will review and followup shortly.
Jon.
--
Computer Architect | Sent from my Fedora powered laptop
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-04-28 0:54 ` Khuong Dinh
(?)
@ 2017-04-28 9:27 ` Marc Zyngier
-1 siblings, 0 replies; 47+ messages in thread
From: Marc Zyngier @ 2017-04-28 9:27 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, msalter, bhelgaas, linux-pci, jcm
Cc: patches, linux-kernel, linux-arm-kernel, rjw, Duc Dang
On 28/04/17 01:54, Khuong Dinh wrote:
> From: Khuong Dinh <kdinh@apm.com>
>
> This patch makes pci-xgene-msi driver ACPI-aware and provides
> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>
> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> ---
> v2:
> - Verify with BIOS version 3.06.25 and 3.07.09
> v1:
> - Initial version
> ---
> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> 1 files changed, 32 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> index f1b633b..00aaa3d 100644
> --- a/drivers/pci/host/pci-xgene-msi.c
> +++ b/drivers/pci/host/pci-xgene-msi.c
> @@ -24,6 +24,7 @@
> #include <linux/pci.h>
> #include <linux/platform_device.h>
> #include <linux/of_pci.h>
> +#include <linux/acpi.h>
>
> #define MSI_IR0 0x000000
> #define MSI_INT0 0x800000
> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> };
>
> struct xgene_msi {
> - struct device_node *node;
> + struct fwnode_handle *fwnode;
> struct irq_domain *inner_domain;
> struct irq_domain *msi_domain;
> u64 msi_addr;
> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> .free = xgene_irq_domain_free,
> };
>
> +#ifdef CONFIG_ACPI
> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> +{
> + return xgene_msi_ctrl.fwnode;
> +}
> +#endif
> +
> static int xgene_allocate_domains(struct xgene_msi *msi)
> {
> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> if (!msi->inner_domain)
> return -ENOMEM;
>
> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> &xgene_msi_domain_info,
> msi->inner_domain);
>
> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> return -ENOMEM;
> }
>
> +#ifdef CONFIG_ACPI
> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> +#endif
> return 0;
> }
>
> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> {},
> };
>
> +#ifdef CONFIG_ACPI
> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> + {"APMC0D0E", 0},
> + { },
> +};
> +#endif
> +
> static int xgene_msi_probe(struct platform_device *pdev)
> {
> struct resource *res;
> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> goto error;
> }
> xgene_msi->msi_addr = res->start;
> - xgene_msi->node = pdev->dev.of_node;
> +
> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> + if (!xgene_msi->fwnode) {
> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
Please provide something other than NULL, such as the base address if
the device. That's quite useful for debugging.
> + if (!xgene_msi->fwnode) {
> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> + rc = ENOMEM;
> + goto error;
> + }
> + }
> +
> xgene_msi->num_cpus = num_possible_cpus();
>
> rc = xgene_msi_init_allocator(xgene_msi);
> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> .driver = {
> .name = "xgene-msi",
> .of_match_table = xgene_msi_match_table,
> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> },
> .probe = xgene_msi_probe,
> .remove = xgene_msi_remove,
>
The code is trivial, but relies on the MSI controller to probe before
the PCI bus. What enforces this? How is it making sure that this is not
going to break in the next kernel release? As far as I can tell, there
is no explicit dependency between the two, making it the whole thing
extremely fragile.
Thanks,
M.
--
Jazz is not dead. It just smells funny...
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 9:27 ` Marc Zyngier
0 siblings, 0 replies; 47+ messages in thread
From: Marc Zyngier @ 2017-04-28 9:27 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, msalter, bhelgaas, linux-pci, jcm
Cc: rjw, Duc Dang, patches, linux-kernel, linux-arm-kernel
On 28/04/17 01:54, Khuong Dinh wrote:
> From: Khuong Dinh <kdinh@apm.com>
>
> This patch makes pci-xgene-msi driver ACPI-aware and provides
> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>
> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> ---
> v2:
> - Verify with BIOS version 3.06.25 and 3.07.09
> v1:
> - Initial version
> ---
> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> 1 files changed, 32 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> index f1b633b..00aaa3d 100644
> --- a/drivers/pci/host/pci-xgene-msi.c
> +++ b/drivers/pci/host/pci-xgene-msi.c
> @@ -24,6 +24,7 @@
> #include <linux/pci.h>
> #include <linux/platform_device.h>
> #include <linux/of_pci.h>
> +#include <linux/acpi.h>
>
> #define MSI_IR0 0x000000
> #define MSI_INT0 0x800000
> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> };
>
> struct xgene_msi {
> - struct device_node *node;
> + struct fwnode_handle *fwnode;
> struct irq_domain *inner_domain;
> struct irq_domain *msi_domain;
> u64 msi_addr;
> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> .free = xgene_irq_domain_free,
> };
>
> +#ifdef CONFIG_ACPI
> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> +{
> + return xgene_msi_ctrl.fwnode;
> +}
> +#endif
> +
> static int xgene_allocate_domains(struct xgene_msi *msi)
> {
> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> if (!msi->inner_domain)
> return -ENOMEM;
>
> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> &xgene_msi_domain_info,
> msi->inner_domain);
>
> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> return -ENOMEM;
> }
>
> +#ifdef CONFIG_ACPI
> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> +#endif
> return 0;
> }
>
> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> {},
> };
>
> +#ifdef CONFIG_ACPI
> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> + {"APMC0D0E", 0},
> + { },
> +};
> +#endif
> +
> static int xgene_msi_probe(struct platform_device *pdev)
> {
> struct resource *res;
> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> goto error;
> }
> xgene_msi->msi_addr = res->start;
> - xgene_msi->node = pdev->dev.of_node;
> +
> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> + if (!xgene_msi->fwnode) {
> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
Please provide something other than NULL, such as the base address if
the device. That's quite useful for debugging.
> + if (!xgene_msi->fwnode) {
> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> + rc = ENOMEM;
> + goto error;
> + }
> + }
> +
> xgene_msi->num_cpus = num_possible_cpus();
>
> rc = xgene_msi_init_allocator(xgene_msi);
> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> .driver = {
> .name = "xgene-msi",
> .of_match_table = xgene_msi_match_table,
> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> },
> .probe = xgene_msi_probe,
> .remove = xgene_msi_remove,
>
The code is trivial, but relies on the MSI controller to probe before
the PCI bus. What enforces this? How is it making sure that this is not
going to break in the next kernel release? As far as I can tell, there
is no explicit dependency between the two, making it the whole thing
extremely fragile.
Thanks,
M.
--
Jazz is not dead. It just smells funny...
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 9:27 ` Marc Zyngier
0 siblings, 0 replies; 47+ messages in thread
From: Marc Zyngier @ 2017-04-28 9:27 UTC (permalink / raw)
To: linux-arm-kernel
On 28/04/17 01:54, Khuong Dinh wrote:
> From: Khuong Dinh <kdinh@apm.com>
>
> This patch makes pci-xgene-msi driver ACPI-aware and provides
> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>
> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> ---
> v2:
> - Verify with BIOS version 3.06.25 and 3.07.09
> v1:
> - Initial version
> ---
> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> 1 files changed, 32 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> index f1b633b..00aaa3d 100644
> --- a/drivers/pci/host/pci-xgene-msi.c
> +++ b/drivers/pci/host/pci-xgene-msi.c
> @@ -24,6 +24,7 @@
> #include <linux/pci.h>
> #include <linux/platform_device.h>
> #include <linux/of_pci.h>
> +#include <linux/acpi.h>
>
> #define MSI_IR0 0x000000
> #define MSI_INT0 0x800000
> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> };
>
> struct xgene_msi {
> - struct device_node *node;
> + struct fwnode_handle *fwnode;
> struct irq_domain *inner_domain;
> struct irq_domain *msi_domain;
> u64 msi_addr;
> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> .free = xgene_irq_domain_free,
> };
>
> +#ifdef CONFIG_ACPI
> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> +{
> + return xgene_msi_ctrl.fwnode;
> +}
> +#endif
> +
> static int xgene_allocate_domains(struct xgene_msi *msi)
> {
> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> if (!msi->inner_domain)
> return -ENOMEM;
>
> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> &xgene_msi_domain_info,
> msi->inner_domain);
>
> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> return -ENOMEM;
> }
>
> +#ifdef CONFIG_ACPI
> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> +#endif
> return 0;
> }
>
> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> {},
> };
>
> +#ifdef CONFIG_ACPI
> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> + {"APMC0D0E", 0},
> + { },
> +};
> +#endif
> +
> static int xgene_msi_probe(struct platform_device *pdev)
> {
> struct resource *res;
> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> goto error;
> }
> xgene_msi->msi_addr = res->start;
> - xgene_msi->node = pdev->dev.of_node;
> +
> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> + if (!xgene_msi->fwnode) {
> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
Please provide something other than NULL, such as the base address if
the device. That's quite useful for debugging.
> + if (!xgene_msi->fwnode) {
> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> + rc = ENOMEM;
> + goto error;
> + }
> + }
> +
> xgene_msi->num_cpus = num_possible_cpus();
>
> rc = xgene_msi_init_allocator(xgene_msi);
> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> .driver = {
> .name = "xgene-msi",
> .of_match_table = xgene_msi_match_table,
> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> },
> .probe = xgene_msi_probe,
> .remove = xgene_msi_remove,
>
The code is trivial, but relies on the MSI controller to probe before
the PCI bus. What enforces this? How is it making sure that this is not
going to break in the next kernel release? As far as I can tell, there
is no explicit dependency between the two, making it the whole thing
extremely fragile.
Thanks,
M.
--
Jazz is not dead. It just smells funny...
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-04-28 5:11 ` Jon Masters
(?)
@ 2017-04-28 16:36 ` Jon Masters
-1 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-04-28 16:36 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas,
linux-pci
Cc: patches, linux-kernel, linux-arm-kernel, rjw, Duc Dang, Jon Masters
[-- Attachment #1: Type: text/plain, Size: 995 bytes --]
On 04/28/2017 01:11 AM, Jon Masters wrote:
> On 04/27/2017 08:54 PM, Khuong Dinh wrote:
>> From: Khuong Dinh <kdinh@apm.com>
>>
>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>
>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> Signed-off-by: Duc Dang <dhdang@apm.com>
>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>
> Thanks. Currently testing. Will review and followup shortly.
Attaching boot logs before and after. I tested with an Intel e1000.
Before the patch I get fallback interrupts. Afterward, I get X-Gene MSI
interrupts working. Please followup with answers to Marc's question
about how you propose to guarantee boottime ordering of binding.
I urgently require that AppliedMicro complete upstreaming of this. It
should have been upstream a year ago. You're not the only ones with
patches like this left around - I'm coming for everyone else, too.
Tested-by: Jon Masters <jcm@jonmasters.org>
[-- Attachment #2: dmesg_4.11.0-rc8.xgene_msix1+_nomsi.txt --]
[-- Type: text/plain, Size: 39241 bytes --]
[ 0.000000] Booting Linux on physical CPU 0x0
[ 0.000000] Linux version 4.11.0-rc8.xgene_msix1+ (root@mustang4.bos.jonmasters.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-14) (GCC) ) #1 SMP Fri Apr 28 04:44:20 EDT 2017
[ 0.000000] Boot CPU: AArch64 Processor [500f0001]
[ 0.000000] earlycon: uart8250 at MMIO32 0x000000001c020000 (options '')
[ 0.000000] bootconsole [uart8250] enabled
[ 0.000000] efi: Getting EFI parameters from FDT:
[ 0.000000] efi: EFI v2.40 by X-Gene Mustang Board EFI Aug 19 2016 18:02:24
[ 0.000000] efi: ACPI=0x43fa82d000 ACPI 2.0=0x43fa82d014 SMBIOS 3.0=0x43fa82b000 ESRT=0x43ff053d18
[ 0.000000] esrt: Reserving ESRT space from 0x00000043ff053d18 to 0x00000043ff053d78.
[ 0.000000] cma: Reserved 512 MiB at 0x00000040e0000000
[ 0.000000] ACPI: Early table checksum verification disabled
[ 0.000000] ACPI: RSDP 0x00000043FA82D014 000024 (v02 APM )
[ 0.000000] ACPI: XSDT 0x00000043FA82C0E8 000084 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: FACP 0x00000043FA822000 00010C (v05 APM XGENE 00000003 INTL 20140724)
[ 0.000000] ACPI: DSDT 0x00000043FA823000 005837 (v05 APM APM88xxx 00000001 INTL 20140724)
[ 0.000000] ACPI: DBG2 0x00000043FA829000 0000AA (v00 APMC0D XGENEDBG 00000000 INTL 20140724)
[ 0.000000] ACPI: GTDT 0x00000043FA820000 000060 (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: MCFG 0x00000043FA81F000 00003C (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: SPCR 0x00000043FA81E000 000050 (v02 APMC0D XGENESPC 00000000 INTL 20140724)
[ 0.000000] ACPI: SSDT 0x00000043FA81D000 00002D (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: BERT 0x00000043FA81C000 000030 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: HEST 0x00000043FA81B000 0002A8 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: APIC 0x00000043FA81A000 0002A4 (v03 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA819000 000063 (v02 REDHAT MACADDRS 00000001 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA818000 000032 (v02 REDHAT UARTCLKS 00000001 01000013)
[ 0.000000] ACPI: PCCT 0x00000043FA817000 000300 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SPCR: console: uart,mmio,0x1c020000,115200
[ 0.000000] ACPI: NUMA: Failed to initialise from firmware
[ 0.000000] NUMA: Faking a node at [mem 0x0000000000000000-0x00000043ffffffff]
[ 0.000000] NUMA: Adding memblock [0x4000000000 - 0x40001fffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x4000200000 - 0x43fa7cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43fa7d0000 - 0x43faa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43faa20000 - 0x43ff9cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ff9d0000 - 0x43ffa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ffa20000 - 0x43ffffffff] on node 0
[ 0.000000] NUMA: Initmem setup node 0 [mem 0x4000000000-0x43ffffffff]
[ 0.000000] NUMA: NODE_DATA [mem 0x43ffffe200-0x43ffffffff]
[ 0.000000] Zone ranges:
[ 0.000000] DMA [mem 0x0000004000000000-0x00000040ffffffff]
[ 0.000000] Normal [mem 0x0000004100000000-0x00000043ffffffff]
[ 0.000000] Movable zone start for each node
[ 0.000000] Early memory node ranges
[ 0.000000] node 0: [mem 0x0000004000000000-0x00000040001fffff]
[ 0.000000] node 0: [mem 0x0000004000200000-0x00000043fa7cffff]
[ 0.000000] node 0: [mem 0x00000043fa7d0000-0x00000043faa1ffff]
[ 0.000000] node 0: [mem 0x00000043faa20000-0x00000043ff9cffff]
[ 0.000000] node 0: [mem 0x00000043ff9d0000-0x00000043ffa1ffff]
[ 0.000000] node 0: [mem 0x00000043ffa20000-0x00000043ffffffff]
[ 0.000000] Initmem setup node 0 [mem 0x0000004000000000-0x00000043ffffffff]
[ 0.000000] On node 0 totalpages: 262144
[ 0.000000] DMA zone: 64 pages used for memmap
[ 0.000000] DMA zone: 0 pages reserved
[ 0.000000] DMA zone: 65536 pages, LIFO batch:1
[ 0.000000] Normal zone: 192 pages used for memmap
[ 0.000000] Normal zone: 196608 pages, LIFO batch:1
[ 0.000000] psci: is not implemented in ACPI.
[ 0.000000] percpu: Embedded 3 pages/cpu @ffff8003ffdf0000 s124312 r8192 d64104 u196608
[ 0.000000] pcpu-alloc: s124312 r8192 d64104 u196608 alloc=3*65536
[ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3 [0] 4 [0] 5 [0] 6 [0] 7
[ 0.000000] Detected PIPT I-cache on CPU0
[ 0.000000] Built 1 zonelists in Node order, mobility grouping on. Total pages: 261888
[ 0.000000] Policy zone: Normal
[ 0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.11.0-rc8.xgene_msix1+ root=/dev/mapper/rhelp_dhcp--171-root ro crashkernel=auto rd.lvm.lv=rhelp_dhcp-171/root rd.lvm.lv=rhelp_dhcp-171/swap LANG=en_US.UTF-8 acpi=force earlycon=uart8250,mmio32,0x1c020000 console=ttyS0,115200
[ 0.000000] PID hash table entries: 4096 (order: -1, 32768 bytes)
[ 0.000000] software IO TLB [mem 0x40dbff0000-0x40dfff0000] (64MB) mapped at [ffff8000dbff0000-ffff8000dffeffff]
[ 0.000000] Memory: 16109888K/16777216K available (7420K kernel code, 1354K rwdata, 3200K rodata, 1216K init, 7286K bss, 143040K reserved, 524288K cma-reserved)
[ 0.000000] Virtual kernel memory layout:
[ 0.000000] modules : 0xffff000000000000 - 0xffff000008000000 ( 128 MB)
[ 0.000000] vmalloc : 0xffff000008000000 - 0xffff7bdfffff0000 (126847 GB)
[ 0.000000] .text : 0xffff000008080000 - 0xffff0000087c0000 ( 7424 KB)
[ 0.000000] .rodata : 0xffff0000087c0000 - 0xffff000008af0000 ( 3264 KB)
[ 0.000000] .init : 0xffff000008af0000 - 0xffff000008c20000 ( 1216 KB)
[ 0.000000] .data : 0xffff000008c20000 - 0xffff000008d72a00 ( 1355 KB)
[ 0.000000] .bss : 0xffff000008d72a00 - 0xffff0000094902d4 ( 7287 KB)
[ 0.000000] fixed : 0xffff7fdffe7d0000 - 0xffff7fdffec00000 ( 4288 KB)
[ 0.000000] PCI I/O : 0xffff7fdffee00000 - 0xffff7fdfffe00000 ( 16 MB)
[ 0.000000] vmemmap : 0xffff7fe000000000 - 0xffff800000000000 ( 128 GB maximum)
[ 0.000000] 0xffff7fe000000000 - 0xffff7fe001000000 ( 16 MB actual)
[ 0.000000] memory : 0xffff800000000000 - 0xffff800400000000 ( 16384 MB)
[ 0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=8, Nodes=1
[ 0.000000] Hierarchical RCU implementation.
[ 0.000000] Build-time adjustment of leaf fanout to 64.
[ 0.000000] RCU restricting CPUs from NR_CPUS=4096 to nr_cpu_ids=8.
[ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=8
[ 0.000000] NR_IRQS:64 nr_irqs:64 0
[ 0.000000] GIC: Using split EOI/Deactivate mode
[ 0.000000] GICv3: No distributor detected at @ffff000008050000, giving up
[ 0.000000] arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys).
[ 0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns
[ 0.000002] sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns
[ 0.008382] Console: colour dummy device 80x25
[ 0.013031] Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000)
[ 0.023774] pid_max: default: 32768 minimum: 301
[ 0.028553] ACPI: Core revision 20170119
[ 0.036208] ACPI Error: Method parse/execution failed [\_SB.ET00._STA] (Node ffff8003c0195c08), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.049249] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.057204] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.063972] ACPI Error: Method parse/execution failed [\_SB.ET01._STA] (Node ffff8003c0195eb0), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.077125] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.084966] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.092086] ACPI: 4 ACPI AML tables successfully acquired and loaded
[ 0.098656] Security Framework initialized
[ 0.102953] SELinux: Initializing.
[ 0.106548] SELinux: Starting in permissive mode
[ 0.106747] Dentry cache hash table entries: 2097152 (order: 8, 16777216 bytes)
[ 0.116984] Inode-cache hash table entries: 1048576 (order: 7, 8388608 bytes)
[ 0.125658] Mount-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.132758] Mountpoint-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.140711] ftrace: allocating 26814 entries in 7 pages
[ 0.205140] ASID allocator initialised with 65536 entries
[ 0.211274] Remapping and enabling EFI services.
[ 0.216029] EFI remap 0x0000000010510000 => 0000000020000000
[ 0.222058] EFI remap 0x0000000010548000 => 0000000020018000
[ 0.228081] EFI remap 0x0000000017000000 => 0000000020020000
[ 0.234066] EFI remap 0x000000001c025000 => 0000000020035000
[ 0.240084] UEFI Runtime regions are not aligned to 64 KB -- buggy firmware?
[ 0.240089] ------------[ cut here ]------------
[ 0.252134] WARNING: CPU: 0 PID: 1 at arch/arm64/kernel/efi.c:34 efi_create_mapping+0x4c/0x104
[ 0.261010] Modules linked in:
[ 0.265699] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc8.xgene_msix1+ #1
[ 0.273313] Hardware name: APM X-Gene Mustang board (DT)
[ 0.278750] task: ffff8003c1bc0000 task.stack: ffff8003c1c00000
[ 0.284827] PC is at efi_create_mapping+0x4c/0x104
[ 0.289772] LR is at efi_create_mapping+0x4c/0x104
[ 0.294664] pc : [<ffff000008af51f0>] lr : [<ffff000008af51f0>] pstate: 60000045
[ 0.302261] sp : ffff8003c1c03d20
[ 0.305710] x29: ffff8003c1c03d20 x28: 0000000000000000
[ 0.311183] x27: ffff000008c53000 x26: ffff000008d429c8
[ 0.316639] x25: ffff000008d42000 x24: 0000000000000000
[ 0.322102] x23: ffff000008a4c8f0 x22: ffff000008c60000
[ 0.327600] x21: ffff000008c53000 x20: ffff000008d429c8
[ 0.333038] x19: ffff8003f96ab348 x18: 000000003eca74e7
[ 0.338485] x17: 0000000000000000 x16: 0000000000000000
[ 0.343992] x15: ffff000008c53000 x14: 000000004e2b17ac
[ 0.349464] x13: ffff0000089ca500 x12: 0000000000040b00
[ 0.354988] x11: ffff8003c1c039e0 x10: ffff8003c1c039e0
[ 0.360435] x9 : 0000000000000073 x8 : 6620796767756220
[ 0.365898] x7 : 2d2d20424b203436 x6 : ffff000009274718
[ 0.371362] x5 : ffff8003ffdf7530 x4 : 0000000000000000
[ 0.376834] x3 : 0000000000000000 x2 : 0000000000040b00
[ 0.382316] x1 : 0000000000040b00 x0 : 0000000000000040
[ 0.389363] ---[ end trace 4a6935def63e1e16 ]---
[ 0.394126] Call trace:
[ 0.396626] Exception stack(0xffff8003c1c03b30 to 0xffff8003c1c03c60)
[ 0.403246] 3b20: ffff8003f96ab348 0001000000000000
[ 0.411302] 3b40: ffff8003c1c03d20 ffff000008af51f0 0000000060000045 ffff000008d429c8
[ 0.419331] 3b60: 0000000000000000 0000000000000000 ffff8003c1c03d20 ffff8003c1c03d20
[ 0.427377] 3b80: ffff8003c1c03ce0 00000000ffffffc8 ffff8003c1c03be0 ffff00000812f6b8
[ 0.435449] 3ba0: ffff000008c53000 00000000ffffffc8 ffff8003c1c03c00 ffff0000089be038
[ 0.443487] 3bc0: ffff8003c1c03d20 ffff8003c1c03d20 ffff8003c1c03ce0 00000000ffffffc8
[ 0.451585] 3be0: ffff8003c1c03c70 0000000000040b00 0000000000000040 0000000000040b00
[ 0.459666] 3c00: 0000000000040b00 0000000000000000 0000000000000000 ffff8003ffdf7530
[ 0.467729] 3c20: ffff000009274718 2d2d20424b203436 6620796767756220 0000000000000073
[ 0.475784] 3c40: ffff8003c1c039e0 ffff8003c1c039e0 0000000000040b00 ffff0000089ca500
[ 0.483840] [<ffff000008af51f0>] efi_create_mapping+0x4c/0x104
[ 0.489858] [<ffff000008b3db98>] arm_enable_runtime_services+0x154/0x238
[ 0.496817] [<ffff000008083698>] do_one_initcall+0x54/0x15c
[ 0.502634] [<ffff000008af0d18>] kernel_init_freeable+0xe4/0x25c
[ 0.508816] [<ffff0000087ae430>] kernel_init+0x18/0x108
[ 0.514202] [<ffff0000080833c0>] ret_from_fork+0x10/0x50
[ 0.519621] EFI remap 0x00000043fa7df000 => 000000002004f000
[ 0.525659] EFI remap 0x00000043fa82e000 => 000000002006e000
[ 0.531673] EFI remap 0x00000043ff9d6000 => 0000000020266000
[ 0.537700] EFI remap 0x00000043ff9ea000 => 000000002027a000
[ 0.543899] smp: Bringing up secondary CPUs ...
[ 0.548738] Detected PIPT I-cache on CPU1
[ 0.548766] CPU1: Booted secondary processor [500f0001]
[ 0.548987] Detected PIPT I-cache on CPU2
[ 0.549008] CPU2: Booted secondary processor [500f0001]
[ 0.549225] Detected PIPT I-cache on CPU3
[ 0.549239] CPU3: Booted secondary processor [500f0001]
[ 0.549432] Detected PIPT I-cache on CPU4
[ 0.549452] CPU4: Booted secondary processor [500f0001]
[ 0.549653] Detected PIPT I-cache on CPU5
[ 0.549667] CPU5: Booted secondary processor [500f0001]
[ 0.549861] Detected PIPT I-cache on CPU6
[ 0.549882] CPU6: Booted secondary processor [500f0001]
[ 0.550086] Detected PIPT I-cache on CPU7
[ 0.550100] CPU7: Booted secondary processor [500f0001]
[ 0.550142] smp: Brought up 1 node, 8 CPUs
[ 0.621892] SMP: Total of 8 processors activated.
[ 0.626806] CPU features: detected feature: 32-bit EL0 Support
[ 0.632944] CPU: All CPU(s) started at EL2
[ 0.637984] devtmpfs: initialized
[ 0.641749] SMBIOS 3.0.0 present.
[ 0.645224] DMI: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.15 Aug 19 2016
[ 0.654642] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[ 0.664967] futex hash table entries: 2048 (order: 2, 262144 bytes)
[ 0.671981] pinctrl core: initialized pinctrl subsystem
[ 0.677705] NET: Registered protocol family 16
[ 0.682714] cpuidle: using governor menu
[ 0.686960] Detected 8 PCC Subspaces
[ 0.690784] Registering PCC driver as Mailbox controller
[ 0.696410] vdso: 2 pages (1 code @ ffff0000087e0000, 1 data @ ffff000008c40000)
[ 0.704152] hw-breakpoint: found 4 breakpoint and 4 watchpoint registers.
[ 0.713765] DMA: preallocated 256 KiB pool for atomic allocations
[ 0.720154] ACPI: bus type PCI registered
[ 0.724382] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
[ 0.731202] Serial: AMBA PL011 UART driver
[ 0.737465] HugeTLB registered 512 MB page size, pre-allocated 0 pages
[ 0.744898] ACPI: Added _OSI(Module Device)
[ 0.749274] ACPI: Added _OSI(Processor Device)
[ 0.753882] ACPI: Added _OSI(3.0 _SCP Extensions)
[ 0.758804] ACPI: Added _OSI(Processor Aggregator Device)
[ 0.773904] ACPI: Interpreter enabled
[ 0.777753] ACPI: Using GIC for interrupt routing
[ 0.782685] ACPI: MCFG table detected, 1 entries
[ 0.787684] HEST: Table parsing has been initialized.
[ 0.795411] ACPI: Power Resource [SCVR] (on)
[ 0.830232] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff])
[ 0.836749] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.845539] acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug PME AER PCIeCapability]
[ 0.853915] acpi PNP0A08:00: MCFG quirk: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] with xgene_v1_pcie_ecam_ops
[ 0.866664] acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0xe0d0000000-0xe0dfffffff] not reserved in ACPI namespace
[ 0.877881] acpi PNP0A08:00: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff]
[ 0.886116] Remapped I/O 0x000000e010000000 to [io 0x0000-0xffff window]
[ 0.893327] PCI host bridge to bus 0000:00
[ 0.897628] pci_bus 0000:00: root bus resource [io 0x0000-0xffff window] (bus address [0x10000000-0x1000ffff])
[ 0.908283] pci_bus 0000:00: root bus resource [mem 0xe040000000-0xe07fffffff window] (bus address [0x40000000-0x7fffffff])
[ 0.919920] pci_bus 0000:00: root bus resource [mem 0xf000000000-0xffffffffff window]
[ 0.928114] pci_bus 0000:00: root bus resource [bus 00-ff]
[ 0.933821] pci 0000:00:00.0: [10e8:e004] type 01 class 0x060400
[ 0.933899] pci 0000:00:00.0: supports D1 D2
[ 0.934059] pci 0000:01:00.0: [8086:10d3] type 00 class 0x020000
[ 0.934092] pci 0000:01:00.0: reg 0x10: [mem 0xe040080000-0xe04009ffff]
[ 0.934108] pci 0000:01:00.0: reg 0x14: [mem 0xe040000000-0xe04007ffff]
[ 0.934125] pci 0000:01:00.0: reg 0x18: [io 0x0000-0x001f]
[ 0.934141] pci 0000:01:00.0: reg 0x1c: [mem 0xe0400a0000-0xe0400a3fff]
[ 0.934186] pci 0000:01:00.0: reg 0x30: [mem 0xfffc0000-0xffffffff pref]
[ 0.934326] pci 0000:01:00.0: PME# supported from D0 D3hot D3cold
[ 0.963877] pci 0000:00:00.0: BAR 14: assigned [mem 0xe040000000-0xe0400fffff]
[ 0.971433] pci 0000:00:00.0: BAR 13: assigned [io 0x1000-0x1fff]
[ 0.977908] pci 0000:01:00.0: BAR 1: assigned [mem 0xe040000000-0xe04007ffff]
[ 0.985417] pci 0000:01:00.0: BAR 6: assigned [mem 0xe040080000-0xe0400bffff pref]
[ 0.993304] pci 0000:01:00.0: BAR 0: assigned [mem 0xe0400c0000-0xe0400dffff]
[ 1.000812] pci 0000:01:00.0: BAR 3: assigned [mem 0xe0400e0000-0xe0400e3fff]
[ 1.008298] pci 0000:01:00.0: BAR 2: assigned [io 0x1000-0x101f]
[ 1.014701] pci 0000:00:00.0: PCI bridge to [bus 01]
[ 1.019893] pci 0000:00:00.0: bridge window [io 0x1000-0x1fff]
[ 1.026276] pci 0000:00:00.0: bridge window [mem 0xe040000000-0xe0400fffff]
[ 1.034419] vgaarb: loaded
[ 1.037393] SCSI subsystem initialized
[ 1.041366] libata version 3.00 loaded.
[ 1.041403] ACPI: bus type USB registered
[ 1.045616] usbcore: registered new interface driver usbfs
[ 1.051373] usbcore: registered new interface driver hub
[ 1.056947] usbcore: registered new device driver usb
[ 1.062355] Registered efivars operations
[ 1.066820] NetLabel: Initializing
[ 1.070418] NetLabel: domain hash size = 128
[ 1.074997] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO
[ 1.080980] NetLabel: unlabeled traffic allowed by default
[ 1.086945] clocksource: Switched to clocksource arch_sys_counter
[ 1.111811] VFS: Disk quotas dquot_6.6.0
[ 1.115980] VFS: Dquot-cache hash table entries: 8192 (order 0, 65536 bytes)
[ 1.123569] pnp: PnP ACPI init
[ 1.127405] pnp: PnP ACPI: found 0 devices
[ 1.134249] NET: Registered protocol family 2
[ 1.139069] TCP established hash table entries: 131072 (order: 4, 1048576 bytes)
[ 1.147156] TCP bind hash table entries: 65536 (order: 4, 1048576 bytes)
[ 1.154513] TCP: Hash tables configured (established 131072 bind 65536)
[ 1.161541] UDP hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.168029] UDP-Lite hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.175030] NET: Registered protocol family 1
[ 1.179607] PCI: CLS 64 bytes, default 128
[ 1.179708] Unpacking initramfs...
[ 1.863109] Freeing initrd memory: 32640K
[ 1.867699] kvm [1]: 8-bit VMID
[ 1.870944] kvm [1]: IDMAP page: 40009b7000
[ 1.875301] kvm [1]: HYP VA range: 800000000000:ffffffffffff
[ 1.881276] kvm [1]: Hyp mode initialized successfully
[ 1.886624] kvm [1]: GICV region size/alignment is unsafe, using trapping (reduced performance)
[ 1.895748] kvm [1]: vgic-v2@780cf000
[ 1.899646] kvm [1]: vgic interrupt IRQ1
[ 1.903757] kvm [1]: virtual timer IRQ4
[ 1.923421] audit: initializing netlink subsys (disabled)
[ 1.929159] audit: type=2000 audit(1.470:1): state=initialized audit_enabled=0 res=1
[ 1.929303] Initialise system trusted keyrings
[ 1.929373] workingset: timestamp_bits=44 max_order=18 bucket_order=0
[ 1.931445] zbud: loaded
[ 1.931829] SELinux: Registering netfilter hooks
[ 2.011199] NET: Registered protocol family 38
[ 2.015847] Key type asymmetric registered
[ 2.020145] Asymmetric key parser 'x509' registered
[ 2.025283] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
[ 2.033081] io scheduler noop registered
[ 2.037167] io scheduler deadline registered (default)
[ 2.042536] io scheduler cfq registered
[ 2.046565] io scheduler mq-deadline registered
[ 2.051474] atomic64_test: passed
[ 2.055189] xgene-gpio APMC0D14:00: X-Gene GPIO driver registered.
[ 2.062003] pcieport 0000:00:00.0: AER enabled with IRQ 83
[ 2.067787] pcieport 0000:00:00.0: Signaling PME with IRQ 83
[ 2.073972] input: Power Button as /devices/LNXSYSTM:00/PNP0C0C:00/input/input0
[ 2.081619] ACPI: Power Button [PWRB]
[ 2.085922] GHES: Failed to map GSI to IRQ for generic hardware error source: 2
[ 2.093593] GHES: probe of GHES.2 failed with error 82
[ 2.098936] GHES: Failed to map GSI to IRQ for generic hardware error source: 3
[ 2.106588] GHES: probe of GHES.3 failed with error 82
[ 2.111936] GHES: Failed to map GSI to IRQ for generic hardware error source: 4
[ 2.119608] GHES: probe of GHES.4 failed with error 82
[ 2.125011] GHES: Failed to map GSI to IRQ for generic hardware error source: 5
[ 2.132659] GHES: probe of GHES.5 failed with error 82
[ 2.138017] GHES: Failed to map GSI to IRQ for generic hardware error source: 6
[ 2.145633] GHES: probe of GHES.6 failed with error 82
[ 2.151042] GHES: Failed to map GSI to IRQ for generic hardware error source: 7
[ 2.158724] GHES: probe of GHES.7 failed with error 82
[ 2.164126] GHES: Failed to map GSI to IRQ for generic hardware error source: 8
[ 2.171797] GHES: probe of GHES.8 failed with error 82
[ 2.177160] GHES: Failed to map GSI to IRQ for generic hardware error source: 9
[ 2.184845] GHES: probe of GHES.9 failed with error 82
[ 2.190233] GHES: Failed to enable APEI firmware first mode.
[ 2.196297] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[ 2.203376] console [ttyS0] disabled
[ 2.227246] APMC0D08:00: ttyS0 at MMIO 0x1c020000 (irq = 22, base_baud = 3125000) is a U6_16550A
[ 2.236527] console [ttyS0] enabled
[ 2.243657] bootconsole [uart8250] disabled
[ 2.272506] APMC0D08:01: ttyS1 at MMIO 0x1c021000 (irq = 23, base_baud = 3125000) is a U6_16550A
[ 2.281671] cacheinfo: Unable to detect cache hierarchy for CPU 0
[ 2.287755] rdac: device handler registered
[ 2.291963] hp_sw: device handler registered
[ 2.296209] emc: device handler registered
[ 2.300380] alua: device handler registered
[ 2.304688] libphy: Fixed MDIO Bus: probed
[ 2.308849] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
[ 2.315346] ehci-pci: EHCI PCI platform driver
[ 2.319790] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver
[ 2.325941] ohci-pci: OHCI PCI platform driver
[ 2.330377] uhci_hcd: USB Universal Host Controller Interface driver
[ 2.336788] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.341980] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 1
[ 2.349425] xhci-hcd 808622B7:00: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.357756] xhci-hcd 808622B7:00: irq 29, io mem 0x19000000
[ 2.363370] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.370130] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.377318] usb usb1: Product: xHCI Host Controller
[ 2.382169] usb usb1: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.389010] usb usb1: SerialNumber: 808622B7:00
[ 2.393727] hub 1-0:1.0: USB hub found
[ 2.397479] hub 1-0:1.0: 1 port detected
[ 2.401480] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.406677] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 2
[ 2.413986] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.422078] usb usb2: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.428836] usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.436020] usb usb2: Product: xHCI Host Controller
[ 2.440875] usb usb2: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.447720] usb usb2: SerialNumber: 808622B7:00
[ 2.452398] hub 2-0:1.0: USB hub found
[ 2.456137] hub 2-0:1.0: 1 port detected
[ 2.460168] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.465356] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 3
[ 2.472759] xhci-hcd 808622B7:01: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.481087] xhci-hcd 808622B7:01: irq 30, io mem 0x19800000
[ 2.486684] usb usb3: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.493443] usb usb3: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.500630] usb usb3: Product: xHCI Host Controller
[ 2.505482] usb usb3: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.512327] usb usb3: SerialNumber: 808622B7:01
[ 2.517049] hub 3-0:1.0: USB hub found
[ 2.520789] hub 3-0:1.0: 1 port detected
[ 2.524790] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.529976] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 4
[ 2.537284] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.545370] usb usb4: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.552128] usb usb4: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.559317] usb usb4: Product: xHCI Host Controller
[ 2.564168] usb usb4: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.571013] usb usb4: SerialNumber: 808622B7:01
[ 2.575716] hub 4-0:1.0: USB hub found
[ 2.579464] hub 4-0:1.0: 1 port detected
[ 2.583505] usbcore: registered new interface driver usbserial
[ 2.589324] usbcore: registered new interface driver usbserial_generic
[ 2.595826] usbserial: USB Serial support registered for generic
[ 2.601884] mousedev: PS/2 mouse device common for all mice
[ 2.607620] rtc-efi rtc-efi: rtc core: registered rtc-efi as rtc0
[ 2.613906] EFI Variables Facility v0.08 2004-May-17
[ 2.619081] hidraw: raw HID events driver (C) Jiri Kosina
[ 2.624514] usbcore: registered new interface driver usbhid
[ 2.630067] usbhid: USB HID core driver
[ 2.633953] drop_monitor: Initializing network drop monitor service
[ 2.640280] Initializing XFRM netlink socket
[ 2.644691] NET: Registered protocol family 10
[ 2.649496] Segment Routing with IPv6
[ 2.653160] NET: Registered protocol family 17
[ 2.657841] registered taskstats version 1
[ 2.661921] Loading compiled-in X.509 certificates
[ 2.669794] alg: No test for pkcs1pad(rsa,sha1) (pkcs1pad(rsa-generic,sha1))
[ 2.678237] Loaded X.509 cert 'Build time autogenerated kernel key: 98aa1b1964f4052ba920878b6f8348e178459895'
[ 2.688204] zswap: loaded using pool lzo/zbud
[ 2.713270] Key type big_key registered
[ 2.717168] BERT: [Firmware Bug]: table invalid.
[ 2.721839] rtc-efi rtc-efi: setting system clock to 2017-04-28 16:00:41 UTC (1493395241)
[ 2.757884] Freeing unused kernel memory: 1216K
[ 2.767418] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.774589] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.781211] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.789666] systemd[1]: systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL +XZ -LZ4 -SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
[ 2.807890] systemd[1]: Detected architecture arm64.
[ 2.812833] systemd[1]: Running in initial RAM disk.
[ 2.847031] systemd[1]: Set hostname to <localhost.localdomain>.
[ 2.979499] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.986233] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.993080] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.000078] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.010094] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.017557] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.025329] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.087174] systemd[1]: Reached target Swap.
[ 3.091495] systemd[1]: Starting Swap.
[ 3.117069] systemd[1]: Reached target Timers.
[ 3.121551] systemd[1]: Starting Timers.
[ 3.157062] systemd[1]: Reached target Local File Systems.
[ 3.162579] systemd[1]: Starting Local File Systems.
[ 4.087490] device-mapper: uevent: version 1.0.3
[ 4.092424] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel@redhat.com
[ 4.400761] libphy: APM X-Gene MDIO bus: probed
[ 4.444412] pps_core: LinuxPPS API ver. 1 registered
[ 4.449417] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[ 4.450547] xgene-ahci APMC0D0D:00: skip clock and PHY initialization
[ 4.450561] xgene-ahci APMC0D0D:00: controller can't do NCQ, turning off CAP_NCQ
[ 4.450629] xgene-ahci APMC0D0D:00: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.450643] xgene-ahci APMC0D0D:00: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.450670] xgene-ahci APMC0D0D:00: port 0 is not capable of FBS
[ 4.491029] sdhci: Secure Digital Host Controller Interface driver
[ 4.491033] sdhci: Copyright(c) Pierre Ossman
[ 4.512024] PTP clock support registered
[ 4.512754] xgene-ahci APMC0D0D:00: port 1 is not capable of FBS
[ 4.525820] scsi host0: xgene-ahci
[ 4.529855] scsi host1: xgene-ahci
[ 4.533507] ata1: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x100 irq 27
[ 4.534007] mmc0: SDHCI controller on ACPI [APMC0D0C:00] using PIO
[ 4.547673] ata2: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x180 irq 27
[ 4.548138] xgene-enet APMC0D05:00: clocks have been setup already
[ 4.553893] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k
[ 4.553894] e1000e: Copyright(c) 1999 - 2015 Intel Corporation.
[ 4.554194] e1000e 0000:01:00.0: enabling device (0000 -> 0002)
[ 4.554351] e1000e 0000:01:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[ 4.554379] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): Failed to initialize MSI-X interrupts. Falling back to MSI interrupts.
[ 4.554394] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): Failed to initialize MSI interrupts. Falling back to legacy interrupts.
[ 4.617549] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): registered PHC clock
[ 4.647217] xgene-enet APMC0D30:00: clocks have been setup already
[ 4.665238] xgene-ahci APMC0D0D:01: skip clock and PHY initialization
[ 4.671721] xgene-ahci APMC0D0D:01: controller can't do NCQ, turning off CAP_NCQ
[ 4.679178] xgene-ahci APMC0D0D:01: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.688066] xgene-ahci APMC0D0D:01: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.693143] e1000e 0000:01:00.0 eth2: (PCI Express:2.5GT/s:Width x1) 68:05:ca:46:ce:08
[ 4.693146] e1000e 0000:01:00.0 eth2: Intel(R) PRO/1000 Network Connection
[ 4.693162] e1000e 0000:01:00.0 eth2: MAC: 3, PHY: 8, PBA No: E46981-008
[ 4.696649] e1000e 0000:01:00.0 enp1s0: renamed from eth2
[ 4.722971] xgene-ahci APMC0D0D:01: port 0 is not capable of FBS
[ 4.729089] xgene-ahci APMC0D0D:01: port 1 is not capable of FBS
[ 4.736197] scsi host2: xgene-ahci
[ 4.740144] scsi host3: xgene-ahci
[ 4.743793] ata3: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x100 irq 28
[ 4.751735] ata4: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x180 irq 28
[ 4.857119] xgene-enet APMC0D30:01: clocks have been setup already
[ 4.874076] xgene-enet APMC0D31:00: clocks have been setup already
[ 5.088466] ata3: SATA link down (SStatus 0 SControl 4300)
[ 5.099014] ata4: SATA link down (SStatus 0 SControl 4300)
[ 5.217012] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.223552] ata1.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.230179] ata1.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.236797] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.243335] ata2.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.249961] ata2.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.256600] ata1.00: configured for UDMA/133
[ 5.261153] ata2.00: configured for UDMA/133
[ 5.261162] scsi 0:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.357513] scsi 1:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.357514] sd 0:0:0:0: [sda] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.357539] sd 0:0:0:0: [sda] Write Protect is off
[ 5.357542] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[ 5.357575] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.360081] sda: sda1 sda2 sda3
[ 5.360645] sd 0:0:0:0: [sda] Attached SCSI disk
[ 5.487441] sd 1:0:0:0: [sdb] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.494939] sd 1:0:0:0: [sdb] Write Protect is off
[ 5.499723] sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00
[ 5.499757] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.511144] sdb: sdb1
[ 5.513907] sd 1:0:0:0: [sdb] Attached SCSI disk
[ 5.962036] random: fast init done
[ 6.345919] SGI XFS with ACLs, security attributes, no debug enabled
[ 6.356572] XFS (dm-0): Mounting V5 Filesystem
[ 6.398617] XFS (dm-0): Ending clean mount
[ 6.973554] systemd-journald[198]: Received SIGTERM from PID 1 (systemd).
[ 6.989018] systemd: 20 output lines suppressed due to ratelimiting
[ 7.019839] audit: type=1404 audit(1493395245.790:2): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295
[ 7.106438] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.137061] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.216426] SELinux: 8 users, 14 roles, 4990 types, 304 bools, 1 sens, 1024 cats
[ 7.216433] SELinux: 94 classes, 105786 rules
[ 7.223909] SELinux: Class sctp_socket not defined in policy.
[ 7.229741] SELinux: Class icmp_socket not defined in policy.
[ 7.235543] SELinux: Class ax25_socket not defined in policy.
[ 7.241349] SELinux: Class ipx_socket not defined in policy.
[ 7.247069] SELinux: Class netrom_socket not defined in policy.
[ 7.253043] SELinux: Class atmpvc_socket not defined in policy.
[ 7.259021] SELinux: Class x25_socket not defined in policy.
[ 7.264736] SELinux: Class rose_socket not defined in policy.
[ 7.270540] SELinux: Class decnet_socket not defined in policy.
[ 7.276513] SELinux: Class atmsvc_socket not defined in policy.
[ 7.282490] SELinux: Class rds_socket not defined in policy.
[ 7.288207] SELinux: Class irda_socket not defined in policy.
[ 7.294008] SELinux: Class pppox_socket not defined in policy.
[ 7.299898] SELinux: Class llc_socket not defined in policy.
[ 7.305613] SELinux: Class can_socket not defined in policy.
[ 7.311330] SELinux: Class tipc_socket not defined in policy.
[ 7.317135] SELinux: Class bluetooth_socket not defined in policy.
[ 7.323368] SELinux: Class iucv_socket not defined in policy.
[ 7.329172] SELinux: Class rxrpc_socket not defined in policy.
[ 7.335059] SELinux: Class isdn_socket not defined in policy.
[ 7.340863] SELinux: Class phonet_socket not defined in policy.
[ 7.346836] SELinux: Class ieee802154_socket not defined in policy.
[ 7.353159] SELinux: Class caif_socket not defined in policy.
[ 7.358965] SELinux: Class alg_socket not defined in policy.
[ 7.364679] SELinux: Class nfc_socket not defined in policy.
[ 7.370397] SELinux: Class vsock_socket not defined in policy.
[ 7.376284] SELinux: Class kcm_socket not defined in policy.
[ 7.382002] SELinux: Class qipcrtr_socket not defined in policy.
[ 7.388066] SELinux: Class smc_socket not defined in policy.
[ 7.393780] SELinux: the above unknown classes and permissions will be allowed
[ 7.400979] SELinux: Completing initialization.
[ 7.400980] SELinux: Setting up existing superblocks.
[ 7.450804] audit: type=1403 audit(1493395246.220:3): policy loaded auid=4294967295 ses=4294967295
[ 7.455100] systemd[1]: Successfully loaded SELinux policy in 435.279ms.
[ 7.470770] systemd[1]: RTC configured in localtime, applying delta of -240 minutes to system time.
[ 7.497588] ip_tables: (C) 2000-2006 Netfilter Core Team
[ 7.502902] systemd[1]: Inserted module 'ip_tables'
[ 7.552579] systemd[1]: Relabelled /dev and /run in 42.787ms.
[ 8.005378] systemd-journald[616]: Received request to flush runtime journal from PID 1
[ 8.167863] xgene-slimpro-mbox APMC0D01:00: APM X-Gene SLIMpro MailBox registered
[ 8.172785] RPC: Registered named UNIX socket transport module.
[ 8.172790] RPC: Registered udp transport module.
[ 8.172794] RPC: Registered tcp transport module.
[ 8.172797] RPC: Registered tcp NFSv4.1 backchannel transport module.
[ 8.197228] xgene-rng APMC0D18:00: Couldn't get the clock for RNG
[ 8.204225] xgene-gpio-sb APMC0D15:00: Support 22 gpios, 6 irqs start from pin 8
[ 8.213102] xgene-gpio-sb APMC0D15:00: X-Gene GPIO Standby driver registered
[ 8.223510] xgene-slimpro-i2c APMC0D40:00: i2c mailbox channel request failed
[ 8.289837] sd 0:0:0:0: Attached scsi generic sg0 type 0
[ 8.295454] sd 1:0:0:0: Attached scsi generic sg1 type 0
[ 8.387796] xgene-slimpro-hwmon APMC0D29:00: APM X-Gene SoC HW monitor driver registered
[ 8.518644] Adding 8388544k swap on /dev/mapper/rhelp_dhcp--171-swap. Priority:-1 extents:1 across:8388544k SSFS
[ 8.531009] XFS (sda2): Mounting V5 Filesystem
[ 8.551661] XFS (sda2): Ending clean mount
[ 8.640881] Installing knfsd (copyright (C) 1996 okir@monad.swb.de).
[ 8.702746] XFS (dm-2): Mounting V5 Filesystem
[ 8.721803] XFS (dm-2): Ending clean mount
[ 9.540607] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.637501] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.650581] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.656789] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.669293] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.675510] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.687727] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.693923] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.706099] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 9.712318] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 9.730562] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.737469] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.744174] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.751066] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.757808] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.687474] xgene-enet APMC0D30:00 eth1: Link is Down
[ 10.692580] xgene-enet APMC0D05:00 eth0: Link is Down
[ 10.767324] xgene-enet APMC0D30:01 eth2: Link is Down
[ 12.418373] e1000e: enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
[ 12.426374] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready
[ 13.808268] xgene-enet APMC0D05:00 eth0: Link is Up - 1Gbps/Full - flow control off
[ 13.815925] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 108.537236] random: crng init done
[-- Attachment #3: dmesg_4.11.0-rc8.xgene_msix2+_msi.txt --]
[-- Type: text/plain, Size: 39010 bytes --]
[ 0.000000] Booting Linux on physical CPU 0x0
[ 0.000000] Linux version 4.11.0-rc8.xgene_msix2+ (root@mustang4.bos.jonmasters.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-14) (GCC) ) #2 SMP Fri Apr 28 12:06:14 EDT 2017
[ 0.000000] Boot CPU: AArch64 Processor [500f0001]
[ 0.000000] earlycon: uart8250 at MMIO32 0x000000001c020000 (options '')
[ 0.000000] bootconsole [uart8250] enabled
[ 0.000000] efi: Getting EFI parameters from FDT:
[ 0.000000] efi: EFI v2.40 by X-Gene Mustang Board EFI Aug 19 2016 18:02:24
[ 0.000000] efi: ACPI=0x43fa82d000 ACPI 2.0=0x43fa82d014 SMBIOS 3.0=0x43fa82b000 ESRT=0x43ff053d18
[ 0.000000] esrt: Reserving ESRT space from 0x00000043ff053d18 to 0x00000043ff053d78.
