* [PATCH v2 0/7] hw/cxl: RAS error emulation and injection
@ 2023-01-20 14:24 ` Jonathan Cameron
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
v2: Thanks to Mike Maslenkin for review.
- Fix wrong parameter type to ct3d_qmp_cor_err_to_cxl()
- Rework use of CXLError local variable in ct3d_reg_write() to improve
code readability.
CXL error reporting is complex. This series only covers the protocol
related errors reported via PCIE AER - Ira Weiny has posted support for
Event log based injection and I will post an update of Poison list injection
shortly. My proposal is to upstream this one first, followed by Ira's Event
Log series, then finally the Poison List handling. That is based on likely
order of Linux kernel support (the support for this type of error reporting
went in during the recent merge window, the others are still under review).
Note we may propose other non error related features in between!
The current revisions of all the error injection can be found at:
https://gitlab.com/jic23/qemu/-/tree/cxl-2023-01-11
In order to test the kernel support for RAS error handling, I previously
provided this series via gitlab, enabling David Jiang's kernel patches
to be tested.
Now that Linux kernel support is upstream, this series is proposing the
support for upstream inclusion in QEMU. Note that support for Multiple
Header Recording has been added to QEMU the meantime and a kernel
patch to use that feature sent out.
https://lore.kernel.org/linux-cxl/20230113154058.16227-1-Jonathan.Cameron@huawei.com/T/#t
There are two generic PCI AER precursor feature additions.
1) The PCI_ERR_UCOR_MASK register has not been implemented until now
and is necessary for correct emulation.
2) The routing for AER errors, via existing AER error injection, only
covered one of two paths given in the PCIe base specification,
unfortunately not the one used by the Linux kernel CXL support.
The use of MSI for the CXL root ports, both makes sense from the point
of view of how it may well be implemented, and works around the documented
lack of PCI interrupt routing in i386/q35. I have a hack that lets
us correctly route those interrupts but don't currently plan to post it.
The actual CXL error injection uses a new QMP interface as documented
in the final patch description. The existing AER error injection
internals are reused though it's HMP interface is not.
Injection via QMP:
{ "execute": "qmp_capabilities" }
...
{ "execute": "cxl-inject-uncorrectable-errors",
"arguments": {
"path": "/machine/peripheral/cxl-pmem0",
"errors": [
{
"type": "cache-address-parity",
"header": [ 3, 4]
},
{
"type": "cache-data-parity",
"header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
},
{
"type": "internal",
"header": [ 1, 2, 4]
}
]
}}
...
{ "execute": "cxl-inject-correctable-error",
"arguments": {
"path": "/machine/peripheral/cxl-pmem0",
"type": "physical",
"header": [ 3, 4]
} }
Based on top of:
https://lore.kernel.org/all/20230112102644.27830-1-Jonathan.Cameron@huawei.com/
[PATCH v2 0/8] hw/cxl: CXL emulation cleanups and minor fixes for upstream
Jonathan Cameron (7):
hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register
hw/pci/aer: Add missing routing for AER errors
hw/pci-bridge/cxl_root_port: Wire up AER
hw/pci-bridge/cxl_root_port: Wire up MSI
hw/mem/cxl-type3: Add AER extended capability
hw/pci/aer: Make PCIE AER error injection facility available for other
emulation to use.
hw/mem/cxl_type3: Add CXL RAS Error Injection Support.
hw/cxl/cxl-component-utils.c | 4 +-
hw/mem/cxl_type3.c | 303 +++++++++++++++++++++++++++++++++
hw/mem/cxl_type3_stubs.c | 10 ++
hw/mem/meson.build | 2 +
hw/pci-bridge/cxl_root_port.c | 64 +++++++
hw/pci/pci-internal.h | 1 -
hw/pci