[ 0.000000] cma: Reserved 512 MiB at 0x00000040e0000000
[ 0.000000] ACPI: Early table checksum verification disabled
[ 0.000000] ACPI: RSDP 0x00000043FA82D014 000024 (v02 APM )
[ 0.000000] ACPI: XSDT 0x00000043FA82C0E8 000084 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: FACP 0x00000043FA822000 00010C (v05 APM XGENE 00000003 INTL 20140724)
[ 0.000000] ACPI: DSDT 0x00000043FA823000 005837 (v05 APM APM88xxx 00000001 INTL 20140724)
[ 0.000000] ACPI: DBG2 0x00000043FA829000 0000AA (v00 APMC0D XGENEDBG 00000000 INTL 20140724)
[ 0.000000] ACPI: GTDT 0x00000043FA820000 000060 (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: MCFG 0x00000043FA81F000 00003C (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: SPCR 0x00000043FA81E000 000050 (v02 APMC0D XGENESPC 00000000 INTL 20140724)
[ 0.000000] ACPI: SSDT 0x00000043FA81D000 00002D (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: BERT 0x00000043FA81C000 000030 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: HEST 0x00000043FA81B000 0002A8 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: APIC 0x00000043FA81A000 0002A4 (v03 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA819000 000063 (v02 REDHAT MACADDRS 00000001 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA818000 000032 (v02 REDHAT UARTCLKS 00000001 01000013)
[ 0.000000] ACPI: PCCT 0x00000043FA817000 000300 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SPCR: console: uart,mmio,0x1c020000,115200
[ 0.000000] ACPI: NUMA: Failed to initialise from firmware
[ 0.000000] NUMA: Faking a node at [mem 0x0000000000000000-0x00000043ffffffff]
[ 0.000000] NUMA: Adding memblock [0x4000000000 - 0x40001fffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x4000200000 - 0x43fa7cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43fa7d0000 - 0x43faa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43faa20000 - 0x43ff9cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ff9d0000 - 0x43ffa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ffa20000 - 0x43ffffffff] on node 0
[ 0.000000] NUMA: Initmem setup node 0 [mem 0x4000000000-0x43ffffffff]
[ 0.000000] NUMA: NODE_DATA [mem 0x43ffffe200-0x43ffffffff]
[ 0.000000] Zone ranges:
[ 0.000000] DMA [mem 0x0000004000000000-0x00000040ffffffff]
[ 0.000000] Normal [mem 0x0000004100000000-0x00000043ffffffff]
[ 0.000000] Movable zone start for each node
[ 0.000000] Early memory node ranges
[ 0.000000] node 0: [mem 0x0000004000000000-0x00000040001fffff]
[ 0.000000] node 0: [mem 0x0000004000200000-0x00000043fa7cffff]
[ 0.000000] node 0: [mem 0x00000043fa7d0000-0x00000043faa1ffff]
[ 0.000000] node 0: [mem 0x00000043faa20000-0x00000043ff9cffff]
[ 0.000000] node 0: [mem 0x00000043ff9d0000-0x00000043ffa1ffff]
[ 0.000000] node 0: [mem 0x00000043ffa20000-0x00000043ffffffff]
[ 0.000000] Initmem setup node 0 [mem 0x0000004000000000-0x00000043ffffffff]
[ 0.000000] On node 0 totalpages: 262144
[ 0.000000] DMA zone: 64 pages used for memmap
[ 0.000000] DMA zone: 0 pages reserved
[ 0.000000] DMA zone: 65536 pages, LIFO batch:1
[ 0.000000] Normal zone: 192 pages used for memmap
[ 0.000000] Normal zone: 196608 pages, LIFO batch:1
[ 0.000000] psci: is not implemented in ACPI.
[ 0.000000] percpu: Embedded 3 pages/cpu @ffff8003ffdf0000 s124312 r8192 d64104 u196608
[ 0.000000] pcpu-alloc: s124312 r8192 d64104 u196608 alloc=3*65536
[ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3 [0] 4 [0] 5 [0] 6 [0] 7
[ 0.000000] Detected PIPT I-cache on CPU0
[ 0.000000] Built 1 zonelists in Node order, mobility grouping on. Total pages: 261888
[ 0.000000] Policy zone: Normal
[ 0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.11.0-rc8.xgene_msix2+ root=/dev/mapper/rhelp_dhcp--171-root ro crashkernel=auto rd.lvm.lv=rhelp_dhcp-171/root rd.lvm.lv=rhelp_dhcp-171/swap LANG=en_US.UTF-8 acpi=force earlycon=uart8250,mmio32,0x1c020000 console=ttyS0,115200
[ 0.000000] PID hash table entries: 4096 (order: -1, 32768 bytes)
[ 0.000000] software IO TLB [mem 0x40dbff0000-0x40dfff0000] (64MB) mapped at [ffff8000dbff0000-ffff8000dffeffff]
[ 0.000000] Memory: 16109888K/16777216K available (7420K kernel code, 1354K rwdata, 3200K rodata, 1216K init, 7286K bss, 143040K reserved, 524288K cma-reserved)
[ 0.000000] Virtual kernel memory layout:
[ 0.000000] modules : 0xffff000000000000 - 0xffff000008000000 ( 128 MB)
[ 0.000000] vmalloc : 0xffff000008000000 - 0xffff7bdfffff0000 (126847 GB)
[ 0.000000] .text : 0xffff000008080000 - 0xffff0000087c0000 ( 7424 KB)
[ 0.000000] .rodata : 0xffff0000087c0000 - 0xffff000008af0000 ( 3264 KB)
[ 0.000000] .init : 0xffff000008af0000 - 0xffff000008c20000 ( 1216 KB)
[ 0.000000] .data : 0xffff000008c20000 - 0xffff000008d72a00 ( 1355 KB)
[ 0.000000] .bss : 0xffff000008d72a00 - 0xffff0000094902d4 ( 7287 KB)
[ 0.000000] fixed : 0xffff7fdffe7d0000 - 0xffff7fdffec00000 ( 4288 KB)
[ 0.000000] PCI I/O : 0xffff7fdffee00000 - 0xffff7fdfffe00000 ( 16 MB)
[ 0.000000] vmemmap : 0xffff7fe000000000 - 0xffff800000000000 ( 128 GB maximum)
[ 0.000000] 0xffff7fe000000000 - 0xffff7fe001000000 ( 16 MB actual)
[ 0.000000] memory : 0xffff800000000000 - 0xffff800400000000 ( 16384 MB)
[ 0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=8, Nodes=1
[ 0.000000] Hierarchical RCU implementation.
[ 0.000000] Build-time adjustment of leaf fanout to 64.
[ 0.000000] RCU restricting CPUs from NR_CPUS=4096 to nr_cpu_ids=8.
[ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=8
[ 0.000000] NR_IRQS:64 nr_irqs:64 0
[ 0.000000] GIC: Using split EOI/Deactivate mode
[ 0.000000] GICv3: No distributor detected at @ffff000008050000, giving up
[ 0.000000] arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys).
[ 0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns
[ 0.000002] sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns
[ 0.008379] Console: colour dummy device 80x25
[ 0.012950] Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000)
[ 0.023693] pid_max: default: 32768 minimum: 301
[ 0.028471] ACPI: Core revision 20170119
[ 0.036063] ACPI Error: Method parse/execution failed [\_SB.ET00._STA] (Node ffff8003c0195c08), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.049121] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.056989] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.063808] ACPI Error: Method parse/execution failed [\_SB.ET01._STA] (Node ffff8003c0195eb0), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.076952] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.084871] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.092015] ACPI: 4 ACPI AML tables successfully acquired and loaded
[ 0.098567] Security Framework initialized
[ 0.102838] SELinux: Initializing.
[ 0.106408] SELinux: Starting in permissive mode
[ 0.106607] Dentry cache hash table entries: 2097152 (order: 8, 16777216 bytes)
[ 0.116844] Inode-cache hash table entries: 1048576 (order: 7, 8388608 bytes)
[ 0.125552] Mount-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.132626] Mountpoint-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.140570] ftrace: allocating 26815 entries in 7 pages
[ 0.203748] ASID allocator initialised with 65536 entries
[ 0.209911] Remapping and enabling EFI services.
[ 0.214705] EFI remap 0x0000000010510000 => 0000000020000000
[ 0.220700] EFI remap 0x0000000010548000 => 0000000020018000
[ 0.226689] EFI remap 0x0000000017000000 => 0000000020020000
[ 0.232657] EFI remap 0x000000001c025000 => 0000000020035000
[ 0.238736] UEFI Runtime regions are not aligned to 64 KB -- buggy firmware?
[ 0.238741] ------------[ cut here ]------------
[ 0.250727] WARNING: CPU: 0 PID: 1 at arch/arm64/kernel/efi.c:34 efi_create_mapping+0x4c/0x104
[ 0.259532] Modules linked in:
[ 0.264214] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc8.xgene_msix2+ #2
[ 0.271827] Hardware name: APM X-Gene Mustang board (DT)
[ 0.277238] task: ffff8003c1bc0000 task.stack: ffff8003c1c00000
[ 0.283324] PC is at efi_create_mapping+0x4c/0x104
[ 0.288269] LR is at efi_create_mapping+0x4c/0x104
[ 0.293127] pc : [<ffff000008af51f0>] lr : [<ffff000008af51f0>] pstate: 60000045
[ 0.300784] sp : ffff8003c1c03d20
[ 0.304207] x29: ffff8003c1c03d20 x28: 0000000000000000
[ 0.309655] x27: ffff000008c53000 x26: ffff000008d429c8
[ 0.315110] x25: ffff000008d42000 x24: 0000000000000000
[ 0.320556] x23: ffff000008a4c910 x22: ffff000008c60000
[ 0.325994] x21: ffff000008c53000 x20: ffff000008d429c8
[ 0.331484] x19: ffff8003f96ab348 x18: 00000000b6a2110e
[ 0.337008] x17: 0000000000000000 x16: 0000000000000000
[ 0.342437] x15: ffff000008c53000 x14: 00000000cb9aed1d
[ 0.347944] x13: ffff0000089ca500 x12: 0000000000040b00
[ 0.353425] x11: ffff8003c1c039e0 x10: ffff8003c1c039e0
[ 0.358889] x9 : 0000000000000073 x8 : 6620796767756220
[ 0.364352] x7 : 2d2d20424b203436 x6 : ffff000009274718
[ 0.369825] x5 : ffff8003ffdf7530 x4 : 0000000000000000
[ 0.375280] x3 : 0000000000000000 x2 : 0000000000040b00
[ 0.380769] x1 : 0000000000040b00 x0 : 0000000000000040
[ 0.387782] ---[ end trace 9580655578fda7c1 ]---
[ 0.392519] Call trace:
[ 0.395045] Exception stack(0xffff8003c1c03b30 to 0xffff8003c1c03c60)
[ 0.401735] 3b20: ffff8003f96ab348 0001000000000000
[ 0.409730] 3b40: ffff8003c1c03d20 ffff000008af51f0 0000000060000045 ffff000008d429c8
[ 0.417828] 3b60: 0000000000000000 0000000000000000 ffff8003c1c03d20 ffff8003c1c03d20
[ 0.425857] 3b80: ffff8003c1c03ce0 00000000ffffffc8 ffff8003c1c03be0 ffff00000812f6b8
[ 0.433946] 3ba0: ffff000008c53000 00000000ffffffc8 ffff8003c1c03c00 ffff0000089be038
[ 0.442053] 3bc0: ffff8003c1c03d20 ffff8003c1c03d20 ffff8003c1c03ce0 00000000ffffffc8
[ 0.450126] 3be0: ffff8003c1c03c70 0000000000040b00 0000000000000040 0000000000040b00
[ 0.458207] 3c00: 0000000000040b00 0000000000000000 0000000000000000 ffff8003ffdf7530
[ 0.466227] 3c20: ffff000009274718 2d2d20424b203436 6620796767756220 0000000000000073
[ 0.474282] 3c40: ffff8003c1c039e0 ffff8003c1c039e0 0000000000040b00 ffff0000089ca500
[ 0.482321] [<ffff000008af51f0>] efi_create_mapping+0x4c/0x104
[ 0.488270] [<ffff000008b3db98>] arm_enable_runtime_services+0x154/0x238
[ 0.495238] [<ffff000008083698>] do_one_initcall+0x54/0x15c
[ 0.501030] [<ffff000008af0d18>] kernel_init_freeable+0xe4/0x25c
[ 0.507221] [<ffff0000087ae488>] kernel_init+0x18/0x108
[ 0.512606] [<ffff0000080833c0>] ret_from_fork+0x10/0x50
[ 0.518077] EFI remap 0x00000043fa7df000 => 000000002004f000
[ 0.524089] EFI remap 0x00000043fa82e000 => 000000002006e000
[ 0.530085] EFI remap 0x00000043ff9d6000 => 0000000020266000
[ 0.536148] EFI remap 0x00000043ff9ea000 => 000000002027a000
[ 0.542320] smp: Bringing up secondary CPUs ...
[ 0.547164] Detected PIPT I-cache on CPU1
[ 0.547192] CPU1: Booted secondary processor [500f0001]
[ 0.547411] Detected PIPT I-cache on CPU2
[ 0.547433] CPU2: Booted secondary processor [500f0001]
[ 0.547646] Detected PIPT I-cache on CPU3
[ 0.547660] CPU3: Booted secondary processor [500f0001]
[ 0.547854] Detected PIPT I-cache on CPU4
[ 0.547875] CPU4: Booted secondary processor [500f0001]
[ 0.548076] Detected PIPT I-cache on CPU5
[ 0.548090] CPU5: Booted secondary processor [500f0001]
[ 0.548281] Detected PIPT I-cache on CPU6
[ 0.548302] CPU6: Booted secondary processor [500f0001]
[ 0.548505] Detected PIPT I-cache on CPU7
[ 0.548519] CPU7: Booted secondary processor [500f0001]
[ 0.548562] smp: Brought up 1 node, 8 CPUs
[ 0.620493] SMP: Total of 8 processors activated.
[ 0.625425] CPU features: detected feature: 32-bit EL0 Support
[ 0.631511] CPU: All CPU(s) started at EL2
[ 0.636531] devtmpfs: initialized
[ 0.640271] SMBIOS 3.0.0 present.
[ 0.643745] DMI: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.15 Aug 19 2016
[ 0.653110] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[ 0.663392] futex hash table entries: 2048 (order: 2, 262144 bytes)
[ 0.670383] pinctrl core: initialized pinctrl subsystem
[ 0.676091] NET: Registered protocol family 16
[ 0.681082] cpuidle: using governor menu
[ 0.685276] Detected 8 PCC Subspaces
[ 0.689040] Registering PCC driver as Mailbox controller
[ 0.694649] vdso: 2 pages (1 code @ ffff0000087e0000, 1 data @ ffff000008c40000)
[ 0.702382] hw-breakpoint: found 4 breakpoint and 4 watchpoint registers.
[ 0.711945] DMA: preallocated 256 KiB pool for atomic allocations
[ 0.718368] ACPI: bus type PCI registered
[ 0.722572] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
[ 0.729340] Serial: AMBA PL011 UART driver
[ 0.735597] HugeTLB registered 512 MB page size, pre-allocated 0 pages
[ 0.743073] ACPI: Added _OSI(Module Device)
[ 0.747432] ACPI: Added _OSI(Processor Device)
[ 0.752083] ACPI: Added _OSI(3.0 _SCP Extensions)
[ 0.757005] ACPI: Added _OSI(Processor Aggregator Device)
[ 0.772090] ACPI: Interpreter enabled
[ 0.775913] ACPI: Using GIC for interrupt routing
[ 0.780864] ACPI: MCFG table detected, 1 entries
[ 0.785858] HEST: Table parsing has been initialized.
[ 0.793565] ACPI: Power Resource [SCVR] (on)
[ 0.828517] xgene-msi APMC0D0E:00: APM X-Gene PCIe MSI driver loaded
[ 0.835285] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff])
[ 0.841791] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.850585] acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug PME AER PCIeCapability]
[ 0.859017] acpi PNP0A08:00: MCFG quirk: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] with xgene_v1_pcie_ecam_ops
[ 0.871645] acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0xe0d0000000-0xe0dfffffff] not reserved in ACPI namespace
[ 0.882851] acpi PNP0A08:00: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff]
[ 0.891009] Remapped I/O 0x000000e010000000 to [io 0x0000-0xffff window]
[ 0.898264] PCI host bridge to bus 0000:00
[ 0.902574] pci_bus 0000:00: root bus resource [io 0x0000-0xffff window] (bus address [0x10000000-0x1000ffff])
[ 0.913134] pci_bus 0000:00: root bus resource [mem 0xe040000000-0xe07fffffff window] (bus address [0x40000000-0x7fffffff])
[ 0.924779] pci_bus 0000:00: root bus resource [mem 0xf000000000-0xffffffffff window]
[ 0.932964] pci_bus 0000:00: root bus resource [bus 00-ff]
[ 0.938697] pci 0000:00:00.0: [10e8:e004] type 01 class 0x060400
[ 0.938776] pci 0000:00:00.0: supports D1 D2
[ 0.938937] pci 0000:01:00.0: [8086:10d3] type 00 class 0x020000
[ 0.938970] pci 0000:01:00.0: reg 0x10: [mem 0xe040080000-0xe04009ffff]
[ 0.938986] pci 0000:01:00.0: reg 0x14: [mem 0xe040000000-0xe04007ffff]
[ 0.939003] pci 0000:01:00.0: reg 0x18: [io 0x0000-0x001f]
[ 0.939019] pci 0000:01:00.0: reg 0x1c: [mem 0xe0400a0000-0xe0400a3fff]
[ 0.939065] pci 0000:01:00.0: reg 0x30: [mem 0xfffc0000-0xffffffff pref]
[ 0.939205] pci 0000:01:00.0: PME# supported from D0 D3hot D3cold
[ 0.968751] pci 0000:00:00.0: BAR 14: assigned [mem 0xe040000000-0xe0400fffff]
[ 0.976281] pci 0000:00:00.0: BAR 13: assigned [io 0x1000-0x1fff]
[ 0.982748] pci 0000:01:00.0: BAR 1: assigned [mem 0xe040000000-0xe04007ffff]
[ 0.990222] pci 0000:01:00.0: BAR 6: assigned [mem 0xe040080000-0xe0400bffff pref]
[ 0.998135] pci 0000:01:00.0: BAR 0: assigned [mem 0xe0400c0000-0xe0400dffff]
[ 1.005583] pci 0000:01:00.0: BAR 3: assigned [mem 0xe0400e0000-0xe0400e3fff]
[ 1.013044] pci 0000:01:00.0: BAR 2: assigned [io 0x1000-0x101f]
[ 1.019447] pci 0000:00:00.0: PCI bridge to [bus 01]
[ 1.024681] pci 0000:00:00.0: bridge window [io 0x1000-0x1fff]
[ 1.031099] pci 0000:00:00.0: bridge window [mem 0xe040000000-0xe0400fffff]
[ 1.039197] vgaarb: loaded
[ 1.042181] SCSI subsystem initialized
[ 1.046203] libata version 3.00 loaded.
[ 1.046240] ACPI: bus type USB registered
[ 1.050444] usbcore: registered new interface driver usbfs
[ 1.056204] usbcore: registered new interface driver hub
[ 1.061821] usbcore: registered new device driver usb
[ 1.067216] Registered efivars operations
[ 1.071684] NetLabel: Initializing
[ 1.075256] NetLabel: domain hash size = 128
[ 1.079777] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO
[ 1.085747] NetLabel: unlabeled traffic allowed by default
[ 1.091744] clocksource: Switched to clocksource arch_sys_counter
[ 1.116637] VFS: Disk quotas dquot_6.6.0
[ 1.120816] VFS: Dquot-cache hash table entries: 8192 (order 0, 65536 bytes)
[ 1.128374] pnp: PnP ACPI init
[ 1.132213] pnp: PnP ACPI: found 0 devices
[ 1.139026] NET: Registered protocol family 2
[ 1.143824] TCP established hash table entries: 131072 (order: 4, 1048576 bytes)
[ 1.151892] TCP bind hash table entries: 65536 (order: 4, 1048576 bytes)
[ 1.159275] TCP: Hash tables configured (established 131072 bind 65536)
[ 1.166293] UDP hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.172798] UDP-Lite hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.179867] NET: Registered protocol family 1
[ 1.184479] PCI: CLS 64 bytes, default 128
[ 1.184578] Unpacking initramfs...
[ 1.868130] Freeing initrd memory: 32640K
[ 1.872713] kvm [1]: 8-bit VMID
[ 1.875992] kvm [1]: IDMAP page: 40009b7000
[ 1.880358] kvm [1]: HYP VA range: 800000000000:ffffffffffff
[ 1.886375] kvm [1]: Hyp mode initialized successfully
[ 1.891755] kvm [1]: GICV region size/alignment is unsafe, using trapping (reduced performance)
[ 1.900915] kvm [1]: vgic-v2@780cf000
[ 1.904807] kvm [1]: vgic interrupt IRQ1
[ 1.908909] kvm [1]: virtual timer IRQ4
[ 1.928681] audit: initializing netlink subsys (disabled)
[ 1.934412] audit: type=2000 audit(1.470:1): state=initialized audit_enabled=0 res=1
[ 1.934540] Initialise system trusted keyrings
[ 1.934609] workingset: timestamp_bits=44 max_order=18 bucket_order=0
[ 1.936691] zbud: loaded
[ 1.937062] SELinux: Registering netfilter hooks
[ 2.016448] NET: Registered protocol family 38
[ 2.021088] Key type asymmetric registered
[ 2.025420] Asymmetric key parser 'x509' registered
[ 2.030583] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
[ 2.038391] io scheduler noop registered
[ 2.042467] io scheduler deadline registered (default)
[ 2.047931] io scheduler cfq registered
[ 2.051950] io scheduler mq-deadline registered
[ 2.056870] atomic64_test: passed
[ 2.060612] xgene-gpio APMC0D14:00: X-Gene GPIO driver registered.
[ 2.067387] pcieport 0000:00:00.0: AER enabled with IRQ 83
[ 2.073172] pcieport 0000:00:00.0: Signaling PME with IRQ 83
[ 2.079355] input: Power Button as /devices/LNXSYSTM:00/PNP0C0C:00/input/input0
[ 2.087026] ACPI: Power Button [PWRB]
[ 2.091368] GHES: Failed to map GSI to IRQ for generic hardware error source: 2
[ 2.099018] GHES: probe of GHES.2 failed with error 82
[ 2.104386] GHES: Failed to map GSI to IRQ for generic hardware error source: 3
[ 2.112068] GHES: probe of GHES.3 failed with error 82
[ 2.117497] GHES: Failed to map GSI to IRQ for generic hardware error source: 4
[ 2.125101] GHES: probe of GHES.4 failed with error 82
[ 2.130452] GHES: Failed to map GSI to IRQ for generic hardware error source: 5
[ 2.138036] GHES: probe of GHES.5 failed with error 82
[ 2.143461] GHES: Failed to map GSI to IRQ for generic hardware error source: 6
[ 2.151121] GHES: probe of GHES.6 failed with error 82
[ 2.156510] GHES: Failed to map GSI to IRQ for generic hardware error source: 7
[ 2.164189] GHES: probe of GHES.7 failed with error 82
[ 2.169574] GHES: Failed to map GSI to IRQ for generic hardware error source: 8
[ 2.177249] GHES: probe of GHES.8 failed with error 82
[ 2.182604] GHES: Failed to map GSI to IRQ for generic hardware error source: 9
[ 2.190284] GHES: probe of GHES.9 failed with error 82
[ 2.195693] GHES: Failed to enable APEI firmware first mode.
[ 2.201771] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[ 2.208814] console [ttyS0] disabled
[ 2.232712] APMC0D08:00: ttyS0 at MMIO 0x1c020000 (irq = 22, base_baud = 3125000) is a U6_16550A
[ 2.241919] console [ttyS0] enabled
[ 2.249007] bootconsole [uart8250] disabled
[ 2.277886] APMC0D08:01: ttyS1 at MMIO 0x1c021000 (irq = 23, base_baud = 3125000) is a U6_16550A
[ 2.287044] cacheinfo: Unable to detect cache hierarchy for CPU 0
[ 2.293126] rdac: device handler registered
[ 2.297348] hp_sw: device handler registered
[ 2.301595] emc: device handler registered
[ 2.305766] alua: device handler registered
[ 2.310077] libphy: Fixed MDIO Bus: probed
[ 2.314237] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
[ 2.320734] ehci-pci: EHCI PCI platform driver
[ 2.325177] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver
[ 2.331327] ohci-pci: OHCI PCI platform driver
[ 2.335764] uhci_hcd: USB Universal Host Controller Interface driver
[ 2.342182] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.347366] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 1
[ 2.354792] xhci-hcd 808622B7:00: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.363122] xhci-hcd 808622B7:00: irq 29, io mem 0x19000000
[ 2.368735] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.375494] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.382682] usb usb1: Product: xHCI Host Controller
[ 2.387533] usb usb1: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.394375] usb usb1: SerialNumber: 808622B7:00
[ 2.399097] hub 1-0:1.0: USB hub found
[ 2.402848] hub 1-0:1.0: 1 port detected
[ 2.406851] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.412034] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 2
[ 2.419332] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.427425] usb usb2: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.434184] usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.441368] usb usb2: Product: xHCI Host Controller
[ 2.446223] usb usb2: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.453069] usb usb2: SerialNumber: 808622B7:00
[ 2.457761] hub 2-0:1.0: USB hub found
[ 2.461501] hub 2-0:1.0: 1 port detected
[ 2.465534] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.470715] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 3
[ 2.478113] xhci-hcd 808622B7:01: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.486441] xhci-hcd 808622B7:01: irq 30, io mem 0x19800000
[ 2.492041] usb usb3: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.498793] usb usb3: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.505984] usb usb3: Product: xHCI Host Controller
[ 2.510836] usb usb3: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.517682] usb usb3: SerialNumber: 808622B7:01
[ 2.522397] hub 3-0:1.0: USB hub found
[ 2.526137] hub 3-0:1.0: 1 port detected
[ 2.530122] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.535305] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 4
[ 2.542607] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.550692] usb usb4: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.557451] usb usb4: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.564640] usb usb4: Product: xHCI Host Controller
[ 2.569491] usb usb4: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.576334] usb usb4: SerialNumber: 808622B7:01
[ 2.581037] hub 4-0:1.0: USB hub found
[ 2.584788] hub 4-0:1.0: 1 port detected
[ 2.588828] usbcore: registered new interface driver usbserial
[ 2.594649] usbcore: registered new interface driver usbserial_generic
[ 2.601150] usbserial: USB Serial support registered for generic
[ 2.607208] mousedev: PS/2 mouse device common for all mice
[ 2.612948] rtc-efi rtc-efi: rtc core: registered rtc-efi as rtc0
[ 2.619241] EFI Variables Facility v0.08 2004-May-17
[ 2.624405] hidraw: raw HID events driver (C) Jiri Kosina
[ 2.629836] usbcore: registered new interface driver usbhid
[ 2.635388] usbhid: USB HID core driver
[ 2.639274] drop_monitor: Initializing network drop monitor service
[ 2.645602] Initializing XFRM netlink socket
[ 2.650006] NET: Registered protocol family 10
[ 2.654800] Segment Routing with IPv6
[ 2.658462] NET: Registered protocol family 17
[ 2.663153] registered taskstats version 1
[ 2.667231] Loading compiled-in X.509 certificates
[ 2.675102] alg: No test for pkcs1pad(rsa,sha1) (pkcs1pad(rsa-generic,sha1))
[ 2.683548] Loaded X.509 cert 'Build time autogenerated kernel key: 98aa1b1964f4052ba920878b6f8348e178459895'
[ 2.693521] zswap: loaded using pool lzo/zbud
[ 2.717703] Key type big_key registered
[ 2.721594] BERT: [Firmware Bug]: table invalid.
[ 2.726272] rtc-efi rtc-efi: setting system clock to 2017-04-28 16:24:34 UTC (1493396674)
[ 2.759039] Freeing unused kernel memory: 1216K
[ 2.768550] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.775743] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.782366] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.790811] systemd[1]: systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL +XZ -LZ4 -SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
[ 2.809044] systemd[1]: Detected architecture arm64.
[ 2.814004] systemd[1]: Running in initial RAM disk.
[ 2.851832] systemd[1]: Set hostname to <localhost.localdomain>.
[ 2.970013] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.976652] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.983295] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.989979] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.997677] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.004524] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.011427] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.051810] systemd[1]: Reached target Timers.
[ 3.056250] systemd[1]: Starting Timers.
[ 3.091786] systemd[1]: Created slice Root Slice.
[ 3.096483] systemd[1]: Starting Root Slice.
[ 3.131780] systemd[1]: Listening on udev Kernel Socket.
[ 3.137083] systemd[1]: Starting udev Kernel Socket.
[ 4.102501] device-mapper: uevent: version 1.0.3
[ 4.107414] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel@redhat.com
[ 4.412199] xgene-ahci APMC0D0D:00: skip clock and PHY initialization
[ 4.418619] xgene-ahci APMC0D0D:00: controller can't do NCQ, turning off CAP_NCQ
[ 4.426125] xgene-ahci APMC0D0D:00: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.434987] xgene-ahci APMC0D0D:00: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.434998] xgene-ahci APMC0D0D:00: port 0 is not capable of FBS
[ 4.470653] pps_core: LinuxPPS API ver. 1 registered
[ 4.475669] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[ 4.491986] xgene-ahci APMC0D0D:00: port 1 is not capable of FBS
[ 4.507826] PTP clock support registered
[ 4.521307] sdhci: Secure Digital Host Controller Interface driver
[ 4.527512] sdhci: Copyright(c) Pierre Ossman
[ 4.577139] mmc0: SDHCI controller on ACPI [APMC0D0C:00] using PIO
[ 4.612188] scsi host0: xgene-ahci
[ 4.615838] scsi host1: xgene-ahci
[ 4.619314] ata1: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x100 irq 27
[ 4.621598] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k
[ 4.621602] e1000e: Copyright(c) 1999 - 2015 Intel Corporation.
[ 4.621841] e1000e 0000:01:00.0: enabling device (0000 -> 0002)
[ 4.622056] e1000e 0000:01:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[ 4.654091] ata2: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x180 irq 27
[ 4.662326] xgene-ahci APMC0D0D:01: skip clock and PHY initialization
[ 4.668755] xgene-ahci APMC0D0D:01: controller can't do NCQ, turning off CAP_NCQ
[ 4.676203] xgene-ahci APMC0D0D:01: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.685069] xgene-ahci APMC0D0D:01: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.693201] xgene-ahci APMC0D0D:01: port 0 is not capable of FBS
[ 4.693880] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): registered PHC clock
[ 4.706842] xgene-ahci APMC0D0D:01: port 1 is not capable of FBS
[ 4.714376] scsi host2: xgene-ahci
[ 4.718239] scsi host3: xgene-ahci
[ 4.721961] ata3: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x100 irq 28
[ 4.729850] ata4: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x180 irq 28
[ 4.738769] libphy: APM X-Gene MDIO bus: probed
[ 4.791711] e1000e 0000:01:00.0 eth0: (PCI Express:2.5GT/s:Width x1) 68:05:ca:46:ce:08
[ 4.799622] e1000e 0000:01:00.0 eth0: Intel(R) PRO/1000 Network Connection
[ 4.806511] e1000e 0000:01:00.0 eth0: MAC: 3, PHY: 8, PBA No: E46981-008
[ 4.808607] xgene-enet APMC0D05:00: clocks have been setup already
[ 4.821883] e1000e 0000:01:00.0 enp1s0: renamed from eth0
[ 5.011899] xgene-enet APMC0D30:00: clocks have been setup already
[ 5.063242] ata3: SATA link down (SStatus 0 SControl 4300)
[ 5.068752] ata4: SATA link down (SStatus 0 SControl 4300)
[ 5.121895] xgene-enet APMC0D30:01: clocks have been setup already
[ 5.136261] xgene-enet APMC0D31:00: clocks have been setup already
[ 5.331836] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.338118] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.344486] ata2.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.351098] ata2.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.357813] ata1.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.364432] ata1.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.371105] ata2.00: configured for UDMA/133
[ 5.375470] ata1.00: configured for UDMA/133
[ 5.380021] scsi 0:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.482145] sd 0:0:0:0: [sda] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.482607] scsi 1:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.498165] sd 0:0:0:0: [sda] Write Protect is off
[ 5.502945] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[ 5.502986] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.514451] sda: sda1 sda2 sda3
[ 5.518153] sd 0:0:0:0: [sda] Attached SCSI disk
[ 5.602396] sd 1:0:0:0: [sdb] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.609881] sd 1:0:0:0: [sdb] Write Protect is off
[ 5.614681] sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00
[ 5.614739] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.626189] sdb: sdb1
[ 5.628980] sd 1:0:0:0: [sdb] Attached SCSI disk
[ 6.113010] random: fast init done
[ 6.461233] SGI XFS with ACLs, security attributes, no debug enabled
[ 6.471638] XFS (dm-0): Mounting V5 Filesystem
[ 6.501907] XFS (dm-0): Ending clean mount
[ 7.216272] systemd-journald[201]: Received SIGTERM from PID 1 (systemd).
[ 7.263513] systemd: 20 output lines suppressed due to ratelimiting
[ 7.301701] audit: type=1404 audit(1493396679.060:2): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295
[ 7.410911] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.516206] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.718524] SELinux: 8 users, 14 roles, 4990 types, 304 bools, 1 sens, 1024 cats
[ 7.718542] SELinux: 94 classes, 105786 rules
[ 7.741338] SELinux: Class sctp_socket not defined in policy.
[ 7.747177] SELinux: Class icmp_socket not defined in policy.
[ 7.753000] SELinux: Class ax25_socket not defined in policy.
[ 7.758809] SELinux: Class ipx_socket not defined in policy.
[ 7.764542] SELinux: Class netrom_socket not defined in policy.
[ 7.770524] SELinux: Class atmpvc_socket not defined in policy.
[ 7.776515] SELinux: Class x25_socket not defined in policy.
[ 7.782247] SELinux: Class rose_socket not defined in policy.
[ 7.788056] SELinux: Class decnet_socket not defined in policy.
[ 7.794048] SELinux: Class atmsvc_socket not defined in policy.
[ 7.800028] SELinux: Class rds_socket not defined in policy.
[ 7.805760] SELinux: Class irda_socket not defined in policy.
[ 7.811568] SELinux: Class pppox_socket not defined in policy.
[ 7.817473] SELinux: Class llc_socket not defined in policy.
[ 7.823204] SELinux: Class can_socket not defined in policy.
[ 7.828926] SELinux: Class tipc_socket not defined in policy.
[ 7.834744] SELinux: Class bluetooth_socket not defined in policy.
[ 7.840984] SELinux: Class iucv_socket not defined in policy.
[ 7.846802] SELinux: Class rxrpc_socket not defined in policy.
[ 7.852707] SELinux: Class isdn_socket not defined in policy.
[ 7.858514] SELinux: Class phonet_socket not defined in policy.
[ 7.864504] SELinux: Class ieee802154_socket not defined in policy.
[ 7.870831] SELinux: Class caif_socket not defined in policy.
[ 7.876649] SELinux: Class alg_socket not defined in policy.
[ 7.882381] SELinux: Class nfc_socket not defined in policy.
[ 7.888103] SELinux: Class vsock_socket not defined in policy.
[ 7.894008] SELinux: Class kcm_socket not defined in policy.
[ 7.899730] SELinux: Class qipcrtr_socket not defined in policy.
[ 7.905822] SELinux: Class smc_socket not defined in policy.
[ 7.911545] SELinux: the above unknown classes and permissions will be allowed
[ 7.918772] SELinux: Completing initialization.
[ 7.918776] SELinux: Setting up existing superblocks.
[ 7.993509] audit: type=1403 audit(1493396679.760:3): policy loaded auid=4294967295 ses=4294967295
[ 8.002331] systemd[1]: Successfully loaded SELinux policy in 701.052ms.
[ 8.029233] systemd[1]: RTC configured in localtime, applying delta of -240 minutes to system time.
[ 8.076262] ip_tables: (C) 2000-2006 Netfilter Core Team
[ 8.081619] systemd[1]: Inserted module 'ip_tables'
[ 8.199798] systemd[1]: Relabelled /dev and /run in 108.385ms.
[ 8.666128] systemd-journald[618]: Received request to flush runtime journal from PID 1
[ 8.793125] xgene-slimpro-i2c APMC0D40:00: i2c mailbox channel request failed
[ 8.809689] RPC: Registered named UNIX socket transport module.
[ 8.815634] RPC: Registered udp transport module.
[ 8.820316] RPC: Registered tcp transport module.
[ 8.825018] RPC: Registered tcp NFSv4.1 backchannel transport module.
[ 8.872038] sd 0:0:0:0: Attached scsi generic sg0 type 0
[ 8.877502] sd 1:0:0:0: Attached scsi generic sg1 type 0
[ 8.883774] xgene-slimpro-hwmon APMC0D29:00: APM X-Gene SoC HW monitor driver registered
[ 8.892777] xgene-rng APMC0D18:00: Couldn't get the clock for RNG
[ 8.899408] xgene-slimpro-mbox APMC0D01:00: APM X-Gene SLIMpro MailBox registered
[ 8.910481] xgene-gpio-sb APMC0D15:00: Support 22 gpios, 6 irqs start from pin 8
[ 8.918470] xgene-gpio-sb APMC0D15:00: X-Gene GPIO Standby driver registered
[ 9.044885] Installing knfsd (copyright (C) 1996 okir@monad.swb.de).
[ 9.118751] XFS (sda2): Mounting V5 Filesystem
[ 9.125864] Adding 8388544k swap on /dev/mapper/rhelp_dhcp--171-swap. Priority:-1 extents:1 across:8388544k SSFS
[ 9.141179] XFS (sda2): Ending clean mount
[ 9.286221] XFS (dm-2): Mounting V5 Filesystem
[ 9.305151] XFS (dm-2): Ending clean mount
[ 10.242361] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.339596] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.367672] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.374758] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.403220] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.410389] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.437608] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.444895] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.471938] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.478809] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.562700] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.571438] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.580052] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.588624] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.597191] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 11.412166] xgene-enet APMC0D05:00 eth0: Link is Down
[ 11.492472] xgene-enet APMC0D30:00 eth1: Link is Down
[ 11.497600] xgene-enet APMC0D30:01 eth2: Link is Down
[ 14.533303] e1000e: enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
[ 14.541337] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready
[ 14.548556] xgene-enet APMC0D05:00 eth0: Link is Up - 1Gbps/Full - flow control off
[ 14.556244] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 16:36 ` Jon Masters
0 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-04-28 16:36 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas,
linux-pci
Cc: patches, Duc Dang, rjw, linux-kernel, Jon Masters, linux-arm-kernel
[-- Attachment #1: Type: text/plain, Size: 995 bytes --]
On 04/28/2017 01:11 AM, Jon Masters wrote:
> On 04/27/2017 08:54 PM, Khuong Dinh wrote:
>> From: Khuong Dinh <kdinh@apm.com>
>>
>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>
>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> Signed-off-by: Duc Dang <dhdang@apm.com>
>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>
> Thanks. Currently testing. Will review and followup shortly.
Attaching boot logs before and after. I tested with an Intel e1000.
Before the patch I get fallback interrupts. Afterward, I get X-Gene MSI
interrupts working. Please followup with answers to Marc's question
about how you propose to guarantee boottime ordering of binding.
I urgently require that AppliedMicro complete upstreaming of this. It
should have been upstream a year ago. You're not the only ones with
patches like this left around - I'm coming for everyone else, too.
Tested-by: Jon Masters <jcm@jonmasters.org>
[-- Attachment #2: dmesg_4.11.0-rc8.xgene_msix1+_nomsi.txt --]
[-- Type: text/plain, Size: 39241 bytes --]
[ 0.000000] Booting Linux on physical CPU 0x0
[ 0.000000] Linux version 4.11.0-rc8.xgene_msix1+ (root@mustang4.bos.jonmasters.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-14) (GCC) ) #1 SMP Fri Apr 28 04:44:20 EDT 2017
[ 0.000000] Boot CPU: AArch64 Processor [500f0001]
[ 0.000000] earlycon: uart8250 at MMIO32 0x000000001c020000 (options '')
[ 0.000000] bootconsole [uart8250] enabled
[ 0.000000] efi: Getting EFI parameters from FDT:
[ 0.000000] efi: EFI v2.40 by X-Gene Mustang Board EFI Aug 19 2016 18:02:24
[ 0.000000] efi: ACPI=0x43fa82d000 ACPI 2.0=0x43fa82d014 SMBIOS 3.0=0x43fa82b000 ESRT=0x43ff053d18
[ 0.000000] esrt: Reserving ESRT space from 0x00000043ff053d18 to 0x00000043ff053d78.
[ 0.000000] cma: Reserved 512 MiB at 0x00000040e0000000
[ 0.000000] ACPI: Early table checksum verification disabled
[ 0.000000] ACPI: RSDP 0x00000043FA82D014 000024 (v02 APM )
[ 0.000000] ACPI: XSDT 0x00000043FA82C0E8 000084 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: FACP 0x00000043FA822000 00010C (v05 APM XGENE 00000003 INTL 20140724)
[ 0.000000] ACPI: DSDT 0x00000043FA823000 005837 (v05 APM APM88xxx 00000001 INTL 20140724)
[ 0.000000] ACPI: DBG2 0x00000043FA829000 0000AA (v00 APMC0D XGENEDBG 00000000 INTL 20140724)
[ 0.000000] ACPI: GTDT 0x00000043FA820000 000060 (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: MCFG 0x00000043FA81F000 00003C (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: SPCR 0x00000043FA81E000 000050 (v02 APMC0D XGENESPC 00000000 INTL 20140724)
[ 0.000000] ACPI: SSDT 0x00000043FA81D000 00002D (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: BERT 0x00000043FA81C000 000030 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: HEST 0x00000043FA81B000 0002A8 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: APIC 0x00000043FA81A000 0002A4 (v03 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA819000 000063 (v02 REDHAT MACADDRS 00000001 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA818000 000032 (v02 REDHAT UARTCLKS 00000001 01000013)
[ 0.000000] ACPI: PCCT 0x00000043FA817000 000300 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SPCR: console: uart,mmio,0x1c020000,115200
[ 0.000000] ACPI: NUMA: Failed to initialise from firmware
[ 0.000000] NUMA: Faking a node at [mem 0x0000000000000000-0x00000043ffffffff]
[ 0.000000] NUMA: Adding memblock [0x4000000000 - 0x40001fffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x4000200000 - 0x43fa7cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43fa7d0000 - 0x43faa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43faa20000 - 0x43ff9cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ff9d0000 - 0x43ffa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ffa20000 - 0x43ffffffff] on node 0
[ 0.000000] NUMA: Initmem setup node 0 [mem 0x4000000000-0x43ffffffff]
[ 0.000000] NUMA: NODE_DATA [mem 0x43ffffe200-0x43ffffffff]
[ 0.000000] Zone ranges:
[ 0.000000] DMA [mem 0x0000004000000000-0x00000040ffffffff]
[ 0.000000] Normal [mem 0x0000004100000000-0x00000043ffffffff]
[ 0.000000] Movable zone start for each node
[ 0.000000] Early memory node ranges
[ 0.000000] node 0: [mem 0x0000004000000000-0x00000040001fffff]
[ 0.000000] node 0: [mem 0x0000004000200000-0x00000043fa7cffff]
[ 0.000000] node 0: [mem 0x00000043fa7d0000-0x00000043faa1ffff]
[ 0.000000] node 0: [mem 0x00000043faa20000-0x00000043ff9cffff]
[ 0.000000] node 0: [mem 0x00000043ff9d0000-0x00000043ffa1ffff]
[ 0.000000] node 0: [mem 0x00000043ffa20000-0x00000043ffffffff]
[ 0.000000] Initmem setup node 0 [mem 0x0000004000000000-0x00000043ffffffff]
[ 0.000000] On node 0 totalpages: 262144
[ 0.000000] DMA zone: 64 pages used for memmap
[ 0.000000] DMA zone: 0 pages reserved
[ 0.000000] DMA zone: 65536 pages, LIFO batch:1
[ 0.000000] Normal zone: 192 pages used for memmap
[ 0.000000] Normal zone: 196608 pages, LIFO batch:1
[ 0.000000] psci: is not implemented in ACPI.
[ 0.000000] percpu: Embedded 3 pages/cpu @ffff8003ffdf0000 s124312 r8192 d64104 u196608
[ 0.000000] pcpu-alloc: s124312 r8192 d64104 u196608 alloc=3*65536
[ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3 [0] 4 [0] 5 [0] 6 [0] 7
[ 0.000000] Detected PIPT I-cache on CPU0
[ 0.000000] Built 1 zonelists in Node order, mobility grouping on. Total pages: 261888
[ 0.000000] Policy zone: Normal
[ 0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.11.0-rc8.xgene_msix1+ root=/dev/mapper/rhelp_dhcp--171-root ro crashkernel=auto rd.lvm.lv=rhelp_dhcp-171/root rd.lvm.lv=rhelp_dhcp-171/swap LANG=en_US.UTF-8 acpi=force earlycon=uart8250,mmio32,0x1c020000 console=ttyS0,115200
[ 0.000000] PID hash table entries: 4096 (order: -1, 32768 bytes)
[ 0.000000] software IO TLB [mem 0x40dbff0000-0x40dfff0000] (64MB) mapped at [ffff8000dbff0000-ffff8000dffeffff]
[ 0.000000] Memory: 16109888K/16777216K available (7420K kernel code, 1354K rwdata, 3200K rodata, 1216K init, 7286K bss, 143040K reserved, 524288K cma-reserved)
[ 0.000000] Virtual kernel memory layout:
[ 0.000000] modules : 0xffff000000000000 - 0xffff000008000000 ( 128 MB)
[ 0.000000] vmalloc : 0xffff000008000000 - 0xffff7bdfffff0000 (126847 GB)
[ 0.000000] .text : 0xffff000008080000 - 0xffff0000087c0000 ( 7424 KB)
[ 0.000000] .rodata : 0xffff0000087c0000 - 0xffff000008af0000 ( 3264 KB)
[ 0.000000] .init : 0xffff000008af0000 - 0xffff000008c20000 ( 1216 KB)
[ 0.000000] .data : 0xffff000008c20000 - 0xffff000008d72a00 ( 1355 KB)
[ 0.000000] .bss : 0xffff000008d72a00 - 0xffff0000094902d4 ( 7287 KB)
[ 0.000000] fixed : 0xffff7fdffe7d0000 - 0xffff7fdffec00000 ( 4288 KB)
[ 0.000000] PCI I/O : 0xffff7fdffee00000 - 0xffff7fdfffe00000 ( 16 MB)
[ 0.000000] vmemmap : 0xffff7fe000000000 - 0xffff800000000000 ( 128 GB maximum)
[ 0.000000] 0xffff7fe000000000 - 0xffff7fe001000000 ( 16 MB actual)
[ 0.000000] memory : 0xffff800000000000 - 0xffff800400000000 ( 16384 MB)
[ 0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=8, Nodes=1
[ 0.000000] Hierarchical RCU implementation.
[ 0.000000] Build-time adjustment of leaf fanout to 64.
[ 0.000000] RCU restricting CPUs from NR_CPUS=4096 to nr_cpu_ids=8.
[ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=8
[ 0.000000] NR_IRQS:64 nr_irqs:64 0
[ 0.000000] GIC: Using split EOI/Deactivate mode
[ 0.000000] GICv3: No distributor detected at @ffff000008050000, giving up
[ 0.000000] arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys).
[ 0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns
[ 0.000002] sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns
[ 0.008382] Console: colour dummy device 80x25
[ 0.013031] Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000)
[ 0.023774] pid_max: default: 32768 minimum: 301
[ 0.028553] ACPI: Core revision 20170119
[ 0.036208] ACPI Error: Method parse/execution failed [\_SB.ET00._STA] (Node ffff8003c0195c08), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.049249] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.057204] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.063972] ACPI Error: Method parse/execution failed [\_SB.ET01._STA] (Node ffff8003c0195eb0), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.077125] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.084966] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.092086] ACPI: 4 ACPI AML tables successfully acquired and loaded
[ 0.098656] Security Framework initialized
[ 0.102953] SELinux: Initializing.
[ 0.106548] SELinux: Starting in permissive mode
[ 0.106747] Dentry cache hash table entries: 2097152 (order: 8, 16777216 bytes)
[ 0.116984] Inode-cache hash table entries: 1048576 (order: 7, 8388608 bytes)
[ 0.125658] Mount-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.132758] Mountpoint-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.140711] ftrace: allocating 26814 entries in 7 pages
[ 0.205140] ASID allocator initialised with 65536 entries
[ 0.211274] Remapping and enabling EFI services.
[ 0.216029] EFI remap 0x0000000010510000 => 0000000020000000
[ 0.222058] EFI remap 0x0000000010548000 => 0000000020018000
[ 0.228081] EFI remap 0x0000000017000000 => 0000000020020000
[ 0.234066] EFI remap 0x000000001c025000 => 0000000020035000
[ 0.240084] UEFI Runtime regions are not aligned to 64 KB -- buggy firmware?
[ 0.240089] ------------[ cut here ]------------
[ 0.252134] WARNING: CPU: 0 PID: 1 at arch/arm64/kernel/efi.c:34 efi_create_mapping+0x4c/0x104
[ 0.261010] Modules linked in:
[ 0.265699] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc8.xgene_msix1+ #1
[ 0.273313] Hardware name: APM X-Gene Mustang board (DT)
[ 0.278750] task: ffff8003c1bc0000 task.stack: ffff8003c1c00000
[ 0.284827] PC is at efi_create_mapping+0x4c/0x104
[ 0.289772] LR is at efi_create_mapping+0x4c/0x104
[ 0.294664] pc : [<ffff000008af51f0>] lr : [<ffff000008af51f0>] pstate: 60000045
[ 0.302261] sp : ffff8003c1c03d20
[ 0.305710] x29: ffff8003c1c03d20 x28: 0000000000000000
[ 0.311183] x27: ffff000008c53000 x26: ffff000008d429c8
[ 0.316639] x25: ffff000008d42000 x24: 0000000000000000
[ 0.322102] x23: ffff000008a4c8f0 x22: ffff000008c60000
[ 0.327600] x21: ffff000008c53000 x20: ffff000008d429c8
[ 0.333038] x19: ffff8003f96ab348 x18: 000000003eca74e7
[ 0.338485] x17: 0000000000000000 x16: 0000000000000000
[ 0.343992] x15: ffff000008c53000 x14: 000000004e2b17ac
[ 0.349464] x13: ffff0000089ca500 x12: 0000000000040b00
[ 0.354988] x11: ffff8003c1c039e0 x10: ffff8003c1c039e0
[ 0.360435] x9 : 0000000000000073 x8 : 6620796767756220
[ 0.365898] x7 : 2d2d20424b203436 x6 : ffff000009274718
[ 0.371362] x5 : ffff8003ffdf7530 x4 : 0000000000000000
[ 0.376834] x3 : 0000000000000000 x2 : 0000000000040b00
[ 0.382316] x1 : 0000000000040b00 x0 : 0000000000000040
[ 0.389363] ---[ end trace 4a6935def63e1e16 ]---
[ 0.394126] Call trace:
[ 0.396626] Exception stack(0xffff8003c1c03b30 to 0xffff8003c1c03c60)
[ 0.403246] 3b20: ffff8003f96ab348 0001000000000000
[ 0.411302] 3b40: ffff8003c1c03d20 ffff000008af51f0 0000000060000045 ffff000008d429c8
[ 0.419331] 3b60: 0000000000000000 0000000000000000 ffff8003c1c03d20 ffff8003c1c03d20
[ 0.427377] 3b80: ffff8003c1c03ce0 00000000ffffffc8 ffff8003c1c03be0 ffff00000812f6b8
[ 0.435449] 3ba0: ffff000008c53000 00000000ffffffc8 ffff8003c1c03c00 ffff0000089be038
[ 0.443487] 3bc0: ffff8003c1c03d20 ffff8003c1c03d20 ffff8003c1c03ce0 00000000ffffffc8
[ 0.451585] 3be0: ffff8003c1c03c70 0000000000040b00 0000000000000040 0000000000040b00
[ 0.459666] 3c00: 0000000000040b00 0000000000000000 0000000000000000 ffff8003ffdf7530
[ 0.467729] 3c20: ffff000009274718 2d2d20424b203436 6620796767756220 0000000000000073
[ 0.475784] 3c40: ffff8003c1c039e0 ffff8003c1c039e0 0000000000040b00 ffff0000089ca500
[ 0.483840] [<ffff000008af51f0>] efi_create_mapping+0x4c/0x104
[ 0.489858] [<ffff000008b3db98>] arm_enable_runtime_services+0x154/0x238
[ 0.496817] [<ffff000008083698>] do_one_initcall+0x54/0x15c
[ 0.502634] [<ffff000008af0d18>] kernel_init_freeable+0xe4/0x25c
[ 0.508816] [<ffff0000087ae430>] kernel_init+0x18/0x108
[ 0.514202] [<ffff0000080833c0>] ret_from_fork+0x10/0x50
[ 0.519621] EFI remap 0x00000043fa7df000 => 000000002004f000
[ 0.525659] EFI remap 0x00000043fa82e000 => 000000002006e000
[ 0.531673] EFI remap 0x00000043ff9d6000 => 0000000020266000
[ 0.537700] EFI remap 0x00000043ff9ea000 => 000000002027a000
[ 0.543899] smp: Bringing up secondary CPUs ...
[ 0.548738] Detected PIPT I-cache on CPU1
[ 0.548766] CPU1: Booted secondary processor [500f0001]
[ 0.548987] Detected PIPT I-cache on CPU2
[ 0.549008] CPU2: Booted secondary processor [500f0001]
[ 0.549225] Detected PIPT I-cache on CPU3
[ 0.549239] CPU3: Booted secondary processor [500f0001]
[ 0.549432] Detected PIPT I-cache on CPU4
[ 0.549452] CPU4: Booted secondary processor [500f0001]
[ 0.549653] Detected PIPT I-cache on CPU5
[ 0.549667] CPU5: Booted secondary processor [500f0001]
[ 0.549861] Detected PIPT I-cache on CPU6
[ 0.549882] CPU6: Booted secondary processor [500f0001]
[ 0.550086] Detected PIPT I-cache on CPU7
[ 0.550100] CPU7: Booted secondary processor [500f0001]
[ 0.550142] smp: Brought up 1 node, 8 CPUs
[ 0.621892] SMP: Total of 8 processors activated.
[ 0.626806] CPU features: detected feature: 32-bit EL0 Support
[ 0.632944] CPU: All CPU(s) started at EL2
[ 0.637984] devtmpfs: initialized
[ 0.641749] SMBIOS 3.0.0 present.
[ 0.645224] DMI: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.15 Aug 19 2016
[ 0.654642] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[ 0.664967] futex hash table entries: 2048 (order: 2, 262144 bytes)
[ 0.671981] pinctrl core: initialized pinctrl subsystem
[ 0.677705] NET: Registered protocol family 16
[ 0.682714] cpuidle: using governor menu
[ 0.686960] Detected 8 PCC Subspaces
[ 0.690784] Registering PCC driver as Mailbox controller
[ 0.696410] vdso: 2 pages (1 code @ ffff0000087e0000, 1 data @ ffff000008c40000)
[ 0.704152] hw-breakpoint: found 4 breakpoint and 4 watchpoint registers.
[ 0.713765] DMA: preallocated 256 KiB pool for atomic allocations
[ 0.720154] ACPI: bus type PCI registered
[ 0.724382] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
[ 0.731202] Serial: AMBA PL011 UART driver
[ 0.737465] HugeTLB registered 512 MB page size, pre-allocated 0 pages
[ 0.744898] ACPI: Added _OSI(Module Device)
[ 0.749274] ACPI: Added _OSI(Processor Device)
[ 0.753882] ACPI: Added _OSI(3.0 _SCP Extensions)
[ 0.758804] ACPI: Added _OSI(Processor Aggregator Device)
[ 0.773904] ACPI: Interpreter enabled
[ 0.777753] ACPI: Using GIC for interrupt routing
[ 0.782685] ACPI: MCFG table detected, 1 entries
[ 0.787684] HEST: Table parsing has been initialized.
[ 0.795411] ACPI: Power Resource [SCVR] (on)
[ 0.830232] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff])
[ 0.836749] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.845539] acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug PME AER PCIeCapability]
[ 0.853915] acpi PNP0A08:00: MCFG quirk: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] with xgene_v1_pcie_ecam_ops
[ 0.866664] acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0xe0d0000000-0xe0dfffffff] not reserved in ACPI namespace
[ 0.877881] acpi PNP0A08:00: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff]
[ 0.886116] Remapped I/O 0x000000e010000000 to [io 0x0000-0xffff window]
[ 0.893327] PCI host bridge to bus 0000:00
[ 0.897628] pci_bus 0000:00: root bus resource [io 0x0000-0xffff window] (bus address [0x10000000-0x1000ffff])
[ 0.908283] pci_bus 0000:00: root bus resource [mem 0xe040000000-0xe07fffffff window] (bus address [0x40000000-0x7fffffff])
[ 0.919920] pci_bus 0000:00: root bus resource [mem 0xf000000000-0xffffffffff window]
[ 0.928114] pci_bus 0000:00: root bus resource [bus 00-ff]
[ 0.933821] pci 0000:00:00.0: [10e8:e004] type 01 class 0x060400
[ 0.933899] pci 0000:00:00.0: supports D1 D2
[ 0.934059] pci 0000:01:00.0: [8086:10d3] type 00 class 0x020000
[ 0.934092] pci 0000:01:00.0: reg 0x10: [mem 0xe040080000-0xe04009ffff]
[ 0.934108] pci 0000:01:00.0: reg 0x14: [mem 0xe040000000-0xe04007ffff]
[ 0.934125] pci 0000:01:00.0: reg 0x18: [io 0x0000-0x001f]
[ 0.934141] pci 0000:01:00.0: reg 0x1c: [mem 0xe0400a0000-0xe0400a3fff]
[ 0.934186] pci 0000:01:00.0: reg 0x30: [mem 0xfffc0000-0xffffffff pref]
[ 0.934326] pci 0000:01:00.0: PME# supported from D0 D3hot D3cold
[ 0.963877] pci 0000:00:00.0: BAR 14: assigned [mem 0xe040000000-0xe0400fffff]
[ 0.971433] pci 0000:00:00.0: BAR 13: assigned [io 0x1000-0x1fff]
[ 0.977908] pci 0000:01:00.0: BAR 1: assigned [mem 0xe040000000-0xe04007ffff]
[ 0.985417] pci 0000:01:00.0: BAR 6: assigned [mem 0xe040080000-0xe0400bffff pref]
[ 0.993304] pci 0000:01:00.0: BAR 0: assigned [mem 0xe0400c0000-0xe0400dffff]
[ 1.000812] pci 0000:01:00.0: BAR 3: assigned [mem 0xe0400e0000-0xe0400e3fff]
[ 1.008298] pci 0000:01:00.0: BAR 2: assigned [io 0x1000-0x101f]
[ 1.014701] pci 0000:00:00.0: PCI bridge to [bus 01]
[ 1.019893] pci 0000:00:00.0: bridge window [io 0x1000-0x1fff]
[ 1.026276] pci 0000:00:00.0: bridge window [mem 0xe040000000-0xe0400fffff]
[ 1.034419] vgaarb: loaded
[ 1.037393] SCSI subsystem initialized
[ 1.041366] libata version 3.00 loaded.
[ 1.041403] ACPI: bus type USB registered
[ 1.045616] usbcore: registered new interface driver usbfs
[ 1.051373] usbcore: registered new interface driver hub
[ 1.056947] usbcore: registered new device driver usb
[ 1.062355] Registered efivars operations
[ 1.066820] NetLabel: Initializing
[ 1.070418] NetLabel: domain hash size = 128
[ 1.074997] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO
[ 1.080980] NetLabel: unlabeled traffic allowed by default
[ 1.086945] clocksource: Switched to clocksource arch_sys_counter
[ 1.111811] VFS: Disk quotas dquot_6.6.0
[ 1.115980] VFS: Dquot-cache hash table entries: 8192 (order 0, 65536 bytes)
[ 1.123569] pnp: PnP ACPI init
[ 1.127405] pnp: PnP ACPI: found 0 devices
[ 1.134249] NET: Registered protocol family 2
[ 1.139069] TCP established hash table entries: 131072 (order: 4, 1048576 bytes)
[ 1.147156] TCP bind hash table entries: 65536 (order: 4, 1048576 bytes)
[ 1.154513] TCP: Hash tables configured (established 131072 bind 65536)
[ 1.161541] UDP hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.168029] UDP-Lite hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.175030] NET: Registered protocol family 1
[ 1.179607] PCI: CLS 64 bytes, default 128
[ 1.179708] Unpacking initramfs...
[ 1.863109] Freeing initrd memory: 32640K
[ 1.867699] kvm [1]: 8-bit VMID
[ 1.870944] kvm [1]: IDMAP page: 40009b7000
[ 1.875301] kvm [1]: HYP VA range: 800000000000:ffffffffffff
[ 1.881276] kvm [1]: Hyp mode initialized successfully
[ 1.886624] kvm [1]: GICV region size/alignment is unsafe, using trapping (reduced performance)
[ 1.895748] kvm [1]: vgic-v2@780cf000
[ 1.899646] kvm [1]: vgic interrupt IRQ1
[ 1.903757] kvm [1]: virtual timer IRQ4
[ 1.923421] audit: initializing netlink subsys (disabled)
[ 1.929159] audit: type=2000 audit(1.470:1): state=initialized audit_enabled=0 res=1
[ 1.929303] Initialise system trusted keyrings
[ 1.929373] workingset: timestamp_bits=44 max_order=18 bucket_order=0
[ 1.931445] zbud: loaded
[ 1.931829] SELinux: Registering netfilter hooks
[ 2.011199] NET: Registered protocol family 38
[ 2.015847] Key type asymmetric registered
[ 2.020145] Asymmetric key parser 'x509' registered
[ 2.025283] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
[ 2.033081] io scheduler noop registered
[ 2.037167] io scheduler deadline registered (default)
[ 2.042536] io scheduler cfq registered
[ 2.046565] io scheduler mq-deadline registered
[ 2.051474] atomic64_test: passed
[ 2.055189] xgene-gpio APMC0D14:00: X-Gene GPIO driver registered.
[ 2.062003] pcieport 0000:00:00.0: AER enabled with IRQ 83
[ 2.067787] pcieport 0000:00:00.0: Signaling PME with IRQ 83
[ 2.073972] input: Power Button as /devices/LNXSYSTM:00/PNP0C0C:00/input/input0
[ 2.081619] ACPI: Power Button [PWRB]
[ 2.085922] GHES: Failed to map GSI to IRQ for generic hardware error source: 2
[ 2.093593] GHES: probe of GHES.2 failed with error 82
[ 2.098936] GHES: Failed to map GSI to IRQ for generic hardware error source: 3
[ 2.106588] GHES: probe of GHES.3 failed with error 82
[ 2.111936] GHES: Failed to map GSI to IRQ for generic hardware error source: 4
[ 2.119608] GHES: probe of GHES.4 failed with error 82
[ 2.125011] GHES: Failed to map GSI to IRQ for generic hardware error source: 5
[ 2.132659] GHES: probe of GHES.5 failed with error 82
[ 2.138017] GHES: Failed to map GSI to IRQ for generic hardware error source: 6
[ 2.145633] GHES: probe of GHES.6 failed with error 82
[ 2.151042] GHES: Failed to map GSI to IRQ for generic hardware error source: 7
[ 2.158724] GHES: probe of GHES.7 failed with error 82
[ 2.164126] GHES: Failed to map GSI to IRQ for generic hardware error source: 8
[ 2.171797] GHES: probe of GHES.8 failed with error 82
[ 2.177160] GHES: Failed to map GSI to IRQ for generic hardware error source: 9
[ 2.184845] GHES: probe of GHES.9 failed with error 82
[ 2.190233] GHES: Failed to enable APEI firmware first mode.
[ 2.196297] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[ 2.203376] console [ttyS0] disabled
[ 2.227246] APMC0D08:00: ttyS0 at MMIO 0x1c020000 (irq = 22, base_baud = 3125000) is a U6_16550A
[ 2.236527] console [ttyS0] enabled
[ 2.243657] bootconsole [uart8250] disabled
[ 2.272506] APMC0D08:01: ttyS1 at MMIO 0x1c021000 (irq = 23, base_baud = 3125000) is a U6_16550A
[ 2.281671] cacheinfo: Unable to detect cache hierarchy for CPU 0
[ 2.287755] rdac: device handler registered
[ 2.291963] hp_sw: device handler registered
[ 2.296209] emc: device handler registered
[ 2.300380] alua: device handler registered
[ 2.304688] libphy: Fixed MDIO Bus: probed
[ 2.308849] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
[ 2.315346] ehci-pci: EHCI PCI platform driver
[ 2.319790] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver
[ 2.325941] ohci-pci: OHCI PCI platform driver
[ 2.330377] uhci_hcd: USB Universal Host Controller Interface driver
[ 2.336788] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.341980] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 1
[ 2.349425] xhci-hcd 808622B7:00: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.357756] xhci-hcd 808622B7:00: irq 29, io mem 0x19000000
[ 2.363370] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.370130] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.377318] usb usb1: Product: xHCI Host Controller
[ 2.382169] usb usb1: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.389010] usb usb1: SerialNumber: 808622B7:00
[ 2.393727] hub 1-0:1.0: USB hub found
[ 2.397479] hub 1-0:1.0: 1 port detected
[ 2.401480] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.406677] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 2
[ 2.413986] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.422078] usb usb2: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.428836] usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.436020] usb usb2: Product: xHCI Host Controller
[ 2.440875] usb usb2: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.447720] usb usb2: SerialNumber: 808622B7:00
[ 2.452398] hub 2-0:1.0: USB hub found
[ 2.456137] hub 2-0:1.0: 1 port detected
[ 2.460168] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.465356] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 3
[ 2.472759] xhci-hcd 808622B7:01: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.481087] xhci-hcd 808622B7:01: irq 30, io mem 0x19800000
[ 2.486684] usb usb3: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.493443] usb usb3: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.500630] usb usb3: Product: xHCI Host Controller
[ 2.505482] usb usb3: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.512327] usb usb3: SerialNumber: 808622B7:01
[ 2.517049] hub 3-0:1.0: USB hub found
[ 2.520789] hub 3-0:1.0: 1 port detected
[ 2.524790] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.529976] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 4
[ 2.537284] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.545370] usb usb4: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.552128] usb usb4: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.559317] usb usb4: Product: xHCI Host Controller
[ 2.564168] usb usb4: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.571013] usb usb4: SerialNumber: 808622B7:01
[ 2.575716] hub 4-0:1.0: USB hub found
[ 2.579464] hub 4-0:1.0: 1 port detected
[ 2.583505] usbcore: registered new interface driver usbserial
[ 2.589324] usbcore: registered new interface driver usbserial_generic
[ 2.595826] usbserial: USB Serial support registered for generic
[ 2.601884] mousedev: PS/2 mouse device common for all mice
[ 2.607620] rtc-efi rtc-efi: rtc core: registered rtc-efi as rtc0
[ 2.613906] EFI Variables Facility v0.08 2004-May-17
[ 2.619081] hidraw: raw HID events driver (C) Jiri Kosina
[ 2.624514] usbcore: registered new interface driver usbhid
[ 2.630067] usbhid: USB HID core driver
[ 2.633953] drop_monitor: Initializing network drop monitor service
[ 2.640280] Initializing XFRM netlink socket
[ 2.644691] NET: Registered protocol family 10
[ 2.649496] Segment Routing with IPv6
[ 2.653160] NET: Registered protocol family 17
[ 2.657841] registered taskstats version 1
[ 2.661921] Loading compiled-in X.509 certificates
[ 2.669794] alg: No test for pkcs1pad(rsa,sha1) (pkcs1pad(rsa-generic,sha1))
[ 2.678237] Loaded X.509 cert 'Build time autogenerated kernel key: 98aa1b1964f4052ba920878b6f8348e178459895'
[ 2.688204] zswap: loaded using pool lzo/zbud
[ 2.713270] Key type big_key registered
[ 2.717168] BERT: [Firmware Bug]: table invalid.
[ 2.721839] rtc-efi rtc-efi: setting system clock to 2017-04-28 16:00:41 UTC (1493395241)
[ 2.757884] Freeing unused kernel memory: 1216K
[ 2.767418] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.774589] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.781211] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.789666] systemd[1]: systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL +XZ -LZ4 -SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
[ 2.807890] systemd[1]: Detected architecture arm64.
[ 2.812833] systemd[1]: Running in initial RAM disk.
[ 2.847031] systemd[1]: Set hostname to <localhost.localdomain>.
[ 2.979499] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.986233] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.993080] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.000078] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.010094] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.017557] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.025329] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.087174] systemd[1]: Reached target Swap.
[ 3.091495] systemd[1]: Starting Swap.
[ 3.117069] systemd[1]: Reached target Timers.
[ 3.121551] systemd[1]: Starting Timers.
[ 3.157062] systemd[1]: Reached target Local File Systems.
[ 3.162579] systemd[1]: Starting Local File Systems.
[ 4.087490] device-mapper: uevent: version 1.0.3
[ 4.092424] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel@redhat.com
[ 4.400761] libphy: APM X-Gene MDIO bus: probed
[ 4.444412] pps_core: LinuxPPS API ver. 1 registered
[ 4.449417] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[ 4.450547] xgene-ahci APMC0D0D:00: skip clock and PHY initialization
[ 4.450561] xgene-ahci APMC0D0D:00: controller can't do NCQ, turning off CAP_NCQ
[ 4.450629] xgene-ahci APMC0D0D:00: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.450643] xgene-ahci APMC0D0D:00: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.450670] xgene-ahci APMC0D0D:00: port 0 is not capable of FBS
[ 4.491029] sdhci: Secure Digital Host Controller Interface driver
[ 4.491033] sdhci: Copyright(c) Pierre Ossman
[ 4.512024] PTP clock support registered
[ 4.512754] xgene-ahci APMC0D0D:00: port 1 is not capable of FBS
[ 4.525820] scsi host0: xgene-ahci
[ 4.529855] scsi host1: xgene-ahci
[ 4.533507] ata1: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x100 irq 27
[ 4.534007] mmc0: SDHCI controller on ACPI [APMC0D0C:00] using PIO
[ 4.547673] ata2: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x180 irq 27
[ 4.548138] xgene-enet APMC0D05:00: clocks have been setup already
[ 4.553893] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k
[ 4.553894] e1000e: Copyright(c) 1999 - 2015 Intel Corporation.
[ 4.554194] e1000e 0000:01:00.0: enabling device (0000 -> 0002)
[ 4.554351] e1000e 0000:01:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[ 4.554379] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): Failed to initialize MSI-X interrupts. Falling back to MSI interrupts.
[ 4.554394] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): Failed to initialize MSI interrupts. Falling back to legacy interrupts.
[ 4.617549] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): registered PHC clock
[ 4.647217] xgene-enet APMC0D30:00: clocks have been setup already
[ 4.665238] xgene-ahci APMC0D0D:01: skip clock and PHY initialization
[ 4.671721] xgene-ahci APMC0D0D:01: controller can't do NCQ, turning off CAP_NCQ
[ 4.679178] xgene-ahci APMC0D0D:01: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.688066] xgene-ahci APMC0D0D:01: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.693143] e1000e 0000:01:00.0 eth2: (PCI Express:2.5GT/s:Width x1) 68:05:ca:46:ce:08
[ 4.693146] e1000e 0000:01:00.0 eth2: Intel(R) PRO/1000 Network Connection
[ 4.693162] e1000e 0000:01:00.0 eth2: MAC: 3, PHY: 8, PBA No: E46981-008
[ 4.696649] e1000e 0000:01:00.0 enp1s0: renamed from eth2
[ 4.722971] xgene-ahci APMC0D0D:01: port 0 is not capable of FBS
[ 4.729089] xgene-ahci APMC0D0D:01: port 1 is not capable of FBS
[ 4.736197] scsi host2: xgene-ahci
[ 4.740144] scsi host3: xgene-ahci
[ 4.743793] ata3: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x100 irq 28
[ 4.751735] ata4: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x180 irq 28
[ 4.857119] xgene-enet APMC0D30:01: clocks have been setup already
[ 4.874076] xgene-enet APMC0D31:00: clocks have been setup already
[ 5.088466] ata3: SATA link down (SStatus 0 SControl 4300)
[ 5.099014] ata4: SATA link down (SStatus 0 SControl 4300)
[ 5.217012] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.223552] ata1.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.230179] ata1.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.236797] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.243335] ata2.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.249961] ata2.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.256600] ata1.00: configured for UDMA/133
[ 5.261153] ata2.00: configured for UDMA/133
[ 5.261162] scsi 0:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.357513] scsi 1:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.357514] sd 0:0:0:0: [sda] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.357539] sd 0:0:0:0: [sda] Write Protect is off
[ 5.357542] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[ 5.357575] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.360081] sda: sda1 sda2 sda3
[ 5.360645] sd 0:0:0:0: [sda] Attached SCSI disk
[ 5.487441] sd 1:0:0:0: [sdb] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.494939] sd 1:0:0:0: [sdb] Write Protect is off
[ 5.499723] sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00
[ 5.499757] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.511144] sdb: sdb1
[ 5.513907] sd 1:0:0:0: [sdb] Attached SCSI disk
[ 5.962036] random: fast init done
[ 6.345919] SGI XFS with ACLs, security attributes, no debug enabled
[ 6.356572] XFS (dm-0): Mounting V5 Filesystem
[ 6.398617] XFS (dm-0): Ending clean mount
[ 6.973554] systemd-journald[198]: Received SIGTERM from PID 1 (systemd).
[ 6.989018] systemd: 20 output lines suppressed due to ratelimiting
[ 7.019839] audit: type=1404 audit(1493395245.790:2): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295
[ 7.106438] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.137061] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.216426] SELinux: 8 users, 14 roles, 4990 types, 304 bools, 1 sens, 1024 cats
[ 7.216433] SELinux: 94 classes, 105786 rules
[ 7.223909] SELinux: Class sctp_socket not defined in policy.
[ 7.229741] SELinux: Class icmp_socket not defined in policy.
[ 7.235543] SELinux: Class ax25_socket not defined in policy.
[ 7.241349] SELinux: Class ipx_socket not defined in policy.
[ 7.247069] SELinux: Class netrom_socket not defined in policy.
[ 7.253043] SELinux: Class atmpvc_socket not defined in policy.
[ 7.259021] SELinux: Class x25_socket not defined in policy.
[ 7.264736] SELinux: Class rose_socket not defined in policy.
[ 7.270540] SELinux: Class decnet_socket not defined in policy.
[ 7.276513] SELinux: Class atmsvc_socket not defined in policy.
[ 7.282490] SELinux: Class rds_socket not defined in policy.
[ 7.288207] SELinux: Class irda_socket not defined in policy.
[ 7.294008] SELinux: Class pppox_socket not defined in policy.
[ 7.299898] SELinux: Class llc_socket not defined in policy.
[ 7.305613] SELinux: Class can_socket not defined in policy.
[ 7.311330] SELinux: Class tipc_socket not defined in policy.
[ 7.317135] SELinux: Class bluetooth_socket not defined in policy.
[ 7.323368] SELinux: Class iucv_socket not defined in policy.
[ 7.329172] SELinux: Class rxrpc_socket not defined in policy.
[ 7.335059] SELinux: Class isdn_socket not defined in policy.
[ 7.340863] SELinux: Class phonet_socket not defined in policy.
[ 7.346836] SELinux: Class ieee802154_socket not defined in policy.
[ 7.353159] SELinux: Class caif_socket not defined in policy.
[ 7.358965] SELinux: Class alg_socket not defined in policy.
[ 7.364679] SELinux: Class nfc_socket not defined in policy.
[ 7.370397] SELinux: Class vsock_socket not defined in policy.
[ 7.376284] SELinux: Class kcm_socket not defined in policy.
[ 7.382002] SELinux: Class qipcrtr_socket not defined in policy.
[ 7.388066] SELinux: Class smc_socket not defined in policy.
[ 7.393780] SELinux: the above unknown classes and permissions will be allowed
[ 7.400979] SELinux: Completing initialization.
[ 7.400980] SELinux: Setting up existing superblocks.
[ 7.450804] audit: type=1403 audit(1493395246.220:3): policy loaded auid=4294967295 ses=4294967295
[ 7.455100] systemd[1]: Successfully loaded SELinux policy in 435.279ms.
[ 7.470770] systemd[1]: RTC configured in localtime, applying delta of -240 minutes to system time.
[ 7.497588] ip_tables: (C) 2000-2006 Netfilter Core Team
[ 7.502902] systemd[1]: Inserted module 'ip_tables'
[ 7.552579] systemd[1]: Relabelled /dev and /run in 42.787ms.
[ 8.005378] systemd-journald[616]: Received request to flush runtime journal from PID 1
[ 8.167863] xgene-slimpro-mbox APMC0D01:00: APM X-Gene SLIMpro MailBox registered
[ 8.172785] RPC: Registered named UNIX socket transport module.
[ 8.172790] RPC: Registered udp transport module.
[ 8.172794] RPC: Registered tcp transport module.
[ 8.172797] RPC: Registered tcp NFSv4.1 backchannel transport module.
[ 8.197228] xgene-rng APMC0D18:00: Couldn't get the clock for RNG
[ 8.204225] xgene-gpio-sb APMC0D15:00: Support 22 gpios, 6 irqs start from pin 8
[ 8.213102] xgene-gpio-sb APMC0D15:00: X-Gene GPIO Standby driver registered
[ 8.223510] xgene-slimpro-i2c APMC0D40:00: i2c mailbox channel request failed
[ 8.289837] sd 0:0:0:0: Attached scsi generic sg0 type 0
[ 8.295454] sd 1:0:0:0: Attached scsi generic sg1 type 0
[ 8.387796] xgene-slimpro-hwmon APMC0D29:00: APM X-Gene SoC HW monitor driver registered
[ 8.518644] Adding 8388544k swap on /dev/mapper/rhelp_dhcp--171-swap. Priority:-1 extents:1 across:8388544k SSFS
[ 8.531009] XFS (sda2): Mounting V5 Filesystem
[ 8.551661] XFS (sda2): Ending clean mount
[ 8.640881] Installing knfsd (copyright (C) 1996 okir@monad.swb.de).
[ 8.702746] XFS (dm-2): Mounting V5 Filesystem
[ 8.721803] XFS (dm-2): Ending clean mount
[ 9.540607] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.637501] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.650581] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.656789] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.669293] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.675510] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.687727] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.693923] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.706099] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 9.712318] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 9.730562] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.737469] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.744174] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.751066] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.757808] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.687474] xgene-enet APMC0D30:00 eth1: Link is Down
[ 10.692580] xgene-enet APMC0D05:00 eth0: Link is Down
[ 10.767324] xgene-enet APMC0D30:01 eth2: Link is Down
[ 12.418373] e1000e: enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
[ 12.426374] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready
[ 13.808268] xgene-enet APMC0D05:00 eth0: Link is Up - 1Gbps/Full - flow control off
[ 13.815925] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 108.537236] random: crng init done
[-- Attachment #3: dmesg_4.11.0-rc8.xgene_msix2+_msi.txt --]
[-- Type: text/plain, Size: 39010 bytes --]
[ 0.000000] Booting Linux on physical CPU 0x0
[ 0.000000] Linux version 4.11.0-rc8.xgene_msix2+ (root@mustang4.bos.jonmasters.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-14) (GCC) ) #2 SMP Fri Apr 28 12:06:14 EDT 2017
[ 0.000000] Boot CPU: AArch64 Processor [500f0001]
[ 0.000000] earlycon: uart8250 at MMIO32 0x000000001c020000 (options '')
[ 0.000000] bootconsole [uart8250] enabled
[ 0.000000] efi: Getting EFI parameters from FDT:
[ 0.000000] efi: EFI v2.40 by X-Gene Mustang Board EFI Aug 19 2016 18:02:24
[ 0.000000] efi: ACPI=0x43fa82d000 ACPI 2.0=0x43fa82d014 SMBIOS 3.0=0x43fa82b000 ESRT=0x43ff053d18
[ 0.000000] esrt: Reserving ESRT space from 0x00000043ff053d18 to 0x00000043ff053d78.
[ 0.000000] cma: Reserved 512 MiB at 0x00000040e0000000
[ 0.000000] ACPI: Early table checksum verification disabled
[ 0.000000] ACPI: RSDP 0x00000043FA82D014 000024 (v02 APM )
[ 0.000000] ACPI: XSDT 0x00000043FA82C0E8 000084 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: FACP 0x00000043FA822000 00010C (v05 APM XGENE 00000003 INTL 20140724)
[ 0.000000] ACPI: DSDT 0x00000043FA823000 005837 (v05 APM APM88xxx 00000001 INTL 20140724)
[ 0.000000] ACPI: DBG2 0x00000043FA829000 0000AA (v00 APMC0D XGENEDBG 00000000 INTL 20140724)
[ 0.000000] ACPI: GTDT 0x00000043FA820000 000060 (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: MCFG 0x00000043FA81F000 00003C (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: SPCR 0x00000043FA81E000 000050 (v02 APMC0D XGENESPC 00000000 INTL 20140724)
[ 0.000000] ACPI: SSDT 0x00000043FA81D000 00002D (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: BERT 0x00000043FA81C000 000030 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: HEST 0x00000043FA81B000 0002A8 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: APIC 0x00000043FA81A000 0002A4 (v03 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA819000 000063 (v02 REDHAT MACADDRS 00000001 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA818000 000032 (v02 REDHAT UARTCLKS 00000001 01000013)
[ 0.000000] ACPI: PCCT 0x00000043FA817000 000300 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SPCR: console: uart,mmio,0x1c020000,115200
[ 0.000000] ACPI: NUMA: Failed to initialise from firmware
[ 0.000000] NUMA: Faking a node at [mem 0x0000000000000000-0x00000043ffffffff]
[ 0.000000] NUMA: Adding memblock [0x4000000000 - 0x40001fffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x4000200000 - 0x43fa7cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43fa7d0000 - 0x43faa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43faa20000 - 0x43ff9cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ff9d0000 - 0x43ffa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ffa20000 - 0x43ffffffff] on node 0
[ 0.000000] NUMA: Initmem setup node 0 [mem 0x4000000000-0x43ffffffff]
[ 0.000000] NUMA: NODE_DATA [mem 0x43ffffe200-0x43ffffffff]
[ 0.000000] Zone ranges:
[ 0.000000] DMA [mem 0x0000004000000000-0x00000040ffffffff]
[ 0.000000] Normal [mem 0x0000004100000000-0x00000043ffffffff]
[ 0.000000] Movable zone start for each node
[ 0.000000] Early memory node ranges
[ 0.000000] node 0: [mem 0x0000004000000000-0x00000040001fffff]
[ 0.000000] node 0: [mem 0x0000004000200000-0x00000043fa7cffff]
[ 0.000000] node 0: [mem 0x00000043fa7d0000-0x00000043faa1ffff]
[ 0.000000] node 0: [mem 0x00000043faa20000-0x00000043ff9cffff]
[ 0.000000] node 0: [mem 0x00000043ff9d0000-0x00000043ffa1ffff]
[ 0.000000] node 0: [mem 0x00000043ffa20000-0x00000043ffffffff]
[ 0.000000] Initmem setup node 0 [mem 0x0000004000000000-0x00000043ffffffff]
[ 0.000000] On node 0 totalpages: 262144
[ 0.000000] DMA zone: 64 pages used for memmap
[ 0.000000] DMA zone: 0 pages reserved
[ 0.000000] DMA zone: 65536 pages, LIFO batch:1
[ 0.000000] Normal zone: 192 pages used for memmap
[ 0.000000] Normal zone: 196608 pages, LIFO batch:1
[ 0.000000] psci: is not implemented in ACPI.
[ 0.000000] percpu: Embedded 3 pages/cpu @ffff8003ffdf0000 s124312 r8192 d64104 u196608
[ 0.000000] pcpu-alloc: s124312 r8192 d64104 u196608 alloc=3*65536
[ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3 [0] 4 [0] 5 [0] 6 [0] 7
[ 0.000000] Detected PIPT I-cache on CPU0
[ 0.000000] Built 1 zonelists in Node order, mobility grouping on. Total pages: 261888
[ 0.000000] Policy zone: Normal
[ 0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.11.0-rc8.xgene_msix2+ root=/dev/mapper/rhelp_dhcp--171-root ro crashkernel=auto rd.lvm.lv=rhelp_dhcp-171/root rd.lvm.lv=rhelp_dhcp-171/swap LANG=en_US.UTF-8 acpi=force earlycon=uart8250,mmio32,0x1c020000 console=ttyS0,115200
[ 0.000000] PID hash table entries: 4096 (order: -1, 32768 bytes)
[ 0.000000] software IO TLB [mem 0x40dbff0000-0x40dfff0000] (64MB) mapped at [ffff8000dbff0000-ffff8000dffeffff]
[ 0.000000] Memory: 16109888K/16777216K available (7420K kernel code, 1354K rwdata, 3200K rodata, 1216K init, 7286K bss, 143040K reserved, 524288K cma-reserved)
[ 0.000000] Virtual kernel memory layout:
[ 0.000000] modules : 0xffff000000000000 - 0xffff000008000000 ( 128 MB)
[ 0.000000] vmalloc : 0xffff000008000000 - 0xffff7bdfffff0000 (126847 GB)
[ 0.000000] .text : 0xffff000008080000 - 0xffff0000087c0000 ( 7424 KB)
[ 0.000000] .rodata : 0xffff0000087c0000 - 0xffff000008af0000 ( 3264 KB)
[ 0.000000] .init : 0xffff000008af0000 - 0xffff000008c20000 ( 1216 KB)
[ 0.000000] .data : 0xffff000008c20000 - 0xffff000008d72a00 ( 1355 KB)
[ 0.000000] .bss : 0xffff000008d72a00 - 0xffff0000094902d4 ( 7287 KB)
[ 0.000000] fixed : 0xffff7fdffe7d0000 - 0xffff7fdffec00000 ( 4288 KB)
[ 0.000000] PCI I/O : 0xffff7fdffee00000 - 0xffff7fdfffe00000 ( 16 MB)
[ 0.000000] vmemmap : 0xffff7fe000000000 - 0xffff800000000000 ( 128 GB maximum)
[ 0.000000] 0xffff7fe000000000 - 0xffff7fe001000000 ( 16 MB actual)
[ 0.000000] memory : 0xffff800000000000 - 0xffff800400000000 ( 16384 MB)
[ 0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=8, Nodes=1
[ 0.000000] Hierarchical RCU implementation.
[ 0.000000] Build-time adjustment of leaf fanout to 64.
[ 0.000000] RCU restricting CPUs from NR_CPUS=4096 to nr_cpu_ids=8.
[ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=8
[ 0.000000] NR_IRQS:64 nr_irqs:64 0
[ 0.000000] GIC: Using split EOI/Deactivate mode
[ 0.000000] GICv3: No distributor detected at @ffff000008050000, giving up
[ 0.000000] arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys).
[ 0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns
[ 0.000002] sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns
[ 0.008379] Console: colour dummy device 80x25
[ 0.012950] Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000)
[ 0.023693] pid_max: default: 32768 minimum: 301
[ 0.028471] ACPI: Core revision 20170119
[ 0.036063] ACPI Error: Method parse/execution failed [\_SB.ET00._STA] (Node ffff8003c0195c08), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.049121] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.056989] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.063808] ACPI Error: Method parse/execution failed [\_SB.ET01._STA] (Node ffff8003c0195eb0), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.076952] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.084871] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.092015] ACPI: 4 ACPI AML tables successfully acquired and loaded
[ 0.098567] Security Framework initialized
[ 0.102838] SELinux: Initializing.
[ 0.106408] SELinux: Starting in permissive mode
[ 0.106607] Dentry cache hash table entries: 2097152 (order: 8, 16777216 bytes)
[ 0.116844] Inode-cache hash table entries: 1048576 (order: 7, 8388608 bytes)
[ 0.125552] Mount-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.132626] Mountpoint-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.140570] ftrace: allocating 26815 entries in 7 pages
[ 0.203748] ASID allocator initialised with 65536 entries
[ 0.209911] Remapping and enabling EFI services.
[ 0.214705] EFI remap 0x0000000010510000 => 0000000020000000
[ 0.220700] EFI remap 0x0000000010548000 => 0000000020018000
[ 0.226689] EFI remap 0x0000000017000000 => 0000000020020000
[ 0.232657] EFI remap 0x000000001c025000 => 0000000020035000
[ 0.238736] UEFI Runtime regions are not aligned to 64 KB -- buggy firmware?
[ 0.238741] ------------[ cut here ]------------
[ 0.250727] WARNING: CPU: 0 PID: 1 at arch/arm64/kernel/efi.c:34 efi_create_mapping+0x4c/0x104
[ 0.259532] Modules linked in:
[ 0.264214] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc8.xgene_msix2+ #2
[ 0.271827] Hardware name: APM X-Gene Mustang board (DT)
[ 0.277238] task: ffff8003c1bc0000 task.stack: ffff8003c1c00000
[ 0.283324] PC is at efi_create_mapping+0x4c/0x104
[ 0.288269] LR is at efi_create_mapping+0x4c/0x104
[ 0.293127] pc : [<ffff000008af51f0>] lr : [<ffff000008af51f0>] pstate: 60000045
[ 0.300784] sp : ffff8003c1c03d20
[ 0.304207] x29: ffff8003c1c03d20 x28: 0000000000000000
[ 0.309655] x27: ffff000008c53000 x26: ffff000008d429c8
[ 0.315110] x25: ffff000008d42000 x24: 0000000000000000
[ 0.320556] x23: ffff000008a4c910 x22: ffff000008c60000
[ 0.325994] x21: ffff000008c53000 x20: ffff000008d429c8
[ 0.331484] x19: ffff8003f96ab348 x18: 00000000b6a2110e
[ 0.337008] x17: 0000000000000000 x16: 0000000000000000
[ 0.342437] x15: ffff000008c53000 x14: 00000000cb9aed1d
[ 0.347944] x13: ffff0000089ca500 x12: 0000000000040b00
[ 0.353425] x11: ffff8003c1c039e0 x10: ffff8003c1c039e0
[ 0.358889] x9 : 0000000000000073 x8 : 6620796767756220
[ 0.364352] x7 : 2d2d20424b203436 x6 : ffff000009274718
[ 0.369825] x5 : ffff8003ffdf7530 x4 : 0000000000000000
[ 0.375280] x3 : 0000000000000000 x2 : 0000000000040b00
[ 0.380769] x1 : 0000000000040b00 x0 : 0000000000000040
[ 0.387782] ---[ end trace 9580655578fda7c1 ]---
[ 0.392519] Call trace:
[ 0.395045] Exception stack(0xffff8003c1c03b30 to 0xffff8003c1c03c60)
[ 0.401735] 3b20: ffff8003f96ab348 0001000000000000
[ 0.409730] 3b40: ffff8003c1c03d20 ffff000008af51f0 0000000060000045 ffff000008d429c8
[ 0.417828] 3b60: 0000000000000000 0000000000000000 ffff8003c1c03d20 ffff8003c1c03d20
[ 0.425857] 3b80: ffff8003c1c03ce0 00000000ffffffc8 ffff8003c1c03be0 ffff00000812f6b8
[ 0.433946] 3ba0: ffff000008c53000 00000000ffffffc8 ffff8003c1c03c00 ffff0000089be038
[ 0.442053] 3bc0: ffff8003c1c03d20 ffff8003c1c03d20 ffff8003c1c03ce0 00000000ffffffc8
[ 0.450126] 3be0: ffff8003c1c03c70 0000000000040b00 0000000000000040 0000000000040b00
[ 0.458207] 3c00: 0000000000040b00 0000000000000000 0000000000000000 ffff8003ffdf7530
[ 0.466227] 3c20: ffff000009274718 2d2d20424b203436 6620796767756220 0000000000000073
[ 0.474282] 3c40: ffff8003c1c039e0 ffff8003c1c039e0 0000000000040b00 ffff0000089ca500
[ 0.482321] [<ffff000008af51f0>] efi_create_mapping+0x4c/0x104
[ 0.488270] [<ffff000008b3db98>] arm_enable_runtime_services+0x154/0x238
[ 0.495238] [<ffff000008083698>] do_one_initcall+0x54/0x15c
[ 0.501030] [<ffff000008af0d18>] kernel_init_freeable+0xe4/0x25c
[ 0.507221] [<ffff0000087ae488>] kernel_init+0x18/0x108
[ 0.512606] [<ffff0000080833c0>] ret_from_fork+0x10/0x50
[ 0.518077] EFI remap 0x00000043fa7df000 => 000000002004f000
[ 0.524089] EFI remap 0x00000043fa82e000 => 000000002006e000
[ 0.530085] EFI remap 0x00000043ff9d6000 => 0000000020266000
[ 0.536148] EFI remap 0x00000043ff9ea000 => 000000002027a000
[ 0.542320] smp: Bringing up secondary CPUs ...
[ 0.547164] Detected PIPT I-cache on CPU1
[ 0.547192] CPU1: Booted secondary processor [500f0001]
[ 0.547411] Detected PIPT I-cache on CPU2
[ 0.547433] CPU2: Booted secondary processor [500f0001]
[ 0.547646] Detected PIPT I-cache on CPU3
[ 0.547660] CPU3: Booted secondary processor [500f0001]
[ 0.547854] Detected PIPT I-cache on CPU4
[ 0.547875] CPU4: Booted secondary processor [500f0001]
[ 0.548076] Detected PIPT I-cache on CPU5
[ 0.548090] CPU5: Booted secondary processor [500f0001]
[ 0.548281] Detected PIPT I-cache on CPU6
[ 0.548302] CPU6: Booted secondary processor [500f0001]
[ 0.548505] Detected PIPT I-cache on CPU7
[ 0.548519] CPU7: Booted secondary processor [500f0001]
[ 0.548562] smp: Brought up 1 node, 8 CPUs
[ 0.620493] SMP: Total of 8 processors activated.
[ 0.625425] CPU features: detected feature: 32-bit EL0 Support
[ 0.631511] CPU: All CPU(s) started at EL2
[ 0.636531] devtmpfs: initialized
[ 0.640271] SMBIOS 3.0.0 present.
[ 0.643745] DMI: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.15 Aug 19 2016
[ 0.653110] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[ 0.663392] futex hash table entries: 2048 (order: 2, 262144 bytes)
[ 0.670383] pinctrl core: initialized pinctrl subsystem
[ 0.676091] NET: Registered protocol family 16
[ 0.681082] cpuidle: using governor menu
[ 0.685276] Detected 8 PCC Subspaces
[ 0.689040] Registering PCC driver as Mailbox controller
[ 0.694649] vdso: 2 pages (1 code @ ffff0000087e0000, 1 data @ ffff000008c40000)
[ 0.702382] hw-breakpoint: found 4 breakpoint and 4 watchpoint registers.
[ 0.711945] DMA: preallocated 256 KiB pool for atomic allocations
[ 0.718368] ACPI: bus type PCI registered
[ 0.722572] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
[ 0.729340] Serial: AMBA PL011 UART driver
[ 0.735597] HugeTLB registered 512 MB page size, pre-allocated 0 pages
[ 0.743073] ACPI: Added _OSI(Module Device)
[ 0.747432] ACPI: Added _OSI(Processor Device)
[ 0.752083] ACPI: Added _OSI(3.0 _SCP Extensions)
[ 0.757005] ACPI: Added _OSI(Processor Aggregator Device)
[ 0.772090] ACPI: Interpreter enabled
[ 0.775913] ACPI: Using GIC for interrupt routing
[ 0.780864] ACPI: MCFG table detected, 1 entries
[ 0.785858] HEST: Table parsing has been initialized.
[ 0.793565] ACPI: Power Resource [SCVR] (on)
[ 0.828517] xgene-msi APMC0D0E:00: APM X-Gene PCIe MSI driver loaded
[ 0.835285] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff])
[ 0.841791] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.850585] acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug PME AER PCIeCapability]
[ 0.859017] acpi PNP0A08:00: MCFG quirk: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] with xgene_v1_pcie_ecam_ops
[ 0.871645] acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0xe0d0000000-0xe0dfffffff] not reserved in ACPI namespace
[ 0.882851] acpi PNP0A08:00: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff]
[ 0.891009] Remapped I/O 0x000000e010000000 to [io 0x0000-0xffff window]
[ 0.898264] PCI host bridge to bus 0000:00
[ 0.902574] pci_bus 0000:00: root bus resource [io 0x0000-0xffff window] (bus address [0x10000000-0x1000ffff])
[ 0.913134] pci_bus 0000:00: root bus resource [mem 0xe040000000-0xe07fffffff window] (bus address [0x40000000-0x7fffffff])
[ 0.924779] pci_bus 0000:00: root bus resource [mem 0xf000000000-0xffffffffff window]
[ 0.932964] pci_bus 0000:00: root bus resource [bus 00-ff]
[ 0.938697] pci 0000:00:00.0: [10e8:e004] type 01 class 0x060400
[ 0.938776] pci 0000:00:00.0: supports D1 D2
[ 0.938937] pci 0000:01:00.0: [8086:10d3] type 00 class 0x020000
[ 0.938970] pci 0000:01:00.0: reg 0x10: [mem 0xe040080000-0xe04009ffff]
[ 0.938986] pci 0000:01:00.0: reg 0x14: [mem 0xe040000000-0xe04007ffff]
[ 0.939003] pci 0000:01:00.0: reg 0x18: [io 0x0000-0x001f]
[ 0.939019] pci 0000:01:00.0: reg 0x1c: [mem 0xe0400a0000-0xe0400a3fff]
[ 0.939065] pci 0000:01:00.0: reg 0x30: [mem 0xfffc0000-0xffffffff pref]
[ 0.939205] pci 0000:01:00.0: PME# supported from D0 D3hot D3cold
[ 0.968751] pci 0000:00:00.0: BAR 14: assigned [mem 0xe040000000-0xe0400fffff]
[ 0.976281] pci 0000:00:00.0: BAR 13: assigned [io 0x1000-0x1fff]
[ 0.982748] pci 0000:01:00.0: BAR 1: assigned [mem 0xe040000000-0xe04007ffff]
[ 0.990222] pci 0000:01:00.0: BAR 6: assigned [mem 0xe040080000-0xe0400bffff pref]
[ 0.998135] pci 0000:01:00.0: BAR 0: assigned [mem 0xe0400c0000-0xe0400dffff]
[ 1.005583] pci 0000:01:00.0: BAR 3: assigned [mem 0xe0400e0000-0xe0400e3fff]
[ 1.013044] pci 0000:01:00.0: BAR 2: assigned [io 0x1000-0x101f]
[ 1.019447] pci 0000:00:00.0: PCI bridge to [bus 01]
[ 1.024681] pci 0000:00:00.0: bridge window [io 0x1000-0x1fff]
[ 1.031099] pci 0000:00:00.0: bridge window [mem 0xe040000000-0xe0400fffff]
[ 1.039197] vgaarb: loaded
[ 1.042181] SCSI subsystem initialized
[ 1.046203] libata version 3.00 loaded.
[ 1.046240] ACPI: bus type USB registered
[ 1.050444] usbcore: registered new interface driver usbfs
[ 1.056204] usbcore: registered new interface driver hub
[ 1.061821] usbcore: registered new device driver usb
[ 1.067216] Registered efivars operations
[ 1.071684] NetLabel: Initializing
[ 1.075256] NetLabel: domain hash size = 128
[ 1.079777] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO
[ 1.085747] NetLabel: unlabeled traffic allowed by default
[ 1.091744] clocksource: Switched to clocksource arch_sys_counter
[ 1.116637] VFS: Disk quotas dquot_6.6.0
[ 1.120816] VFS: Dquot-cache hash table entries: 8192 (order 0, 65536 bytes)
[ 1.128374] pnp: PnP ACPI init
[ 1.132213] pnp: PnP ACPI: found 0 devices
[ 1.139026] NET: Registered protocol family 2
[ 1.143824] TCP established hash table entries: 131072 (order: 4, 1048576 bytes)
[ 1.151892] TCP bind hash table entries: 65536 (order: 4, 1048576 bytes)
[ 1.159275] TCP: Hash tables configured (established 131072 bind 65536)
[ 1.166293] UDP hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.172798] UDP-Lite hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.179867] NET: Registered protocol family 1
[ 1.184479] PCI: CLS 64 bytes, default 128
[ 1.184578] Unpacking initramfs...
[ 1.868130] Freeing initrd memory: 32640K
[ 1.872713] kvm [1]: 8-bit VMID
[ 1.875992] kvm [1]: IDMAP page: 40009b7000
[ 1.880358] kvm [1]: HYP VA range: 800000000000:ffffffffffff
[ 1.886375] kvm [1]: Hyp mode initialized successfully
[ 1.891755] kvm [1]: GICV region size/alignment is unsafe, using trapping (reduced performance)
[ 1.900915] kvm [1]: vgic-v2@780cf000
[ 1.904807] kvm [1]: vgic interrupt IRQ1
[ 1.908909] kvm [1]: virtual timer IRQ4
[ 1.928681] audit: initializing netlink subsys (disabled)
[ 1.934412] audit: type=2000 audit(1.470:1): state=initialized audit_enabled=0 res=1
[ 1.934540] Initialise system trusted keyrings
[ 1.934609] workingset: timestamp_bits=44 max_order=18 bucket_order=0
[ 1.936691] zbud: loaded
[ 1.937062] SELinux: Registering netfilter hooks
[ 2.016448] NET: Registered protocol family 38
[ 2.021088] Key type asymmetric registered
[ 2.025420] Asymmetric key parser 'x509' registered
[ 2.030583] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
[ 2.038391] io scheduler noop registered
[ 2.042467] io scheduler deadline registered (default)
[ 2.047931] io scheduler cfq registered
[ 2.051950] io scheduler mq-deadline registered
[ 2.056870] atomic64_test: passed
[ 2.060612] xgene-gpio APMC0D14:00: X-Gene GPIO driver registered.
[ 2.067387] pcieport 0000:00:00.0: AER enabled with IRQ 83
[ 2.073172] pcieport 0000:00:00.0: Signaling PME with IRQ 83
[ 2.079355] input: Power Button as /devices/LNXSYSTM:00/PNP0C0C:00/input/input0
[ 2.087026] ACPI: Power Button [PWRB]
[ 2.091368] GHES: Failed to map GSI to IRQ for generic hardware error source: 2
[ 2.099018] GHES: probe of GHES.2 failed with error 82
[ 2.104386] GHES: Failed to map GSI to IRQ for generic hardware error source: 3
[ 2.112068] GHES: probe of GHES.3 failed with error 82
[ 2.117497] GHES: Failed to map GSI to IRQ for generic hardware error source: 4
[ 2.125101] GHES: probe of GHES.4 failed with error 82
[ 2.130452] GHES: Failed to map GSI to IRQ for generic hardware error source: 5
[ 2.138036] GHES: probe of GHES.5 failed with error 82
[ 2.143461] GHES: Failed to map GSI to IRQ for generic hardware error source: 6
[ 2.151121] GHES: probe of GHES.6 failed with error 82
[ 2.156510] GHES: Failed to map GSI to IRQ for generic hardware error source: 7
[ 2.164189] GHES: probe of GHES.7 failed with error 82
[ 2.169574] GHES: Failed to map GSI to IRQ for generic hardware error source: 8
[ 2.177249] GHES: probe of GHES.8 failed with error 82
[ 2.182604] GHES: Failed to map GSI to IRQ for generic hardware error source: 9
[ 2.190284] GHES: probe of GHES.9 failed with error 82
[ 2.195693] GHES: Failed to enable APEI firmware first mode.
[ 2.201771] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[ 2.208814] console [ttyS0] disabled
[ 2.232712] APMC0D08:00: ttyS0 at MMIO 0x1c020000 (irq = 22, base_baud = 3125000) is a U6_16550A
[ 2.241919] console [ttyS0] enabled
[ 2.249007] bootconsole [uart8250] disabled
[ 2.277886] APMC0D08:01: ttyS1 at MMIO 0x1c021000 (irq = 23, base_baud = 3125000) is a U6_16550A
[ 2.287044] cacheinfo: Unable to detect cache hierarchy for CPU 0
[ 2.293126] rdac: device handler registered
[ 2.297348] hp_sw: device handler registered
[ 2.301595] emc: device handler registered
[ 2.305766] alua: device handler registered
[ 2.310077] libphy: Fixed MDIO Bus: probed
[ 2.314237] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
[ 2.320734] ehci-pci: EHCI PCI platform driver
[ 2.325177] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver
[ 2.331327] ohci-pci: OHCI PCI platform driver
[ 2.335764] uhci_hcd: USB Universal Host Controller Interface driver
[ 2.342182] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.347366] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 1
[ 2.354792] xhci-hcd 808622B7:00: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.363122] xhci-hcd 808622B7:00: irq 29, io mem 0x19000000
[ 2.368735] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.375494] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.382682] usb usb1: Product: xHCI Host Controller
[ 2.387533] usb usb1: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.394375] usb usb1: SerialNumber: 808622B7:00
[ 2.399097] hub 1-0:1.0: USB hub found
[ 2.402848] hub 1-0:1.0: 1 port detected
[ 2.406851] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.412034] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 2
[ 2.419332] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.427425] usb usb2: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.434184] usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.441368] usb usb2: Product: xHCI Host Controller
[ 2.446223] usb usb2: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.453069] usb usb2: SerialNumber: 808622B7:00
[ 2.457761] hub 2-0:1.0: USB hub found
[ 2.461501] hub 2-0:1.0: 1 port detected
[ 2.465534] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.470715] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 3
[ 2.478113] xhci-hcd 808622B7:01: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.486441] xhci-hcd 808622B7:01: irq 30, io mem 0x19800000
[ 2.492041] usb usb3: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.498793] usb usb3: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.505984] usb usb3: Product: xHCI Host Controller
[ 2.510836] usb usb3: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.517682] usb usb3: SerialNumber: 808622B7:01
[ 2.522397] hub 3-0:1.0: USB hub found
[ 2.526137] hub 3-0:1.0: 1 port detected
[ 2.530122] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.535305] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 4
[ 2.542607] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.550692] usb usb4: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.557451] usb usb4: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.564640] usb usb4: Product: xHCI Host Controller
[ 2.569491] usb usb4: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.576334] usb usb4: SerialNumber: 808622B7:01
[ 2.581037] hub 4-0:1.0: USB hub found
[ 2.584788] hub 4-0:1.0: 1 port detected
[ 2.588828] usbcore: registered new interface driver usbserial
[ 2.594649] usbcore: registered new interface driver usbserial_generic
[ 2.601150] usbserial: USB Serial support registered for generic
[ 2.607208] mousedev: PS/2 mouse device common for all mice
[ 2.612948] rtc-efi rtc-efi: rtc core: registered rtc-efi as rtc0
[ 2.619241] EFI Variables Facility v0.08 2004-May-17
[ 2.624405] hidraw: raw HID events driver (C) Jiri Kosina
[ 2.629836] usbcore: registered new interface driver usbhid
[ 2.635388] usbhid: USB HID core driver
[ 2.639274] drop_monitor: Initializing network drop monitor service
[ 2.645602] Initializing XFRM netlink socket
[ 2.650006] NET: Registered protocol family 10
[ 2.654800] Segment Routing with IPv6
[ 2.658462] NET: Registered protocol family 17
[ 2.663153] registered taskstats version 1
[ 2.667231] Loading compiled-in X.509 certificates
[ 2.675102] alg: No test for pkcs1pad(rsa,sha1) (pkcs1pad(rsa-generic,sha1))
[ 2.683548] Loaded X.509 cert 'Build time autogenerated kernel key: 98aa1b1964f4052ba920878b6f8348e178459895'
[ 2.693521] zswap: loaded using pool lzo/zbud
[ 2.717703] Key type big_key registered
[ 2.721594] BERT: [Firmware Bug]: table invalid.
[ 2.726272] rtc-efi rtc-efi: setting system clock to 2017-04-28 16:24:34 UTC (1493396674)
[ 2.759039] Freeing unused kernel memory: 1216K
[ 2.768550] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.775743] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.782366] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.790811] systemd[1]: systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL +XZ -LZ4 -SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
[ 2.809044] systemd[1]: Detected architecture arm64.
[ 2.814004] systemd[1]: Running in initial RAM disk.
[ 2.851832] systemd[1]: Set hostname to <localhost.localdomain>.
[ 2.970013] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.976652] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.983295] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.989979] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.997677] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.004524] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.011427] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.051810] systemd[1]: Reached target Timers.
[ 3.056250] systemd[1]: Starting Timers.
[ 3.091786] systemd[1]: Created slice Root Slice.
[ 3.096483] systemd[1]: Starting Root Slice.
[ 3.131780] systemd[1]: Listening on udev Kernel Socket.
[ 3.137083] systemd[1]: Starting udev Kernel Socket.
[ 4.102501] device-mapper: uevent: version 1.0.3
[ 4.107414] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel@redhat.com
[ 4.412199] xgene-ahci APMC0D0D:00: skip clock and PHY initialization
[ 4.418619] xgene-ahci APMC0D0D:00: controller can't do NCQ, turning off CAP_NCQ
[ 4.426125] xgene-ahci APMC0D0D:00: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.434987] xgene-ahci APMC0D0D:00: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.434998] xgene-ahci APMC0D0D:00: port 0 is not capable of FBS
[ 4.470653] pps_core: LinuxPPS API ver. 1 registered
[ 4.475669] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[ 4.491986] xgene-ahci APMC0D0D:00: port 1 is not capable of FBS
[ 4.507826] PTP clock support registered
[ 4.521307] sdhci: Secure Digital Host Controller Interface driver
[ 4.527512] sdhci: Copyright(c) Pierre Ossman
[ 4.577139] mmc0: SDHCI controller on ACPI [APMC0D0C:00] using PIO
[ 4.612188] scsi host0: xgene-ahci
[ 4.615838] scsi host1: xgene-ahci
[ 4.619314] ata1: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x100 irq 27
[ 4.621598] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k
[ 4.621602] e1000e: Copyright(c) 1999 - 2015 Intel Corporation.
[ 4.621841] e1000e 0000:01:00.0: enabling device (0000 -> 0002)
[ 4.622056] e1000e 0000:01:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[ 4.654091] ata2: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x180 irq 27
[ 4.662326] xgene-ahci APMC0D0D:01: skip clock and PHY initialization
[ 4.668755] xgene-ahci APMC0D0D:01: controller can't do NCQ, turning off CAP_NCQ
[ 4.676203] xgene-ahci APMC0D0D:01: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.685069] xgene-ahci APMC0D0D:01: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.693201] xgene-ahci APMC0D0D:01: port 0 is not capable of FBS
[ 4.693880] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): registered PHC clock
[ 4.706842] xgene-ahci APMC0D0D:01: port 1 is not capable of FBS
[ 4.714376] scsi host2: xgene-ahci
[ 4.718239] scsi host3: xgene-ahci
[ 4.721961] ata3: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x100 irq 28
[ 4.729850] ata4: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x180 irq 28
[ 4.738769] libphy: APM X-Gene MDIO bus: probed
[ 4.791711] e1000e 0000:01:00.0 eth0: (PCI Express:2.5GT/s:Width x1) 68:05:ca:46:ce:08
[ 4.799622] e1000e 0000:01:00.0 eth0: Intel(R) PRO/1000 Network Connection
[ 4.806511] e1000e 0000:01:00.0 eth0: MAC: 3, PHY: 8, PBA No: E46981-008
[ 4.808607] xgene-enet APMC0D05:00: clocks have been setup already
[ 4.821883] e1000e 0000:01:00.0 enp1s0: renamed from eth0
[ 5.011899] xgene-enet APMC0D30:00: clocks have been setup already
[ 5.063242] ata3: SATA link down (SStatus 0 SControl 4300)
[ 5.068752] ata4: SATA link down (SStatus 0 SControl 4300)
[ 5.121895] xgene-enet APMC0D30:01: clocks have been setup already
[ 5.136261] xgene-enet APMC0D31:00: clocks have been setup already
[ 5.331836] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.338118] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.344486] ata2.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.351098] ata2.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.357813] ata1.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.364432] ata1.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.371105] ata2.00: configured for UDMA/133
[ 5.375470] ata1.00: configured for UDMA/133
[ 5.380021] scsi 0:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.482145] sd 0:0:0:0: [sda] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.482607] scsi 1:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.498165] sd 0:0:0:0: [sda] Write Protect is off
[ 5.502945] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[ 5.502986] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.514451] sda: sda1 sda2 sda3
[ 5.518153] sd 0:0:0:0: [sda] Attached SCSI disk
[ 5.602396] sd 1:0:0:0: [sdb] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.609881] sd 1:0:0:0: [sdb] Write Protect is off
[ 5.614681] sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00
[ 5.614739] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.626189] sdb: sdb1
[ 5.628980] sd 1:0:0:0: [sdb] Attached SCSI disk
[ 6.113010] random: fast init done
[ 6.461233] SGI XFS with ACLs, security attributes, no debug enabled
[ 6.471638] XFS (dm-0): Mounting V5 Filesystem
[ 6.501907] XFS (dm-0): Ending clean mount
[ 7.216272] systemd-journald[201]: Received SIGTERM from PID 1 (systemd).
[ 7.263513] systemd: 20 output lines suppressed due to ratelimiting
[ 7.301701] audit: type=1404 audit(1493396679.060:2): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295
[ 7.410911] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.516206] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.718524] SELinux: 8 users, 14 roles, 4990 types, 304 bools, 1 sens, 1024 cats
[ 7.718542] SELinux: 94 classes, 105786 rules
[ 7.741338] SELinux: Class sctp_socket not defined in policy.
[ 7.747177] SELinux: Class icmp_socket not defined in policy.
[ 7.753000] SELinux: Class ax25_socket not defined in policy.
[ 7.758809] SELinux: Class ipx_socket not defined in policy.
[ 7.764542] SELinux: Class netrom_socket not defined in policy.
[ 7.770524] SELinux: Class atmpvc_socket not defined in policy.
[ 7.776515] SELinux: Class x25_socket not defined in policy.
[ 7.782247] SELinux: Class rose_socket not defined in policy.
[ 7.788056] SELinux: Class decnet_socket not defined in policy.
[ 7.794048] SELinux: Class atmsvc_socket not defined in policy.
[ 7.800028] SELinux: Class rds_socket not defined in policy.
[ 7.805760] SELinux: Class irda_socket not defined in policy.
[ 7.811568] SELinux: Class pppox_socket not defined in policy.
[ 7.817473] SELinux: Class llc_socket not defined in policy.
[ 7.823204] SELinux: Class can_socket not defined in policy.
[ 7.828926] SELinux: Class tipc_socket not defined in policy.
[ 7.834744] SELinux: Class bluetooth_socket not defined in policy.
[ 7.840984] SELinux: Class iucv_socket not defined in policy.
[ 7.846802] SELinux: Class rxrpc_socket not defined in policy.
[ 7.852707] SELinux: Class isdn_socket not defined in policy.
[ 7.858514] SELinux: Class phonet_socket not defined in policy.
[ 7.864504] SELinux: Class ieee802154_socket not defined in policy.
[ 7.870831] SELinux: Class caif_socket not defined in policy.
[ 7.876649] SELinux: Class alg_socket not defined in policy.
[ 7.882381] SELinux: Class nfc_socket not defined in policy.
[ 7.888103] SELinux: Class vsock_socket not defined in policy.
[ 7.894008] SELinux: Class kcm_socket not defined in policy.
[ 7.899730] SELinux: Class qipcrtr_socket not defined in policy.
[ 7.905822] SELinux: Class smc_socket not defined in policy.
[ 7.911545] SELinux: the above unknown classes and permissions will be allowed
[ 7.918772] SELinux: Completing initialization.
[ 7.918776] SELinux: Setting up existing superblocks.
[ 7.993509] audit: type=1403 audit(1493396679.760:3): policy loaded auid=4294967295 ses=4294967295
[ 8.002331] systemd[1]: Successfully loaded SELinux policy in 701.052ms.
[ 8.029233] systemd[1]: RTC configured in localtime, applying delta of -240 minutes to system time.
[ 8.076262] ip_tables: (C) 2000-2006 Netfilter Core Team
[ 8.081619] systemd[1]: Inserted module 'ip_tables'
[ 8.199798] systemd[1]: Relabelled /dev and /run in 108.385ms.
[ 8.666128] systemd-journald[618]: Received request to flush runtime journal from PID 1
[ 8.793125] xgene-slimpro-i2c APMC0D40:00: i2c mailbox channel request failed
[ 8.809689] RPC: Registered named UNIX socket transport module.
[ 8.815634] RPC: Registered udp transport module.
[ 8.820316] RPC: Registered tcp transport module.
[ 8.825018] RPC: Registered tcp NFSv4.1 backchannel transport module.
[ 8.872038] sd 0:0:0:0: Attached scsi generic sg0 type 0
[ 8.877502] sd 1:0:0:0: Attached scsi generic sg1 type 0
[ 8.883774] xgene-slimpro-hwmon APMC0D29:00: APM X-Gene SoC HW monitor driver registered
[ 8.892777] xgene-rng APMC0D18:00: Couldn't get the clock for RNG
[ 8.899408] xgene-slimpro-mbox APMC0D01:00: APM X-Gene SLIMpro MailBox registered
[ 8.910481] xgene-gpio-sb APMC0D15:00: Support 22 gpios, 6 irqs start from pin 8
[ 8.918470] xgene-gpio-sb APMC0D15:00: X-Gene GPIO Standby driver registered
[ 9.044885] Installing knfsd (copyright (C) 1996 okir@monad.swb.de).
[ 9.118751] XFS (sda2): Mounting V5 Filesystem
[ 9.125864] Adding 8388544k swap on /dev/mapper/rhelp_dhcp--171-swap. Priority:-1 extents:1 across:8388544k SSFS
[ 9.141179] XFS (sda2): Ending clean mount
[ 9.286221] XFS (dm-2): Mounting V5 Filesystem
[ 9.305151] XFS (dm-2): Ending clean mount
[ 10.242361] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.339596] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.367672] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.374758] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.403220] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.410389] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.437608] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.444895] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.471938] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.478809] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.562700] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.571438] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.580052] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.588624] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.597191] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 11.412166] xgene-enet APMC0D05:00 eth0: Link is Down
[ 11.492472] xgene-enet APMC0D30:00 eth1: Link is Down
[ 11.497600] xgene-enet APMC0D30:01 eth2: Link is Down
[ 14.533303] e1000e: enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
[ 14.541337] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready
[ 14.548556] xgene-enet APMC0D05:00 eth0: Link is Up - 1Gbps/Full - flow control off
[ 14.556244] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[-- Attachment #4: Type: text/plain, Size: 176 bytes --]
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-04-28 16:36 ` Jon Masters
0 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-04-28 16:36 UTC (permalink / raw)
To: linux-arm-kernel
On 04/28/2017 01:11 AM, Jon Masters wrote:
> On 04/27/2017 08:54 PM, Khuong Dinh wrote:
>> From: Khuong Dinh <kdinh@apm.com>
>>
>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>
>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> Signed-off-by: Duc Dang <dhdang@apm.com>
>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>
> Thanks. Currently testing. Will review and followup shortly.
Attaching boot logs before and after. I tested with an Intel e1000.
Before the patch I get fallback interrupts. Afterward, I get X-Gene MSI
interrupts working. Please followup with answers to Marc's question
about how you propose to guarantee boottime ordering of binding.
I urgently require that AppliedMicro complete upstreaming of this. It
should have been upstream a year ago. You're not the only ones with
patches like this left around - I'm coming for everyone else, too.
Tested-by: Jon Masters <jcm@jonmasters.org>
-------------- next part --------------
[ 0.000000] Booting Linux on physical CPU 0x0
[ 0.000000] Linux version 4.11.0-rc8.xgene_msix1+ (root at mustang4.bos.jonmasters.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-14) (GCC) ) #1 SMP Fri Apr 28 04:44:20 EDT 2017
[ 0.000000] Boot CPU: AArch64 Processor [500f0001]
[ 0.000000] earlycon: uart8250 at MMIO32 0x000000001c020000 (options '')
[ 0.000000] bootconsole [uart8250] enabled
[ 0.000000] efi: Getting EFI parameters from FDT:
[ 0.000000] efi: EFI v2.40 by X-Gene Mustang Board EFI Aug 19 2016 18:02:24
[ 0.000000] efi: ACPI=0x43fa82d000 ACPI 2.0=0x43fa82d014 SMBIOS 3.0=0x43fa82b000 ESRT=0x43ff053d18
[ 0.000000] esrt: Reserving ESRT space from 0x00000043ff053d18 to 0x00000043ff053d78.
[ 0.000000] cma: Reserved 512 MiB at 0x00000040e0000000
[ 0.000000] ACPI: Early table checksum verification disabled
[ 0.000000] ACPI: RSDP 0x00000043FA82D014 000024 (v02 APM )
[ 0.000000] ACPI: XSDT 0x00000043FA82C0E8 000084 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: FACP 0x00000043FA822000 00010C (v05 APM XGENE 00000003 INTL 20140724)
[ 0.000000] ACPI: DSDT 0x00000043FA823000 005837 (v05 APM APM88xxx 00000001 INTL 20140724)
[ 0.000000] ACPI: DBG2 0x00000043FA829000 0000AA (v00 APMC0D XGENEDBG 00000000 INTL 20140724)
[ 0.000000] ACPI: GTDT 0x00000043FA820000 000060 (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: MCFG 0x00000043FA81F000 00003C (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: SPCR 0x00000043FA81E000 000050 (v02 APMC0D XGENESPC 00000000 INTL 20140724)
[ 0.000000] ACPI: SSDT 0x00000043FA81D000 00002D (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: BERT 0x00000043FA81C000 000030 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: HEST 0x00000043FA81B000 0002A8 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: APIC 0x00000043FA81A000 0002A4 (v03 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA819000 000063 (v02 REDHAT MACADDRS 00000001 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA818000 000032 (v02 REDHAT UARTCLKS 00000001 01000013)
[ 0.000000] ACPI: PCCT 0x00000043FA817000 000300 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SPCR: console: uart,mmio,0x1c020000,115200
[ 0.000000] ACPI: NUMA: Failed to initialise from firmware
[ 0.000000] NUMA: Faking a node at [mem 0x0000000000000000-0x00000043ffffffff]
[ 0.000000] NUMA: Adding memblock [0x4000000000 - 0x40001fffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x4000200000 - 0x43fa7cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43fa7d0000 - 0x43faa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43faa20000 - 0x43ff9cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ff9d0000 - 0x43ffa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ffa20000 - 0x43ffffffff] on node 0
[ 0.000000] NUMA: Initmem setup node 0 [mem 0x4000000000-0x43ffffffff]
[ 0.000000] NUMA: NODE_DATA [mem 0x43ffffe200-0x43ffffffff]
[ 0.000000] Zone ranges:
[ 0.000000] DMA [mem 0x0000004000000000-0x00000040ffffffff]
[ 0.000000] Normal [mem 0x0000004100000000-0x00000043ffffffff]
[ 0.000000] Movable zone start for each node
[ 0.000000] Early memory node ranges
[ 0.000000] node 0: [mem 0x0000004000000000-0x00000040001fffff]
[ 0.000000] node 0: [mem 0x0000004000200000-0x00000043fa7cffff]
[ 0.000000] node 0: [mem 0x00000043fa7d0000-0x00000043faa1ffff]
[ 0.000000] node 0: [mem 0x00000043faa20000-0x00000043ff9cffff]
[ 0.000000] node 0: [mem 0x00000043ff9d0000-0x00000043ffa1ffff]
[ 0.000000] node 0: [mem 0x00000043ffa20000-0x00000043ffffffff]
[ 0.000000] Initmem setup node 0 [mem 0x0000004000000000-0x00000043ffffffff]
[ 0.000000] On node 0 totalpages: 262144
[ 0.000000] DMA zone: 64 pages used for memmap
[ 0.000000] DMA zone: 0 pages reserved
[ 0.000000] DMA zone: 65536 pages, LIFO batch:1
[ 0.000000] Normal zone: 192 pages used for memmap
[ 0.000000] Normal zone: 196608 pages, LIFO batch:1
[ 0.000000] psci: is not implemented in ACPI.
[ 0.000000] percpu: Embedded 3 pages/cpu @ffff8003ffdf0000 s124312 r8192 d64104 u196608
[ 0.000000] pcpu-alloc: s124312 r8192 d64104 u196608 alloc=3*65536
[ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3 [0] 4 [0] 5 [0] 6 [0] 7
[ 0.000000] Detected PIPT I-cache on CPU0
[ 0.000000] Built 1 zonelists in Node order, mobility grouping on. Total pages: 261888
[ 0.000000] Policy zone: Normal
[ 0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.11.0-rc8.xgene_msix1+ root=/dev/mapper/rhelp_dhcp--171-root ro crashkernel=auto rd.lvm.lv=rhelp_dhcp-171/root rd.lvm.lv=rhelp_dhcp-171/swap LANG=en_US.UTF-8 acpi=force earlycon=uart8250,mmio32,0x1c020000 console=ttyS0,115200
[ 0.000000] PID hash table entries: 4096 (order: -1, 32768 bytes)
[ 0.000000] software IO TLB [mem 0x40dbff0000-0x40dfff0000] (64MB) mapped at [ffff8000dbff0000-ffff8000dffeffff]
[ 0.000000] Memory: 16109888K/16777216K available (7420K kernel code, 1354K rwdata, 3200K rodata, 1216K init, 7286K bss, 143040K reserved, 524288K cma-reserved)
[ 0.000000] Virtual kernel memory layout:
[ 0.000000] modules : 0xffff000000000000 - 0xffff000008000000 ( 128 MB)
[ 0.000000] vmalloc : 0xffff000008000000 - 0xffff7bdfffff0000 (126847 GB)
[ 0.000000] .text : 0xffff000008080000 - 0xffff0000087c0000 ( 7424 KB)
[ 0.000000] .rodata : 0xffff0000087c0000 - 0xffff000008af0000 ( 3264 KB)
[ 0.000000] .init : 0xffff000008af0000 - 0xffff000008c20000 ( 1216 KB)
[ 0.000000] .data : 0xffff000008c20000 - 0xffff000008d72a00 ( 1355 KB)
[ 0.000000] .bss : 0xffff000008d72a00 - 0xffff0000094902d4 ( 7287 KB)
[ 0.000000] fixed : 0xffff7fdffe7d0000 - 0xffff7fdffec00000 ( 4288 KB)
[ 0.000000] PCI I/O : 0xffff7fdffee00000 - 0xffff7fdfffe00000 ( 16 MB)
[ 0.000000] vmemmap : 0xffff7fe000000000 - 0xffff800000000000 ( 128 GB maximum)
[ 0.000000] 0xffff7fe000000000 - 0xffff7fe001000000 ( 16 MB actual)
[ 0.000000] memory : 0xffff800000000000 - 0xffff800400000000 ( 16384 MB)
[ 0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=8, Nodes=1
[ 0.000000] Hierarchical RCU implementation.
[ 0.000000] Build-time adjustment of leaf fanout to 64.
[ 0.000000] RCU restricting CPUs from NR_CPUS=4096 to nr_cpu_ids=8.
[ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=8
[ 0.000000] NR_IRQS:64 nr_irqs:64 0
[ 0.000000] GIC: Using split EOI/Deactivate mode
[ 0.000000] GICv3: No distributor detected at @ffff000008050000, giving up
[ 0.000000] arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys).
[ 0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns
[ 0.000002] sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns
[ 0.008382] Console: colour dummy device 80x25
[ 0.013031] Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000)
[ 0.023774] pid_max: default: 32768 minimum: 301
[ 0.028553] ACPI: Core revision 20170119
[ 0.036208] ACPI Error: Method parse/execution failed [\_SB.ET00._STA] (Node ffff8003c0195c08), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.049249] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.057204] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.063972] ACPI Error: Method parse/execution failed [\_SB.ET01._STA] (Node ffff8003c0195eb0), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.077125] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.084966] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.092086] ACPI: 4 ACPI AML tables successfully acquired and loaded
[ 0.098656] Security Framework initialized
[ 0.102953] SELinux: Initializing.
[ 0.106548] SELinux: Starting in permissive mode
[ 0.106747] Dentry cache hash table entries: 2097152 (order: 8, 16777216 bytes)
[ 0.116984] Inode-cache hash table entries: 1048576 (order: 7, 8388608 bytes)
[ 0.125658] Mount-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.132758] Mountpoint-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.140711] ftrace: allocating 26814 entries in 7 pages
[ 0.205140] ASID allocator initialised with 65536 entries
[ 0.211274] Remapping and enabling EFI services.
[ 0.216029] EFI remap 0x0000000010510000 => 0000000020000000
[ 0.222058] EFI remap 0x0000000010548000 => 0000000020018000
[ 0.228081] EFI remap 0x0000000017000000 => 0000000020020000
[ 0.234066] EFI remap 0x000000001c025000 => 0000000020035000
[ 0.240084] UEFI Runtime regions are not aligned to 64 KB -- buggy firmware?
[ 0.240089] ------------[ cut here ]------------
[ 0.252134] WARNING: CPU: 0 PID: 1 at arch/arm64/kernel/efi.c:34 efi_create_mapping+0x4c/0x104
[ 0.261010] Modules linked in:
[ 0.265699] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc8.xgene_msix1+ #1
[ 0.273313] Hardware name: APM X-Gene Mustang board (DT)
[ 0.278750] task: ffff8003c1bc0000 task.stack: ffff8003c1c00000
[ 0.284827] PC is at efi_create_mapping+0x4c/0x104
[ 0.289772] LR is at efi_create_mapping+0x4c/0x104
[ 0.294664] pc : [<ffff000008af51f0>] lr : [<ffff000008af51f0>] pstate: 60000045
[ 0.302261] sp : ffff8003c1c03d20
[ 0.305710] x29: ffff8003c1c03d20 x28: 0000000000000000
[ 0.311183] x27: ffff000008c53000 x26: ffff000008d429c8
[ 0.316639] x25: ffff000008d42000 x24: 0000000000000000
[ 0.322102] x23: ffff000008a4c8f0 x22: ffff000008c60000
[ 0.327600] x21: ffff000008c53000 x20: ffff000008d429c8
[ 0.333038] x19: ffff8003f96ab348 x18: 000000003eca74e7
[ 0.338485] x17: 0000000000000000 x16: 0000000000000000
[ 0.343992] x15: ffff000008c53000 x14: 000000004e2b17ac
[ 0.349464] x13: ffff0000089ca500 x12: 0000000000040b00
[ 0.354988] x11: ffff8003c1c039e0 x10: ffff8003c1c039e0
[ 0.360435] x9 : 0000000000000073 x8 : 6620796767756220
[ 0.365898] x7 : 2d2d20424b203436 x6 : ffff000009274718
[ 0.371362] x5 : ffff8003ffdf7530 x4 : 0000000000000000
[ 0.376834] x3 : 0000000000000000 x2 : 0000000000040b00
[ 0.382316] x1 : 0000000000040b00 x0 : 0000000000000040
[ 0.389363] ---[ end trace 4a6935def63e1e16 ]---
[ 0.394126] Call trace:
[ 0.396626] Exception stack(0xffff8003c1c03b30 to 0xffff8003c1c03c60)
[ 0.403246] 3b20: ffff8003f96ab348 0001000000000000
[ 0.411302] 3b40: ffff8003c1c03d20 ffff000008af51f0 0000000060000045 ffff000008d429c8
[ 0.419331] 3b60: 0000000000000000 0000000000000000 ffff8003c1c03d20 ffff8003c1c03d20
[ 0.427377] 3b80: ffff8003c1c03ce0 00000000ffffffc8 ffff8003c1c03be0 ffff00000812f6b8
[ 0.435449] 3ba0: ffff000008c53000 00000000ffffffc8 ffff8003c1c03c00 ffff0000089be038
[ 0.443487] 3bc0: ffff8003c1c03d20 ffff8003c1c03d20 ffff8003c1c03ce0 00000000ffffffc8
[ 0.451585] 3be0: ffff8003c1c03c70 0000000000040b00 0000000000000040 0000000000040b00
[ 0.459666] 3c00: 0000000000040b00 0000000000000000 0000000000000000 ffff8003ffdf7530
[ 0.467729] 3c20: ffff000009274718 2d2d20424b203436 6620796767756220 0000000000000073
[ 0.475784] 3c40: ffff8003c1c039e0 ffff8003c1c039e0 0000000000040b00 ffff0000089ca500
[ 0.483840] [<ffff000008af51f0>] efi_create_mapping+0x4c/0x104
[ 0.489858] [<ffff000008b3db98>] arm_enable_runtime_services+0x154/0x238
[ 0.496817] [<ffff000008083698>] do_one_initcall+0x54/0x15c
[ 0.502634] [<ffff000008af0d18>] kernel_init_freeable+0xe4/0x25c
[ 0.508816] [<ffff0000087ae430>] kernel_init+0x18/0x108
[ 0.514202] [<ffff0000080833c0>] ret_from_fork+0x10/0x50
[ 0.519621] EFI remap 0x00000043fa7df000 => 000000002004f000
[ 0.525659] EFI remap 0x00000043fa82e000 => 000000002006e000
[ 0.531673] EFI remap 0x00000043ff9d6000 => 0000000020266000
[ 0.537700] EFI remap 0x00000043ff9ea000 => 000000002027a000
[ 0.543899] smp: Bringing up secondary CPUs ...
[ 0.548738] Detected PIPT I-cache on CPU1
[ 0.548766] CPU1: Booted secondary processor [500f0001]
[ 0.548987] Detected PIPT I-cache on CPU2
[ 0.549008] CPU2: Booted secondary processor [500f0001]
[ 0.549225] Detected PIPT I-cache on CPU3
[ 0.549239] CPU3: Booted secondary processor [500f0001]
[ 0.549432] Detected PIPT I-cache on CPU4
[ 0.549452] CPU4: Booted secondary processor [500f0001]
[ 0.549653] Detected PIPT I-cache on CPU5
[ 0.549667] CPU5: Booted secondary processor [500f0001]
[ 0.549861] Detected PIPT I-cache on CPU6
[ 0.549882] CPU6: Booted secondary processor [500f0001]
[ 0.550086] Detected PIPT I-cache on CPU7
[ 0.550100] CPU7: Booted secondary processor [500f0001]
[ 0.550142] smp: Brought up 1 node, 8 CPUs
[ 0.621892] SMP: Total of 8 processors activated.
[ 0.626806] CPU features: detected feature: 32-bit EL0 Support
[ 0.632944] CPU: All CPU(s) started at EL2
[ 0.637984] devtmpfs: initialized
[ 0.641749] SMBIOS 3.0.0 present.
[ 0.645224] DMI: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.15 Aug 19 2016
[ 0.654642] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[ 0.664967] futex hash table entries: 2048 (order: 2, 262144 bytes)
[ 0.671981] pinctrl core: initialized pinctrl subsystem
[ 0.677705] NET: Registered protocol family 16
[ 0.682714] cpuidle: using governor menu
[ 0.686960] Detected 8 PCC Subspaces
[ 0.690784] Registering PCC driver as Mailbox controller
[ 0.696410] vdso: 2 pages (1 code @ ffff0000087e0000, 1 data @ ffff000008c40000)
[ 0.704152] hw-breakpoint: found 4 breakpoint and 4 watchpoint registers.
[ 0.713765] DMA: preallocated 256 KiB pool for atomic allocations
[ 0.720154] ACPI: bus type PCI registered
[ 0.724382] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
[ 0.731202] Serial: AMBA PL011 UART driver
[ 0.737465] HugeTLB registered 512 MB page size, pre-allocated 0 pages
[ 0.744898] ACPI: Added _OSI(Module Device)
[ 0.749274] ACPI: Added _OSI(Processor Device)
[ 0.753882] ACPI: Added _OSI(3.0 _SCP Extensions)
[ 0.758804] ACPI: Added _OSI(Processor Aggregator Device)
[ 0.773904] ACPI: Interpreter enabled
[ 0.777753] ACPI: Using GIC for interrupt routing
[ 0.782685] ACPI: MCFG table detected, 1 entries
[ 0.787684] HEST: Table parsing has been initialized.
[ 0.795411] ACPI: Power Resource [SCVR] (on)
[ 0.830232] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff])
[ 0.836749] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.845539] acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug PME AER PCIeCapability]
[ 0.853915] acpi PNP0A08:00: MCFG quirk: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] with xgene_v1_pcie_ecam_ops
[ 0.866664] acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0xe0d0000000-0xe0dfffffff] not reserved in ACPI namespace
[ 0.877881] acpi PNP0A08:00: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff]
[ 0.886116] Remapped I/O 0x000000e010000000 to [io 0x0000-0xffff window]
[ 0.893327] PCI host bridge to bus 0000:00
[ 0.897628] pci_bus 0000:00: root bus resource [io 0x0000-0xffff window] (bus address [0x10000000-0x1000ffff])
[ 0.908283] pci_bus 0000:00: root bus resource [mem 0xe040000000-0xe07fffffff window] (bus address [0x40000000-0x7fffffff])
[ 0.919920] pci_bus 0000:00: root bus resource [mem 0xf000000000-0xffffffffff window]
[ 0.928114] pci_bus 0000:00: root bus resource [bus 00-ff]
[ 0.933821] pci 0000:00:00.0: [10e8:e004] type 01 class 0x060400
[ 0.933899] pci 0000:00:00.0: supports D1 D2
[ 0.934059] pci 0000:01:00.0: [8086:10d3] type 00 class 0x020000
[ 0.934092] pci 0000:01:00.0: reg 0x10: [mem 0xe040080000-0xe04009ffff]
[ 0.934108] pci 0000:01:00.0: reg 0x14: [mem 0xe040000000-0xe04007ffff]
[ 0.934125] pci 0000:01:00.0: reg 0x18: [io 0x0000-0x001f]
[ 0.934141] pci 0000:01:00.0: reg 0x1c: [mem 0xe0400a0000-0xe0400a3fff]
[ 0.934186] pci 0000:01:00.0: reg 0x30: [mem 0xfffc0000-0xffffffff pref]
[ 0.934326] pci 0000:01:00.0: PME# supported from D0 D3hot D3cold
[ 0.963877] pci 0000:00:00.0: BAR 14: assigned [mem 0xe040000000-0xe0400fffff]
[ 0.971433] pci 0000:00:00.0: BAR 13: assigned [io 0x1000-0x1fff]
[ 0.977908] pci 0000:01:00.0: BAR 1: assigned [mem 0xe040000000-0xe04007ffff]
[ 0.985417] pci 0000:01:00.0: BAR 6: assigned [mem 0xe040080000-0xe0400bffff pref]
[ 0.993304] pci 0000:01:00.0: BAR 0: assigned [mem 0xe0400c0000-0xe0400dffff]
[ 1.000812] pci 0000:01:00.0: BAR 3: assigned [mem 0xe0400e0000-0xe0400e3fff]
[ 1.008298] pci 0000:01:00.0: BAR 2: assigned [io 0x1000-0x101f]
[ 1.014701] pci 0000:00:00.0: PCI bridge to [bus 01]
[ 1.019893] pci 0000:00:00.0: bridge window [io 0x1000-0x1fff]
[ 1.026276] pci 0000:00:00.0: bridge window [mem 0xe040000000-0xe0400fffff]
[ 1.034419] vgaarb: loaded
[ 1.037393] SCSI subsystem initialized
[ 1.041366] libata version 3.00 loaded.
[ 1.041403] ACPI: bus type USB registered
[ 1.045616] usbcore: registered new interface driver usbfs
[ 1.051373] usbcore: registered new interface driver hub
[ 1.056947] usbcore: registered new device driver usb
[ 1.062355] Registered efivars operations
[ 1.066820] NetLabel: Initializing
[ 1.070418] NetLabel: domain hash size = 128
[ 1.074997] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO
[ 1.080980] NetLabel: unlabeled traffic allowed by default
[ 1.086945] clocksource: Switched to clocksource arch_sys_counter
[ 1.111811] VFS: Disk quotas dquot_6.6.0
[ 1.115980] VFS: Dquot-cache hash table entries: 8192 (order 0, 65536 bytes)
[ 1.123569] pnp: PnP ACPI init
[ 1.127405] pnp: PnP ACPI: found 0 devices
[ 1.134249] NET: Registered protocol family 2
[ 1.139069] TCP established hash table entries: 131072 (order: 4, 1048576 bytes)
[ 1.147156] TCP bind hash table entries: 65536 (order: 4, 1048576 bytes)
[ 1.154513] TCP: Hash tables configured (established 131072 bind 65536)
[ 1.161541] UDP hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.168029] UDP-Lite hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.175030] NET: Registered protocol family 1
[ 1.179607] PCI: CLS 64 bytes, default 128
[ 1.179708] Unpacking initramfs...
[ 1.863109] Freeing initrd memory: 32640K
[ 1.867699] kvm [1]: 8-bit VMID
[ 1.870944] kvm [1]: IDMAP page: 40009b7000
[ 1.875301] kvm [1]: HYP VA range: 800000000000:ffffffffffff
[ 1.881276] kvm [1]: Hyp mode initialized successfully
[ 1.886624] kvm [1]: GICV region size/alignment is unsafe, using trapping (reduced performance)
[ 1.895748] kvm [1]: vgic-v2 at 780cf000
[ 1.899646] kvm [1]: vgic interrupt IRQ1
[ 1.903757] kvm [1]: virtual timer IRQ4
[ 1.923421] audit: initializing netlink subsys (disabled)
[ 1.929159] audit: type=2000 audit(1.470:1): state=initialized audit_enabled=0 res=1
[ 1.929303] Initialise system trusted keyrings
[ 1.929373] workingset: timestamp_bits=44 max_order=18 bucket_order=0
[ 1.931445] zbud: loaded
[ 1.931829] SELinux: Registering netfilter hooks
[ 2.011199] NET: Registered protocol family 38
[ 2.015847] Key type asymmetric registered
[ 2.020145] Asymmetric key parser 'x509' registered
[ 2.025283] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
[ 2.033081] io scheduler noop registered
[ 2.037167] io scheduler deadline registered (default)
[ 2.042536] io scheduler cfq registered
[ 2.046565] io scheduler mq-deadline registered
[ 2.051474] atomic64_test: passed
[ 2.055189] xgene-gpio APMC0D14:00: X-Gene GPIO driver registered.
[ 2.062003] pcieport 0000:00:00.0: AER enabled with IRQ 83
[ 2.067787] pcieport 0000:00:00.0: Signaling PME with IRQ 83
[ 2.073972] input: Power Button as /devices/LNXSYSTM:00/PNP0C0C:00/input/input0
[ 2.081619] ACPI: Power Button [PWRB]
[ 2.085922] GHES: Failed to map GSI to IRQ for generic hardware error source: 2
[ 2.093593] GHES: probe of GHES.2 failed with error 82
[ 2.098936] GHES: Failed to map GSI to IRQ for generic hardware error source: 3
[ 2.106588] GHES: probe of GHES.3 failed with error 82
[ 2.111936] GHES: Failed to map GSI to IRQ for generic hardware error source: 4
[ 2.119608] GHES: probe of GHES.4 failed with error 82
[ 2.125011] GHES: Failed to map GSI to IRQ for generic hardware error source: 5
[ 2.132659] GHES: probe of GHES.5 failed with error 82
[ 2.138017] GHES: Failed to map GSI to IRQ for generic hardware error source: 6
[ 2.145633] GHES: probe of GHES.6 failed with error 82
[ 2.151042] GHES: Failed to map GSI to IRQ for generic hardware error source: 7
[ 2.158724] GHES: probe of GHES.7 failed with error 82
[ 2.164126] GHES: Failed to map GSI to IRQ for generic hardware error source: 8
[ 2.171797] GHES: probe of GHES.8 failed with error 82
[ 2.177160] GHES: Failed to map GSI to IRQ for generic hardware error source: 9
[ 2.184845] GHES: probe of GHES.9 failed with error 82
[ 2.190233] GHES: Failed to enable APEI firmware first mode.
[ 2.196297] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[ 2.203376] console [ttyS0] disabled
[ 2.227246] APMC0D08:00: ttyS0 at MMIO 0x1c020000 (irq = 22, base_baud = 3125000) is a U6_16550A
[ 2.236527] console [ttyS0] enabled
[ 2.243657] bootconsole [uart8250] disabled
[ 2.272506] APMC0D08:01: ttyS1 at MMIO 0x1c021000 (irq = 23, base_baud = 3125000) is a U6_16550A
[ 2.281671] cacheinfo: Unable to detect cache hierarchy for CPU 0
[ 2.287755] rdac: device handler registered
[ 2.291963] hp_sw: device handler registered
[ 2.296209] emc: device handler registered
[ 2.300380] alua: device handler registered
[ 2.304688] libphy: Fixed MDIO Bus: probed
[ 2.308849] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
[ 2.315346] ehci-pci: EHCI PCI platform driver
[ 2.319790] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver
[ 2.325941] ohci-pci: OHCI PCI platform driver
[ 2.330377] uhci_hcd: USB Universal Host Controller Interface driver
[ 2.336788] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.341980] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 1
[ 2.349425] xhci-hcd 808622B7:00: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.357756] xhci-hcd 808622B7:00: irq 29, io mem 0x19000000
[ 2.363370] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.370130] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.377318] usb usb1: Product: xHCI Host Controller
[ 2.382169] usb usb1: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.389010] usb usb1: SerialNumber: 808622B7:00
[ 2.393727] hub 1-0:1.0: USB hub found
[ 2.397479] hub 1-0:1.0: 1 port detected
[ 2.401480] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.406677] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 2
[ 2.413986] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.422078] usb usb2: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.428836] usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.436020] usb usb2: Product: xHCI Host Controller
[ 2.440875] usb usb2: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.447720] usb usb2: SerialNumber: 808622B7:00
[ 2.452398] hub 2-0:1.0: USB hub found
[ 2.456137] hub 2-0:1.0: 1 port detected
[ 2.460168] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.465356] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 3
[ 2.472759] xhci-hcd 808622B7:01: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.481087] xhci-hcd 808622B7:01: irq 30, io mem 0x19800000
[ 2.486684] usb usb3: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.493443] usb usb3: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.500630] usb usb3: Product: xHCI Host Controller
[ 2.505482] usb usb3: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.512327] usb usb3: SerialNumber: 808622B7:01
[ 2.517049] hub 3-0:1.0: USB hub found
[ 2.520789] hub 3-0:1.0: 1 port detected
[ 2.524790] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.529976] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 4
[ 2.537284] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.545370] usb usb4: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.552128] usb usb4: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.559317] usb usb4: Product: xHCI Host Controller
[ 2.564168] usb usb4: Manufacturer: Linux 4.11.0-rc8.xgene_msix1+ xhci-hcd
[ 2.571013] usb usb4: SerialNumber: 808622B7:01
[ 2.575716] hub 4-0:1.0: USB hub found
[ 2.579464] hub 4-0:1.0: 1 port detected
[ 2.583505] usbcore: registered new interface driver usbserial
[ 2.589324] usbcore: registered new interface driver usbserial_generic
[ 2.595826] usbserial: USB Serial support registered for generic
[ 2.601884] mousedev: PS/2 mouse device common for all mice
[ 2.607620] rtc-efi rtc-efi: rtc core: registered rtc-efi as rtc0
[ 2.613906] EFI Variables Facility v0.08 2004-May-17
[ 2.619081] hidraw: raw HID events driver (C) Jiri Kosina
[ 2.624514] usbcore: registered new interface driver usbhid
[ 2.630067] usbhid: USB HID core driver
[ 2.633953] drop_monitor: Initializing network drop monitor service
[ 2.640280] Initializing XFRM netlink socket
[ 2.644691] NET: Registered protocol family 10
[ 2.649496] Segment Routing with IPv6
[ 2.653160] NET: Registered protocol family 17
[ 2.657841] registered taskstats version 1
[ 2.661921] Loading compiled-in X.509 certificates
[ 2.669794] alg: No test for pkcs1pad(rsa,sha1) (pkcs1pad(rsa-generic,sha1))
[ 2.678237] Loaded X.509 cert 'Build time autogenerated kernel key: 98aa1b1964f4052ba920878b6f8348e178459895'
[ 2.688204] zswap: loaded using pool lzo/zbud
[ 2.713270] Key type big_key registered
[ 2.717168] BERT: [Firmware Bug]: table invalid.
[ 2.721839] rtc-efi rtc-efi: setting system clock to 2017-04-28 16:00:41 UTC (1493395241)
[ 2.757884] Freeing unused kernel memory: 1216K
[ 2.767418] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.774589] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.781211] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.789666] systemd[1]: systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL +XZ -LZ4 -SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
[ 2.807890] systemd[1]: Detected architecture arm64.
[ 2.812833] systemd[1]: Running in initial RAM disk.
[ 2.847031] systemd[1]: Set hostname to <localhost.localdomain>.
[ 2.979499] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.986233] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.993080] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.000078] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.010094] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.017557] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.025329] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.087174] systemd[1]: Reached target Swap.
[ 3.091495] systemd[1]: Starting Swap.
[ 3.117069] systemd[1]: Reached target Timers.
[ 3.121551] systemd[1]: Starting Timers.
[ 3.157062] systemd[1]: Reached target Local File Systems.
[ 3.162579] systemd[1]: Starting Local File Systems.
[ 4.087490] device-mapper: uevent: version 1.0.3
[ 4.092424] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel at redhat.com
[ 4.400761] libphy: APM X-Gene MDIO bus: probed
[ 4.444412] pps_core: LinuxPPS API ver. 1 registered
[ 4.449417] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[ 4.450547] xgene-ahci APMC0D0D:00: skip clock and PHY initialization
[ 4.450561] xgene-ahci APMC0D0D:00: controller can't do NCQ, turning off CAP_NCQ
[ 4.450629] xgene-ahci APMC0D0D:00: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.450643] xgene-ahci APMC0D0D:00: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.450670] xgene-ahci APMC0D0D:00: port 0 is not capable of FBS
[ 4.491029] sdhci: Secure Digital Host Controller Interface driver
[ 4.491033] sdhci: Copyright(c) Pierre Ossman
[ 4.512024] PTP clock support registered
[ 4.512754] xgene-ahci APMC0D0D:00: port 1 is not capable of FBS
[ 4.525820] scsi host0: xgene-ahci
[ 4.529855] scsi host1: xgene-ahci
[ 4.533507] ata1: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x100 irq 27
[ 4.534007] mmc0: SDHCI controller on ACPI [APMC0D0C:00] using PIO
[ 4.547673] ata2: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x180 irq 27
[ 4.548138] xgene-enet APMC0D05:00: clocks have been setup already
[ 4.553893] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k
[ 4.553894] e1000e: Copyright(c) 1999 - 2015 Intel Corporation.
[ 4.554194] e1000e 0000:01:00.0: enabling device (0000 -> 0002)
[ 4.554351] e1000e 0000:01:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[ 4.554379] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): Failed to initialize MSI-X interrupts. Falling back to MSI interrupts.
[ 4.554394] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): Failed to initialize MSI interrupts. Falling back to legacy interrupts.
[ 4.617549] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): registered PHC clock
[ 4.647217] xgene-enet APMC0D30:00: clocks have been setup already
[ 4.665238] xgene-ahci APMC0D0D:01: skip clock and PHY initialization
[ 4.671721] xgene-ahci APMC0D0D:01: controller can't do NCQ, turning off CAP_NCQ
[ 4.679178] xgene-ahci APMC0D0D:01: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.688066] xgene-ahci APMC0D0D:01: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.693143] e1000e 0000:01:00.0 eth2: (PCI Express:2.5GT/s:Width x1) 68:05:ca:46:ce:08
[ 4.693146] e1000e 0000:01:00.0 eth2: Intel(R) PRO/1000 Network Connection
[ 4.693162] e1000e 0000:01:00.0 eth2: MAC: 3, PHY: 8, PBA No: E46981-008
[ 4.696649] e1000e 0000:01:00.0 enp1s0: renamed from eth2
[ 4.722971] xgene-ahci APMC0D0D:01: port 0 is not capable of FBS
[ 4.729089] xgene-ahci APMC0D0D:01: port 1 is not capable of FBS
[ 4.736197] scsi host2: xgene-ahci
[ 4.740144] scsi host3: xgene-ahci
[ 4.743793] ata3: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x100 irq 28
[ 4.751735] ata4: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x180 irq 28
[ 4.857119] xgene-enet APMC0D30:01: clocks have been setup already
[ 4.874076] xgene-enet APMC0D31:00: clocks have been setup already
[ 5.088466] ata3: SATA link down (SStatus 0 SControl 4300)
[ 5.099014] ata4: SATA link down (SStatus 0 SControl 4300)
[ 5.217012] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.223552] ata1.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.230179] ata1.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.236797] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.243335] ata2.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.249961] ata2.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.256600] ata1.00: configured for UDMA/133
[ 5.261153] ata2.00: configured for UDMA/133
[ 5.261162] scsi 0:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.357513] scsi 1:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.357514] sd 0:0:0:0: [sda] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.357539] sd 0:0:0:0: [sda] Write Protect is off
[ 5.357542] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[ 5.357575] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.360081] sda: sda1 sda2 sda3
[ 5.360645] sd 0:0:0:0: [sda] Attached SCSI disk
[ 5.487441] sd 1:0:0:0: [sdb] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.494939] sd 1:0:0:0: [sdb] Write Protect is off
[ 5.499723] sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00
[ 5.499757] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.511144] sdb: sdb1
[ 5.513907] sd 1:0:0:0: [sdb] Attached SCSI disk
[ 5.962036] random: fast init done
[ 6.345919] SGI XFS with ACLs, security attributes, no debug enabled
[ 6.356572] XFS (dm-0): Mounting V5 Filesystem
[ 6.398617] XFS (dm-0): Ending clean mount
[ 6.973554] systemd-journald[198]: Received SIGTERM from PID 1 (systemd).
[ 6.989018] systemd: 20 output lines suppressed due to ratelimiting
[ 7.019839] audit: type=1404 audit(1493395245.790:2): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295
[ 7.106438] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.137061] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.216426] SELinux: 8 users, 14 roles, 4990 types, 304 bools, 1 sens, 1024 cats
[ 7.216433] SELinux: 94 classes, 105786 rules
[ 7.223909] SELinux: Class sctp_socket not defined in policy.
[ 7.229741] SELinux: Class icmp_socket not defined in policy.
[ 7.235543] SELinux: Class ax25_socket not defined in policy.
[ 7.241349] SELinux: Class ipx_socket not defined in policy.
[ 7.247069] SELinux: Class netrom_socket not defined in policy.
[ 7.253043] SELinux: Class atmpvc_socket not defined in policy.
[ 7.259021] SELinux: Class x25_socket not defined in policy.
[ 7.264736] SELinux: Class rose_socket not defined in policy.
[ 7.270540] SELinux: Class decnet_socket not defined in policy.
[ 7.276513] SELinux: Class atmsvc_socket not defined in policy.
[ 7.282490] SELinux: Class rds_socket not defined in policy.
[ 7.288207] SELinux: Class irda_socket not defined in policy.
[ 7.294008] SELinux: Class pppox_socket not defined in policy.
[ 7.299898] SELinux: Class llc_socket not defined in policy.
[ 7.305613] SELinux: Class can_socket not defined in policy.
[ 7.311330] SELinux: Class tipc_socket not defined in policy.
[ 7.317135] SELinux: Class bluetooth_socket not defined in policy.
[ 7.323368] SELinux: Class iucv_socket not defined in policy.
[ 7.329172] SELinux: Class rxrpc_socket not defined in policy.
[ 7.335059] SELinux: Class isdn_socket not defined in policy.
[ 7.340863] SELinux: Class phonet_socket not defined in policy.
[ 7.346836] SELinux: Class ieee802154_socket not defined in policy.
[ 7.353159] SELinux: Class caif_socket not defined in policy.
[ 7.358965] SELinux: Class alg_socket not defined in policy.
[ 7.364679] SELinux: Class nfc_socket not defined in policy.
[ 7.370397] SELinux: Class vsock_socket not defined in policy.
[ 7.376284] SELinux: Class kcm_socket not defined in policy.
[ 7.382002] SELinux: Class qipcrtr_socket not defined in policy.
[ 7.388066] SELinux: Class smc_socket not defined in policy.
[ 7.393780] SELinux: the above unknown classes and permissions will be allowed
[ 7.400979] SELinux: Completing initialization.
[ 7.400980] SELinux: Setting up existing superblocks.
[ 7.450804] audit: type=1403 audit(1493395246.220:3): policy loaded auid=4294967295 ses=4294967295
[ 7.455100] systemd[1]: Successfully loaded SELinux policy in 435.279ms.
[ 7.470770] systemd[1]: RTC configured in localtime, applying delta of -240 minutes to system time.
[ 7.497588] ip_tables: (C) 2000-2006 Netfilter Core Team
[ 7.502902] systemd[1]: Inserted module 'ip_tables'
[ 7.552579] systemd[1]: Relabelled /dev and /run in 42.787ms.
[ 8.005378] systemd-journald[616]: Received request to flush runtime journal from PID 1
[ 8.167863] xgene-slimpro-mbox APMC0D01:00: APM X-Gene SLIMpro MailBox registered
[ 8.172785] RPC: Registered named UNIX socket transport module.
[ 8.172790] RPC: Registered udp transport module.
[ 8.172794] RPC: Registered tcp transport module.
[ 8.172797] RPC: Registered tcp NFSv4.1 backchannel transport module.
[ 8.197228] xgene-rng APMC0D18:00: Couldn't get the clock for RNG
[ 8.204225] xgene-gpio-sb APMC0D15:00: Support 22 gpios, 6 irqs start from pin 8
[ 8.213102] xgene-gpio-sb APMC0D15:00: X-Gene GPIO Standby driver registered
[ 8.223510] xgene-slimpro-i2c APMC0D40:00: i2c mailbox channel request failed
[ 8.289837] sd 0:0:0:0: Attached scsi generic sg0 type 0
[ 8.295454] sd 1:0:0:0: Attached scsi generic sg1 type 0
[ 8.387796] xgene-slimpro-hwmon APMC0D29:00: APM X-Gene SoC HW monitor driver registered
[ 8.518644] Adding 8388544k swap on /dev/mapper/rhelp_dhcp--171-swap. Priority:-1 extents:1 across:8388544k SSFS
[ 8.531009] XFS (sda2): Mounting V5 Filesystem
[ 8.551661] XFS (sda2): Ending clean mount
[ 8.640881] Installing knfsd (copyright (C) 1996 okir at monad.swb.de).
[ 8.702746] XFS (dm-2): Mounting V5 Filesystem
[ 8.721803] XFS (dm-2): Ending clean mount
[ 9.540607] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.637501] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.650581] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.656789] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.669293] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.675510] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.687727] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.693923] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.706099] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 9.712318] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 9.730562] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 9.737469] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 9.744174] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 9.751066] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 9.757808] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.687474] xgene-enet APMC0D30:00 eth1: Link is Down
[ 10.692580] xgene-enet APMC0D05:00 eth0: Link is Down
[ 10.767324] xgene-enet APMC0D30:01 eth2: Link is Down
[ 12.418373] e1000e: enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
[ 12.426374] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready
[ 13.808268] xgene-enet APMC0D05:00 eth0: Link is Up - 1Gbps/Full - flow control off
[ 13.815925] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 108.537236] random: crng init done
-------------- next part --------------
[ 0.000000] Booting Linux on physical CPU 0x0
[ 0.000000] Linux version 4.11.0-rc8.xgene_msix2+ (root at mustang4.bos.jonmasters.org) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-14) (GCC) ) #2 SMP Fri Apr 28 12:06:14 EDT 2017
[ 0.000000] Boot CPU: AArch64 Processor [500f0001]
[ 0.000000] earlycon: uart8250 at MMIO32 0x000000001c020000 (options '')
[ 0.000000] bootconsole [uart8250] enabled
[ 0.000000] efi: Getting EFI parameters from FDT:
[ 0.000000] efi: EFI v2.40 by X-Gene Mustang Board EFI Aug 19 2016 18:02:24
[ 0.000000] efi: ACPI=0x43fa82d000 ACPI 2.0=0x43fa82d014 SMBIOS 3.0=0x43fa82b000 ESRT=0x43ff053d18
[ 0.000000] esrt: Reserving ESRT space from 0x00000043ff053d18 to 0x00000043ff053d78.
[ 0.000000] cma: Reserved 512 MiB at 0x00000040e0000000
[ 0.000000] ACPI: Early table checksum verification disabled
[ 0.000000] ACPI: RSDP 0x00000043FA82D014 000024 (v02 APM )
[ 0.000000] ACPI: XSDT 0x00000043FA82C0E8 000084 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: FACP 0x00000043FA822000 00010C (v05 APM XGENE 00000003 INTL 20140724)
[ 0.000000] ACPI: DSDT 0x00000043FA823000 005837 (v05 APM APM88xxx 00000001 INTL 20140724)
[ 0.000000] ACPI: DBG2 0x00000043FA829000 0000AA (v00 APMC0D XGENEDBG 00000000 INTL 20140724)
[ 0.000000] ACPI: GTDT 0x00000043FA820000 000060 (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: MCFG 0x00000043FA81F000 00003C (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: SPCR 0x00000043FA81E000 000050 (v02 APMC0D XGENESPC 00000000 INTL 20140724)
[ 0.000000] ACPI: SSDT 0x00000043FA81D000 00002D (v02 APM XGENE 00000001 INTL 20140724)
[ 0.000000] ACPI: BERT 0x00000043FA81C000 000030 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: HEST 0x00000043FA81B000 0002A8 (v01 APM XGENE 00000002 INTL 20140724)
[ 0.000000] ACPI: APIC 0x00000043FA81A000 0002A4 (v03 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA819000 000063 (v02 REDHAT MACADDRS 00000001 01000013)
[ 0.000000] ACPI: SSDT 0x00000043FA818000 000032 (v02 REDHAT UARTCLKS 00000001 01000013)
[ 0.000000] ACPI: PCCT 0x00000043FA817000 000300 (v01 APM XGENE 00000003 01000013)
[ 0.000000] ACPI: SPCR: console: uart,mmio,0x1c020000,115200
[ 0.000000] ACPI: NUMA: Failed to initialise from firmware
[ 0.000000] NUMA: Faking a node at [mem 0x0000000000000000-0x00000043ffffffff]
[ 0.000000] NUMA: Adding memblock [0x4000000000 - 0x40001fffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x4000200000 - 0x43fa7cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43fa7d0000 - 0x43faa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43faa20000 - 0x43ff9cffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ff9d0000 - 0x43ffa1ffff] on node 0
[ 0.000000] NUMA: Adding memblock [0x43ffa20000 - 0x43ffffffff] on node 0
[ 0.000000] NUMA: Initmem setup node 0 [mem 0x4000000000-0x43ffffffff]
[ 0.000000] NUMA: NODE_DATA [mem 0x43ffffe200-0x43ffffffff]
[ 0.000000] Zone ranges:
[ 0.000000] DMA [mem 0x0000004000000000-0x00000040ffffffff]
[ 0.000000] Normal [mem 0x0000004100000000-0x00000043ffffffff]
[ 0.000000] Movable zone start for each node
[ 0.000000] Early memory node ranges
[ 0.000000] node 0: [mem 0x0000004000000000-0x00000040001fffff]
[ 0.000000] node 0: [mem 0x0000004000200000-0x00000043fa7cffff]
[ 0.000000] node 0: [mem 0x00000043fa7d0000-0x00000043faa1ffff]
[ 0.000000] node 0: [mem 0x00000043faa20000-0x00000043ff9cffff]
[ 0.000000] node 0: [mem 0x00000043ff9d0000-0x00000043ffa1ffff]
[ 0.000000] node 0: [mem 0x00000043ffa20000-0x00000043ffffffff]
[ 0.000000] Initmem setup node 0 [mem 0x0000004000000000-0x00000043ffffffff]
[ 0.000000] On node 0 totalpages: 262144
[ 0.000000] DMA zone: 64 pages used for memmap
[ 0.000000] DMA zone: 0 pages reserved
[ 0.000000] DMA zone: 65536 pages, LIFO batch:1
[ 0.000000] Normal zone: 192 pages used for memmap
[ 0.000000] Normal zone: 196608 pages, LIFO batch:1
[ 0.000000] psci: is not implemented in ACPI.
[ 0.000000] percpu: Embedded 3 pages/cpu @ffff8003ffdf0000 s124312 r8192 d64104 u196608
[ 0.000000] pcpu-alloc: s124312 r8192 d64104 u196608 alloc=3*65536
[ 0.000000] pcpu-alloc: [0] 0 [0] 1 [0] 2 [0] 3 [0] 4 [0] 5 [0] 6 [0] 7
[ 0.000000] Detected PIPT I-cache on CPU0
[ 0.000000] Built 1 zonelists in Node order, mobility grouping on. Total pages: 261888
[ 0.000000] Policy zone: Normal
[ 0.000000] Kernel command line: BOOT_IMAGE=/vmlinuz-4.11.0-rc8.xgene_msix2+ root=/dev/mapper/rhelp_dhcp--171-root ro crashkernel=auto rd.lvm.lv=rhelp_dhcp-171/root rd.lvm.lv=rhelp_dhcp-171/swap LANG=en_US.UTF-8 acpi=force earlycon=uart8250,mmio32,0x1c020000 console=ttyS0,115200
[ 0.000000] PID hash table entries: 4096 (order: -1, 32768 bytes)
[ 0.000000] software IO TLB [mem 0x40dbff0000-0x40dfff0000] (64MB) mapped at [ffff8000dbff0000-ffff8000dffeffff]
[ 0.000000] Memory: 16109888K/16777216K available (7420K kernel code, 1354K rwdata, 3200K rodata, 1216K init, 7286K bss, 143040K reserved, 524288K cma-reserved)
[ 0.000000] Virtual kernel memory layout:
[ 0.000000] modules : 0xffff000000000000 - 0xffff000008000000 ( 128 MB)
[ 0.000000] vmalloc : 0xffff000008000000 - 0xffff7bdfffff0000 (126847 GB)
[ 0.000000] .text : 0xffff000008080000 - 0xffff0000087c0000 ( 7424 KB)
[ 0.000000] .rodata : 0xffff0000087c0000 - 0xffff000008af0000 ( 3264 KB)
[ 0.000000] .init : 0xffff000008af0000 - 0xffff000008c20000 ( 1216 KB)
[ 0.000000] .data : 0xffff000008c20000 - 0xffff000008d72a00 ( 1355 KB)
[ 0.000000] .bss : 0xffff000008d72a00 - 0xffff0000094902d4 ( 7287 KB)
[ 0.000000] fixed : 0xffff7fdffe7d0000 - 0xffff7fdffec00000 ( 4288 KB)
[ 0.000000] PCI I/O : 0xffff7fdffee00000 - 0xffff7fdfffe00000 ( 16 MB)
[ 0.000000] vmemmap : 0xffff7fe000000000 - 0xffff800000000000 ( 128 GB maximum)
[ 0.000000] 0xffff7fe000000000 - 0xffff7fe001000000 ( 16 MB actual)
[ 0.000000] memory : 0xffff800000000000 - 0xffff800400000000 ( 16384 MB)
[ 0.000000] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=8, Nodes=1
[ 0.000000] Hierarchical RCU implementation.
[ 0.000000] Build-time adjustment of leaf fanout to 64.
[ 0.000000] RCU restricting CPUs from NR_CPUS=4096 to nr_cpu_ids=8.
[ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=8
[ 0.000000] NR_IRQS:64 nr_irqs:64 0
[ 0.000000] GIC: Using split EOI/Deactivate mode
[ 0.000000] GICv3: No distributor detected at @ffff000008050000, giving up
[ 0.000000] arm_arch_timer: Architected cp15 timer(s) running at 50.00MHz (phys).
[ 0.000000] clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0xb8812736b, max_idle_ns: 440795202655 ns
[ 0.000002] sched_clock: 56 bits at 50MHz, resolution 20ns, wraps every 4398046511100ns
[ 0.008379] Console: colour dummy device 80x25
[ 0.012950] Calibrating delay loop (skipped), value calculated using timer frequency.. 100.00 BogoMIPS (lpj=500000)
[ 0.023693] pid_max: default: 32768 minimum: 301
[ 0.028471] ACPI: Core revision 20170119
[ 0.036063] ACPI Error: Method parse/execution failed [\_SB.ET00._STA] (Node ffff8003c0195c08), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.049121] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.056989] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.063808] ACPI Error: Method parse/execution failed [\_SB.ET01._STA] (Node ffff8003c0195eb0), AE_CTRL_PARSE_CONTINUE (20170119/psparse-543)
[ 0.076952] ACPI Error: Invalid zero thread count in method (20170119/dsmethod-790)
[ 0.084871] ACPI Error: Invalid OwnerId: 0x00 (20170119/utownerid-186)
[ 0.092015] ACPI: 4 ACPI AML tables successfully acquired and loaded
[ 0.098567] Security Framework initialized
[ 0.102838] SELinux: Initializing.
[ 0.106408] SELinux: Starting in permissive mode
[ 0.106607] Dentry cache hash table entries: 2097152 (order: 8, 16777216 bytes)
[ 0.116844] Inode-cache hash table entries: 1048576 (order: 7, 8388608 bytes)
[ 0.125552] Mount-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.132626] Mountpoint-cache hash table entries: 32768 (order: 2, 262144 bytes)
[ 0.140570] ftrace: allocating 26815 entries in 7 pages
[ 0.203748] ASID allocator initialised with 65536 entries
[ 0.209911] Remapping and enabling EFI services.
[ 0.214705] EFI remap 0x0000000010510000 => 0000000020000000
[ 0.220700] EFI remap 0x0000000010548000 => 0000000020018000
[ 0.226689] EFI remap 0x0000000017000000 => 0000000020020000
[ 0.232657] EFI remap 0x000000001c025000 => 0000000020035000
[ 0.238736] UEFI Runtime regions are not aligned to 64 KB -- buggy firmware?
[ 0.238741] ------------[ cut here ]------------
[ 0.250727] WARNING: CPU: 0 PID: 1 at arch/arm64/kernel/efi.c:34 efi_create_mapping+0x4c/0x104
[ 0.259532] Modules linked in:
[ 0.264214] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.11.0-rc8.xgene_msix2+ #2
[ 0.271827] Hardware name: APM X-Gene Mustang board (DT)
[ 0.277238] task: ffff8003c1bc0000 task.stack: ffff8003c1c00000
[ 0.283324] PC is at efi_create_mapping+0x4c/0x104
[ 0.288269] LR is at efi_create_mapping+0x4c/0x104
[ 0.293127] pc : [<ffff000008af51f0>] lr : [<ffff000008af51f0>] pstate: 60000045
[ 0.300784] sp : ffff8003c1c03d20
[ 0.304207] x29: ffff8003c1c03d20 x28: 0000000000000000
[ 0.309655] x27: ffff000008c53000 x26: ffff000008d429c8
[ 0.315110] x25: ffff000008d42000 x24: 0000000000000000
[ 0.320556] x23: ffff000008a4c910 x22: ffff000008c60000
[ 0.325994] x21: ffff000008c53000 x20: ffff000008d429c8
[ 0.331484] x19: ffff8003f96ab348 x18: 00000000b6a2110e
[ 0.337008] x17: 0000000000000000 x16: 0000000000000000
[ 0.342437] x15: ffff000008c53000 x14: 00000000cb9aed1d
[ 0.347944] x13: ffff0000089ca500 x12: 0000000000040b00
[ 0.353425] x11: ffff8003c1c039e0 x10: ffff8003c1c039e0
[ 0.358889] x9 : 0000000000000073 x8 : 6620796767756220
[ 0.364352] x7 : 2d2d20424b203436 x6 : ffff000009274718
[ 0.369825] x5 : ffff8003ffdf7530 x4 : 0000000000000000
[ 0.375280] x3 : 0000000000000000 x2 : 0000000000040b00
[ 0.380769] x1 : 0000000000040b00 x0 : 0000000000000040
[ 0.387782] ---[ end trace 9580655578fda7c1 ]---
[ 0.392519] Call trace:
[ 0.395045] Exception stack(0xffff8003c1c03b30 to 0xffff8003c1c03c60)
[ 0.401735] 3b20: ffff8003f96ab348 0001000000000000
[ 0.409730] 3b40: ffff8003c1c03d20 ffff000008af51f0 0000000060000045 ffff000008d429c8
[ 0.417828] 3b60: 0000000000000000 0000000000000000 ffff8003c1c03d20 ffff8003c1c03d20
[ 0.425857] 3b80: ffff8003c1c03ce0 00000000ffffffc8 ffff8003c1c03be0 ffff00000812f6b8
[ 0.433946] 3ba0: ffff000008c53000 00000000ffffffc8 ffff8003c1c03c00 ffff0000089be038
[ 0.442053] 3bc0: ffff8003c1c03d20 ffff8003c1c03d20 ffff8003c1c03ce0 00000000ffffffc8
[ 0.450126] 3be0: ffff8003c1c03c70 0000000000040b00 0000000000000040 0000000000040b00
[ 0.458207] 3c00: 0000000000040b00 0000000000000000 0000000000000000 ffff8003ffdf7530
[ 0.466227] 3c20: ffff000009274718 2d2d20424b203436 6620796767756220 0000000000000073
[ 0.474282] 3c40: ffff8003c1c039e0 ffff8003c1c039e0 0000000000040b00 ffff0000089ca500
[ 0.482321] [<ffff000008af51f0>] efi_create_mapping+0x4c/0x104
[ 0.488270] [<ffff000008b3db98>] arm_enable_runtime_services+0x154/0x238
[ 0.495238] [<ffff000008083698>] do_one_initcall+0x54/0x15c
[ 0.501030] [<ffff000008af0d18>] kernel_init_freeable+0xe4/0x25c
[ 0.507221] [<ffff0000087ae488>] kernel_init+0x18/0x108
[ 0.512606] [<ffff0000080833c0>] ret_from_fork+0x10/0x50
[ 0.518077] EFI remap 0x00000043fa7df000 => 000000002004f000
[ 0.524089] EFI remap 0x00000043fa82e000 => 000000002006e000
[ 0.530085] EFI remap 0x00000043ff9d6000 => 0000000020266000
[ 0.536148] EFI remap 0x00000043ff9ea000 => 000000002027a000
[ 0.542320] smp: Bringing up secondary CPUs ...
[ 0.547164] Detected PIPT I-cache on CPU1
[ 0.547192] CPU1: Booted secondary processor [500f0001]
[ 0.547411] Detected PIPT I-cache on CPU2
[ 0.547433] CPU2: Booted secondary processor [500f0001]
[ 0.547646] Detected PIPT I-cache on CPU3
[ 0.547660] CPU3: Booted secondary processor [500f0001]
[ 0.547854] Detected PIPT I-cache on CPU4
[ 0.547875] CPU4: Booted secondary processor [500f0001]
[ 0.548076] Detected PIPT I-cache on CPU5
[ 0.548090] CPU5: Booted secondary processor [500f0001]
[ 0.548281] Detected PIPT I-cache on CPU6
[ 0.548302] CPU6: Booted secondary processor [500f0001]
[ 0.548505] Detected PIPT I-cache on CPU7
[ 0.548519] CPU7: Booted secondary processor [500f0001]
[ 0.548562] smp: Brought up 1 node, 8 CPUs
[ 0.620493] SMP: Total of 8 processors activated.
[ 0.625425] CPU features: detected feature: 32-bit EL0 Support
[ 0.631511] CPU: All CPU(s) started at EL2
[ 0.636531] devtmpfs: initialized
[ 0.640271] SMBIOS 3.0.0 present.
[ 0.643745] DMI: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.15 Aug 19 2016
[ 0.653110] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604462750000 ns
[ 0.663392] futex hash table entries: 2048 (order: 2, 262144 bytes)
[ 0.670383] pinctrl core: initialized pinctrl subsystem
[ 0.676091] NET: Registered protocol family 16
[ 0.681082] cpuidle: using governor menu
[ 0.685276] Detected 8 PCC Subspaces
[ 0.689040] Registering PCC driver as Mailbox controller
[ 0.694649] vdso: 2 pages (1 code @ ffff0000087e0000, 1 data @ ffff000008c40000)
[ 0.702382] hw-breakpoint: found 4 breakpoint and 4 watchpoint registers.
[ 0.711945] DMA: preallocated 256 KiB pool for atomic allocations
[ 0.718368] ACPI: bus type PCI registered
[ 0.722572] acpiphp: ACPI Hot Plug PCI Controller Driver version: 0.5
[ 0.729340] Serial: AMBA PL011 UART driver
[ 0.735597] HugeTLB registered 512 MB page size, pre-allocated 0 pages
[ 0.743073] ACPI: Added _OSI(Module Device)
[ 0.747432] ACPI: Added _OSI(Processor Device)
[ 0.752083] ACPI: Added _OSI(3.0 _SCP Extensions)
[ 0.757005] ACPI: Added _OSI(Processor Aggregator Device)
[ 0.772090] ACPI: Interpreter enabled
[ 0.775913] ACPI: Using GIC for interrupt routing
[ 0.780864] ACPI: MCFG table detected, 1 entries
[ 0.785858] HEST: Table parsing has been initialized.
[ 0.793565] ACPI: Power Resource [SCVR] (on)
[ 0.828517] xgene-msi APMC0D0E:00: APM X-Gene PCIe MSI driver loaded
[ 0.835285] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff])
[ 0.841791] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.850585] acpi PNP0A08:00: _OSC: OS now controls [PCIeHotplug PME AER PCIeCapability]
[ 0.859017] acpi PNP0A08:00: MCFG quirk: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff] with xgene_v1_pcie_ecam_ops
[ 0.871645] acpi PNP0A08:00: [Firmware Bug]: ECAM area [mem 0xe0d0000000-0xe0dfffffff] not reserved in ACPI namespace
[ 0.882851] acpi PNP0A08:00: ECAM at [mem 0xe0d0000000-0xe0dfffffff] for [bus 00-ff]
[ 0.891009] Remapped I/O 0x000000e010000000 to [io 0x0000-0xffff window]
[ 0.898264] PCI host bridge to bus 0000:00
[ 0.902574] pci_bus 0000:00: root bus resource [io 0x0000-0xffff window] (bus address [0x10000000-0x1000ffff])
[ 0.913134] pci_bus 0000:00: root bus resource [mem 0xe040000000-0xe07fffffff window] (bus address [0x40000000-0x7fffffff])
[ 0.924779] pci_bus 0000:00: root bus resource [mem 0xf000000000-0xffffffffff window]
[ 0.932964] pci_bus 0000:00: root bus resource [bus 00-ff]
[ 0.938697] pci 0000:00:00.0: [10e8:e004] type 01 class 0x060400
[ 0.938776] pci 0000:00:00.0: supports D1 D2
[ 0.938937] pci 0000:01:00.0: [8086:10d3] type 00 class 0x020000
[ 0.938970] pci 0000:01:00.0: reg 0x10: [mem 0xe040080000-0xe04009ffff]
[ 0.938986] pci 0000:01:00.0: reg 0x14: [mem 0xe040000000-0xe04007ffff]
[ 0.939003] pci 0000:01:00.0: reg 0x18: [io 0x0000-0x001f]
[ 0.939019] pci 0000:01:00.0: reg 0x1c: [mem 0xe0400a0000-0xe0400a3fff]
[ 0.939065] pci 0000:01:00.0: reg 0x30: [mem 0xfffc0000-0xffffffff pref]
[ 0.939205] pci 0000:01:00.0: PME# supported from D0 D3hot D3cold
[ 0.968751] pci 0000:00:00.0: BAR 14: assigned [mem 0xe040000000-0xe0400fffff]
[ 0.976281] pci 0000:00:00.0: BAR 13: assigned [io 0x1000-0x1fff]
[ 0.982748] pci 0000:01:00.0: BAR 1: assigned [mem 0xe040000000-0xe04007ffff]
[ 0.990222] pci 0000:01:00.0: BAR 6: assigned [mem 0xe040080000-0xe0400bffff pref]
[ 0.998135] pci 0000:01:00.0: BAR 0: assigned [mem 0xe0400c0000-0xe0400dffff]
[ 1.005583] pci 0000:01:00.0: BAR 3: assigned [mem 0xe0400e0000-0xe0400e3fff]
[ 1.013044] pci 0000:01:00.0: BAR 2: assigned [io 0x1000-0x101f]
[ 1.019447] pci 0000:00:00.0: PCI bridge to [bus 01]
[ 1.024681] pci 0000:00:00.0: bridge window [io 0x1000-0x1fff]
[ 1.031099] pci 0000:00:00.0: bridge window [mem 0xe040000000-0xe0400fffff]
[ 1.039197] vgaarb: loaded
[ 1.042181] SCSI subsystem initialized
[ 1.046203] libata version 3.00 loaded.
[ 1.046240] ACPI: bus type USB registered
[ 1.050444] usbcore: registered new interface driver usbfs
[ 1.056204] usbcore: registered new interface driver hub
[ 1.061821] usbcore: registered new device driver usb
[ 1.067216] Registered efivars operations
[ 1.071684] NetLabel: Initializing
[ 1.075256] NetLabel: domain hash size = 128
[ 1.079777] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO
[ 1.085747] NetLabel: unlabeled traffic allowed by default
[ 1.091744] clocksource: Switched to clocksource arch_sys_counter
[ 1.116637] VFS: Disk quotas dquot_6.6.0
[ 1.120816] VFS: Dquot-cache hash table entries: 8192 (order 0, 65536 bytes)
[ 1.128374] pnp: PnP ACPI init
[ 1.132213] pnp: PnP ACPI: found 0 devices
[ 1.139026] NET: Registered protocol family 2
[ 1.143824] TCP established hash table entries: 131072 (order: 4, 1048576 bytes)
[ 1.151892] TCP bind hash table entries: 65536 (order: 4, 1048576 bytes)
[ 1.159275] TCP: Hash tables configured (established 131072 bind 65536)
[ 1.166293] UDP hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.172798] UDP-Lite hash table entries: 8192 (order: 2, 262144 bytes)
[ 1.179867] NET: Registered protocol family 1
[ 1.184479] PCI: CLS 64 bytes, default 128
[ 1.184578] Unpacking initramfs...
[ 1.868130] Freeing initrd memory: 32640K
[ 1.872713] kvm [1]: 8-bit VMID
[ 1.875992] kvm [1]: IDMAP page: 40009b7000
[ 1.880358] kvm [1]: HYP VA range: 800000000000:ffffffffffff
[ 1.886375] kvm [1]: Hyp mode initialized successfully
[ 1.891755] kvm [1]: GICV region size/alignment is unsafe, using trapping (reduced performance)
[ 1.900915] kvm [1]: vgic-v2 at 780cf000
[ 1.904807] kvm [1]: vgic interrupt IRQ1
[ 1.908909] kvm [1]: virtual timer IRQ4
[ 1.928681] audit: initializing netlink subsys (disabled)
[ 1.934412] audit: type=2000 audit(1.470:1): state=initialized audit_enabled=0 res=1
[ 1.934540] Initialise system trusted keyrings
[ 1.934609] workingset: timestamp_bits=44 max_order=18 bucket_order=0
[ 1.936691] zbud: loaded
[ 1.937062] SELinux: Registering netfilter hooks
[ 2.016448] NET: Registered protocol family 38
[ 2.021088] Key type asymmetric registered
[ 2.025420] Asymmetric key parser 'x509' registered
[ 2.030583] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251)
[ 2.038391] io scheduler noop registered
[ 2.042467] io scheduler deadline registered (default)
[ 2.047931] io scheduler cfq registered
[ 2.051950] io scheduler mq-deadline registered
[ 2.056870] atomic64_test: passed
[ 2.060612] xgene-gpio APMC0D14:00: X-Gene GPIO driver registered.
[ 2.067387] pcieport 0000:00:00.0: AER enabled with IRQ 83
[ 2.073172] pcieport 0000:00:00.0: Signaling PME with IRQ 83
[ 2.079355] input: Power Button as /devices/LNXSYSTM:00/PNP0C0C:00/input/input0
[ 2.087026] ACPI: Power Button [PWRB]
[ 2.091368] GHES: Failed to map GSI to IRQ for generic hardware error source: 2
[ 2.099018] GHES: probe of GHES.2 failed with error 82
[ 2.104386] GHES: Failed to map GSI to IRQ for generic hardware error source: 3
[ 2.112068] GHES: probe of GHES.3 failed with error 82
[ 2.117497] GHES: Failed to map GSI to IRQ for generic hardware error source: 4
[ 2.125101] GHES: probe of GHES.4 failed with error 82
[ 2.130452] GHES: Failed to map GSI to IRQ for generic hardware error source: 5
[ 2.138036] GHES: probe of GHES.5 failed with error 82
[ 2.143461] GHES: Failed to map GSI to IRQ for generic hardware error source: 6
[ 2.151121] GHES: probe of GHES.6 failed with error 82
[ 2.156510] GHES: Failed to map GSI to IRQ for generic hardware error source: 7
[ 2.164189] GHES: probe of GHES.7 failed with error 82
[ 2.169574] GHES: Failed to map GSI to IRQ for generic hardware error source: 8
[ 2.177249] GHES: probe of GHES.8 failed with error 82
[ 2.182604] GHES: Failed to map GSI to IRQ for generic hardware error source: 9
[ 2.190284] GHES: probe of GHES.9 failed with error 82
[ 2.195693] GHES: Failed to enable APEI firmware first mode.
[ 2.201771] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled
[ 2.208814] console [ttyS0] disabled
[ 2.232712] APMC0D08:00: ttyS0 at MMIO 0x1c020000 (irq = 22, base_baud = 3125000) is a U6_16550A
[ 2.241919] console [ttyS0] enabled
[ 2.249007] bootconsole [uart8250] disabled
[ 2.277886] APMC0D08:01: ttyS1 at MMIO 0x1c021000 (irq = 23, base_baud = 3125000) is a U6_16550A
[ 2.287044] cacheinfo: Unable to detect cache hierarchy for CPU 0
[ 2.293126] rdac: device handler registered
[ 2.297348] hp_sw: device handler registered
[ 2.301595] emc: device handler registered
[ 2.305766] alua: device handler registered
[ 2.310077] libphy: Fixed MDIO Bus: probed
[ 2.314237] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver
[ 2.320734] ehci-pci: EHCI PCI platform driver
[ 2.325177] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver
[ 2.331327] ohci-pci: OHCI PCI platform driver
[ 2.335764] uhci_hcd: USB Universal Host Controller Interface driver
[ 2.342182] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.347366] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 1
[ 2.354792] xhci-hcd 808622B7:00: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.363122] xhci-hcd 808622B7:00: irq 29, io mem 0x19000000
[ 2.368735] usb usb1: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.375494] usb usb1: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.382682] usb usb1: Product: xHCI Host Controller
[ 2.387533] usb usb1: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.394375] usb usb1: SerialNumber: 808622B7:00
[ 2.399097] hub 1-0:1.0: USB hub found
[ 2.402848] hub 1-0:1.0: 1 port detected
[ 2.406851] xhci-hcd 808622B7:00: xHCI Host Controller
[ 2.412034] xhci-hcd 808622B7:00: new USB bus registered, assigned bus number 2
[ 2.419332] usb usb2: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.427425] usb usb2: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.434184] usb usb2: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.441368] usb usb2: Product: xHCI Host Controller
[ 2.446223] usb usb2: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.453069] usb usb2: SerialNumber: 808622B7:00
[ 2.457761] hub 2-0:1.0: USB hub found
[ 2.461501] hub 2-0:1.0: 1 port detected
[ 2.465534] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.470715] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 3
[ 2.478113] xhci-hcd 808622B7:01: hcc params 0x0220f06d hci version 0x100 quirks 0x00010010
[ 2.486441] xhci-hcd 808622B7:01: irq 30, io mem 0x19800000
[ 2.492041] usb usb3: New USB device found, idVendor=1d6b, idProduct=0002
[ 2.498793] usb usb3: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.505984] usb usb3: Product: xHCI Host Controller
[ 2.510836] usb usb3: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.517682] usb usb3: SerialNumber: 808622B7:01
[ 2.522397] hub 3-0:1.0: USB hub found
[ 2.526137] hub 3-0:1.0: 1 port detected
[ 2.530122] xhci-hcd 808622B7:01: xHCI Host Controller
[ 2.535305] xhci-hcd 808622B7:01: new USB bus registered, assigned bus number 4
[ 2.542607] usb usb4: We don't know the algorithms for LPM for this host, disabling LPM.
[ 2.550692] usb usb4: New USB device found, idVendor=1d6b, idProduct=0003
[ 2.557451] usb usb4: New USB device strings: Mfr=3, Product=2, SerialNumber=1
[ 2.564640] usb usb4: Product: xHCI Host Controller
[ 2.569491] usb usb4: Manufacturer: Linux 4.11.0-rc8.xgene_msix2+ xhci-hcd
[ 2.576334] usb usb4: SerialNumber: 808622B7:01
[ 2.581037] hub 4-0:1.0: USB hub found
[ 2.584788] hub 4-0:1.0: 1 port detected
[ 2.588828] usbcore: registered new interface driver usbserial
[ 2.594649] usbcore: registered new interface driver usbserial_generic
[ 2.601150] usbserial: USB Serial support registered for generic
[ 2.607208] mousedev: PS/2 mouse device common for all mice
[ 2.612948] rtc-efi rtc-efi: rtc core: registered rtc-efi as rtc0
[ 2.619241] EFI Variables Facility v0.08 2004-May-17
[ 2.624405] hidraw: raw HID events driver (C) Jiri Kosina
[ 2.629836] usbcore: registered new interface driver usbhid
[ 2.635388] usbhid: USB HID core driver
[ 2.639274] drop_monitor: Initializing network drop monitor service
[ 2.645602] Initializing XFRM netlink socket
[ 2.650006] NET: Registered protocol family 10
[ 2.654800] Segment Routing with IPv6
[ 2.658462] NET: Registered protocol family 17
[ 2.663153] registered taskstats version 1
[ 2.667231] Loading compiled-in X.509 certificates
[ 2.675102] alg: No test for pkcs1pad(rsa,sha1) (pkcs1pad(rsa-generic,sha1))
[ 2.683548] Loaded X.509 cert 'Build time autogenerated kernel key: 98aa1b1964f4052ba920878b6f8348e178459895'
[ 2.693521] zswap: loaded using pool lzo/zbud
[ 2.717703] Key type big_key registered
[ 2.721594] BERT: [Firmware Bug]: table invalid.
[ 2.726272] rtc-efi rtc-efi: setting system clock to 2017-04-28 16:24:34 UTC (1493396674)
[ 2.759039] Freeing unused kernel memory: 1216K
[ 2.768550] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.775743] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.782366] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.790811] systemd[1]: systemd 219 running in system mode. (+PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP +LIBCRYPTSETUP +GCRYPT +GNUTLS +ACL +XZ -LZ4 -SECCOMP +BLKID +ELFUTILS +KMOD +IDN)
[ 2.809044] systemd[1]: Detected architecture arm64.
[ 2.814004] systemd[1]: Running in initial RAM disk.
[ 2.851832] systemd[1]: Set hostname to <localhost.localdomain>.
[ 2.970013] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.976652] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.983295] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.989979] random: systemd: uninitialized urandom read (16 bytes read)
[ 2.997677] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.004524] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.011427] random: systemd: uninitialized urandom read (16 bytes read)
[ 3.051810] systemd[1]: Reached target Timers.
[ 3.056250] systemd[1]: Starting Timers.
[ 3.091786] systemd[1]: Created slice Root Slice.
[ 3.096483] systemd[1]: Starting Root Slice.
[ 3.131780] systemd[1]: Listening on udev Kernel Socket.
[ 3.137083] systemd[1]: Starting udev Kernel Socket.
[ 4.102501] device-mapper: uevent: version 1.0.3
[ 4.107414] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel at redhat.com
[ 4.412199] xgene-ahci APMC0D0D:00: skip clock and PHY initialization
[ 4.418619] xgene-ahci APMC0D0D:00: controller can't do NCQ, turning off CAP_NCQ
[ 4.426125] xgene-ahci APMC0D0D:00: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.434987] xgene-ahci APMC0D0D:00: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.434998] xgene-ahci APMC0D0D:00: port 0 is not capable of FBS
[ 4.470653] pps_core: LinuxPPS API ver. 1 registered
[ 4.475669] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[ 4.491986] xgene-ahci APMC0D0D:00: port 1 is not capable of FBS
[ 4.507826] PTP clock support registered
[ 4.521307] sdhci: Secure Digital Host Controller Interface driver
[ 4.527512] sdhci: Copyright(c) Pierre Ossman
[ 4.577139] mmc0: SDHCI controller on ACPI [APMC0D0C:00] using PIO
[ 4.612188] scsi host0: xgene-ahci
[ 4.615838] scsi host1: xgene-ahci
[ 4.619314] ata1: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x100 irq 27
[ 4.621598] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k
[ 4.621602] e1000e: Copyright(c) 1999 - 2015 Intel Corporation.
[ 4.621841] e1000e 0000:01:00.0: enabling device (0000 -> 0002)
[ 4.622056] e1000e 0000:01:00.0: Interrupt Throttling Rate (ints/sec) set to dynamic conservative mode
[ 4.654091] ata2: SATA max UDMA/133 mmio [mem 0x1a400000-0x1a400fff] port 0x180 irq 27
[ 4.662326] xgene-ahci APMC0D0D:01: skip clock and PHY initialization
[ 4.668755] xgene-ahci APMC0D0D:01: controller can't do NCQ, turning off CAP_NCQ
[ 4.676203] xgene-ahci APMC0D0D:01: AHCI 0001.0300 32 slots 2 ports 6 Gbps 0x3 impl platform mode
[ 4.685069] xgene-ahci APMC0D0D:01: flags: 64bit sntf pm only pmp fbs pio slum part ccc
[ 4.693201] xgene-ahci APMC0D0D:01: port 0 is not capable of FBS
[ 4.693880] e1000e 0000:01:00.0 0000:01:00.0 (uninitialized): registered PHC clock
[ 4.706842] xgene-ahci APMC0D0D:01: port 1 is not capable of FBS
[ 4.714376] scsi host2: xgene-ahci
[ 4.718239] scsi host3: xgene-ahci
[ 4.721961] ata3: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x100 irq 28
[ 4.729850] ata4: SATA max UDMA/133 mmio [mem 0x1a800000-0x1a800fff] port 0x180 irq 28
[ 4.738769] libphy: APM X-Gene MDIO bus: probed
[ 4.791711] e1000e 0000:01:00.0 eth0: (PCI Express:2.5GT/s:Width x1) 68:05:ca:46:ce:08
[ 4.799622] e1000e 0000:01:00.0 eth0: Intel(R) PRO/1000 Network Connection
[ 4.806511] e1000e 0000:01:00.0 eth0: MAC: 3, PHY: 8, PBA No: E46981-008
[ 4.808607] xgene-enet APMC0D05:00: clocks have been setup already
[ 4.821883] e1000e 0000:01:00.0 enp1s0: renamed from eth0
[ 5.011899] xgene-enet APMC0D30:00: clocks have been setup already
[ 5.063242] ata3: SATA link down (SStatus 0 SControl 4300)
[ 5.068752] ata4: SATA link down (SStatus 0 SControl 4300)
[ 5.121895] xgene-enet APMC0D30:01: clocks have been setup already
[ 5.136261] xgene-enet APMC0D31:00: clocks have been setup already
[ 5.331836] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.338118] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 4300)
[ 5.344486] ata2.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.351098] ata2.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.357813] ata1.00: ATA-8: Corsair Neutron GTX SSD, M311, max UDMA/133
[ 5.364432] ata1.00: 234441648 sectors, multi 1: LBA48 NCQ (depth 0/32)
[ 5.371105] ata2.00: configured for UDMA/133
[ 5.375470] ata1.00: configured for UDMA/133
[ 5.380021] scsi 0:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.482145] sd 0:0:0:0: [sda] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.482607] scsi 1:0:0:0: Direct-Access ATA Corsair Neutron M311 PQ: 0 ANSI: 5
[ 5.498165] sd 0:0:0:0: [sda] Write Protect is off
[ 5.502945] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00
[ 5.502986] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.514451] sda: sda1 sda2 sda3
[ 5.518153] sd 0:0:0:0: [sda] Attached SCSI disk
[ 5.602396] sd 1:0:0:0: [sdb] 234441648 512-byte logical blocks: (120 GB/112 GiB)
[ 5.609881] sd 1:0:0:0: [sdb] Write Protect is off
[ 5.614681] sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00
[ 5.614739] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[ 5.626189] sdb: sdb1
[ 5.628980] sd 1:0:0:0: [sdb] Attached SCSI disk
[ 6.113010] random: fast init done
[ 6.461233] SGI XFS with ACLs, security attributes, no debug enabled
[ 6.471638] XFS (dm-0): Mounting V5 Filesystem
[ 6.501907] XFS (dm-0): Ending clean mount
[ 7.216272] systemd-journald[201]: Received SIGTERM from PID 1 (systemd).
[ 7.263513] systemd: 20 output lines suppressed due to ratelimiting
[ 7.301701] audit: type=1404 audit(1493396679.060:2): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295
[ 7.410911] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.516206] SELinux: 32768 avtab hash slots, 105786 rules.
[ 7.718524] SELinux: 8 users, 14 roles, 4990 types, 304 bools, 1 sens, 1024 cats
[ 7.718542] SELinux: 94 classes, 105786 rules
[ 7.741338] SELinux: Class sctp_socket not defined in policy.
[ 7.747177] SELinux: Class icmp_socket not defined in policy.
[ 7.753000] SELinux: Class ax25_socket not defined in policy.
[ 7.758809] SELinux: Class ipx_socket not defined in policy.
[ 7.764542] SELinux: Class netrom_socket not defined in policy.
[ 7.770524] SELinux: Class atmpvc_socket not defined in policy.
[ 7.776515] SELinux: Class x25_socket not defined in policy.
[ 7.782247] SELinux: Class rose_socket not defined in policy.
[ 7.788056] SELinux: Class decnet_socket not defined in policy.
[ 7.794048] SELinux: Class atmsvc_socket not defined in policy.
[ 7.800028] SELinux: Class rds_socket not defined in policy.
[ 7.805760] SELinux: Class irda_socket not defined in policy.
[ 7.811568] SELinux: Class pppox_socket not defined in policy.
[ 7.817473] SELinux: Class llc_socket not defined in policy.
[ 7.823204] SELinux: Class can_socket not defined in policy.
[ 7.828926] SELinux: Class tipc_socket not defined in policy.
[ 7.834744] SELinux: Class bluetooth_socket not defined in policy.
[ 7.840984] SELinux: Class iucv_socket not defined in policy.
[ 7.846802] SELinux: Class rxrpc_socket not defined in policy.
[ 7.852707] SELinux: Class isdn_socket not defined in policy.
[ 7.858514] SELinux: Class phonet_socket not defined in policy.
[ 7.864504] SELinux: Class ieee802154_socket not defined in policy.
[ 7.870831] SELinux: Class caif_socket not defined in policy.
[ 7.876649] SELinux: Class alg_socket not defined in policy.
[ 7.882381] SELinux: Class nfc_socket not defined in policy.
[ 7.888103] SELinux: Class vsock_socket not defined in policy.
[ 7.894008] SELinux: Class kcm_socket not defined in policy.
[ 7.899730] SELinux: Class qipcrtr_socket not defined in policy.
[ 7.905822] SELinux: Class smc_socket not defined in policy.
[ 7.911545] SELinux: the above unknown classes and permissions will be allowed
[ 7.918772] SELinux: Completing initialization.
[ 7.918776] SELinux: Setting up existing superblocks.
[ 7.993509] audit: type=1403 audit(1493396679.760:3): policy loaded auid=4294967295 ses=4294967295
[ 8.002331] systemd[1]: Successfully loaded SELinux policy in 701.052ms.
[ 8.029233] systemd[1]: RTC configured in localtime, applying delta of -240 minutes to system time.
[ 8.076262] ip_tables: (C) 2000-2006 Netfilter Core Team
[ 8.081619] systemd[1]: Inserted module 'ip_tables'
[ 8.199798] systemd[1]: Relabelled /dev and /run in 108.385ms.
[ 8.666128] systemd-journald[618]: Received request to flush runtime journal from PID 1
[ 8.793125] xgene-slimpro-i2c APMC0D40:00: i2c mailbox channel request failed
[ 8.809689] RPC: Registered named UNIX socket transport module.
[ 8.815634] RPC: Registered udp transport module.
[ 8.820316] RPC: Registered tcp transport module.
[ 8.825018] RPC: Registered tcp NFSv4.1 backchannel transport module.
[ 8.872038] sd 0:0:0:0: Attached scsi generic sg0 type 0
[ 8.877502] sd 1:0:0:0: Attached scsi generic sg1 type 0
[ 8.883774] xgene-slimpro-hwmon APMC0D29:00: APM X-Gene SoC HW monitor driver registered
[ 8.892777] xgene-rng APMC0D18:00: Couldn't get the clock for RNG
[ 8.899408] xgene-slimpro-mbox APMC0D01:00: APM X-Gene SLIMpro MailBox registered
[ 8.910481] xgene-gpio-sb APMC0D15:00: Support 22 gpios, 6 irqs start from pin 8
[ 8.918470] xgene-gpio-sb APMC0D15:00: X-Gene GPIO Standby driver registered
[ 9.044885] Installing knfsd (copyright (C) 1996 okir at monad.swb.de).
[ 9.118751] XFS (sda2): Mounting V5 Filesystem
[ 9.125864] Adding 8388544k swap on /dev/mapper/rhelp_dhcp--171-swap. Priority:-1 extents:1 across:8388544k SSFS
[ 9.141179] XFS (sda2): Ending clean mount
[ 9.286221] XFS (dm-2): Mounting V5 Filesystem
[ 9.305151] XFS (dm-2): Ending clean mount
[ 10.242361] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.339596] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.367672] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.374758] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.403220] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.410389] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.437608] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.444895] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.471938] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.478809] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 10.562700] IPv6: ADDRCONF(NETDEV_UP): enp1s0: link is not ready
[ 10.571438] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 10.580052] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
[ 10.588624] IPv6: ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 10.597191] IPv6: ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 11.412166] xgene-enet APMC0D05:00 eth0: Link is Down
[ 11.492472] xgene-enet APMC0D30:00 eth1: Link is Down
[ 11.497600] xgene-enet APMC0D30:01 eth2: Link is Down
[ 14.533303] e1000e: enp1s0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
[ 14.541337] IPv6: ADDRCONF(NETDEV_CHANGE): enp1s0: link becomes ready
[ 14.548556] xgene-enet APMC0D05:00 eth0: Link is Up - 1Gbps/Full - flow control off
[ 14.556244] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-04-28 16:36 ` Jon Masters
(?)
@ 2017-05-04 23:24 ` Jon Masters
-1 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-05-04 23:24 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas,
linux-pci
Cc: patches, linux-kernel, linux-arm-kernel, rjw, Duc Dang, Jon Masters
On 04/28/2017 12:36 PM, Jon Masters wrote:
> On 04/28/2017 01:11 AM, Jon Masters wrote:
>> On 04/27/2017 08:54 PM, Khuong Dinh wrote:
>>> From: Khuong Dinh <kdinh@apm.com>
>>>
>>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>>
>>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>>> Signed-off-by: Duc Dang <dhdang@apm.com>
>>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>>
>> Thanks. Currently testing. Will review and followup shortly.
>
> Attaching boot logs before and after. I tested with an Intel e1000.
> Before the patch I get fallback interrupts. Afterward, I get X-Gene MSI
> interrupts working. Please followup with answers to Marc's question
> about how you propose to guarantee boottime ordering of binding.
Any word?
Jon.
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-04 23:24 ` Jon Masters
0 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-05-04 23:24 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas,
linux-pci
Cc: patches, Duc Dang, rjw, linux-kernel, Jon Masters, linux-arm-kernel
On 04/28/2017 12:36 PM, Jon Masters wrote:
> On 04/28/2017 01:11 AM, Jon Masters wrote:
>> On 04/27/2017 08:54 PM, Khuong Dinh wrote:
>>> From: Khuong Dinh <kdinh@apm.com>
>>>
>>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>>
>>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>>> Signed-off-by: Duc Dang <dhdang@apm.com>
>>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>>
>> Thanks. Currently testing. Will review and followup shortly.
>
> Attaching boot logs before and after. I tested with an Intel e1000.
> Before the patch I get fallback interrupts. Afterward, I get X-Gene MSI
> interrupts working. Please followup with answers to Marc's question
> about how you propose to guarantee boottime ordering of binding.
Any word?
Jon.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-04 23:24 ` Jon Masters
0 siblings, 0 replies; 47+ messages in thread
From: Jon Masters @ 2017-05-04 23:24 UTC (permalink / raw)
To: linux-arm-kernel
On 04/28/2017 12:36 PM, Jon Masters wrote:
> On 04/28/2017 01:11 AM, Jon Masters wrote:
>> On 04/27/2017 08:54 PM, Khuong Dinh wrote:
>>> From: Khuong Dinh <kdinh@apm.com>
>>>
>>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>>
>>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>>> Signed-off-by: Duc Dang <dhdang@apm.com>
>>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>>
>> Thanks. Currently testing. Will review and followup shortly.
>
> Attaching boot logs before and after. I tested with an Intel e1000.
> Before the patch I get fallback interrupts. Afterward, I get X-Gene MSI
> interrupts working. Please followup with answers to Marc's question
> about how you propose to guarantee boottime ordering of binding.
Any word?
Jon.
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-04-28 9:27 ` Marc Zyngier
(?)
@ 2017-05-05 0:36 ` Khuong Dinh
-1 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-05-05 0:36 UTC (permalink / raw)
To: Marc Zyngier
Cc: Lorenzo Pieralisi, msalter, Bjorn Helgaas, linux-pci, jcm,
patches, linux-kernel, linux-arm-kernel, rjw, Duc Dang
Hi Marc,
There's no explicit dependency between pcie driver and msi controller.
The current solution that we did is relying on the node ordering in BIOS.
ACPI 5.0 introduced _DEP method to assign a higher priority in
start ordering.
This method could be applied in case of msi and pcie are the same
level of subsys_init (in ACPI boot).
However, PCIe driver has not supported for this dependency check yet.
How do you think about this solution.
Best regards,
Khuong
On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> On 28/04/17 01:54, Khuong Dinh wrote:
>> From: Khuong Dinh <kdinh@apm.com>
>>
>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>
>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> Signed-off-by: Duc Dang <dhdang@apm.com>
>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> ---
>> v2:
>> - Verify with BIOS version 3.06.25 and 3.07.09
>> v1:
>> - Initial version
>> ---
>> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> 1 files changed, 32 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> index f1b633b..00aaa3d 100644
>> --- a/drivers/pci/host/pci-xgene-msi.c
>> +++ b/drivers/pci/host/pci-xgene-msi.c
>> @@ -24,6 +24,7 @@
>> #include <linux/pci.h>
>> #include <linux/platform_device.h>
>> #include <linux/of_pci.h>
>> +#include <linux/acpi.h>
>>
>> #define MSI_IR0 0x000000
>> #define MSI_INT0 0x800000
>> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> };
>>
>> struct xgene_msi {
>> - struct device_node *node;
>> + struct fwnode_handle *fwnode;
>> struct irq_domain *inner_domain;
>> struct irq_domain *msi_domain;
>> u64 msi_addr;
>> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> .free = xgene_irq_domain_free,
>> };
>>
>> +#ifdef CONFIG_ACPI
>> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> +{
>> + return xgene_msi_ctrl.fwnode;
>> +}
>> +#endif
>> +
>> static int xgene_allocate_domains(struct xgene_msi *msi)
>> {
>> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> if (!msi->inner_domain)
>> return -ENOMEM;
>>
>> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> &xgene_msi_domain_info,
>> msi->inner_domain);
>>
>> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> return -ENOMEM;
>> }
>>
>> +#ifdef CONFIG_ACPI
>> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> +#endif
>> return 0;
>> }
>>
>> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> {},
>> };
>>
>> +#ifdef CONFIG_ACPI
>> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> + {"APMC0D0E", 0},
>> + { },
>> +};
>> +#endif
>> +
>> static int xgene_msi_probe(struct platform_device *pdev)
>> {
>> struct resource *res;
>> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> goto error;
>> }
>> xgene_msi->msi_addr = res->start;
>> - xgene_msi->node = pdev->dev.of_node;
>> +
>> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> + if (!xgene_msi->fwnode) {
>> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>
> Please provide something other than NULL, such as the base address if
> the device. That's quite useful for debugging.
>
>> + if (!xgene_msi->fwnode) {
>> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> + rc = ENOMEM;
>> + goto error;
>> + }
>> + }
>> +
>> xgene_msi->num_cpus = num_possible_cpus();
>>
>> rc = xgene_msi_init_allocator(xgene_msi);
>> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> .driver = {
>> .name = "xgene-msi",
>> .of_match_table = xgene_msi_match_table,
>> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> },
>> .probe = xgene_msi_probe,
>> .remove = xgene_msi_remove,
>>
>
> The code is trivial, but relies on the MSI controller to probe before
> the PCI bus. What enforces this? How is it making sure that this is not
> going to break in the next kernel release? As far as I can tell, there
> is no explicit dependency between the two, making it the whole thing
> extremely fragile.
>
> Thanks,
>
> M.
> --
> Jazz is not dead. It just smells funny...
--
CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
for the sole use of the intended recipient(s) and contains information that
is confidential and proprietary to Applied Micro Circuits Corporation or
its subsidiaries. It is to be used solely for the purpose of furthering the
parties' business relationship. All unauthorized review, use, disclosure or
distribution is prohibited. If you are not the intended recipient, please
contact the sender by reply e-mail and destroy all copies of the original
message.
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-05 0:36 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-05-05 0:36 UTC (permalink / raw)
To: Marc Zyngier
Cc: rjw, Lorenzo Pieralisi, Duc Dang, linux-pci, msalter, patches,
linux-kernel, jcm, Bjorn Helgaas, linux-arm-kernel
Hi Marc,
There's no explicit dependency between pcie driver and msi controller.
The current solution that we did is relying on the node ordering in BIOS.
ACPI 5.0 introduced _DEP method to assign a higher priority in
start ordering.
This method could be applied in case of msi and pcie are the same
level of subsys_init (in ACPI boot).
However, PCIe driver has not supported for this dependency check yet.
How do you think about this solution.
Best regards,
Khuong
On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> On 28/04/17 01:54, Khuong Dinh wrote:
>> From: Khuong Dinh <kdinh@apm.com>
>>
>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>
>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> Signed-off-by: Duc Dang <dhdang@apm.com>
>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> ---
>> v2:
>> - Verify with BIOS version 3.06.25 and 3.07.09
>> v1:
>> - Initial version
>> ---
>> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> 1 files changed, 32 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> index f1b633b..00aaa3d 100644
>> --- a/drivers/pci/host/pci-xgene-msi.c
>> +++ b/drivers/pci/host/pci-xgene-msi.c
>> @@ -24,6 +24,7 @@
>> #include <linux/pci.h>
>> #include <linux/platform_device.h>
>> #include <linux/of_pci.h>
>> +#include <linux/acpi.h>
>>
>> #define MSI_IR0 0x000000
>> #define MSI_INT0 0x800000
>> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> };
>>
>> struct xgene_msi {
>> - struct device_node *node;
>> + struct fwnode_handle *fwnode;
>> struct irq_domain *inner_domain;
>> struct irq_domain *msi_domain;
>> u64 msi_addr;
>> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> .free = xgene_irq_domain_free,
>> };
>>
>> +#ifdef CONFIG_ACPI
>> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> +{
>> + return xgene_msi_ctrl.fwnode;
>> +}
>> +#endif
>> +
>> static int xgene_allocate_domains(struct xgene_msi *msi)
>> {
>> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> if (!msi->inner_domain)
>> return -ENOMEM;
>>
>> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> &xgene_msi_domain_info,
>> msi->inner_domain);
>>
>> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> return -ENOMEM;
>> }
>>
>> +#ifdef CONFIG_ACPI
>> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> +#endif
>> return 0;
>> }
>>
>> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> {},
>> };
>>
>> +#ifdef CONFIG_ACPI
>> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> + {"APMC0D0E", 0},
>> + { },
>> +};
>> +#endif
>> +
>> static int xgene_msi_probe(struct platform_device *pdev)
>> {
>> struct resource *res;
>> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> goto error;
>> }
>> xgene_msi->msi_addr = res->start;
>> - xgene_msi->node = pdev->dev.of_node;
>> +
>> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> + if (!xgene_msi->fwnode) {
>> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>
> Please provide something other than NULL, such as the base address if
> the device. That's quite useful for debugging.
>
>> + if (!xgene_msi->fwnode) {
>> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> + rc = ENOMEM;
>> + goto error;
>> + }
>> + }
>> +
>> xgene_msi->num_cpus = num_possible_cpus();
>>
>> rc = xgene_msi_init_allocator(xgene_msi);
>> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> .driver = {
>> .name = "xgene-msi",
>> .of_match_table = xgene_msi_match_table,
>> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> },
>> .probe = xgene_msi_probe,
>> .remove = xgene_msi_remove,
>>
>
> The code is trivial, but relies on the MSI controller to probe before
> the PCI bus. What enforces this? How is it making sure that this is not
> going to break in the next kernel release? As far as I can tell, there
> is no explicit dependency between the two, making it the whole thing
> extremely fragile.
>
> Thanks,
>
> M.
> --
> Jazz is not dead. It just smells funny...
--
CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
for the sole use of the intended recipient(s) and contains information that
is confidential and proprietary to Applied Micro Circuits Corporation or
its subsidiaries. It is to be used solely for the purpose of furthering the
parties' business relationship. All unauthorized review, use, disclosure or
distribution is prohibited. If you are not the intended recipient, please
contact the sender by reply e-mail and destroy all copies of the original
message.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-05 0:36 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-05-05 0:36 UTC (permalink / raw)
To: linux-arm-kernel
Hi Marc,
There's no explicit dependency between pcie driver and msi controller.
The current solution that we did is relying on the node ordering in BIOS.
ACPI 5.0 introduced _DEP method to assign a higher priority in
start ordering.
This method could be applied in case of msi and pcie are the same
level of subsys_init (in ACPI boot).
However, PCIe driver has not supported for this dependency check yet.
How do you think about this solution.
Best regards,
Khuong
On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> On 28/04/17 01:54, Khuong Dinh wrote:
>> From: Khuong Dinh <kdinh@apm.com>
>>
>> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>
>> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> Signed-off-by: Duc Dang <dhdang@apm.com>
>> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> ---
>> v2:
>> - Verify with BIOS version 3.06.25 and 3.07.09
>> v1:
>> - Initial version
>> ---
>> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> 1 files changed, 32 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> index f1b633b..00aaa3d 100644
>> --- a/drivers/pci/host/pci-xgene-msi.c
>> +++ b/drivers/pci/host/pci-xgene-msi.c
>> @@ -24,6 +24,7 @@
>> #include <linux/pci.h>
>> #include <linux/platform_device.h>
>> #include <linux/of_pci.h>
>> +#include <linux/acpi.h>
>>
>> #define MSI_IR0 0x000000
>> #define MSI_INT0 0x800000
>> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> };
>>
>> struct xgene_msi {
>> - struct device_node *node;
>> + struct fwnode_handle *fwnode;
>> struct irq_domain *inner_domain;
>> struct irq_domain *msi_domain;
>> u64 msi_addr;
>> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> .free = xgene_irq_domain_free,
>> };
>>
>> +#ifdef CONFIG_ACPI
>> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> +{
>> + return xgene_msi_ctrl.fwnode;
>> +}
>> +#endif
>> +
>> static int xgene_allocate_domains(struct xgene_msi *msi)
>> {
>> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> if (!msi->inner_domain)
>> return -ENOMEM;
>>
>> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> &xgene_msi_domain_info,
>> msi->inner_domain);
>>
>> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> return -ENOMEM;
>> }
>>
>> +#ifdef CONFIG_ACPI
>> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> +#endif
>> return 0;
>> }
>>
>> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> {},
>> };
>>
>> +#ifdef CONFIG_ACPI
>> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> + {"APMC0D0E", 0},
>> + { },
>> +};
>> +#endif
>> +
>> static int xgene_msi_probe(struct platform_device *pdev)
>> {
>> struct resource *res;
>> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> goto error;
>> }
>> xgene_msi->msi_addr = res->start;
>> - xgene_msi->node = pdev->dev.of_node;
>> +
>> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> + if (!xgene_msi->fwnode) {
>> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>
> Please provide something other than NULL, such as the base address if
> the device. That's quite useful for debugging.
>
>> + if (!xgene_msi->fwnode) {
>> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> + rc = ENOMEM;
>> + goto error;
>> + }
>> + }
>> +
>> xgene_msi->num_cpus = num_possible_cpus();
>>
>> rc = xgene_msi_init_allocator(xgene_msi);
>> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> .driver = {
>> .name = "xgene-msi",
>> .of_match_table = xgene_msi_match_table,
>> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> },
>> .probe = xgene_msi_probe,
>> .remove = xgene_msi_remove,
>>
>
> The code is trivial, but relies on the MSI controller to probe before
> the PCI bus. What enforces this? How is it making sure that this is not
> going to break in the next kernel release? As far as I can tell, there
> is no explicit dependency between the two, making it the whole thing
> extremely fragile.
>
> Thanks,
>
> M.
> --
> Jazz is not dead. It just smells funny...
--
CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
for the sole use of the intended recipient(s) and contains information that
is confidential and proprietary to Applied Micro Circuits Corporation or
its subsidiaries. It is to be used solely for the purpose of furthering the
parties' business relationship. All unauthorized review, use, disclosure or
distribution is prohibited. If you are not the intended recipient, please
contact the sender by reply e-mail and destroy all copies of the original
message.
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-05-05 0:36 ` Khuong Dinh
(?)
@ 2017-05-05 16:51 ` Lorenzo Pieralisi
-1 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-05-05 16:51 UTC (permalink / raw)
To: Khuong Dinh
Cc: Marc Zyngier, msalter, Bjorn Helgaas, linux-pci, jcm, patches,
linux-kernel, linux-arm-kernel, rjw, Duc Dang
On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
> Hi Marc,
> There's no explicit dependency between pcie driver and msi
> controller. The current solution that we did is relying on the
> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
> higher priority in start ordering. This method could be applied in
> case of msi and pcie are the same level of subsys_init (in ACPI
> boot). However, PCIe driver has not supported for this dependency
> check yet. How do you think about this solution.
First off, you can't post emails with confidentiality notices on
mailing lists so remove it from now onwards.
Secondly, I commented on this before, so you know what my opinion is.
Finally, please execute this command on the vmlinux that "works"
for you:
nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
Even by ordering devices in the ACPI tables (that I abhor) I still do
not understand how this works (I mean without relying on kernel link
order to ensure that the MSI driver is probed before PCI devices are
enumerated in acpi_init()).
Thanks,
Lorenzo
> Best regards,
> Khuong
>
> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> > On 28/04/17 01:54, Khuong Dinh wrote:
> >> From: Khuong Dinh <kdinh@apm.com>
> >>
> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >>
> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >> ---
> >> v2:
> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >> v1:
> >> - Initial version
> >> ---
> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >> index f1b633b..00aaa3d 100644
> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >> @@ -24,6 +24,7 @@
> >> #include <linux/pci.h>
> >> #include <linux/platform_device.h>
> >> #include <linux/of_pci.h>
> >> +#include <linux/acpi.h>
> >>
> >> #define MSI_IR0 0x000000
> >> #define MSI_INT0 0x800000
> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >> };
> >>
> >> struct xgene_msi {
> >> - struct device_node *node;
> >> + struct fwnode_handle *fwnode;
> >> struct irq_domain *inner_domain;
> >> struct irq_domain *msi_domain;
> >> u64 msi_addr;
> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >> .free = xgene_irq_domain_free,
> >> };
> >>
> >> +#ifdef CONFIG_ACPI
> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >> +{
> >> + return xgene_msi_ctrl.fwnode;
> >> +}
> >> +#endif
> >> +
> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >> {
> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> if (!msi->inner_domain)
> >> return -ENOMEM;
> >>
> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >> &xgene_msi_domain_info,
> >> msi->inner_domain);
> >>
> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> return -ENOMEM;
> >> }
> >>
> >> +#ifdef CONFIG_ACPI
> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >> +#endif
> >> return 0;
> >> }
> >>
> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >> {},
> >> };
> >>
> >> +#ifdef CONFIG_ACPI
> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >> + {"APMC0D0E", 0},
> >> + { },
> >> +};
> >> +#endif
> >> +
> >> static int xgene_msi_probe(struct platform_device *pdev)
> >> {
> >> struct resource *res;
> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> goto error;
> >> }
> >> xgene_msi->msi_addr = res->start;
> >> - xgene_msi->node = pdev->dev.of_node;
> >> +
> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >> + if (!xgene_msi->fwnode) {
> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >
> > Please provide something other than NULL, such as the base address if
> > the device. That's quite useful for debugging.
> >
> >> + if (!xgene_msi->fwnode) {
> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >> + rc = ENOMEM;
> >> + goto error;
> >> + }
> >> + }
> >> +
> >> xgene_msi->num_cpus = num_possible_cpus();
> >>
> >> rc = xgene_msi_init_allocator(xgene_msi);
> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> .driver = {
> >> .name = "xgene-msi",
> >> .of_match_table = xgene_msi_match_table,
> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >> },
> >> .probe = xgene_msi_probe,
> >> .remove = xgene_msi_remove,
> >>
> >
> > The code is trivial, but relies on the MSI controller to probe before
> > the PCI bus. What enforces this? How is it making sure that this is not
> > going to break in the next kernel release? As far as I can tell, there
> > is no explicit dependency between the two, making it the whole thing
> > extremely fragile.
> >
> > Thanks,
> >
> > M.
> > --
> > Jazz is not dead. It just smells funny...
>
> --
> CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
> for the sole use of the intended recipient(s) and contains information that
> is confidential and proprietary to Applied Micro Circuits Corporation or
> its subsidiaries. It is to be used solely for the purpose of furthering the
> parties' business relationship. All unauthorized review, use, disclosure or
> distribution is prohibited. If you are not the intended recipient, please
> contact the sender by reply e-mail and destroy all copies of the original
> message.
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-05 16:51 ` Lorenzo Pieralisi
0 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-05-05 16:51 UTC (permalink / raw)
To: Khuong Dinh
Cc: rjw, Duc Dang, Marc Zyngier, linux-pci, msalter, patches,
linux-kernel, jcm, Bjorn Helgaas, linux-arm-kernel
On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
> Hi Marc,
> There's no explicit dependency between pcie driver and msi
> controller. The current solution that we did is relying on the
> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
> higher priority in start ordering. This method could be applied in
> case of msi and pcie are the same level of subsys_init (in ACPI
> boot). However, PCIe driver has not supported for this dependency
> check yet. How do you think about this solution.
First off, you can't post emails with confidentiality notices on
mailing lists so remove it from now onwards.
Secondly, I commented on this before, so you know what my opinion is.
Finally, please execute this command on the vmlinux that "works"
for you:
nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
Even by ordering devices in the ACPI tables (that I abhor) I still do
not understand how this works (I mean without relying on kernel link
order to ensure that the MSI driver is probed before PCI devices are
enumerated in acpi_init()).
Thanks,
Lorenzo
> Best regards,
> Khuong
>
> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> > On 28/04/17 01:54, Khuong Dinh wrote:
> >> From: Khuong Dinh <kdinh@apm.com>
> >>
> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >>
> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >> ---
> >> v2:
> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >> v1:
> >> - Initial version
> >> ---
> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >> index f1b633b..00aaa3d 100644
> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >> @@ -24,6 +24,7 @@
> >> #include <linux/pci.h>
> >> #include <linux/platform_device.h>
> >> #include <linux/of_pci.h>
> >> +#include <linux/acpi.h>
> >>
> >> #define MSI_IR0 0x000000
> >> #define MSI_INT0 0x800000
> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >> };
> >>
> >> struct xgene_msi {
> >> - struct device_node *node;
> >> + struct fwnode_handle *fwnode;
> >> struct irq_domain *inner_domain;
> >> struct irq_domain *msi_domain;
> >> u64 msi_addr;
> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >> .free = xgene_irq_domain_free,
> >> };
> >>
> >> +#ifdef CONFIG_ACPI
> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >> +{
> >> + return xgene_msi_ctrl.fwnode;
> >> +}
> >> +#endif
> >> +
> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >> {
> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> if (!msi->inner_domain)
> >> return -ENOMEM;
> >>
> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >> &xgene_msi_domain_info,
> >> msi->inner_domain);
> >>
> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> return -ENOMEM;
> >> }
> >>
> >> +#ifdef CONFIG_ACPI
> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >> +#endif
> >> return 0;
> >> }
> >>
> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >> {},
> >> };
> >>
> >> +#ifdef CONFIG_ACPI
> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >> + {"APMC0D0E", 0},
> >> + { },
> >> +};
> >> +#endif
> >> +
> >> static int xgene_msi_probe(struct platform_device *pdev)
> >> {
> >> struct resource *res;
> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> goto error;
> >> }
> >> xgene_msi->msi_addr = res->start;
> >> - xgene_msi->node = pdev->dev.of_node;
> >> +
> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >> + if (!xgene_msi->fwnode) {
> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >
> > Please provide something other than NULL, such as the base address if
> > the device. That's quite useful for debugging.
> >
> >> + if (!xgene_msi->fwnode) {
> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >> + rc = ENOMEM;
> >> + goto error;
> >> + }
> >> + }
> >> +
> >> xgene_msi->num_cpus = num_possible_cpus();
> >>
> >> rc = xgene_msi_init_allocator(xgene_msi);
> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> .driver = {
> >> .name = "xgene-msi",
> >> .of_match_table = xgene_msi_match_table,
> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >> },
> >> .probe = xgene_msi_probe,
> >> .remove = xgene_msi_remove,
> >>
> >
> > The code is trivial, but relies on the MSI controller to probe before
> > the PCI bus. What enforces this? How is it making sure that this is not
> > going to break in the next kernel release? As far as I can tell, there
> > is no explicit dependency between the two, making it the whole thing
> > extremely fragile.
> >
> > Thanks,
> >
> > M.
> > --
> > Jazz is not dead. It just smells funny...
>
> --
> CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
> for the sole use of the intended recipient(s) and contains information that
> is confidential and proprietary to Applied Micro Circuits Corporation or
> its subsidiaries. It is to be used solely for the purpose of furthering the
> parties' business relationship. All unauthorized review, use, disclosure or
> distribution is prohibited. If you are not the intended recipient, please
> contact the sender by reply e-mail and destroy all copies of the original
> message.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-05 16:51 ` Lorenzo Pieralisi
0 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-05-05 16:51 UTC (permalink / raw)
To: linux-arm-kernel
On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
> Hi Marc,
> There's no explicit dependency between pcie driver and msi
> controller. The current solution that we did is relying on the
> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
> higher priority in start ordering. This method could be applied in
> case of msi and pcie are the same level of subsys_init (in ACPI
> boot). However, PCIe driver has not supported for this dependency
> check yet. How do you think about this solution.
First off, you can't post emails with confidentiality notices on
mailing lists so remove it from now onwards.
Secondly, I commented on this before, so you know what my opinion is.
Finally, please execute this command on the vmlinux that "works"
for you:
nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
Even by ordering devices in the ACPI tables (that I abhor) I still do
not understand how this works (I mean without relying on kernel link
order to ensure that the MSI driver is probed before PCI devices are
enumerated in acpi_init()).
Thanks,
Lorenzo
> Best regards,
> Khuong
>
> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> > On 28/04/17 01:54, Khuong Dinh wrote:
> >> From: Khuong Dinh <kdinh@apm.com>
> >>
> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >>
> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >> ---
> >> v2:
> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >> v1:
> >> - Initial version
> >> ---
> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >> index f1b633b..00aaa3d 100644
> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >> @@ -24,6 +24,7 @@
> >> #include <linux/pci.h>
> >> #include <linux/platform_device.h>
> >> #include <linux/of_pci.h>
> >> +#include <linux/acpi.h>
> >>
> >> #define MSI_IR0 0x000000
> >> #define MSI_INT0 0x800000
> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >> };
> >>
> >> struct xgene_msi {
> >> - struct device_node *node;
> >> + struct fwnode_handle *fwnode;
> >> struct irq_domain *inner_domain;
> >> struct irq_domain *msi_domain;
> >> u64 msi_addr;
> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >> .free = xgene_irq_domain_free,
> >> };
> >>
> >> +#ifdef CONFIG_ACPI
> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >> +{
> >> + return xgene_msi_ctrl.fwnode;
> >> +}
> >> +#endif
> >> +
> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >> {
> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> if (!msi->inner_domain)
> >> return -ENOMEM;
> >>
> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >> &xgene_msi_domain_info,
> >> msi->inner_domain);
> >>
> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> return -ENOMEM;
> >> }
> >>
> >> +#ifdef CONFIG_ACPI
> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >> +#endif
> >> return 0;
> >> }
> >>
> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >> {},
> >> };
> >>
> >> +#ifdef CONFIG_ACPI
> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >> + {"APMC0D0E", 0},
> >> + { },
> >> +};
> >> +#endif
> >> +
> >> static int xgene_msi_probe(struct platform_device *pdev)
> >> {
> >> struct resource *res;
> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> goto error;
> >> }
> >> xgene_msi->msi_addr = res->start;
> >> - xgene_msi->node = pdev->dev.of_node;
> >> +
> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >> + if (!xgene_msi->fwnode) {
> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >
> > Please provide something other than NULL, such as the base address if
> > the device. That's quite useful for debugging.
> >
> >> + if (!xgene_msi->fwnode) {
> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >> + rc = ENOMEM;
> >> + goto error;
> >> + }
> >> + }
> >> +
> >> xgene_msi->num_cpus = num_possible_cpus();
> >>
> >> rc = xgene_msi_init_allocator(xgene_msi);
> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> .driver = {
> >> .name = "xgene-msi",
> >> .of_match_table = xgene_msi_match_table,
> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >> },
> >> .probe = xgene_msi_probe,
> >> .remove = xgene_msi_remove,
> >>
> >
> > The code is trivial, but relies on the MSI controller to probe before
> > the PCI bus. What enforces this? How is it making sure that this is not
> > going to break in the next kernel release? As far as I can tell, there
> > is no explicit dependency between the two, making it the whole thing
> > extremely fragile.
> >
> > Thanks,
> >
> > M.
> > --
> > Jazz is not dead. It just smells funny...
>
> --
> CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
> for the sole use of the intended recipient(s) and contains information that
> is confidential and proprietary to Applied Micro Circuits Corporation or
> its subsidiaries. It is to be used solely for the purpose of furthering the
> parties' business relationship. All unauthorized review, use, disclosure or
> distribution is prohibited. If you are not the intended recipient, please
> contact the sender by reply e-mail and destroy all copies of the original
> message.
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-05-05 16:51 ` Lorenzo Pieralisi
(?)
@ 2017-05-09 22:55 ` Khuong Dinh
-1 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-05-09 22:55 UTC (permalink / raw)
To: Lorenzo Pieralisi
Cc: Marc Zyngier, msalter, Bjorn Helgaas, linux-pci, jcm, patches,
linux-kernel, linux-arm-kernel, rjw, Duc Dang
Hi Lorenzo,
On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
<lorenzo.pieralisi@arm.com> wrote:
> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>> Hi Marc,
>> There's no explicit dependency between pcie driver and msi
>> controller. The current solution that we did is relying on the
>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>> higher priority in start ordering. This method could be applied in
>> case of msi and pcie are the same level of subsys_init (in ACPI
>> boot). However, PCIe driver has not supported for this dependency
>> check yet. How do you think about this solution.
>
> First off, you can't post emails with confidentiality notices on
> mailing lists so remove it from now onwards.
Fixed
> Secondly, I commented on this before, so you know what my opinion is.
I got your opinion. I'll implement as your suggestion.
> Finally, please execute this command on the vmlinux that "works"
> for you:
>
> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
$ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
ffff000008dab2d8 t __initcall_acpi_init4
ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
Best regards,
Khuong Dinh
> Even by ordering devices in the ACPI tables (that I abhor) I still do
> not understand how this works (I mean without relying on kernel link
> order to ensure that the MSI driver is probed before PCI devices are
> enumerated in acpi_init()).
>
> Thanks,
> Lorenzo
>
>> Best regards,
>> Khuong
>>
>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>> > On 28/04/17 01:54, Khuong Dinh wrote:
>> >> From: Khuong Dinh <kdinh@apm.com>
>> >>
>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>> >>
>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> >> ---
>> >> v2:
>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>> >> v1:
>> >> - Initial version
>> >> ---
>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>> >>
>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> >> index f1b633b..00aaa3d 100644
>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>> >> @@ -24,6 +24,7 @@
>> >> #include <linux/pci.h>
>> >> #include <linux/platform_device.h>
>> >> #include <linux/of_pci.h>
>> >> +#include <linux/acpi.h>
>> >>
>> >> #define MSI_IR0 0x000000
>> >> #define MSI_INT0 0x800000
>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> >> };
>> >>
>> >> struct xgene_msi {
>> >> - struct device_node *node;
>> >> + struct fwnode_handle *fwnode;
>> >> struct irq_domain *inner_domain;
>> >> struct irq_domain *msi_domain;
>> >> u64 msi_addr;
>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> >> .free = xgene_irq_domain_free,
>> >> };
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> >> +{
>> >> + return xgene_msi_ctrl.fwnode;
>> >> +}
>> >> +#endif
>> >> +
>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> {
>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> if (!msi->inner_domain)
>> >> return -ENOMEM;
>> >>
>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> >> &xgene_msi_domain_info,
>> >> msi->inner_domain);
>> >>
>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> return -ENOMEM;
>> >> }
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> >> +#endif
>> >> return 0;
>> >> }
>> >>
>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> >> {},
>> >> };
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> >> + {"APMC0D0E", 0},
>> >> + { },
>> >> +};
>> >> +#endif
>> >> +
>> >> static int xgene_msi_probe(struct platform_device *pdev)
>> >> {
>> >> struct resource *res;
>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> goto error;
>> >> }
>> >> xgene_msi->msi_addr = res->start;
>> >> - xgene_msi->node = pdev->dev.of_node;
>> >> +
>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> >> + if (!xgene_msi->fwnode) {
>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>> >
>> > Please provide something other than NULL, such as the base address if
>> > the device. That's quite useful for debugging.
>> >
>> >> + if (!xgene_msi->fwnode) {
>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> >> + rc = ENOMEM;
>> >> + goto error;
>> >> + }
>> >> + }
>> >> +
>> >> xgene_msi->num_cpus = num_possible_cpus();
>> >>
>> >> rc = xgene_msi_init_allocator(xgene_msi);
>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> .driver = {
>> >> .name = "xgene-msi",
>> >> .of_match_table = xgene_msi_match_table,
>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> >> },
>> >> .probe = xgene_msi_probe,
>> >> .remove = xgene_msi_remove,
>> >>
>> >
>> > The code is trivial, but relies on the MSI controller to probe before
>> > the PCI bus. What enforces this? How is it making sure that this is not
>> > going to break in the next kernel release? As far as I can tell, there
>> > is no explicit dependency between the two, making it the whole thing
>> > extremely fragile.
>> >
>> > Thanks,
>> >
>> > M.
>> > --
>> > Jazz is not dead. It just smells funny...
>>
>> --
>> CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
>> for the sole use of the intended recipient(s) and contains information that
>> is confidential and proprietary to Applied Micro Circuits Corporation or
>> its subsidiaries. It is to be used solely for the purpose of furthering the
>> parties' business relationship. All unauthorized review, use, disclosure or
>> distribution is prohibited. If you are not the intended recipient, please
>> contact the sender by reply e-mail and destroy all copies of the original
>> message.
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-09 22:55 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-05-09 22:55 UTC (permalink / raw)
To: Lorenzo Pieralisi
Cc: rjw, Duc Dang, Marc Zyngier, linux-pci, msalter, patches,
linux-kernel, jcm, Bjorn Helgaas, linux-arm-kernel
Hi Lorenzo,
On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
<lorenzo.pieralisi@arm.com> wrote:
> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>> Hi Marc,
>> There's no explicit dependency between pcie driver and msi
>> controller. The current solution that we did is relying on the
>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>> higher priority in start ordering. This method could be applied in
>> case of msi and pcie are the same level of subsys_init (in ACPI
>> boot). However, PCIe driver has not supported for this dependency
>> check yet. How do you think about this solution.
>
> First off, you can't post emails with confidentiality notices on
> mailing lists so remove it from now onwards.
Fixed
> Secondly, I commented on this before, so you know what my opinion is.
I got your opinion. I'll implement as your suggestion.
> Finally, please execute this command on the vmlinux that "works"
> for you:
>
> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
$ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
ffff000008dab2d8 t __initcall_acpi_init4
ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
Best regards,
Khuong Dinh
> Even by ordering devices in the ACPI tables (that I abhor) I still do
> not understand how this works (I mean without relying on kernel link
> order to ensure that the MSI driver is probed before PCI devices are
> enumerated in acpi_init()).
>
> Thanks,
> Lorenzo
>
>> Best regards,
>> Khuong
>>
>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>> > On 28/04/17 01:54, Khuong Dinh wrote:
>> >> From: Khuong Dinh <kdinh@apm.com>
>> >>
>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>> >>
>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> >> ---
>> >> v2:
>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>> >> v1:
>> >> - Initial version
>> >> ---
>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>> >>
>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> >> index f1b633b..00aaa3d 100644
>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>> >> @@ -24,6 +24,7 @@
>> >> #include <linux/pci.h>
>> >> #include <linux/platform_device.h>
>> >> #include <linux/of_pci.h>
>> >> +#include <linux/acpi.h>
>> >>
>> >> #define MSI_IR0 0x000000
>> >> #define MSI_INT0 0x800000
>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> >> };
>> >>
>> >> struct xgene_msi {
>> >> - struct device_node *node;
>> >> + struct fwnode_handle *fwnode;
>> >> struct irq_domain *inner_domain;
>> >> struct irq_domain *msi_domain;
>> >> u64 msi_addr;
>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> >> .free = xgene_irq_domain_free,
>> >> };
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> >> +{
>> >> + return xgene_msi_ctrl.fwnode;
>> >> +}
>> >> +#endif
>> >> +
>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> {
>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> if (!msi->inner_domain)
>> >> return -ENOMEM;
>> >>
>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> >> &xgene_msi_domain_info,
>> >> msi->inner_domain);
>> >>
>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> return -ENOMEM;
>> >> }
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> >> +#endif
>> >> return 0;
>> >> }
>> >>
>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> >> {},
>> >> };
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> >> + {"APMC0D0E", 0},
>> >> + { },
>> >> +};
>> >> +#endif
>> >> +
>> >> static int xgene_msi_probe(struct platform_device *pdev)
>> >> {
>> >> struct resource *res;
>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> goto error;
>> >> }
>> >> xgene_msi->msi_addr = res->start;
>> >> - xgene_msi->node = pdev->dev.of_node;
>> >> +
>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> >> + if (!xgene_msi->fwnode) {
>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>> >
>> > Please provide something other than NULL, such as the base address if
>> > the device. That's quite useful for debugging.
>> >
>> >> + if (!xgene_msi->fwnode) {
>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> >> + rc = ENOMEM;
>> >> + goto error;
>> >> + }
>> >> + }
>> >> +
>> >> xgene_msi->num_cpus = num_possible_cpus();
>> >>
>> >> rc = xgene_msi_init_allocator(xgene_msi);
>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> .driver = {
>> >> .name = "xgene-msi",
>> >> .of_match_table = xgene_msi_match_table,
>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> >> },
>> >> .probe = xgene_msi_probe,
>> >> .remove = xgene_msi_remove,
>> >>
>> >
>> > The code is trivial, but relies on the MSI controller to probe before
>> > the PCI bus. What enforces this? How is it making sure that this is not
>> > going to break in the next kernel release? As far as I can tell, there
>> > is no explicit dependency between the two, making it the whole thing
>> > extremely fragile.
>> >
>> > Thanks,
>> >
>> > M.
>> > --
>> > Jazz is not dead. It just smells funny...
>>
>> --
>> CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
>> for the sole use of the intended recipient(s) and contains information that
>> is confidential and proprietary to Applied Micro Circuits Corporation or
>> its subsidiaries. It is to be used solely for the purpose of furthering the
>> parties' business relationship. All unauthorized review, use, disclosure or
>> distribution is prohibited. If you are not the intended recipient, please
>> contact the sender by reply e-mail and destroy all copies of the original
>> message.
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-09 22:55 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-05-09 22:55 UTC (permalink / raw)
To: linux-arm-kernel
Hi Lorenzo,
On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
<lorenzo.pieralisi@arm.com> wrote:
> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>> Hi Marc,
>> There's no explicit dependency between pcie driver and msi
>> controller. The current solution that we did is relying on the
>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>> higher priority in start ordering. This method could be applied in
>> case of msi and pcie are the same level of subsys_init (in ACPI
>> boot). However, PCIe driver has not supported for this dependency
>> check yet. How do you think about this solution.
>
> First off, you can't post emails with confidentiality notices on
> mailing lists so remove it from now onwards.
Fixed
> Secondly, I commented on this before, so you know what my opinion is.
I got your opinion. I'll implement as your suggestion.
> Finally, please execute this command on the vmlinux that "works"
> for you:
>
> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
$ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
ffff000008dab2d8 t __initcall_acpi_init4
ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
Best regards,
Khuong Dinh
> Even by ordering devices in the ACPI tables (that I abhor) I still do
> not understand how this works (I mean without relying on kernel link
> order to ensure that the MSI driver is probed before PCI devices are
> enumerated in acpi_init()).
>
> Thanks,
> Lorenzo
>
>> Best regards,
>> Khuong
>>
>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>> > On 28/04/17 01:54, Khuong Dinh wrote:
>> >> From: Khuong Dinh <kdinh@apm.com>
>> >>
>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>> >>
>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> >> ---
>> >> v2:
>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>> >> v1:
>> >> - Initial version
>> >> ---
>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>> >>
>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> >> index f1b633b..00aaa3d 100644
>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>> >> @@ -24,6 +24,7 @@
>> >> #include <linux/pci.h>
>> >> #include <linux/platform_device.h>
>> >> #include <linux/of_pci.h>
>> >> +#include <linux/acpi.h>
>> >>
>> >> #define MSI_IR0 0x000000
>> >> #define MSI_INT0 0x800000
>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> >> };
>> >>
>> >> struct xgene_msi {
>> >> - struct device_node *node;
>> >> + struct fwnode_handle *fwnode;
>> >> struct irq_domain *inner_domain;
>> >> struct irq_domain *msi_domain;
>> >> u64 msi_addr;
>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> >> .free = xgene_irq_domain_free,
>> >> };
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> >> +{
>> >> + return xgene_msi_ctrl.fwnode;
>> >> +}
>> >> +#endif
>> >> +
>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> {
>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> if (!msi->inner_domain)
>> >> return -ENOMEM;
>> >>
>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> >> &xgene_msi_domain_info,
>> >> msi->inner_domain);
>> >>
>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> return -ENOMEM;
>> >> }
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> >> +#endif
>> >> return 0;
>> >> }
>> >>
>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> >> {},
>> >> };
>> >>
>> >> +#ifdef CONFIG_ACPI
>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> >> + {"APMC0D0E", 0},
>> >> + { },
>> >> +};
>> >> +#endif
>> >> +
>> >> static int xgene_msi_probe(struct platform_device *pdev)
>> >> {
>> >> struct resource *res;
>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> goto error;
>> >> }
>> >> xgene_msi->msi_addr = res->start;
>> >> - xgene_msi->node = pdev->dev.of_node;
>> >> +
>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> >> + if (!xgene_msi->fwnode) {
>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>> >
>> > Please provide something other than NULL, such as the base address if
>> > the device. That's quite useful for debugging.
>> >
>> >> + if (!xgene_msi->fwnode) {
>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> >> + rc = ENOMEM;
>> >> + goto error;
>> >> + }
>> >> + }
>> >> +
>> >> xgene_msi->num_cpus = num_possible_cpus();
>> >>
>> >> rc = xgene_msi_init_allocator(xgene_msi);
>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> .driver = {
>> >> .name = "xgene-msi",
>> >> .of_match_table = xgene_msi_match_table,
>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> >> },
>> >> .probe = xgene_msi_probe,
>> >> .remove = xgene_msi_remove,
>> >>
>> >
>> > The code is trivial, but relies on the MSI controller to probe before
>> > the PCI bus. What enforces this? How is it making sure that this is not
>> > going to break in the next kernel release? As far as I can tell, there
>> > is no explicit dependency between the two, making it the whole thing
>> > extremely fragile.
>> >
>> > Thanks,
>> >
>> > M.
>> > --
>> > Jazz is not dead. It just smells funny...
>>
>> --
>> CONFIDENTIALITY NOTICE: This e-mail message, including any attachments, is
>> for the sole use of the intended recipient(s) and contains information that
>> is confidential and proprietary to Applied Micro Circuits Corporation or
>> its subsidiaries. It is to be used solely for the purpose of furthering the
>> parties' business relationship. All unauthorized review, use, disclosure or
>> distribution is prohibited. If you are not the intended recipient, please
>> contact the sender by reply e-mail and destroy all copies of the original
>> message.
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-04-28 0:54 ` Khuong Dinh
@ 2017-05-31 11:29 ` Jonathan Toppins
-1 siblings, 0 replies; 47+ messages in thread
From: Jonathan Toppins @ 2017-05-31 11:29 UTC (permalink / raw)
To: Khuong Dinh, lorenzo.pieralisi, marc.zyngier, msalter, bhelgaas,
linux-pci, jcm
Cc: patches, Duc Dang, rjw, linux-kernel, linux-arm-kernel
On 04/27/2017 08:54 PM, Khuong Dinh wrote:
> From: Khuong Dinh <kdinh@apm.com>
>
> This patch makes pci-xgene-msi driver ACPI-aware and provides
> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>
> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> ---
> v2:
> - Verify with BIOS version 3.06.25 and 3.07.09
> v1:
> - Initial version
Hoping I just missed it, but was a follow-up (v3) ever provided for
this? It seems this thread ended with the expectation that a v3 would be
provided to address the sequencing issue.
Regards,
-Jon
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-31 11:29 ` Jonathan Toppins
0 siblings, 0 replies; 47+ messages in thread
From: Jonathan Toppins @ 2017-05-31 11:29 UTC (permalink / raw)
To: linux-arm-kernel
On 04/27/2017 08:54 PM, Khuong Dinh wrote:
> From: Khuong Dinh <kdinh@apm.com>
>
> This patch makes pci-xgene-msi driver ACPI-aware and provides
> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>
> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> Signed-off-by: Duc Dang <dhdang@apm.com>
> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> ---
> v2:
> - Verify with BIOS version 3.06.25 and 3.07.09
> v1:
> - Initial version
Hoping I just missed it, but was a follow-up (v3) ever provided for
this? It seems this thread ended with the expectation that a v3 would be
provided to address the sequencing issue.
Regards,
-Jon
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-05-31 11:29 ` Jonathan Toppins
(?)
@ 2017-05-31 16:13 ` Loc Ho
-1 siblings, 0 replies; 47+ messages in thread
From: Loc Ho @ 2017-05-31 16:13 UTC (permalink / raw)
To: jtoppins
Cc: Khuong Dinh, Lorenzo Pieralisi, Marc Zyngier, Mark Salter,
Bjorn Helgaas, linux-pci, Jon Masters, patches, Duc Dang,
Rafael J. Wysocki, Linux Kernel Mailing List, linux-arm-kernel
Hi Jon,
> > From: Khuong Dinh <kdinh@apm.com>
> >
> > This patch makes pci-xgene-msi driver ACPI-aware and provides
> > MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >
> > Signed-off-by: Khuong Dinh <kdinh@apm.com>
> > Signed-off-by: Duc Dang <dhdang@apm.com>
> > Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> > ---
> > v2:
> > - Verify with BIOS version 3.06.25 and 3.07.09
> > v1:
> > - Initial version
>
>
> Hoping I just missed it, but was a follow-up (v3) ever provided for
> this? It seems this thread ended with the expectation that a v3 would be
> provided to address the sequencing issue.
No... You didn't missed anything. Khuong is out last couple days. He
is currently implementing the dependence requirement using the _DED
method. As soon as he get this working correctly, he will post the
next version. Though, there isn't anything wrong with the last version
besides future potential change in the kernel which is very unlikely
to change for the next one or more release of the future kernel.
-Loc
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-31 16:13 ` Loc Ho
0 siblings, 0 replies; 47+ messages in thread
From: Loc Ho @ 2017-05-31 16:13 UTC (permalink / raw)
To: jtoppins
Cc: Rafael J. Wysocki, Lorenzo Pieralisi, Marc Zyngier, linux-pci,
Duc Dang, patches, Linux Kernel Mailing List, Khuong Dinh,
linux-arm-kernel, Mark Salter, Bjorn Helgaas, Jon Masters
Hi Jon,
> > From: Khuong Dinh <kdinh@apm.com>
> >
> > This patch makes pci-xgene-msi driver ACPI-aware and provides
> > MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >
> > Signed-off-by: Khuong Dinh <kdinh@apm.com>
> > Signed-off-by: Duc Dang <dhdang@apm.com>
> > Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> > ---
> > v2:
> > - Verify with BIOS version 3.06.25 and 3.07.09
> > v1:
> > - Initial version
>
>
> Hoping I just missed it, but was a follow-up (v3) ever provided for
> this? It seems this thread ended with the expectation that a v3 would be
> provided to address the sequencing issue.
No... You didn't missed anything. Khuong is out last couple days. He
is currently implementing the dependence requirement using the _DED
method. As soon as he get this working correctly, he will post the
next version. Though, there isn't anything wrong with the last version
besides future potential change in the kernel which is very unlikely
to change for the next one or more release of the future kernel.
-Loc
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-05-31 16:13 ` Loc Ho
0 siblings, 0 replies; 47+ messages in thread
From: Loc Ho @ 2017-05-31 16:13 UTC (permalink / raw)
To: linux-arm-kernel
Hi Jon,
> > From: Khuong Dinh <kdinh@apm.com>
> >
> > This patch makes pci-xgene-msi driver ACPI-aware and provides
> > MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >
> > Signed-off-by: Khuong Dinh <kdinh@apm.com>
> > Signed-off-by: Duc Dang <dhdang@apm.com>
> > Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> > ---
> > v2:
> > - Verify with BIOS version 3.06.25 and 3.07.09
> > v1:
> > - Initial version
>
>
> Hoping I just missed it, but was a follow-up (v3) ever provided for
> this? It seems this thread ended with the expectation that a v3 would be
> provided to address the sequencing issue.
No... You didn't missed anything. Khuong is out last couple days. He
is currently implementing the dependence requirement using the _DED
method. As soon as he get this working correctly, he will post the
next version. Though, there isn't anything wrong with the last version
besides future potential change in the kernel which is very unlikely
to change for the next one or more release of the future kernel.
-Loc
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-05-09 22:55 ` Khuong Dinh
(?)
@ 2017-06-06 16:44 ` Khuong Dinh
-1 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-06 16:44 UTC (permalink / raw)
To: Lorenzo Pieralisi
Cc: Marc Zyngier, msalter, Bjorn Helgaas, linux-pci, jcm, patches,
linux-kernel, linux-arm-kernel, rjw, Duc Dang
Hi Lorenzo,
On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
> Hi Lorenzo,
>
> On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
> <lorenzo.pieralisi@arm.com> wrote:
>> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>>> Hi Marc,
>>> There's no explicit dependency between pcie driver and msi
>>> controller. The current solution that we did is relying on the
>>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>>> higher priority in start ordering. This method could be applied in
>>> case of msi and pcie are the same level of subsys_init (in ACPI
>>> boot). However, PCIe driver has not supported for this dependency
>>> check yet. How do you think about this solution.
>>
>> First off, you can't post emails with confidentiality notices on
>> mailing lists so remove it from now onwards.
>
> Fixed
>
>> Secondly, I commented on this before, so you know what my opinion is.
>
> I got your opinion. I'll implement as your suggestion.
>
Regarding to your previous suggestion to add a hook walking the ACPI
namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
controllers must be probed before PCI.
I have a concern that the MSI namespace “ _SB.MSIX” is not a unique
name and how can we walk the ACPI namespace and use this name to make
MSI probed before PCI.
May you have little bit more information for this or do you have
another suggestion for this?
There’s another solution which makes this simpler and I’d like to
ask your opinion about this.
The solution is to make an hierarchy between MSI and PCI nodes as below:
Device(\_SB.MSIX) {
Device(\_SB.PCI0)
Device(\_SB.PCI1)
……
}
In other word, MSIX node is a parent node of PCI nodes in ACPI table.
In this sense, there’s an explicit dependency between MSI and PCI,
MSI controller must be probed before PCI and it will guarantee not
breaking next kernel releases.
How do you think about this solution.
I also tried to use _DEP method to describe the dependency of PCIe
nodes, but it looks that it does not work as PCI and MSI are handed by
acpi_bus_scan and still having issue when we re-probe PCI.
Thanks,
Khuong Dinh
>> Finally, please execute this command on the vmlinux that "works"
>> for you:
>>
>> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>
> $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> ffff000008dab2d8 t __initcall_acpi_init4
> ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
>
> Best regards,
> Khuong Dinh
>
>> Even by ordering devices in the ACPI tables (that I abhor) I still do
>> not understand how this works (I mean without relying on kernel link
>> order to ensure that the MSI driver is probed before PCI devices are
>> enumerated in acpi_init()).
>>
>> Thanks,
>> Lorenzo
>>
>>> Best regards,
>>> Khuong
>>>
>>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>>> > On 28/04/17 01:54, Khuong Dinh wrote:
>>> >> From: Khuong Dinh <kdinh@apm.com>
>>> >>
>>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>> >>
>>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>>> >> ---
>>> >> v2:
>>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>>> >> v1:
>>> >> - Initial version
>>> >> ---
>>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>>> >>
>>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>>> >> index f1b633b..00aaa3d 100644
>>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>>> >> @@ -24,6 +24,7 @@
>>> >> #include <linux/pci.h>
>>> >> #include <linux/platform_device.h>
>>> >> #include <linux/of_pci.h>
>>> >> +#include <linux/acpi.h>
>>> >>
>>> >> #define MSI_IR0 0x000000
>>> >> #define MSI_INT0 0x800000
>>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>>> >> };
>>> >>
>>> >> struct xgene_msi {
>>> >> - struct device_node *node;
>>> >> + struct fwnode_handle *fwnode;
>>> >> struct irq_domain *inner_domain;
>>> >> struct irq_domain *msi_domain;
>>> >> u64 msi_addr;
>>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>>> >> .free = xgene_irq_domain_free,
>>> >> };
>>> >>
>>> >> +#ifdef CONFIG_ACPI
>>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>>> >> +{
>>> >> + return xgene_msi_ctrl.fwnode;
>>> >> +}
>>> >> +#endif
>>> >> +
>>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>>> >> {
>>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>>> >> if (!msi->inner_domain)
>>> >> return -ENOMEM;
>>> >>
>>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>>> >> &xgene_msi_domain_info,
>>> >> msi->inner_domain);
>>> >>
>>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>>> >> return -ENOMEM;
>>> >> }
>>> >>
>>> >> +#ifdef CONFIG_ACPI
>>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>>> >> +#endif
>>> >> return 0;
>>> >> }
>>> >>
>>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>>> >> {},
>>> >> };
>>> >>
>>> >> +#ifdef CONFIG_ACPI
>>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>>> >> + {"APMC0D0E", 0},
>>> >> + { },
>>> >> +};
>>> >> +#endif
>>> >> +
>>> >> static int xgene_msi_probe(struct platform_device *pdev)
>>> >> {
>>> >> struct resource *res;
>>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>>> >> goto error;
>>> >> }
>>> >> xgene_msi->msi_addr = res->start;
>>> >> - xgene_msi->node = pdev->dev.of_node;
>>> >> +
>>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>>> >> + if (!xgene_msi->fwnode) {
>>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>>> >
>>> > Please provide something other than NULL, such as the base address if
>>> > the device. That's quite useful for debugging.
>>> >
>>> >> + if (!xgene_msi->fwnode) {
>>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>>> >> + rc = ENOMEM;
>>> >> + goto error;
>>> >> + }
>>> >> + }
>>> >> +
>>> >> xgene_msi->num_cpus = num_possible_cpus();
>>> >>
>>> >> rc = xgene_msi_init_allocator(xgene_msi);
>>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>>> >> .driver = {
>>> >> .name = "xgene-msi",
>>> >> .of_match_table = xgene_msi_match_table,
>>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>>> >> },
>>> >> .probe = xgene_msi_probe,
>>> >> .remove = xgene_msi_remove,
>>> >>
>>> >
>>> > The code is trivial, but relies on the MSI controller to probe before
>>> > the PCI bus. What enforces this? How is it making sure that this is not
>>> > going to break in the next kernel release? As far as I can tell, there
>>> > is no explicit dependency between the two, making it the whole thing
>>> > extremely fragile.
>>> >
>>> > Thanks,
>>> >
>>> > M.
>>> > --
>>> > Jazz is not dead. It just smells funny...
>>>
>>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-06 16:44 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-06 16:44 UTC (permalink / raw)
To: Lorenzo Pieralisi
Cc: rjw, Duc Dang, Marc Zyngier, linux-pci, msalter, patches,
linux-kernel, jcm, Bjorn Helgaas, linux-arm-kernel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^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-06 16:44 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-06 16:44 UTC (permalink / raw)
To: linux-arm-kernel
Hi Lorenzo,
On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
> Hi Lorenzo,
>
> On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
> <lorenzo.pieralisi@arm.com> wrote:
>> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>>> Hi Marc,
>>> There's no explicit dependency between pcie driver and msi
>>> controller. The current solution that we did is relying on the
>>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>>> higher priority in start ordering. This method could be applied in
>>> case of msi and pcie are the same level of subsys_init (in ACPI
>>> boot). However, PCIe driver has not supported for this dependency
>>> check yet. How do you think about this solution.
>>
>> First off, you can't post emails with confidentiality notices on
>> mailing lists so remove it from now onwards.
>
> Fixed
>
>> Secondly, I commented on this before, so you know what my opinion is.
>
> I got your opinion. I'll implement as your suggestion.
>
Regarding to your previous suggestion to add a hook walking the ACPI
namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
controllers must be probed before PCI.
I have a concern that the MSI namespace ? _SB.MSIX? is not a unique
name and how can we walk the ACPI namespace and use this name to make
MSI probed before PCI.
May you have little bit more information for this or do you have
another suggestion for this?
There?s another solution which makes this simpler and I?d like to
ask your opinion about this.
The solution is to make an hierarchy between MSI and PCI nodes as below:
Device(\_SB.MSIX) {
Device(\_SB.PCI0)
Device(\_SB.PCI1)
??
}
In other word, MSIX node is a parent node of PCI nodes in ACPI table.
In this sense, there?s an explicit dependency between MSI and PCI,
MSI controller must be probed before PCI and it will guarantee not
breaking next kernel releases.
How do you think about this solution.
I also tried to use _DEP method to describe the dependency of PCIe
nodes, but it looks that it does not work as PCI and MSI are handed by
acpi_bus_scan and still having issue when we re-probe PCI.
Thanks,
Khuong Dinh
>> Finally, please execute this command on the vmlinux that "works"
>> for you:
>>
>> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>
> $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> ffff000008dab2d8 t __initcall_acpi_init4
> ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
>
> Best regards,
> Khuong Dinh
>
>> Even by ordering devices in the ACPI tables (that I abhor) I still do
>> not understand how this works (I mean without relying on kernel link
>> order to ensure that the MSI driver is probed before PCI devices are
>> enumerated in acpi_init()).
>>
>> Thanks,
>> Lorenzo
>>
>>> Best regards,
>>> Khuong
>>>
>>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>>> > On 28/04/17 01:54, Khuong Dinh wrote:
>>> >> From: Khuong Dinh <kdinh@apm.com>
>>> >>
>>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>>> >>
>>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>>> >> ---
>>> >> v2:
>>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>>> >> v1:
>>> >> - Initial version
>>> >> ---
>>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>>> >>
>>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>>> >> index f1b633b..00aaa3d 100644
>>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>>> >> @@ -24,6 +24,7 @@
>>> >> #include <linux/pci.h>
>>> >> #include <linux/platform_device.h>
>>> >> #include <linux/of_pci.h>
>>> >> +#include <linux/acpi.h>
>>> >>
>>> >> #define MSI_IR0 0x000000
>>> >> #define MSI_INT0 0x800000
>>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>>> >> };
>>> >>
>>> >> struct xgene_msi {
>>> >> - struct device_node *node;
>>> >> + struct fwnode_handle *fwnode;
>>> >> struct irq_domain *inner_domain;
>>> >> struct irq_domain *msi_domain;
>>> >> u64 msi_addr;
>>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>>> >> .free = xgene_irq_domain_free,
>>> >> };
>>> >>
>>> >> +#ifdef CONFIG_ACPI
>>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>>> >> +{
>>> >> + return xgene_msi_ctrl.fwnode;
>>> >> +}
>>> >> +#endif
>>> >> +
>>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>>> >> {
>>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>>> >> if (!msi->inner_domain)
>>> >> return -ENOMEM;
>>> >>
>>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>>> >> &xgene_msi_domain_info,
>>> >> msi->inner_domain);
>>> >>
>>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>>> >> return -ENOMEM;
>>> >> }
>>> >>
>>> >> +#ifdef CONFIG_ACPI
>>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>>> >> +#endif
>>> >> return 0;
>>> >> }
>>> >>
>>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>>> >> {},
>>> >> };
>>> >>
>>> >> +#ifdef CONFIG_ACPI
>>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>>> >> + {"APMC0D0E", 0},
>>> >> + { },
>>> >> +};
>>> >> +#endif
>>> >> +
>>> >> static int xgene_msi_probe(struct platform_device *pdev)
>>> >> {
>>> >> struct resource *res;
>>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>>> >> goto error;
>>> >> }
>>> >> xgene_msi->msi_addr = res->start;
>>> >> - xgene_msi->node = pdev->dev.of_node;
>>> >> +
>>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>>> >> + if (!xgene_msi->fwnode) {
>>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>>> >
>>> > Please provide something other than NULL, such as the base address if
>>> > the device. That's quite useful for debugging.
>>> >
>>> >> + if (!xgene_msi->fwnode) {
>>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>>> >> + rc = ENOMEM;
>>> >> + goto error;
>>> >> + }
>>> >> + }
>>> >> +
>>> >> xgene_msi->num_cpus = num_possible_cpus();
>>> >>
>>> >> rc = xgene_msi_init_allocator(xgene_msi);
>>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>>> >> .driver = {
>>> >> .name = "xgene-msi",
>>> >> .of_match_table = xgene_msi_match_table,
>>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>>> >> },
>>> >> .probe = xgene_msi_probe,
>>> >> .remove = xgene_msi_remove,
>>> >>
>>> >
>>> > The code is trivial, but relies on the MSI controller to probe before
>>> > the PCI bus. What enforces this? How is it making sure that this is not
>>> > going to break in the next kernel release? As far as I can tell, there
>>> > is no explicit dependency between the two, making it the whole thing
>>> > extremely fragile.
>>> >
>>> > Thanks,
>>> >
>>> > M.
>>> > --
>>> > Jazz is not dead. It just smells funny...
>>>
>>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-06-06 16:44 ` Khuong Dinh
(?)
@ 2017-06-08 10:39 ` Lorenzo Pieralisi
-1 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-08 10:39 UTC (permalink / raw)
To: Khuong Dinh
Cc: Marc Zyngier, msalter, Bjorn Helgaas, linux-pci, jcm, patches,
linux-kernel, linux-arm-kernel, rjw, Duc Dang
On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote:
> Hi Lorenzo,
>
> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
> > Hi Lorenzo,
> >
> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
> > <lorenzo.pieralisi@arm.com> wrote:
> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
> >>> Hi Marc,
> >>> There's no explicit dependency between pcie driver and msi
> >>> controller. The current solution that we did is relying on the
> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
> >>> higher priority in start ordering. This method could be applied in
> >>> case of msi and pcie are the same level of subsys_init (in ACPI
> >>> boot). However, PCIe driver has not supported for this dependency
> >>> check yet. How do you think about this solution.
> >>
> >> First off, you can't post emails with confidentiality notices on
> >> mailing lists so remove it from now onwards.
> >
> > Fixed
> >
> >> Secondly, I commented on this before, so you know what my opinion is.
> >
> > I got your opinion. I'll implement as your suggestion.
> >
>
> Regarding to your previous suggestion to add a hook walking the ACPI
> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
> controllers must be probed before PCI. I have a concern that the
> MSI namespace “ _SB.MSIX” is not a unique name and how can we walk
> the ACPI namespace and use this name to make MSI probed before PCI.
> May you have little bit more information for this or do you have
> another suggestion for this?
>
> There’s another solution which makes this simpler and I’d like to
> ask your opinion about this.
> The solution is to make an hierarchy between MSI and PCI nodes as below:
> Device(\_SB.MSIX) {
> Device(\_SB.PCI0)
> Device(\_SB.PCI1)
> ……
> }
> In other word, MSIX node is a parent node of PCI nodes in ACPI
> table. In this sense, there’s an explicit dependency between MSI
> and PCI, MSI controller must be probed before PCI and it will
> guarantee not breaking next kernel releases. How do you think about
> this solution.
I think that's a plaster as ineffective as reordering nodes, in short
it is not a solution and you still rely on kernel link ordering, you
can fiddle with ACPI tables as much as you want but that does not change.
> I also tried to use _DEP method to describe the dependency of PCIe
> nodes, but it looks that it does not work as PCI and MSI are handed
> by acpi_bus_scan and still having issue when we re-probe PCI.
That's a tad vague to be frank, anyway it is pretty clear to me that we
have hit a wall. In ACPI there is no way to describe probe dependencies
like the one you have, it is as simple as that, and this MSI issue you
are facing is just an example, there are more eg:
https://www.spinics.net/lists/arm-kernel/msg585825.html
At the end of the day the choice is simple either we accept (and we
maintain because that's the problem) these hacks - and I am not willing
to do that - or we find a way to solve this from a general perspective not
as a point hack.
I can have a look at solving the whole issue but it won't happen
tomorrow.
Thanks,
Lorenzo
> Thanks,
> Khuong Dinh
>
> >> Finally, please execute this command on the vmlinux that "works"
> >> for you:
> >>
> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >
> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> > ffff000008dab2d8 t __initcall_acpi_init4
> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
> >
> > Best regards,
> > Khuong Dinh
> >
> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
> >> not understand how this works (I mean without relying on kernel link
> >> order to ensure that the MSI driver is probed before PCI devices are
> >> enumerated in acpi_init()).
> >>
> >> Thanks,
> >> Lorenzo
> >>
> >>> Best regards,
> >>> Khuong
> >>>
> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
> >>> >> From: Khuong Dinh <kdinh@apm.com>
> >>> >>
> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >>> >>
> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >>> >> ---
> >>> >> v2:
> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >>> >> v1:
> >>> >> - Initial version
> >>> >> ---
> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >>> >>
> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >>> >> index f1b633b..00aaa3d 100644
> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >>> >> @@ -24,6 +24,7 @@
> >>> >> #include <linux/pci.h>
> >>> >> #include <linux/platform_device.h>
> >>> >> #include <linux/of_pci.h>
> >>> >> +#include <linux/acpi.h>
> >>> >>
> >>> >> #define MSI_IR0 0x000000
> >>> >> #define MSI_INT0 0x800000
> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >>> >> };
> >>> >>
> >>> >> struct xgene_msi {
> >>> >> - struct device_node *node;
> >>> >> + struct fwnode_handle *fwnode;
> >>> >> struct irq_domain *inner_domain;
> >>> >> struct irq_domain *msi_domain;
> >>> >> u64 msi_addr;
> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >>> >> .free = xgene_irq_domain_free,
> >>> >> };
> >>> >>
> >>> >> +#ifdef CONFIG_ACPI
> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >>> >> +{
> >>> >> + return xgene_msi_ctrl.fwnode;
> >>> >> +}
> >>> >> +#endif
> >>> >> +
> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >>> >> {
> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >>> >> if (!msi->inner_domain)
> >>> >> return -ENOMEM;
> >>> >>
> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >>> >> &xgene_msi_domain_info,
> >>> >> msi->inner_domain);
> >>> >>
> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >>> >> return -ENOMEM;
> >>> >> }
> >>> >>
> >>> >> +#ifdef CONFIG_ACPI
> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >>> >> +#endif
> >>> >> return 0;
> >>> >> }
> >>> >>
> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >>> >> {},
> >>> >> };
> >>> >>
> >>> >> +#ifdef CONFIG_ACPI
> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >>> >> + {"APMC0D0E", 0},
> >>> >> + { },
> >>> >> +};
> >>> >> +#endif
> >>> >> +
> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
> >>> >> {
> >>> >> struct resource *res;
> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >>> >> goto error;
> >>> >> }
> >>> >> xgene_msi->msi_addr = res->start;
> >>> >> - xgene_msi->node = pdev->dev.of_node;
> >>> >> +
> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >>> >> + if (!xgene_msi->fwnode) {
> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >>> >
> >>> > Please provide something other than NULL, such as the base address if
> >>> > the device. That's quite useful for debugging.
> >>> >
> >>> >> + if (!xgene_msi->fwnode) {
> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >>> >> + rc = ENOMEM;
> >>> >> + goto error;
> >>> >> + }
> >>> >> + }
> >>> >> +
> >>> >> xgene_msi->num_cpus = num_possible_cpus();
> >>> >>
> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >>> >> .driver = {
> >>> >> .name = "xgene-msi",
> >>> >> .of_match_table = xgene_msi_match_table,
> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >>> >> },
> >>> >> .probe = xgene_msi_probe,
> >>> >> .remove = xgene_msi_remove,
> >>> >>
> >>> >
> >>> > The code is trivial, but relies on the MSI controller to probe before
> >>> > the PCI bus. What enforces this? How is it making sure that this is not
> >>> > going to break in the next kernel release? As far as I can tell, there
> >>> > is no explicit dependency between the two, making it the whole thing
> >>> > extremely fragile.
> >>> >
> >>> > Thanks,
> >>> >
> >>> > M.
> >>> > --
> >>> > Jazz is not dead. It just smells funny...
> >>>
> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-08 10:39 ` Lorenzo Pieralisi
0 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-08 10:39 UTC (permalink / raw)
To: Khuong Dinh
Cc: rjw, Duc Dang, Marc Zyngier, linux-pci, msalter, patches,
linux-kernel, jcm, Bjorn Helgaas, linux-arm-kernel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^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-08 10:39 ` Lorenzo Pieralisi
0 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-08 10:39 UTC (permalink / raw)
To: linux-arm-kernel
On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote:
> Hi Lorenzo,
>
> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
> > Hi Lorenzo,
> >
> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
> > <lorenzo.pieralisi@arm.com> wrote:
> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
> >>> Hi Marc,
> >>> There's no explicit dependency between pcie driver and msi
> >>> controller. The current solution that we did is relying on the
> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
> >>> higher priority in start ordering. This method could be applied in
> >>> case of msi and pcie are the same level of subsys_init (in ACPI
> >>> boot). However, PCIe driver has not supported for this dependency
> >>> check yet. How do you think about this solution.
> >>
> >> First off, you can't post emails with confidentiality notices on
> >> mailing lists so remove it from now onwards.
> >
> > Fixed
> >
> >> Secondly, I commented on this before, so you know what my opinion is.
> >
> > I got your opinion. I'll implement as your suggestion.
> >
>
> Regarding to your previous suggestion to add a hook walking the ACPI
> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
> controllers must be probed before PCI. I have a concern that the
> MSI namespace ? _SB.MSIX? is not a unique name and how can we walk
> the ACPI namespace and use this name to make MSI probed before PCI.
> May you have little bit more information for this or do you have
> another suggestion for this?
>
> There?s another solution which makes this simpler and I?d like to
> ask your opinion about this.
> The solution is to make an hierarchy between MSI and PCI nodes as below:
> Device(\_SB.MSIX) {
> Device(\_SB.PCI0)
> Device(\_SB.PCI1)
> ??
> }
> In other word, MSIX node is a parent node of PCI nodes in ACPI
> table. In this sense, there?s an explicit dependency between MSI
> and PCI, MSI controller must be probed before PCI and it will
> guarantee not breaking next kernel releases. How do you think about
> this solution.
I think that's a plaster as ineffective as reordering nodes, in short
it is not a solution and you still rely on kernel link ordering, you
can fiddle with ACPI tables as much as you want but that does not change.
> I also tried to use _DEP method to describe the dependency of PCIe
> nodes, but it looks that it does not work as PCI and MSI are handed
> by acpi_bus_scan and still having issue when we re-probe PCI.
That's a tad vague to be frank, anyway it is pretty clear to me that we
have hit a wall. In ACPI there is no way to describe probe dependencies
like the one you have, it is as simple as that, and this MSI issue you
are facing is just an example, there are more eg:
https://www.spinics.net/lists/arm-kernel/msg585825.html
At the end of the day the choice is simple either we accept (and we
maintain because that's the problem) these hacks - and I am not willing
to do that - or we find a way to solve this from a general perspective not
as a point hack.
I can have a look at solving the whole issue but it won't happen
tomorrow.
Thanks,
Lorenzo
> Thanks,
> Khuong Dinh
>
> >> Finally, please execute this command on the vmlinux that "works"
> >> for you:
> >>
> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >
> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> > ffff000008dab2d8 t __initcall_acpi_init4
> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
> >
> > Best regards,
> > Khuong Dinh
> >
> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
> >> not understand how this works (I mean without relying on kernel link
> >> order to ensure that the MSI driver is probed before PCI devices are
> >> enumerated in acpi_init()).
> >>
> >> Thanks,
> >> Lorenzo
> >>
> >>> Best regards,
> >>> Khuong
> >>>
> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
> >>> >> From: Khuong Dinh <kdinh@apm.com>
> >>> >>
> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >>> >>
> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >>> >> ---
> >>> >> v2:
> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >>> >> v1:
> >>> >> - Initial version
> >>> >> ---
> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >>> >>
> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >>> >> index f1b633b..00aaa3d 100644
> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >>> >> @@ -24,6 +24,7 @@
> >>> >> #include <linux/pci.h>
> >>> >> #include <linux/platform_device.h>
> >>> >> #include <linux/of_pci.h>
> >>> >> +#include <linux/acpi.h>
> >>> >>
> >>> >> #define MSI_IR0 0x000000
> >>> >> #define MSI_INT0 0x800000
> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >>> >> };
> >>> >>
> >>> >> struct xgene_msi {
> >>> >> - struct device_node *node;
> >>> >> + struct fwnode_handle *fwnode;
> >>> >> struct irq_domain *inner_domain;
> >>> >> struct irq_domain *msi_domain;
> >>> >> u64 msi_addr;
> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >>> >> .free = xgene_irq_domain_free,
> >>> >> };
> >>> >>
> >>> >> +#ifdef CONFIG_ACPI
> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >>> >> +{
> >>> >> + return xgene_msi_ctrl.fwnode;
> >>> >> +}
> >>> >> +#endif
> >>> >> +
> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >>> >> {
> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >>> >> if (!msi->inner_domain)
> >>> >> return -ENOMEM;
> >>> >>
> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >>> >> &xgene_msi_domain_info,
> >>> >> msi->inner_domain);
> >>> >>
> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >>> >> return -ENOMEM;
> >>> >> }
> >>> >>
> >>> >> +#ifdef CONFIG_ACPI
> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >>> >> +#endif
> >>> >> return 0;
> >>> >> }
> >>> >>
> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >>> >> {},
> >>> >> };
> >>> >>
> >>> >> +#ifdef CONFIG_ACPI
> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >>> >> + {"APMC0D0E", 0},
> >>> >> + { },
> >>> >> +};
> >>> >> +#endif
> >>> >> +
> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
> >>> >> {
> >>> >> struct resource *res;
> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >>> >> goto error;
> >>> >> }
> >>> >> xgene_msi->msi_addr = res->start;
> >>> >> - xgene_msi->node = pdev->dev.of_node;
> >>> >> +
> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >>> >> + if (!xgene_msi->fwnode) {
> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >>> >
> >>> > Please provide something other than NULL, such as the base address if
> >>> > the device. That's quite useful for debugging.
> >>> >
> >>> >> + if (!xgene_msi->fwnode) {
> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >>> >> + rc = ENOMEM;
> >>> >> + goto error;
> >>> >> + }
> >>> >> + }
> >>> >> +
> >>> >> xgene_msi->num_cpus = num_possible_cpus();
> >>> >>
> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >>> >> .driver = {
> >>> >> .name = "xgene-msi",
> >>> >> .of_match_table = xgene_msi_match_table,
> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >>> >> },
> >>> >> .probe = xgene_msi_probe,
> >>> >> .remove = xgene_msi_remove,
> >>> >>
> >>> >
> >>> > The code is trivial, but relies on the MSI controller to probe before
> >>> > the PCI bus. What enforces this? How is it making sure that this is not
> >>> > going to break in the next kernel release? As far as I can tell, there
> >>> > is no explicit dependency between the two, making it the whole thing
> >>> > extremely fragile.
> >>> >
> >>> > Thanks,
> >>> >
> >>> > M.
> >>> > --
> >>> > Jazz is not dead. It just smells funny...
> >>>
> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-06-08 10:39 ` Lorenzo Pieralisi
(?)
@ 2017-06-13 20:56 ` Khuong Dinh
-1 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-13 20:56 UTC (permalink / raw)
To: Lorenzo Pieralisi
Cc: Marc Zyngier, msalter, Bjorn Helgaas, linux-pci, jcm, patches,
linux-kernel, linux-arm-kernel, rjw, Duc Dang
Hi Lorenzo,
On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi
<lorenzo.pieralisi@arm.com> wrote:
> On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote:
>> Hi Lorenzo,
>>
>> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
>> > Hi Lorenzo,
>> >
>> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
>> > <lorenzo.pieralisi@arm.com> wrote:
>> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>> >>> Hi Marc,
>> >>> There's no explicit dependency between pcie driver and msi
>> >>> controller. The current solution that we did is relying on the
>> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>> >>> higher priority in start ordering. This method could be applied in
>> >>> case of msi and pcie are the same level of subsys_init (in ACPI
>> >>> boot). However, PCIe driver has not supported for this dependency
>> >>> check yet. How do you think about this solution.
>> >>
>> >> First off, you can't post emails with confidentiality notices on
>> >> mailing lists so remove it from now onwards.
>> >
>> > Fixed
>> >
>> >> Secondly, I commented on this before, so you know what my opinion is.
>> >
>> > I got your opinion. I'll implement as your suggestion.
>> >
>>
>> Regarding to your previous suggestion to add a hook walking the ACPI
>> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
>> controllers must be probed before PCI. I have a concern that the
>> MSI namespace “ _SB.MSIX” is not a unique name and how can we walk
>> the ACPI namespace and use this name to make MSI probed before PCI.
>> May you have little bit more information for this or do you have
>> another suggestion for this?
>>
>> There’s another solution which makes this simpler and I’d like to
>> ask your opinion about this.
>> The solution is to make an hierarchy between MSI and PCI nodes as below:
>> Device(\_SB.MSIX) {
>> Device(\_SB.PCI0)
>> Device(\_SB.PCI1)
>> ……
>> }
>> In other word, MSIX node is a parent node of PCI nodes in ACPI
>> table. In this sense, there’s an explicit dependency between MSI
>> and PCI, MSI controller must be probed before PCI and it will
>> guarantee not breaking next kernel releases. How do you think about
>> this solution.
>
> I think that's a plaster as ineffective as reordering nodes, in short
> it is not a solution and you still rely on kernel link ordering, you
> can fiddle with ACPI tables as much as you want but that does not change.
>
>> I also tried to use _DEP method to describe the dependency of PCIe
>> nodes, but it looks that it does not work as PCI and MSI are handed
>> by acpi_bus_scan and still having issue when we re-probe PCI.
>
> That's a tad vague to be frank, anyway it is pretty clear to me that we
> have hit a wall. In ACPI there is no way to describe probe dependencies
> like the one you have, it is as simple as that, and this MSI issue you
> are facing is just an example, there are more eg:
>
> https://www.spinics.net/lists/arm-kernel/msg585825.html
>
> At the end of the day the choice is simple either we accept (and we
> maintain because that's the problem) these hacks - and I am not willing
> to do that - or we find a way to solve this from a general perspective not
> as a point hack.
>
> I can have a look at solving the whole issue but it won't happen
> tomorrow.
Thanks for helping to resolve this general ACPI dependence issue.
I look forward for a version to test with.
Can we separate the general dependence patch from the X-Gene MSI ACPI
enable patch.
This original patch was to enable ACPI support for PCIe MSI.
The dependence issue can be resolved separately.
Can you help to pull in the patch.
Best regards,
Khuong Dinh
> Thanks,
> Lorenzo
>
>> Thanks,
>> Khuong Dinh
>>
>> >> Finally, please execute this command on the vmlinux that "works"
>> >> for you:
>> >>
>> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>> >
>> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>> > ffff000008dab2d8 t __initcall_acpi_init4
>> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
>> >
>> > Best regards,
>> > Khuong Dinh
>> >
>> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
>> >> not understand how this works (I mean without relying on kernel link
>> >> order to ensure that the MSI driver is probed before PCI devices are
>> >> enumerated in acpi_init()).
>> >>
>> >> Thanks,
>> >> Lorenzo
>> >>
>> >>> Best regards,
>> >>> Khuong
>> >>>
>> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
>> >>> >> From: Khuong Dinh <kdinh@apm.com>
>> >>> >>
>> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>> >>> >>
>> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> >>> >> ---
>> >>> >> v2:
>> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>> >>> >> v1:
>> >>> >> - Initial version
>> >>> >> ---
>> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>> >>> >>
>> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> >>> >> index f1b633b..00aaa3d 100644
>> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>> >>> >> @@ -24,6 +24,7 @@
>> >>> >> #include <linux/pci.h>
>> >>> >> #include <linux/platform_device.h>
>> >>> >> #include <linux/of_pci.h>
>> >>> >> +#include <linux/acpi.h>
>> >>> >>
>> >>> >> #define MSI_IR0 0x000000
>> >>> >> #define MSI_INT0 0x800000
>> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> >>> >> };
>> >>> >>
>> >>> >> struct xgene_msi {
>> >>> >> - struct device_node *node;
>> >>> >> + struct fwnode_handle *fwnode;
>> >>> >> struct irq_domain *inner_domain;
>> >>> >> struct irq_domain *msi_domain;
>> >>> >> u64 msi_addr;
>> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> >>> >> .free = xgene_irq_domain_free,
>> >>> >> };
>> >>> >>
>> >>> >> +#ifdef CONFIG_ACPI
>> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> >>> >> +{
>> >>> >> + return xgene_msi_ctrl.fwnode;
>> >>> >> +}
>> >>> >> +#endif
>> >>> >> +
>> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>> >>> >> {
>> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >>> >> if (!msi->inner_domain)
>> >>> >> return -ENOMEM;
>> >>> >>
>> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> >>> >> &xgene_msi_domain_info,
>> >>> >> msi->inner_domain);
>> >>> >>
>> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >>> >> return -ENOMEM;
>> >>> >> }
>> >>> >>
>> >>> >> +#ifdef CONFIG_ACPI
>> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> >>> >> +#endif
>> >>> >> return 0;
>> >>> >> }
>> >>> >>
>> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> >>> >> {},
>> >>> >> };
>> >>> >>
>> >>> >> +#ifdef CONFIG_ACPI
>> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> >>> >> + {"APMC0D0E", 0},
>> >>> >> + { },
>> >>> >> +};
>> >>> >> +#endif
>> >>> >> +
>> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
>> >>> >> {
>> >>> >> struct resource *res;
>> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >>> >> goto error;
>> >>> >> }
>> >>> >> xgene_msi->msi_addr = res->start;
>> >>> >> - xgene_msi->node = pdev->dev.of_node;
>> >>> >> +
>> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> >>> >> + if (!xgene_msi->fwnode) {
>> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>> >>> >
>> >>> > Please provide something other than NULL, such as the base address if
>> >>> > the device. That's quite useful for debugging.
>> >>> >
>> >>> >> + if (!xgene_msi->fwnode) {
>> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> >>> >> + rc = ENOMEM;
>> >>> >> + goto error;
>> >>> >> + }
>> >>> >> + }
>> >>> >> +
>> >>> >> xgene_msi->num_cpus = num_possible_cpus();
>> >>> >>
>> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
>> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >>> >> .driver = {
>> >>> >> .name = "xgene-msi",
>> >>> >> .of_match_table = xgene_msi_match_table,
>> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> >>> >> },
>> >>> >> .probe = xgene_msi_probe,
>> >>> >> .remove = xgene_msi_remove,
>> >>> >>
>> >>> >
>> >>> > The code is trivial, but relies on the MSI controller to probe before
>> >>> > the PCI bus. What enforces this? How is it making sure that this is not
>> >>> > going to break in the next kernel release? As far as I can tell, there
>> >>> > is no explicit dependency between the two, making it the whole thing
>> >>> > extremely fragile.
>> >>> >
>> >>> > Thanks,
>> >>> >
>> >>> > M.
>> >>> > --
>> >>> > Jazz is not dead. It just smells funny...
>> >>>
>> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-13 20:56 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-13 20:56 UTC (permalink / raw)
To: Lorenzo Pieralisi
Cc: rjw, Duc Dang, Marc Zyngier, linux-pci, msalter, patches,
linux-kernel, jcm, Bjorn Helgaas, linux-arm-kernel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^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-13 20:56 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-13 20:56 UTC (permalink / raw)
To: linux-arm-kernel
Hi Lorenzo,
On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi
<lorenzo.pieralisi@arm.com> wrote:
> On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote:
>> Hi Lorenzo,
>>
>> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
>> > Hi Lorenzo,
>> >
>> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
>> > <lorenzo.pieralisi@arm.com> wrote:
>> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>> >>> Hi Marc,
>> >>> There's no explicit dependency between pcie driver and msi
>> >>> controller. The current solution that we did is relying on the
>> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>> >>> higher priority in start ordering. This method could be applied in
>> >>> case of msi and pcie are the same level of subsys_init (in ACPI
>> >>> boot). However, PCIe driver has not supported for this dependency
>> >>> check yet. How do you think about this solution.
>> >>
>> >> First off, you can't post emails with confidentiality notices on
>> >> mailing lists so remove it from now onwards.
>> >
>> > Fixed
>> >
>> >> Secondly, I commented on this before, so you know what my opinion is.
>> >
>> > I got your opinion. I'll implement as your suggestion.
>> >
>>
>> Regarding to your previous suggestion to add a hook walking the ACPI
>> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
>> controllers must be probed before PCI. I have a concern that the
>> MSI namespace ? _SB.MSIX? is not a unique name and how can we walk
>> the ACPI namespace and use this name to make MSI probed before PCI.
>> May you have little bit more information for this or do you have
>> another suggestion for this?
>>
>> There?s another solution which makes this simpler and I?d like to
>> ask your opinion about this.
>> The solution is to make an hierarchy between MSI and PCI nodes as below:
>> Device(\_SB.MSIX) {
>> Device(\_SB.PCI0)
>> Device(\_SB.PCI1)
>> ??
>> }
>> In other word, MSIX node is a parent node of PCI nodes in ACPI
>> table. In this sense, there?s an explicit dependency between MSI
>> and PCI, MSI controller must be probed before PCI and it will
>> guarantee not breaking next kernel releases. How do you think about
>> this solution.
>
> I think that's a plaster as ineffective as reordering nodes, in short
> it is not a solution and you still rely on kernel link ordering, you
> can fiddle with ACPI tables as much as you want but that does not change.
>
>> I also tried to use _DEP method to describe the dependency of PCIe
>> nodes, but it looks that it does not work as PCI and MSI are handed
>> by acpi_bus_scan and still having issue when we re-probe PCI.
>
> That's a tad vague to be frank, anyway it is pretty clear to me that we
> have hit a wall. In ACPI there is no way to describe probe dependencies
> like the one you have, it is as simple as that, and this MSI issue you
> are facing is just an example, there are more eg:
>
> https://www.spinics.net/lists/arm-kernel/msg585825.html
>
> At the end of the day the choice is simple either we accept (and we
> maintain because that's the problem) these hacks - and I am not willing
> to do that - or we find a way to solve this from a general perspective not
> as a point hack.
>
> I can have a look at solving the whole issue but it won't happen
> tomorrow.
Thanks for helping to resolve this general ACPI dependence issue.
I look forward for a version to test with.
Can we separate the general dependence patch from the X-Gene MSI ACPI
enable patch.
This original patch was to enable ACPI support for PCIe MSI.
The dependence issue can be resolved separately.
Can you help to pull in the patch.
Best regards,
Khuong Dinh
> Thanks,
> Lorenzo
>
>> Thanks,
>> Khuong Dinh
>>
>> >> Finally, please execute this command on the vmlinux that "works"
>> >> for you:
>> >>
>> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>> >
>> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>> > ffff000008dab2d8 t __initcall_acpi_init4
>> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
>> >
>> > Best regards,
>> > Khuong Dinh
>> >
>> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
>> >> not understand how this works (I mean without relying on kernel link
>> >> order to ensure that the MSI driver is probed before PCI devices are
>> >> enumerated in acpi_init()).
>> >>
>> >> Thanks,
>> >> Lorenzo
>> >>
>> >>> Best regards,
>> >>> Khuong
>> >>>
>> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
>> >>> >> From: Khuong Dinh <kdinh@apm.com>
>> >>> >>
>> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>> >>> >>
>> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> >>> >> ---
>> >>> >> v2:
>> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>> >>> >> v1:
>> >>> >> - Initial version
>> >>> >> ---
>> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>> >>> >>
>> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> >>> >> index f1b633b..00aaa3d 100644
>> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>> >>> >> @@ -24,6 +24,7 @@
>> >>> >> #include <linux/pci.h>
>> >>> >> #include <linux/platform_device.h>
>> >>> >> #include <linux/of_pci.h>
>> >>> >> +#include <linux/acpi.h>
>> >>> >>
>> >>> >> #define MSI_IR0 0x000000
>> >>> >> #define MSI_INT0 0x800000
>> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> >>> >> };
>> >>> >>
>> >>> >> struct xgene_msi {
>> >>> >> - struct device_node *node;
>> >>> >> + struct fwnode_handle *fwnode;
>> >>> >> struct irq_domain *inner_domain;
>> >>> >> struct irq_domain *msi_domain;
>> >>> >> u64 msi_addr;
>> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> >>> >> .free = xgene_irq_domain_free,
>> >>> >> };
>> >>> >>
>> >>> >> +#ifdef CONFIG_ACPI
>> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> >>> >> +{
>> >>> >> + return xgene_msi_ctrl.fwnode;
>> >>> >> +}
>> >>> >> +#endif
>> >>> >> +
>> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>> >>> >> {
>> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >>> >> if (!msi->inner_domain)
>> >>> >> return -ENOMEM;
>> >>> >>
>> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> >>> >> &xgene_msi_domain_info,
>> >>> >> msi->inner_domain);
>> >>> >>
>> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >>> >> return -ENOMEM;
>> >>> >> }
>> >>> >>
>> >>> >> +#ifdef CONFIG_ACPI
>> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> >>> >> +#endif
>> >>> >> return 0;
>> >>> >> }
>> >>> >>
>> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> >>> >> {},
>> >>> >> };
>> >>> >>
>> >>> >> +#ifdef CONFIG_ACPI
>> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> >>> >> + {"APMC0D0E", 0},
>> >>> >> + { },
>> >>> >> +};
>> >>> >> +#endif
>> >>> >> +
>> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
>> >>> >> {
>> >>> >> struct resource *res;
>> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >>> >> goto error;
>> >>> >> }
>> >>> >> xgene_msi->msi_addr = res->start;
>> >>> >> - xgene_msi->node = pdev->dev.of_node;
>> >>> >> +
>> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> >>> >> + if (!xgene_msi->fwnode) {
>> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>> >>> >
>> >>> > Please provide something other than NULL, such as the base address if
>> >>> > the device. That's quite useful for debugging.
>> >>> >
>> >>> >> + if (!xgene_msi->fwnode) {
>> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> >>> >> + rc = ENOMEM;
>> >>> >> + goto error;
>> >>> >> + }
>> >>> >> + }
>> >>> >> +
>> >>> >> xgene_msi->num_cpus = num_possible_cpus();
>> >>> >>
>> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
>> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >>> >> .driver = {
>> >>> >> .name = "xgene-msi",
>> >>> >> .of_match_table = xgene_msi_match_table,
>> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> >>> >> },
>> >>> >> .probe = xgene_msi_probe,
>> >>> >> .remove = xgene_msi_remove,
>> >>> >>
>> >>> >
>> >>> > The code is trivial, but relies on the MSI controller to probe before
>> >>> > the PCI bus. What enforces this? How is it making sure that this is not
>> >>> > going to break in the next kernel release? As far as I can tell, there
>> >>> > is no explicit dependency between the two, making it the whole thing
>> >>> > extremely fragile.
>> >>> >
>> >>> > Thanks,
>> >>> >
>> >>> > M.
>> >>> > --
>> >>> > Jazz is not dead. It just smells funny...
>> >>>
>> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-06-13 20:56 ` Khuong Dinh
(?)
@ 2017-06-14 12:59 ` Lorenzo Pieralisi
-1 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-14 12:59 UTC (permalink / raw)
To: Khuong Dinh
Cc: Marc Zyngier, msalter, Bjorn Helgaas, linux-pci, jcm, patches,
linux-kernel, linux-arm-kernel, rjw, Duc Dang
On Tue, Jun 13, 2017 at 01:56:44PM -0700, Khuong Dinh wrote:
> Hi Lorenzo,
>
> On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi
> <lorenzo.pieralisi@arm.com> wrote:
> > On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote:
> >> Hi Lorenzo,
> >>
> >> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
> >> > Hi Lorenzo,
> >> >
> >> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
> >> > <lorenzo.pieralisi@arm.com> wrote:
> >> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
> >> >>> Hi Marc,
> >> >>> There's no explicit dependency between pcie driver and msi
> >> >>> controller. The current solution that we did is relying on the
> >> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
> >> >>> higher priority in start ordering. This method could be applied in
> >> >>> case of msi and pcie are the same level of subsys_init (in ACPI
> >> >>> boot). However, PCIe driver has not supported for this dependency
> >> >>> check yet. How do you think about this solution.
> >> >>
> >> >> First off, you can't post emails with confidentiality notices on
> >> >> mailing lists so remove it from now onwards.
> >> >
> >> > Fixed
> >> >
> >> >> Secondly, I commented on this before, so you know what my opinion is.
> >> >
> >> > I got your opinion. I'll implement as your suggestion.
> >> >
> >>
> >> Regarding to your previous suggestion to add a hook walking the ACPI
> >> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
> >> controllers must be probed before PCI. I have a concern that the
> >> MSI namespace “ _SB.MSIX” is not a unique name and how can we walk
> >> the ACPI namespace and use this name to make MSI probed before PCI.
> >> May you have little bit more information for this or do you have
> >> another suggestion for this?
> >>
> >> There’s another solution which makes this simpler and I’d like to
> >> ask your opinion about this.
> >> The solution is to make an hierarchy between MSI and PCI nodes as below:
> >> Device(\_SB.MSIX) {
> >> Device(\_SB.PCI0)
> >> Device(\_SB.PCI1)
> >> ……
> >> }
> >> In other word, MSIX node is a parent node of PCI nodes in ACPI
> >> table. In this sense, there’s an explicit dependency between MSI
> >> and PCI, MSI controller must be probed before PCI and it will
> >> guarantee not breaking next kernel releases. How do you think about
> >> this solution.
> >
> > I think that's a plaster as ineffective as reordering nodes, in short
> > it is not a solution and you still rely on kernel link ordering, you
> > can fiddle with ACPI tables as much as you want but that does not change.
> >
> >> I also tried to use _DEP method to describe the dependency of PCIe
> >> nodes, but it looks that it does not work as PCI and MSI are handed
> >> by acpi_bus_scan and still having issue when we re-probe PCI.
> >
> > That's a tad vague to be frank, anyway it is pretty clear to me that we
> > have hit a wall. In ACPI there is no way to describe probe dependencies
> > like the one you have, it is as simple as that, and this MSI issue you
> > are facing is just an example, there are more eg:
> >
> > https://www.spinics.net/lists/arm-kernel/msg585825.html
> >
> > At the end of the day the choice is simple either we accept (and we
> > maintain because that's the problem) these hacks - and I am not willing
> > to do that - or we find a way to solve this from a general perspective not
> > as a point hack.
> >
> > I can have a look at solving the whole issue but it won't happen
> > tomorrow.
>
> Thanks for helping to resolve this general ACPI dependence issue.
> I look forward for a version to test with.
>
> Can we separate the general dependence patch from the X-Gene MSI ACPI
> enable patch.
> This original patch was to enable ACPI support for PCIe MSI.
> The dependence issue can be resolved separately.
> Can you help to pull in the patch.
Ok, pragmatically the only sane thing you can do is:
(1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init())
that carries out fwnode registration (ie register the fwnode by
searching the MSI HID object - this does not depend on ACPI device
nodes order - you enforce the order by parsing the namespace before
ACPI core does it, at that point in time ACPI objects are created).
Not sure Rafael will be happy to see this code but that's the only
solution that does not rely on ACPI device nodes ordering and kernel
link ordering. You may achieve the same by extending the ACPI APD
code (drivers/acpi/acpi_apd.c) whether that's the best way forward is
to be seen.
(2) You can also add an xgene specific scan handler at arch_init_call()
but given that PCI root bridge is managed through a scan handler too you
would rely on ACPI devices nodes ordering in the DSDT. Let me explain
something for the benefit of everyone reading this thread: I do not want
X-gene ACPI tables to force the kernel to scan devices in any order
whatsover because this would mean we will _never_ be able to change the
scan order if we wanted to lest we trigger regressions. So this is
a non-option in my opinion.
(3) Last option is to register the MSI fwnode either in PCI ECAM quirk
handling or in arm64 before probing the PCI root bus.
I am not keen on (2) and (3) at all, so _if_ we have to find a solution
to this problem (1) is the only option you have for the time being as
far as I am concerned.
Lorenzo
>
> Best regards,
> Khuong Dinh
>
> > Thanks,
> > Lorenzo
> >
> >> Thanks,
> >> Khuong Dinh
> >>
> >> >> Finally, please execute this command on the vmlinux that "works"
> >> >> for you:
> >> >>
> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> >
> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> > ffff000008dab2d8 t __initcall_acpi_init4
> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
> >> >
> >> > Best regards,
> >> > Khuong Dinh
> >> >
> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
> >> >> not understand how this works (I mean without relying on kernel link
> >> >> order to ensure that the MSI driver is probed before PCI devices are
> >> >> enumerated in acpi_init()).
> >> >>
> >> >> Thanks,
> >> >> Lorenzo
> >> >>
> >> >>> Best regards,
> >> >>> Khuong
> >> >>>
> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
> >> >>> >> From: Khuong Dinh <kdinh@apm.com>
> >> >>> >>
> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >> >>> >>
> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >> >>> >> ---
> >> >>> >> v2:
> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >> >>> >> v1:
> >> >>> >> - Initial version
> >> >>> >> ---
> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >> >>> >>
> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >> >>> >> index f1b633b..00aaa3d 100644
> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >> >>> >> @@ -24,6 +24,7 @@
> >> >>> >> #include <linux/pci.h>
> >> >>> >> #include <linux/platform_device.h>
> >> >>> >> #include <linux/of_pci.h>
> >> >>> >> +#include <linux/acpi.h>
> >> >>> >>
> >> >>> >> #define MSI_IR0 0x000000
> >> >>> >> #define MSI_INT0 0x800000
> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >> >>> >> };
> >> >>> >>
> >> >>> >> struct xgene_msi {
> >> >>> >> - struct device_node *node;
> >> >>> >> + struct fwnode_handle *fwnode;
> >> >>> >> struct irq_domain *inner_domain;
> >> >>> >> struct irq_domain *msi_domain;
> >> >>> >> u64 msi_addr;
> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >> >>> >> .free = xgene_irq_domain_free,
> >> >>> >> };
> >> >>> >>
> >> >>> >> +#ifdef CONFIG_ACPI
> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >> >>> >> +{
> >> >>> >> + return xgene_msi_ctrl.fwnode;
> >> >>> >> +}
> >> >>> >> +#endif
> >> >>> >> +
> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >>> >> {
> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >>> >> if (!msi->inner_domain)
> >> >>> >> return -ENOMEM;
> >> >>> >>
> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >> >>> >> &xgene_msi_domain_info,
> >> >>> >> msi->inner_domain);
> >> >>> >>
> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >>> >> return -ENOMEM;
> >> >>> >> }
> >> >>> >>
> >> >>> >> +#ifdef CONFIG_ACPI
> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >> >>> >> +#endif
> >> >>> >> return 0;
> >> >>> >> }
> >> >>> >>
> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >> >>> >> {},
> >> >>> >> };
> >> >>> >>
> >> >>> >> +#ifdef CONFIG_ACPI
> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >> >>> >> + {"APMC0D0E", 0},
> >> >>> >> + { },
> >> >>> >> +};
> >> >>> >> +#endif
> >> >>> >> +
> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
> >> >>> >> {
> >> >>> >> struct resource *res;
> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >>> >> goto error;
> >> >>> >> }
> >> >>> >> xgene_msi->msi_addr = res->start;
> >> >>> >> - xgene_msi->node = pdev->dev.of_node;
> >> >>> >> +
> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >> >>> >
> >> >>> > Please provide something other than NULL, such as the base address if
> >> >>> > the device. That's quite useful for debugging.
> >> >>> >
> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >> >>> >> + rc = ENOMEM;
> >> >>> >> + goto error;
> >> >>> >> + }
> >> >>> >> + }
> >> >>> >> +
> >> >>> >> xgene_msi->num_cpus = num_possible_cpus();
> >> >>> >>
> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >>> >> .driver = {
> >> >>> >> .name = "xgene-msi",
> >> >>> >> .of_match_table = xgene_msi_match_table,
> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >> >>> >> },
> >> >>> >> .probe = xgene_msi_probe,
> >> >>> >> .remove = xgene_msi_remove,
> >> >>> >>
> >> >>> >
> >> >>> > The code is trivial, but relies on the MSI controller to probe before
> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not
> >> >>> > going to break in the next kernel release? As far as I can tell, there
> >> >>> > is no explicit dependency between the two, making it the whole thing
> >> >>> > extremely fragile.
> >> >>> >
> >> >>> > Thanks,
> >> >>> >
> >> >>> > M.
> >> >>> > --
> >> >>> > Jazz is not dead. It just smells funny...
> >> >>>
> >> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-14 12:59 ` Lorenzo Pieralisi
0 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-14 12:59 UTC (permalink / raw)
To: Khuong Dinh
Cc: rjw, Duc Dang, Marc Zyngier, linux-pci, msalter, patches,
linux-kernel, jcm, Bjorn Helgaas, linux-arm-kernel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^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-14 12:59 ` Lorenzo Pieralisi
0 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-14 12:59 UTC (permalink / raw)
To: linux-arm-kernel
On Tue, Jun 13, 2017 at 01:56:44PM -0700, Khuong Dinh wrote:
> Hi Lorenzo,
>
> On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi
> <lorenzo.pieralisi@arm.com> wrote:
> > On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote:
> >> Hi Lorenzo,
> >>
> >> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
> >> > Hi Lorenzo,
> >> >
> >> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
> >> > <lorenzo.pieralisi@arm.com> wrote:
> >> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
> >> >>> Hi Marc,
> >> >>> There's no explicit dependency between pcie driver and msi
> >> >>> controller. The current solution that we did is relying on the
> >> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
> >> >>> higher priority in start ordering. This method could be applied in
> >> >>> case of msi and pcie are the same level of subsys_init (in ACPI
> >> >>> boot). However, PCIe driver has not supported for this dependency
> >> >>> check yet. How do you think about this solution.
> >> >>
> >> >> First off, you can't post emails with confidentiality notices on
> >> >> mailing lists so remove it from now onwards.
> >> >
> >> > Fixed
> >> >
> >> >> Secondly, I commented on this before, so you know what my opinion is.
> >> >
> >> > I got your opinion. I'll implement as your suggestion.
> >> >
> >>
> >> Regarding to your previous suggestion to add a hook walking the ACPI
> >> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
> >> controllers must be probed before PCI. I have a concern that the
> >> MSI namespace ? _SB.MSIX? is not a unique name and how can we walk
> >> the ACPI namespace and use this name to make MSI probed before PCI.
> >> May you have little bit more information for this or do you have
> >> another suggestion for this?
> >>
> >> There?s another solution which makes this simpler and I?d like to
> >> ask your opinion about this.
> >> The solution is to make an hierarchy between MSI and PCI nodes as below:
> >> Device(\_SB.MSIX) {
> >> Device(\_SB.PCI0)
> >> Device(\_SB.PCI1)
> >> ??
> >> }
> >> In other word, MSIX node is a parent node of PCI nodes in ACPI
> >> table. In this sense, there?s an explicit dependency between MSI
> >> and PCI, MSI controller must be probed before PCI and it will
> >> guarantee not breaking next kernel releases. How do you think about
> >> this solution.
> >
> > I think that's a plaster as ineffective as reordering nodes, in short
> > it is not a solution and you still rely on kernel link ordering, you
> > can fiddle with ACPI tables as much as you want but that does not change.
> >
> >> I also tried to use _DEP method to describe the dependency of PCIe
> >> nodes, but it looks that it does not work as PCI and MSI are handed
> >> by acpi_bus_scan and still having issue when we re-probe PCI.
> >
> > That's a tad vague to be frank, anyway it is pretty clear to me that we
> > have hit a wall. In ACPI there is no way to describe probe dependencies
> > like the one you have, it is as simple as that, and this MSI issue you
> > are facing is just an example, there are more eg:
> >
> > https://www.spinics.net/lists/arm-kernel/msg585825.html
> >
> > At the end of the day the choice is simple either we accept (and we
> > maintain because that's the problem) these hacks - and I am not willing
> > to do that - or we find a way to solve this from a general perspective not
> > as a point hack.
> >
> > I can have a look at solving the whole issue but it won't happen
> > tomorrow.
>
> Thanks for helping to resolve this general ACPI dependence issue.
> I look forward for a version to test with.
>
> Can we separate the general dependence patch from the X-Gene MSI ACPI
> enable patch.
> This original patch was to enable ACPI support for PCIe MSI.
> The dependence issue can be resolved separately.
> Can you help to pull in the patch.
Ok, pragmatically the only sane thing you can do is:
(1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init())
that carries out fwnode registration (ie register the fwnode by
searching the MSI HID object - this does not depend on ACPI device
nodes order - you enforce the order by parsing the namespace before
ACPI core does it, at that point in time ACPI objects are created).
Not sure Rafael will be happy to see this code but that's the only
solution that does not rely on ACPI device nodes ordering and kernel
link ordering. You may achieve the same by extending the ACPI APD
code (drivers/acpi/acpi_apd.c) whether that's the best way forward is
to be seen.
(2) You can also add an xgene specific scan handler at arch_init_call()
but given that PCI root bridge is managed through a scan handler too you
would rely on ACPI devices nodes ordering in the DSDT. Let me explain
something for the benefit of everyone reading this thread: I do not want
X-gene ACPI tables to force the kernel to scan devices in any order
whatsover because this would mean we will _never_ be able to change the
scan order if we wanted to lest we trigger regressions. So this is
a non-option in my opinion.
(3) Last option is to register the MSI fwnode either in PCI ECAM quirk
handling or in arm64 before probing the PCI root bus.
I am not keen on (2) and (3) at all, so _if_ we have to find a solution
to this problem (1) is the only option you have for the time being as
far as I am concerned.
Lorenzo
>
> Best regards,
> Khuong Dinh
>
> > Thanks,
> > Lorenzo
> >
> >> Thanks,
> >> Khuong Dinh
> >>
> >> >> Finally, please execute this command on the vmlinux that "works"
> >> >> for you:
> >> >>
> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> >
> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> > ffff000008dab2d8 t __initcall_acpi_init4
> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
> >> >
> >> > Best regards,
> >> > Khuong Dinh
> >> >
> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
> >> >> not understand how this works (I mean without relying on kernel link
> >> >> order to ensure that the MSI driver is probed before PCI devices are
> >> >> enumerated in acpi_init()).
> >> >>
> >> >> Thanks,
> >> >> Lorenzo
> >> >>
> >> >>> Best regards,
> >> >>> Khuong
> >> >>>
> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
> >> >>> >> From: Khuong Dinh <kdinh@apm.com>
> >> >>> >>
> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >> >>> >>
> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >> >>> >> ---
> >> >>> >> v2:
> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >> >>> >> v1:
> >> >>> >> - Initial version
> >> >>> >> ---
> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >> >>> >>
> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >> >>> >> index f1b633b..00aaa3d 100644
> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >> >>> >> @@ -24,6 +24,7 @@
> >> >>> >> #include <linux/pci.h>
> >> >>> >> #include <linux/platform_device.h>
> >> >>> >> #include <linux/of_pci.h>
> >> >>> >> +#include <linux/acpi.h>
> >> >>> >>
> >> >>> >> #define MSI_IR0 0x000000
> >> >>> >> #define MSI_INT0 0x800000
> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >> >>> >> };
> >> >>> >>
> >> >>> >> struct xgene_msi {
> >> >>> >> - struct device_node *node;
> >> >>> >> + struct fwnode_handle *fwnode;
> >> >>> >> struct irq_domain *inner_domain;
> >> >>> >> struct irq_domain *msi_domain;
> >> >>> >> u64 msi_addr;
> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >> >>> >> .free = xgene_irq_domain_free,
> >> >>> >> };
> >> >>> >>
> >> >>> >> +#ifdef CONFIG_ACPI
> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >> >>> >> +{
> >> >>> >> + return xgene_msi_ctrl.fwnode;
> >> >>> >> +}
> >> >>> >> +#endif
> >> >>> >> +
> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >>> >> {
> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >>> >> if (!msi->inner_domain)
> >> >>> >> return -ENOMEM;
> >> >>> >>
> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >> >>> >> &xgene_msi_domain_info,
> >> >>> >> msi->inner_domain);
> >> >>> >>
> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >>> >> return -ENOMEM;
> >> >>> >> }
> >> >>> >>
> >> >>> >> +#ifdef CONFIG_ACPI
> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >> >>> >> +#endif
> >> >>> >> return 0;
> >> >>> >> }
> >> >>> >>
> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >> >>> >> {},
> >> >>> >> };
> >> >>> >>
> >> >>> >> +#ifdef CONFIG_ACPI
> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >> >>> >> + {"APMC0D0E", 0},
> >> >>> >> + { },
> >> >>> >> +};
> >> >>> >> +#endif
> >> >>> >> +
> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
> >> >>> >> {
> >> >>> >> struct resource *res;
> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >>> >> goto error;
> >> >>> >> }
> >> >>> >> xgene_msi->msi_addr = res->start;
> >> >>> >> - xgene_msi->node = pdev->dev.of_node;
> >> >>> >> +
> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >> >>> >
> >> >>> > Please provide something other than NULL, such as the base address if
> >> >>> > the device. That's quite useful for debugging.
> >> >>> >
> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >> >>> >> + rc = ENOMEM;
> >> >>> >> + goto error;
> >> >>> >> + }
> >> >>> >> + }
> >> >>> >> +
> >> >>> >> xgene_msi->num_cpus = num_possible_cpus();
> >> >>> >>
> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >>> >> .driver = {
> >> >>> >> .name = "xgene-msi",
> >> >>> >> .of_match_table = xgene_msi_match_table,
> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >> >>> >> },
> >> >>> >> .probe = xgene_msi_probe,
> >> >>> >> .remove = xgene_msi_remove,
> >> >>> >>
> >> >>> >
> >> >>> > The code is trivial, but relies on the MSI controller to probe before
> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not
> >> >>> > going to break in the next kernel release? As far as I can tell, there
> >> >>> > is no explicit dependency between the two, making it the whole thing
> >> >>> > extremely fragile.
> >> >>> >
> >> >>> > Thanks,
> >> >>> >
> >> >>> > M.
> >> >>> > --
> >> >>> > Jazz is not dead. It just smells funny...
> >> >>>
> >> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-06-14 12:59 ` Lorenzo Pieralisi
(?)
@ 2017-06-14 17:43 ` Khuong Dinh
-1 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-14 17:43 UTC (permalink / raw)
To: Lorenzo Pieralisi
Cc: Marc Zyngier, msalter, Bjorn Helgaas, linux-pci, jcm, patches,
linux-kernel, linux-arm-kernel, rjw
On Wed, Jun 14, 2017 at 5:59 AM, Lorenzo Pieralisi
<lorenzo.pieralisi@arm.com> wrote:
> On Tue, Jun 13, 2017 at 01:56:44PM -0700, Khuong Dinh wrote:
>> Hi Lorenzo,
>>
>> On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi
>> <lorenzo.pieralisi@arm.com> wrote:
>> > On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote:
>> >> Hi Lorenzo,
>> >>
>> >> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
>> >> > Hi Lorenzo,
>> >> >
>> >> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
>> >> > <lorenzo.pieralisi@arm.com> wrote:
>> >> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>> >> >>> Hi Marc,
>> >> >>> There's no explicit dependency between pcie driver and msi
>> >> >>> controller. The current solution that we did is relying on the
>> >> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>> >> >>> higher priority in start ordering. This method could be applied in
>> >> >>> case of msi and pcie are the same level of subsys_init (in ACPI
>> >> >>> boot). However, PCIe driver has not supported for this dependency
>> >> >>> check yet. How do you think about this solution.
>> >> >>
>> >> >> First off, you can't post emails with confidentiality notices on
>> >> >> mailing lists so remove it from now onwards.
>> >> >
>> >> > Fixed
>> >> >
>> >> >> Secondly, I commented on this before, so you know what my opinion is.
>> >> >
>> >> > I got your opinion. I'll implement as your suggestion.
>> >> >
>> >>
>> >> Regarding to your previous suggestion to add a hook walking the ACPI
>> >> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
>> >> controllers must be probed before PCI. I have a concern that the
>> >> MSI namespace “ _SB.MSIX” is not a unique name and how can we walk
>> >> the ACPI namespace and use this name to make MSI probed before PCI.
>> >> May you have little bit more information for this or do you have
>> >> another suggestion for this?
>> >>
>> >> There’s another solution which makes this simpler and I’d like to
>> >> ask your opinion about this.
>> >> The solution is to make an hierarchy between MSI and PCI nodes as below:
>> >> Device(\_SB.MSIX) {
>> >> Device(\_SB.PCI0)
>> >> Device(\_SB.PCI1)
>> >> ……
>> >> }
>> >> In other word, MSIX node is a parent node of PCI nodes in ACPI
>> >> table. In this sense, there’s an explicit dependency between MSI
>> >> and PCI, MSI controller must be probed before PCI and it will
>> >> guarantee not breaking next kernel releases. How do you think about
>> >> this solution.
>> >
>> > I think that's a plaster as ineffective as reordering nodes, in short
>> > it is not a solution and you still rely on kernel link ordering, you
>> > can fiddle with ACPI tables as much as you want but that does not change.
>> >
>> >> I also tried to use _DEP method to describe the dependency of PCIe
>> >> nodes, but it looks that it does not work as PCI and MSI are handed
>> >> by acpi_bus_scan and still having issue when we re-probe PCI.
>> >
>> > That's a tad vague to be frank, anyway it is pretty clear to me that we
>> > have hit a wall. In ACPI there is no way to describe probe dependencies
>> > like the one you have, it is as simple as that, and this MSI issue you
>> > are facing is just an example, there are more eg:
>> >
>> > https://www.spinics.net/lists/arm-kernel/msg585825.html
>> >
>> > At the end of the day the choice is simple either we accept (and we
>> > maintain because that's the problem) these hacks - and I am not willing
>> > to do that - or we find a way to solve this from a general perspective not
>> > as a point hack.
>> >
>> > I can have a look at solving the whole issue but it won't happen
>> > tomorrow.
>>
>> Thanks for helping to resolve this general ACPI dependence issue.
>> I look forward for a version to test with.
>>
>> Can we separate the general dependence patch from the X-Gene MSI ACPI
>> enable patch.
>> This original patch was to enable ACPI support for PCIe MSI.
>> The dependence issue can be resolved separately.
>> Can you help to pull in the patch.
>
> Ok, pragmatically the only sane thing you can do is:
>
> (1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init())
> that carries out fwnode registration (ie register the fwnode by
> searching the MSI HID object - this does not depend on ACPI device
> nodes order - you enforce the order by parsing the namespace before
> ACPI core does it, at that point in time ACPI objects are created).
> Not sure Rafael will be happy to see this code but that's the only
> solution that does not rely on ACPI device nodes ordering and kernel
> link ordering. You may achieve the same by extending the ACPI APD
> code (drivers/acpi/acpi_apd.c) whether that's the best way forward is
> to be seen.
Hi Rafael,
Do you have any idea about this suggestion?
Otherwise, I'll follow this Lorenzo's approach (1).
> (2) You can also add an xgene specific scan handler at arch_init_call()
> but given that PCI root bridge is managed through a scan handler too you
> would rely on ACPI devices nodes ordering in the DSDT. Let me explain
> something for the benefit of everyone reading this thread: I do not want
> X-gene ACPI tables to force the kernel to scan devices in any order
> whatsover because this would mean we will _never_ be able to change the
> scan order if we wanted to lest we trigger regressions. So this is
> a non-option in my opinion.
>
> (3) Last option is to register the MSI fwnode either in PCI ECAM quirk
> handling or in arm64 before probing the PCI root bus.
>
> I am not keen on (2) and (3) at all, so _if_ we have to find a solution
> to this problem (1) is the only option you have for the time being as
> far as I am concerned.
Thank you very much for your suggestions, Lorenzo.
If there's not any other idea, I'll follow your approach (1)
Best regards,
Khuong Dinh
> Lorenzo
>
>>
>> Best regards,
>> Khuong Dinh
>>
>> > Thanks,
>> > Lorenzo
>> >
>> >> Thanks,
>> >> Khuong Dinh
>> >>
>> >> >> Finally, please execute this command on the vmlinux that "works"
>> >> >> for you:
>> >> >>
>> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>> >> >
>> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>> >> > ffff000008dab2d8 t __initcall_acpi_init4
>> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
>> >> >
>> >> > Best regards,
>> >> > Khuong Dinh
>> >> >
>> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
>> >> >> not understand how this works (I mean without relying on kernel link
>> >> >> order to ensure that the MSI driver is probed before PCI devices are
>> >> >> enumerated in acpi_init()).
>> >> >>
>> >> >> Thanks,
>> >> >> Lorenzo
>> >> >>
>> >> >>> Best regards,
>> >> >>> Khuong
>> >> >>>
>> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
>> >> >>> >> From: Khuong Dinh <kdinh@apm.com>
>> >> >>> >>
>> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>> >> >>> >>
>> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> >> >>> >> ---
>> >> >>> >> v2:
>> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>> >> >>> >> v1:
>> >> >>> >> - Initial version
>> >> >>> >> ---
>> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>> >> >>> >>
>> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> >> >>> >> index f1b633b..00aaa3d 100644
>> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>> >> >>> >> @@ -24,6 +24,7 @@
>> >> >>> >> #include <linux/pci.h>
>> >> >>> >> #include <linux/platform_device.h>
>> >> >>> >> #include <linux/of_pci.h>
>> >> >>> >> +#include <linux/acpi.h>
>> >> >>> >>
>> >> >>> >> #define MSI_IR0 0x000000
>> >> >>> >> #define MSI_INT0 0x800000
>> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> >> >>> >> };
>> >> >>> >>
>> >> >>> >> struct xgene_msi {
>> >> >>> >> - struct device_node *node;
>> >> >>> >> + struct fwnode_handle *fwnode;
>> >> >>> >> struct irq_domain *inner_domain;
>> >> >>> >> struct irq_domain *msi_domain;
>> >> >>> >> u64 msi_addr;
>> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> >> >>> >> .free = xgene_irq_domain_free,
>> >> >>> >> };
>> >> >>> >>
>> >> >>> >> +#ifdef CONFIG_ACPI
>> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> >> >>> >> +{
>> >> >>> >> + return xgene_msi_ctrl.fwnode;
>> >> >>> >> +}
>> >> >>> >> +#endif
>> >> >>> >> +
>> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> >>> >> {
>> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> >>> >> if (!msi->inner_domain)
>> >> >>> >> return -ENOMEM;
>> >> >>> >>
>> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> >> >>> >> &xgene_msi_domain_info,
>> >> >>> >> msi->inner_domain);
>> >> >>> >>
>> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> >>> >> return -ENOMEM;
>> >> >>> >> }
>> >> >>> >>
>> >> >>> >> +#ifdef CONFIG_ACPI
>> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> >> >>> >> +#endif
>> >> >>> >> return 0;
>> >> >>> >> }
>> >> >>> >>
>> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> >> >>> >> {},
>> >> >>> >> };
>> >> >>> >>
>> >> >>> >> +#ifdef CONFIG_ACPI
>> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> >> >>> >> + {"APMC0D0E", 0},
>> >> >>> >> + { },
>> >> >>> >> +};
>> >> >>> >> +#endif
>> >> >>> >> +
>> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
>> >> >>> >> {
>> >> >>> >> struct resource *res;
>> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> >>> >> goto error;
>> >> >>> >> }
>> >> >>> >> xgene_msi->msi_addr = res->start;
>> >> >>> >> - xgene_msi->node = pdev->dev.of_node;
>> >> >>> >> +
>> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> >> >>> >> + if (!xgene_msi->fwnode) {
>> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>> >> >>> >
>> >> >>> > Please provide something other than NULL, such as the base address if
>> >> >>> > the device. That's quite useful for debugging.
>> >> >>> >
>> >> >>> >> + if (!xgene_msi->fwnode) {
>> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> >> >>> >> + rc = ENOMEM;
>> >> >>> >> + goto error;
>> >> >>> >> + }
>> >> >>> >> + }
>> >> >>> >> +
>> >> >>> >> xgene_msi->num_cpus = num_possible_cpus();
>> >> >>> >>
>> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
>> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> >>> >> .driver = {
>> >> >>> >> .name = "xgene-msi",
>> >> >>> >> .of_match_table = xgene_msi_match_table,
>> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> >> >>> >> },
>> >> >>> >> .probe = xgene_msi_probe,
>> >> >>> >> .remove = xgene_msi_remove,
>> >> >>> >>
>> >> >>> >
>> >> >>> > The code is trivial, but relies on the MSI controller to probe before
>> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not
>> >> >>> > going to break in the next kernel release? As far as I can tell, there
>> >> >>> > is no explicit dependency between the two, making it the whole thing
>> >> >>> > extremely fragile.
>> >> >>> >
>> >> >>> > Thanks,
>> >> >>> >
>> >> >>> > M.
>> >> >>> > --
>> >> >>> > Jazz is not dead. It just smells funny...
>> >> >>>
>> >> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-14 17:43 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-14 17:43 UTC (permalink / raw)
To: Lorenzo Pieralisi
Cc: rjw, Marc Zyngier, linux-pci, msalter, patches, linux-kernel,
jcm, Bjorn Helgaas, linux-arm-kernel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^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-14 17:43 ` Khuong Dinh
0 siblings, 0 replies; 47+ messages in thread
From: Khuong Dinh @ 2017-06-14 17:43 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, Jun 14, 2017 at 5:59 AM, Lorenzo Pieralisi
<lorenzo.pieralisi@arm.com> wrote:
> On Tue, Jun 13, 2017 at 01:56:44PM -0700, Khuong Dinh wrote:
>> Hi Lorenzo,
>>
>> On Thu, Jun 8, 2017 at 3:39 AM, Lorenzo Pieralisi
>> <lorenzo.pieralisi@arm.com> wrote:
>> > On Tue, Jun 06, 2017 at 09:44:15AM -0700, Khuong Dinh wrote:
>> >> Hi Lorenzo,
>> >>
>> >> On Tue, May 9, 2017 at 3:55 PM, Khuong Dinh <kdinh@apm.com> wrote:
>> >> > Hi Lorenzo,
>> >> >
>> >> > On Fri, May 5, 2017 at 9:51 AM, Lorenzo Pieralisi
>> >> > <lorenzo.pieralisi@arm.com> wrote:
>> >> >> On Thu, May 04, 2017 at 05:36:06PM -0700, Khuong Dinh wrote:
>> >> >>> Hi Marc,
>> >> >>> There's no explicit dependency between pcie driver and msi
>> >> >>> controller. The current solution that we did is relying on the
>> >> >>> node ordering in BIOS. ACPI 5.0 introduced _DEP method to assign a
>> >> >>> higher priority in start ordering. This method could be applied in
>> >> >>> case of msi and pcie are the same level of subsys_init (in ACPI
>> >> >>> boot). However, PCIe driver has not supported for this dependency
>> >> >>> check yet. How do you think about this solution.
>> >> >>
>> >> >> First off, you can't post emails with confidentiality notices on
>> >> >> mailing lists so remove it from now onwards.
>> >> >
>> >> > Fixed
>> >> >
>> >> >> Secondly, I commented on this before, so you know what my opinion is.
>> >> >
>> >> > I got your opinion. I'll implement as your suggestion.
>> >> >
>> >>
>> >> Regarding to your previous suggestion to add a hook walking the ACPI
>> >> namespace before acpi_bus_scan() in acpi_scan_init() to make MSI
>> >> controllers must be probed before PCI. I have a concern that the
>> >> MSI namespace ? _SB.MSIX? is not a unique name and how can we walk
>> >> the ACPI namespace and use this name to make MSI probed before PCI.
>> >> May you have little bit more information for this or do you have
>> >> another suggestion for this?
>> >>
>> >> There?s another solution which makes this simpler and I?d like to
>> >> ask your opinion about this.
>> >> The solution is to make an hierarchy between MSI and PCI nodes as below:
>> >> Device(\_SB.MSIX) {
>> >> Device(\_SB.PCI0)
>> >> Device(\_SB.PCI1)
>> >> ??
>> >> }
>> >> In other word, MSIX node is a parent node of PCI nodes in ACPI
>> >> table. In this sense, there?s an explicit dependency between MSI
>> >> and PCI, MSI controller must be probed before PCI and it will
>> >> guarantee not breaking next kernel releases. How do you think about
>> >> this solution.
>> >
>> > I think that's a plaster as ineffective as reordering nodes, in short
>> > it is not a solution and you still rely on kernel link ordering, you
>> > can fiddle with ACPI tables as much as you want but that does not change.
>> >
>> >> I also tried to use _DEP method to describe the dependency of PCIe
>> >> nodes, but it looks that it does not work as PCI and MSI are handed
>> >> by acpi_bus_scan and still having issue when we re-probe PCI.
>> >
>> > That's a tad vague to be frank, anyway it is pretty clear to me that we
>> > have hit a wall. In ACPI there is no way to describe probe dependencies
>> > like the one you have, it is as simple as that, and this MSI issue you
>> > are facing is just an example, there are more eg:
>> >
>> > https://www.spinics.net/lists/arm-kernel/msg585825.html
>> >
>> > At the end of the day the choice is simple either we accept (and we
>> > maintain because that's the problem) these hacks - and I am not willing
>> > to do that - or we find a way to solve this from a general perspective not
>> > as a point hack.
>> >
>> > I can have a look at solving the whole issue but it won't happen
>> > tomorrow.
>>
>> Thanks for helping to resolve this general ACPI dependence issue.
>> I look forward for a version to test with.
>>
>> Can we separate the general dependence patch from the X-Gene MSI ACPI
>> enable patch.
>> This original patch was to enable ACPI support for PCIe MSI.
>> The dependence issue can be resolved separately.
>> Can you help to pull in the patch.
>
> Ok, pragmatically the only sane thing you can do is:
>
> (1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init())
> that carries out fwnode registration (ie register the fwnode by
> searching the MSI HID object - this does not depend on ACPI device
> nodes order - you enforce the order by parsing the namespace before
> ACPI core does it, at that point in time ACPI objects are created).
> Not sure Rafael will be happy to see this code but that's the only
> solution that does not rely on ACPI device nodes ordering and kernel
> link ordering. You may achieve the same by extending the ACPI APD
> code (drivers/acpi/acpi_apd.c) whether that's the best way forward is
> to be seen.
Hi Rafael,
Do you have any idea about this suggestion?
Otherwise, I'll follow this Lorenzo's approach (1).
> (2) You can also add an xgene specific scan handler at arch_init_call()
> but given that PCI root bridge is managed through a scan handler too you
> would rely on ACPI devices nodes ordering in the DSDT. Let me explain
> something for the benefit of everyone reading this thread: I do not want
> X-gene ACPI tables to force the kernel to scan devices in any order
> whatsover because this would mean we will _never_ be able to change the
> scan order if we wanted to lest we trigger regressions. So this is
> a non-option in my opinion.
>
> (3) Last option is to register the MSI fwnode either in PCI ECAM quirk
> handling or in arm64 before probing the PCI root bus.
>
> I am not keen on (2) and (3) at all, so _if_ we have to find a solution
> to this problem (1) is the only option you have for the time being as
> far as I am concerned.
Thank you very much for your suggestions, Lorenzo.
If there's not any other idea, I'll follow your approach (1)
Best regards,
Khuong Dinh
> Lorenzo
>
>>
>> Best regards,
>> Khuong Dinh
>>
>> > Thanks,
>> > Lorenzo
>> >
>> >> Thanks,
>> >> Khuong Dinh
>> >>
>> >> >> Finally, please execute this command on the vmlinux that "works"
>> >> >> for you:
>> >> >>
>> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>> >> >
>> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
>> >> > ffff000008dab2d8 t __initcall_acpi_init4
>> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
>> >> >
>> >> > Best regards,
>> >> > Khuong Dinh
>> >> >
>> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
>> >> >> not understand how this works (I mean without relying on kernel link
>> >> >> order to ensure that the MSI driver is probed before PCI devices are
>> >> >> enumerated in acpi_init()).
>> >> >>
>> >> >> Thanks,
>> >> >> Lorenzo
>> >> >>
>> >> >>> Best regards,
>> >> >>> Khuong
>> >> >>>
>> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
>> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
>> >> >>> >> From: Khuong Dinh <kdinh@apm.com>
>> >> >>> >>
>> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
>> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
>> >> >>> >>
>> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
>> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
>> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
>> >> >>> >> ---
>> >> >>> >> v2:
>> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
>> >> >>> >> v1:
>> >> >>> >> - Initial version
>> >> >>> >> ---
>> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
>> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
>> >> >>> >>
>> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
>> >> >>> >> index f1b633b..00aaa3d 100644
>> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
>> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
>> >> >>> >> @@ -24,6 +24,7 @@
>> >> >>> >> #include <linux/pci.h>
>> >> >>> >> #include <linux/platform_device.h>
>> >> >>> >> #include <linux/of_pci.h>
>> >> >>> >> +#include <linux/acpi.h>
>> >> >>> >>
>> >> >>> >> #define MSI_IR0 0x000000
>> >> >>> >> #define MSI_INT0 0x800000
>> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
>> >> >>> >> };
>> >> >>> >>
>> >> >>> >> struct xgene_msi {
>> >> >>> >> - struct device_node *node;
>> >> >>> >> + struct fwnode_handle *fwnode;
>> >> >>> >> struct irq_domain *inner_domain;
>> >> >>> >> struct irq_domain *msi_domain;
>> >> >>> >> u64 msi_addr;
>> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
>> >> >>> >> .free = xgene_irq_domain_free,
>> >> >>> >> };
>> >> >>> >>
>> >> >>> >> +#ifdef CONFIG_ACPI
>> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
>> >> >>> >> +{
>> >> >>> >> + return xgene_msi_ctrl.fwnode;
>> >> >>> >> +}
>> >> >>> >> +#endif
>> >> >>> >> +
>> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> >>> >> {
>> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
>> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> >>> >> if (!msi->inner_domain)
>> >> >>> >> return -ENOMEM;
>> >> >>> >>
>> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
>> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
>> >> >>> >> &xgene_msi_domain_info,
>> >> >>> >> msi->inner_domain);
>> >> >>> >>
>> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
>> >> >>> >> return -ENOMEM;
>> >> >>> >> }
>> >> >>> >>
>> >> >>> >> +#ifdef CONFIG_ACPI
>> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
>> >> >>> >> +#endif
>> >> >>> >> return 0;
>> >> >>> >> }
>> >> >>> >>
>> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
>> >> >>> >> {},
>> >> >>> >> };
>> >> >>> >>
>> >> >>> >> +#ifdef CONFIG_ACPI
>> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
>> >> >>> >> + {"APMC0D0E", 0},
>> >> >>> >> + { },
>> >> >>> >> +};
>> >> >>> >> +#endif
>> >> >>> >> +
>> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
>> >> >>> >> {
>> >> >>> >> struct resource *res;
>> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> >>> >> goto error;
>> >> >>> >> }
>> >> >>> >> xgene_msi->msi_addr = res->start;
>> >> >>> >> - xgene_msi->node = pdev->dev.of_node;
>> >> >>> >> +
>> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
>> >> >>> >> + if (!xgene_msi->fwnode) {
>> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
>> >> >>> >
>> >> >>> > Please provide something other than NULL, such as the base address if
>> >> >>> > the device. That's quite useful for debugging.
>> >> >>> >
>> >> >>> >> + if (!xgene_msi->fwnode) {
>> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
>> >> >>> >> + rc = ENOMEM;
>> >> >>> >> + goto error;
>> >> >>> >> + }
>> >> >>> >> + }
>> >> >>> >> +
>> >> >>> >> xgene_msi->num_cpus = num_possible_cpus();
>> >> >>> >>
>> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
>> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
>> >> >>> >> .driver = {
>> >> >>> >> .name = "xgene-msi",
>> >> >>> >> .of_match_table = xgene_msi_match_table,
>> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
>> >> >>> >> },
>> >> >>> >> .probe = xgene_msi_probe,
>> >> >>> >> .remove = xgene_msi_remove,
>> >> >>> >>
>> >> >>> >
>> >> >>> > The code is trivial, but relies on the MSI controller to probe before
>> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not
>> >> >>> > going to break in the next kernel release? As far as I can tell, there
>> >> >>> > is no explicit dependency between the two, making it the whole thing
>> >> >>> > extremely fragile.
>> >> >>> >
>> >> >>> > Thanks,
>> >> >>> >
>> >> >>> > M.
>> >> >>> > --
>> >> >>> > Jazz is not dead. It just smells funny...
>> >> >>>
>> >> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
2017-06-14 17:43 ` Khuong Dinh
(?)
@ 2017-06-15 11:06 ` Lorenzo Pieralisi
-1 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-15 11:06 UTC (permalink / raw)
To: Khuong Dinh
Cc: Marc Zyngier, msalter, Bjorn Helgaas, linux-pci, jcm, patches,
linux-kernel, linux-arm-kernel, rjw
On Wed, Jun 14, 2017 at 10:43:23AM -0700, Khuong Dinh wrote:
[...]
> >> Thanks for helping to resolve this general ACPI dependence issue.
> >> I look forward for a version to test with.
> >>
> >> Can we separate the general dependence patch from the X-Gene MSI ACPI
> >> enable patch.
> >> This original patch was to enable ACPI support for PCIe MSI.
> >> The dependence issue can be resolved separately.
> >> Can you help to pull in the patch.
> >
> > Ok, pragmatically the only sane thing you can do is:
> >
> > (1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init())
> > that carries out fwnode registration (ie register the fwnode by
> > searching the MSI HID object - this does not depend on ACPI device
> > nodes order - you enforce the order by parsing the namespace before
> > ACPI core does it, at that point in time ACPI objects are created).
> > Not sure Rafael will be happy to see this code but that's the only
> > solution that does not rely on ACPI device nodes ordering and kernel
> > link ordering. You may achieve the same by extending the ACPI APD
> > code (drivers/acpi/acpi_apd.c) whether that's the best way forward is
> > to be seen.
>
> Hi Rafael,
> Do you have any idea about this suggestion?
I mentioned drivers/acpi/acpi_apd.c because is in a way "similar" to
what you need but it is not exactly what's needed (I am not sure what
"devices" acpi_apd.c is meant to manage at the moment but certainly not
MSI controllers). In your callback from ACPI code you can't add a scan
handler - you have to get the ACPI device corresponding to the MSI
controller and register the IRQ domain fwnode.
Honestly it is hackish but that's the only approach that is robust
enough to always work (ie it does not depend on ACPI tables nodes order
and it does not depend on kernel link order).
Lorenzo
> Otherwise, I'll follow this Lorenzo's approach (1).
>
> > (2) You can also add an xgene specific scan handler at arch_init_call()
> > but given that PCI root bridge is managed through a scan handler too you
> > would rely on ACPI devices nodes ordering in the DSDT. Let me explain
> > something for the benefit of everyone reading this thread: I do not want
> > X-gene ACPI tables to force the kernel to scan devices in any order
> > whatsover because this would mean we will _never_ be able to change the
> > scan order if we wanted to lest we trigger regressions. So this is
> > a non-option in my opinion.
> >
> > (3) Last option is to register the MSI fwnode either in PCI ECAM quirk
> > handling or in arm64 before probing the PCI root bus.
> >
> > I am not keen on (2) and (3) at all, so _if_ we have to find a solution
> > to this problem (1) is the only option you have for the time being as
> > far as I am concerned.
>
> Thank you very much for your suggestions, Lorenzo.
> If there's not any other idea, I'll follow your approach (1)
>
> Best regards,
> Khuong Dinh
>
> > Lorenzo
> >
> >>
> >> Best regards,
> >> Khuong Dinh
> >>
> >> > Thanks,
> >> > Lorenzo
> >> >
> >> >> Thanks,
> >> >> Khuong Dinh
> >> >>
> >> >> >> Finally, please execute this command on the vmlinux that "works"
> >> >> >> for you:
> >> >> >>
> >> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> >> >
> >> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> >> > ffff000008dab2d8 t __initcall_acpi_init4
> >> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
> >> >> >
> >> >> > Best regards,
> >> >> > Khuong Dinh
> >> >> >
> >> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
> >> >> >> not understand how this works (I mean without relying on kernel link
> >> >> >> order to ensure that the MSI driver is probed before PCI devices are
> >> >> >> enumerated in acpi_init()).
> >> >> >>
> >> >> >> Thanks,
> >> >> >> Lorenzo
> >> >> >>
> >> >> >>> Best regards,
> >> >> >>> Khuong
> >> >> >>>
> >> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> >> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
> >> >> >>> >> From: Khuong Dinh <kdinh@apm.com>
> >> >> >>> >>
> >> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >> >> >>> >>
> >> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >> >> >>> >> ---
> >> >> >>> >> v2:
> >> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >> >> >>> >> v1:
> >> >> >>> >> - Initial version
> >> >> >>> >> ---
> >> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >> >> >>> >>
> >> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> index f1b633b..00aaa3d 100644
> >> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> @@ -24,6 +24,7 @@
> >> >> >>> >> #include <linux/pci.h>
> >> >> >>> >> #include <linux/platform_device.h>
> >> >> >>> >> #include <linux/of_pci.h>
> >> >> >>> >> +#include <linux/acpi.h>
> >> >> >>> >>
> >> >> >>> >> #define MSI_IR0 0x000000
> >> >> >>> >> #define MSI_INT0 0x800000
> >> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> struct xgene_msi {
> >> >> >>> >> - struct device_node *node;
> >> >> >>> >> + struct fwnode_handle *fwnode;
> >> >> >>> >> struct irq_domain *inner_domain;
> >> >> >>> >> struct irq_domain *msi_domain;
> >> >> >>> >> u64 msi_addr;
> >> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >> >> >>> >> .free = xgene_irq_domain_free,
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >> >> >>> >> +{
> >> >> >>> >> + return xgene_msi_ctrl.fwnode;
> >> >> >>> >> +}
> >> >> >>> >> +#endif
> >> >> >>> >> +
> >> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> {
> >> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> if (!msi->inner_domain)
> >> >> >>> >> return -ENOMEM;
> >> >> >>> >>
> >> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >> >> >>> >> &xgene_msi_domain_info,
> >> >> >>> >> msi->inner_domain);
> >> >> >>> >>
> >> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> return -ENOMEM;
> >> >> >>> >> }
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >> >> >>> >> +#endif
> >> >> >>> >> return 0;
> >> >> >>> >> }
> >> >> >>> >>
> >> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >> >> >>> >> {},
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >> >> >>> >> + {"APMC0D0E", 0},
> >> >> >>> >> + { },
> >> >> >>> >> +};
> >> >> >>> >> +#endif
> >> >> >>> >> +
> >> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> {
> >> >> >>> >> struct resource *res;
> >> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> goto error;
> >> >> >>> >> }
> >> >> >>> >> xgene_msi->msi_addr = res->start;
> >> >> >>> >> - xgene_msi->node = pdev->dev.of_node;
> >> >> >>> >> +
> >> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >> >> >>> >
> >> >> >>> > Please provide something other than NULL, such as the base address if
> >> >> >>> > the device. That's quite useful for debugging.
> >> >> >>> >
> >> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >> >> >>> >> + rc = ENOMEM;
> >> >> >>> >> + goto error;
> >> >> >>> >> + }
> >> >> >>> >> + }
> >> >> >>> >> +
> >> >> >>> >> xgene_msi->num_cpus = num_possible_cpus();
> >> >> >>> >>
> >> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
> >> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> .driver = {
> >> >> >>> >> .name = "xgene-msi",
> >> >> >>> >> .of_match_table = xgene_msi_match_table,
> >> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >> >> >>> >> },
> >> >> >>> >> .probe = xgene_msi_probe,
> >> >> >>> >> .remove = xgene_msi_remove,
> >> >> >>> >>
> >> >> >>> >
> >> >> >>> > The code is trivial, but relies on the MSI controller to probe before
> >> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not
> >> >> >>> > going to break in the next kernel release? As far as I can tell, there
> >> >> >>> > is no explicit dependency between the two, making it the whole thing
> >> >> >>> > extremely fragile.
> >> >> >>> >
> >> >> >>> > Thanks,
> >> >> >>> >
> >> >> >>> > M.
> >> >> >>> > --
> >> >> >>> > Jazz is not dead. It just smells funny...
> >> >> >>>
> >> >> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
* Re: [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-15 11:06 ` Lorenzo Pieralisi
0 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-15 11:06 UTC (permalink / raw)
To: Khuong Dinh
Cc: rjw, Marc Zyngier, linux-pci, msalter, patches, linux-kernel,
jcm, Bjorn Helgaas, linux-arm-kernel
On Wed, Jun 14, 2017 at 10:43:23AM -0700, Khuong Dinh wrote:
[...]
> >> Thanks for helping to resolve this general ACPI dependence issue.
> >> I look forward for a version to test with.
> >>
> >> Can we separate the general dependence patch from the X-Gene MSI ACPI
> >> enable patch.
> >> This original patch was to enable ACPI support for PCIe MSI.
> >> The dependence issue can be resolved separately.
> >> Can you help to pull in the patch.
> >
> > Ok, pragmatically the only sane thing you can do is:
> >
> > (1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init())
> > that carries out fwnode registration (ie register the fwnode by
> > searching the MSI HID object - this does not depend on ACPI device
> > nodes order - you enforce the order by parsing the namespace before
> > ACPI core does it, at that point in time ACPI objects are created).
> > Not sure Rafael will be happy to see this code but that's the only
> > solution that does not rely on ACPI device nodes ordering and kernel
> > link ordering. You may achieve the same by extending the ACPI APD
> > code (drivers/acpi/acpi_apd.c) whether that's the best way forward is
> > to be seen.
>
> Hi Rafael,
> Do you have any idea about this suggestion?
I mentioned drivers/acpi/acpi_apd.c because is in a way "similar" to
what you need but it is not exactly what's needed (I am not sure what
"devices" acpi_apd.c is meant to manage at the moment but certainly not
MSI controllers). In your callback from ACPI code you can't add a scan
handler - you have to get the ACPI device corresponding to the MSI
controller and register the IRQ domain fwnode.
Honestly it is hackish but that's the only approach that is robust
enough to always work (ie it does not depend on ACPI tables nodes order
and it does not depend on kernel link order).
Lorenzo
> Otherwise, I'll follow this Lorenzo's approach (1).
>
> > (2) You can also add an xgene specific scan handler at arch_init_call()
> > but given that PCI root bridge is managed through a scan handler too you
> > would rely on ACPI devices nodes ordering in the DSDT. Let me explain
> > something for the benefit of everyone reading this thread: I do not want
> > X-gene ACPI tables to force the kernel to scan devices in any order
> > whatsover because this would mean we will _never_ be able to change the
> > scan order if we wanted to lest we trigger regressions. So this is
> > a non-option in my opinion.
> >
> > (3) Last option is to register the MSI fwnode either in PCI ECAM quirk
> > handling or in arm64 before probing the PCI root bus.
> >
> > I am not keen on (2) and (3) at all, so _if_ we have to find a solution
> > to this problem (1) is the only option you have for the time being as
> > far as I am concerned.
>
> Thank you very much for your suggestions, Lorenzo.
> If there's not any other idea, I'll follow your approach (1)
>
> Best regards,
> Khuong Dinh
>
> > Lorenzo
> >
> >>
> >> Best regards,
> >> Khuong Dinh
> >>
> >> > Thanks,
> >> > Lorenzo
> >> >
> >> >> Thanks,
> >> >> Khuong Dinh
> >> >>
> >> >> >> Finally, please execute this command on the vmlinux that "works"
> >> >> >> for you:
> >> >> >>
> >> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> >> >
> >> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> >> > ffff000008dab2d8 t __initcall_acpi_init4
> >> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
> >> >> >
> >> >> > Best regards,
> >> >> > Khuong Dinh
> >> >> >
> >> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
> >> >> >> not understand how this works (I mean without relying on kernel link
> >> >> >> order to ensure that the MSI driver is probed before PCI devices are
> >> >> >> enumerated in acpi_init()).
> >> >> >>
> >> >> >> Thanks,
> >> >> >> Lorenzo
> >> >> >>
> >> >> >>> Best regards,
> >> >> >>> Khuong
> >> >> >>>
> >> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> >> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
> >> >> >>> >> From: Khuong Dinh <kdinh@apm.com>
> >> >> >>> >>
> >> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >> >> >>> >>
> >> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >> >> >>> >> ---
> >> >> >>> >> v2:
> >> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >> >> >>> >> v1:
> >> >> >>> >> - Initial version
> >> >> >>> >> ---
> >> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >> >> >>> >>
> >> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> index f1b633b..00aaa3d 100644
> >> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> @@ -24,6 +24,7 @@
> >> >> >>> >> #include <linux/pci.h>
> >> >> >>> >> #include <linux/platform_device.h>
> >> >> >>> >> #include <linux/of_pci.h>
> >> >> >>> >> +#include <linux/acpi.h>
> >> >> >>> >>
> >> >> >>> >> #define MSI_IR0 0x000000
> >> >> >>> >> #define MSI_INT0 0x800000
> >> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> struct xgene_msi {
> >> >> >>> >> - struct device_node *node;
> >> >> >>> >> + struct fwnode_handle *fwnode;
> >> >> >>> >> struct irq_domain *inner_domain;
> >> >> >>> >> struct irq_domain *msi_domain;
> >> >> >>> >> u64 msi_addr;
> >> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >> >> >>> >> .free = xgene_irq_domain_free,
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >> >> >>> >> +{
> >> >> >>> >> + return xgene_msi_ctrl.fwnode;
> >> >> >>> >> +}
> >> >> >>> >> +#endif
> >> >> >>> >> +
> >> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> {
> >> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> if (!msi->inner_domain)
> >> >> >>> >> return -ENOMEM;
> >> >> >>> >>
> >> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >> >> >>> >> &xgene_msi_domain_info,
> >> >> >>> >> msi->inner_domain);
> >> >> >>> >>
> >> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> return -ENOMEM;
> >> >> >>> >> }
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >> >> >>> >> +#endif
> >> >> >>> >> return 0;
> >> >> >>> >> }
> >> >> >>> >>
> >> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >> >> >>> >> {},
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >> >> >>> >> + {"APMC0D0E", 0},
> >> >> >>> >> + { },
> >> >> >>> >> +};
> >> >> >>> >> +#endif
> >> >> >>> >> +
> >> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> {
> >> >> >>> >> struct resource *res;
> >> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> goto error;
> >> >> >>> >> }
> >> >> >>> >> xgene_msi->msi_addr = res->start;
> >> >> >>> >> - xgene_msi->node = pdev->dev.of_node;
> >> >> >>> >> +
> >> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >> >> >>> >
> >> >> >>> > Please provide something other than NULL, such as the base address if
> >> >> >>> > the device. That's quite useful for debugging.
> >> >> >>> >
> >> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >> >> >>> >> + rc = ENOMEM;
> >> >> >>> >> + goto error;
> >> >> >>> >> + }
> >> >> >>> >> + }
> >> >> >>> >> +
> >> >> >>> >> xgene_msi->num_cpus = num_possible_cpus();
> >> >> >>> >>
> >> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
> >> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> .driver = {
> >> >> >>> >> .name = "xgene-msi",
> >> >> >>> >> .of_match_table = xgene_msi_match_table,
> >> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >> >> >>> >> },
> >> >> >>> >> .probe = xgene_msi_probe,
> >> >> >>> >> .remove = xgene_msi_remove,
> >> >> >>> >>
> >> >> >>> >
> >> >> >>> > The code is trivial, but relies on the MSI controller to probe before
> >> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not
> >> >> >>> > going to break in the next kernel release? As far as I can tell, there
> >> >> >>> > is no explicit dependency between the two, making it the whole thing
> >> >> >>> > extremely fragile.
> >> >> >>> >
> >> >> >>> > Thanks,
> >> >> >>> >
> >> >> >>> > M.
> >> >> >>> > --
> >> >> >>> > Jazz is not dead. It just smells funny...
> >> >> >>>
> >> >> >>> --
_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
^ permalink raw reply [flat|nested] 47+ messages in thread
* [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1
@ 2017-06-15 11:06 ` Lorenzo Pieralisi
0 siblings, 0 replies; 47+ messages in thread
From: Lorenzo Pieralisi @ 2017-06-15 11:06 UTC (permalink / raw)
To: linux-arm-kernel
On Wed, Jun 14, 2017 at 10:43:23AM -0700, Khuong Dinh wrote:
[...]
> >> Thanks for helping to resolve this general ACPI dependence issue.
> >> I look forward for a version to test with.
> >>
> >> Can we separate the general dependence patch from the X-Gene MSI ACPI
> >> enable patch.
> >> This original patch was to enable ACPI support for PCIe MSI.
> >> The dependence issue can be resolved separately.
> >> Can you help to pull in the patch.
> >
> > Ok, pragmatically the only sane thing you can do is:
> >
> > (1) Add an X-gene specific hook in drivers/acpi/scan.c (acpi_scan_init())
> > that carries out fwnode registration (ie register the fwnode by
> > searching the MSI HID object - this does not depend on ACPI device
> > nodes order - you enforce the order by parsing the namespace before
> > ACPI core does it, at that point in time ACPI objects are created).
> > Not sure Rafael will be happy to see this code but that's the only
> > solution that does not rely on ACPI device nodes ordering and kernel
> > link ordering. You may achieve the same by extending the ACPI APD
> > code (drivers/acpi/acpi_apd.c) whether that's the best way forward is
> > to be seen.
>
> Hi Rafael,
> Do you have any idea about this suggestion?
I mentioned drivers/acpi/acpi_apd.c because is in a way "similar" to
what you need but it is not exactly what's needed (I am not sure what
"devices" acpi_apd.c is meant to manage at the moment but certainly not
MSI controllers). In your callback from ACPI code you can't add a scan
handler - you have to get the ACPI device corresponding to the MSI
controller and register the IRQ domain fwnode.
Honestly it is hackish but that's the only approach that is robust
enough to always work (ie it does not depend on ACPI tables nodes order
and it does not depend on kernel link order).
Lorenzo
> Otherwise, I'll follow this Lorenzo's approach (1).
>
> > (2) You can also add an xgene specific scan handler at arch_init_call()
> > but given that PCI root bridge is managed through a scan handler too you
> > would rely on ACPI devices nodes ordering in the DSDT. Let me explain
> > something for the benefit of everyone reading this thread: I do not want
> > X-gene ACPI tables to force the kernel to scan devices in any order
> > whatsover because this would mean we will _never_ be able to change the
> > scan order if we wanted to lest we trigger regressions. So this is
> > a non-option in my opinion.
> >
> > (3) Last option is to register the MSI fwnode either in PCI ECAM quirk
> > handling or in arm64 before probing the PCI root bus.
> >
> > I am not keen on (2) and (3) at all, so _if_ we have to find a solution
> > to this problem (1) is the only option you have for the time being as
> > far as I am concerned.
>
> Thank you very much for your suggestions, Lorenzo.
> If there's not any other idea, I'll follow your approach (1)
>
> Best regards,
> Khuong Dinh
>
> > Lorenzo
> >
> >>
> >> Best regards,
> >> Khuong Dinh
> >>
> >> > Thanks,
> >> > Lorenzo
> >> >
> >> >> Thanks,
> >> >> Khuong Dinh
> >> >>
> >> >> >> Finally, please execute this command on the vmlinux that "works"
> >> >> >> for you:
> >> >> >>
> >> >> >> nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> >> >
> >> >> > $ nm vmlinux | grep -E '__initcall_(xgene_pcie_msi_init|acpi_init)'
> >> >> > ffff000008dab2d8 t __initcall_acpi_init4
> >> >> > ffff000008dab2c8 t __initcall_xgene_pcie_msi_init4
> >> >> >
> >> >> > Best regards,
> >> >> > Khuong Dinh
> >> >> >
> >> >> >> Even by ordering devices in the ACPI tables (that I abhor) I still do
> >> >> >> not understand how this works (I mean without relying on kernel link
> >> >> >> order to ensure that the MSI driver is probed before PCI devices are
> >> >> >> enumerated in acpi_init()).
> >> >> >>
> >> >> >> Thanks,
> >> >> >> Lorenzo
> >> >> >>
> >> >> >>> Best regards,
> >> >> >>> Khuong
> >> >> >>>
> >> >> >>> On Fri, Apr 28, 2017 at 2:27 AM, Marc Zyngier <marc.zyngier@arm.com> wrote:
> >> >> >>> > On 28/04/17 01:54, Khuong Dinh wrote:
> >> >> >>> >> From: Khuong Dinh <kdinh@apm.com>
> >> >> >>> >>
> >> >> >>> >> This patch makes pci-xgene-msi driver ACPI-aware and provides
> >> >> >>> >> MSI capability for X-Gene v1 PCIe controllers in ACPI boot mode.
> >> >> >>> >>
> >> >> >>> >> Signed-off-by: Khuong Dinh <kdinh@apm.com>
> >> >> >>> >> Signed-off-by: Duc Dang <dhdang@apm.com>
> >> >> >>> >> Acked-by: Marc Zyngier <marc.zyngier@arm.com>
> >> >> >>> >> ---
> >> >> >>> >> v2:
> >> >> >>> >> - Verify with BIOS version 3.06.25 and 3.07.09
> >> >> >>> >> v1:
> >> >> >>> >> - Initial version
> >> >> >>> >> ---
> >> >> >>> >> drivers/pci/host/pci-xgene-msi.c | 35 ++++++++++++++++++++++++++++++++---
> >> >> >>> >> 1 files changed, 32 insertions(+), 3 deletions(-)
> >> >> >>> >>
> >> >> >>> >> diff --git a/drivers/pci/host/pci-xgene-msi.c b/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> index f1b633b..00aaa3d 100644
> >> >> >>> >> --- a/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> +++ b/drivers/pci/host/pci-xgene-msi.c
> >> >> >>> >> @@ -24,6 +24,7 @@
> >> >> >>> >> #include <linux/pci.h>
> >> >> >>> >> #include <linux/platform_device.h>
> >> >> >>> >> #include <linux/of_pci.h>
> >> >> >>> >> +#include <linux/acpi.h>
> >> >> >>> >>
> >> >> >>> >> #define MSI_IR0 0x000000
> >> >> >>> >> #define MSI_INT0 0x800000
> >> >> >>> >> @@ -39,7 +40,7 @@ struct xgene_msi_group {
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> struct xgene_msi {
> >> >> >>> >> - struct device_node *node;
> >> >> >>> >> + struct fwnode_handle *fwnode;
> >> >> >>> >> struct irq_domain *inner_domain;
> >> >> >>> >> struct irq_domain *msi_domain;
> >> >> >>> >> u64 msi_addr;
> >> >> >>> >> @@ -249,6 +250,13 @@ static void xgene_irq_domain_free(struct irq_domain *domain,
> >> >> >>> >> .free = xgene_irq_domain_free,
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> +static struct fwnode_handle *xgene_msi_get_fwnode(struct device *dev)
> >> >> >>> >> +{
> >> >> >>> >> + return xgene_msi_ctrl.fwnode;
> >> >> >>> >> +}
> >> >> >>> >> +#endif
> >> >> >>> >> +
> >> >> >>> >> static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> {
> >> >> >>> >> msi->inner_domain = irq_domain_add_linear(NULL, NR_MSI_VEC,
> >> >> >>> >> @@ -256,7 +264,7 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> if (!msi->inner_domain)
> >> >> >>> >> return -ENOMEM;
> >> >> >>> >>
> >> >> >>> >> - msi->msi_domain = pci_msi_create_irq_domain(of_node_to_fwnode(msi->node),
> >> >> >>> >> + msi->msi_domain = pci_msi_create_irq_domain(msi->fwnode,
> >> >> >>> >> &xgene_msi_domain_info,
> >> >> >>> >> msi->inner_domain);
> >> >> >>> >>
> >> >> >>> >> @@ -265,6 +273,9 @@ static int xgene_allocate_domains(struct xgene_msi *msi)
> >> >> >>> >> return -ENOMEM;
> >> >> >>> >> }
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> + pci_msi_register_fwnode_provider(&xgene_msi_get_fwnode);
> >> >> >>> >> +#endif
> >> >> >>> >> return 0;
> >> >> >>> >> }
> >> >> >>> >>
> >> >> >>> >> @@ -449,6 +460,13 @@ static int xgene_msi_hwirq_free(unsigned int cpu)
> >> >> >>> >> {},
> >> >> >>> >> };
> >> >> >>> >>
> >> >> >>> >> +#ifdef CONFIG_ACPI
> >> >> >>> >> +static const struct acpi_device_id xgene_msi_acpi_ids[] = {
> >> >> >>> >> + {"APMC0D0E", 0},
> >> >> >>> >> + { },
> >> >> >>> >> +};
> >> >> >>> >> +#endif
> >> >> >>> >> +
> >> >> >>> >> static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> {
> >> >> >>> >> struct resource *res;
> >> >> >>> >> @@ -469,7 +487,17 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> goto error;
> >> >> >>> >> }
> >> >> >>> >> xgene_msi->msi_addr = res->start;
> >> >> >>> >> - xgene_msi->node = pdev->dev.of_node;
> >> >> >>> >> +
> >> >> >>> >> + xgene_msi->fwnode = of_node_to_fwnode(pdev->dev.of_node);
> >> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >> >>> >> + xgene_msi->fwnode = irq_domain_alloc_fwnode(NULL);
> >> >> >>> >
> >> >> >>> > Please provide something other than NULL, such as the base address if
> >> >> >>> > the device. That's quite useful for debugging.
> >> >> >>> >
> >> >> >>> >> + if (!xgene_msi->fwnode) {
> >> >> >>> >> + dev_err(&pdev->dev, "Failed to create fwnode\n");
> >> >> >>> >> + rc = ENOMEM;
> >> >> >>> >> + goto error;
> >> >> >>> >> + }
> >> >> >>> >> + }
> >> >> >>> >> +
> >> >> >>> >> xgene_msi->num_cpus = num_possible_cpus();
> >> >> >>> >>
> >> >> >>> >> rc = xgene_msi_init_allocator(xgene_msi);
> >> >> >>> >> @@ -540,6 +568,7 @@ static int xgene_msi_probe(struct platform_device *pdev)
> >> >> >>> >> .driver = {
> >> >> >>> >> .name = "xgene-msi",
> >> >> >>> >> .of_match_table = xgene_msi_match_table,
> >> >> >>> >> + .acpi_match_table = ACPI_PTR(xgene_msi_acpi_ids),
> >> >> >>> >> },
> >> >> >>> >> .probe = xgene_msi_probe,
> >> >> >>> >> .remove = xgene_msi_remove,
> >> >> >>> >>
> >> >> >>> >
> >> >> >>> > The code is trivial, but relies on the MSI controller to probe before
> >> >> >>> > the PCI bus. What enforces this? How is it making sure that this is not
> >> >> >>> > going to break in the next kernel release? As far as I can tell, there
> >> >> >>> > is no explicit dependency between the two, making it the whole thing
> >> >> >>> > extremely fragile.
> >> >> >>> >
> >> >> >>> > Thanks,
> >> >> >>> >
> >> >> >>> > M.
> >> >> >>> > --
> >> >> >>> > Jazz is not dead. It just smells funny...
> >> >> >>>
> >> >> >>> --
^ permalink raw reply [flat|nested] 47+ messages in thread
end of thread, other threads:[~2017-06-15 11:06 UTC | newest]
Thread overview: 47+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-04-28 0:54 [PATCH v2 pci] PCI/MSI: pci-xgene-msi: Enable MSI support in ACPI boot for X-Gene v1 Khuong Dinh
2017-04-28 0:54 ` Khuong Dinh
2017-04-28 0:54 ` Khuong Dinh
2017-04-28 5:11 ` Jon Masters
2017-04-28 5:11 ` Jon Masters
2017-04-28 5:11 ` Jon Masters
2017-04-28 16:36 ` Jon Masters
2017-04-28 16:36 ` Jon Masters
2017-04-28 16:36 ` Jon Masters
2017-05-04 23:24 ` Jon Masters
2017-05-04 23:24 ` Jon Masters
2017-05-04 23:24 ` Jon Masters
2017-04-28 9:27 ` Marc Zyngier
2017-04-28 9:27 ` Marc Zyngier
2017-04-28 9:27 ` Marc Zyngier
2017-05-05 0:36 ` Khuong Dinh
2017-05-05 0:36 ` Khuong Dinh
2017-05-05 0:36 ` Khuong Dinh
2017-05-05 16:51 ` Lorenzo Pieralisi
2017-05-05 16:51 ` Lorenzo Pieralisi
2017-05-05 16:51 ` Lorenzo Pieralisi
2017-05-09 22:55 ` Khuong Dinh
2017-05-09 22:55 ` Khuong Dinh
2017-05-09 22:55 ` Khuong Dinh
2017-06-06 16:44 ` Khuong Dinh
2017-06-06 16:44 ` Khuong Dinh
2017-06-06 16:44 ` Khuong Dinh
2017-06-08 10:39 ` Lorenzo Pieralisi
2017-06-08 10:39 ` Lorenzo Pieralisi
2017-06-08 10:39 ` Lorenzo Pieralisi
2017-06-13 20:56 ` Khuong Dinh
2017-06-13 20:56 ` Khuong Dinh
2017-06-13 20:56 ` Khuong Dinh
2017-06-14 12:59 ` Lorenzo Pieralisi
2017-06-14 12:59 ` Lorenzo Pieralisi
2017-06-14 12:59 ` Lorenzo Pieralisi
2017-06-14 17:43 ` Khuong Dinh
2017-06-14 17:43 ` Khuong Dinh
2017-06-14 17:43 ` Khuong Dinh
2017-06-15 11:06 ` Lorenzo Pieralisi
2017-06-15 11:06 ` Lorenzo Pieralisi
2017-06-15 11:06 ` Lorenzo Pieralisi
2017-05-31 11:29 ` Jonathan Toppins
2017-05-31 11:29 ` Jonathan Toppins
2017-05-31 16:13 ` Loc Ho
2017-05-31 16:13 ` Loc Ho
2017-05-31 16:13 ` Loc Ho
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.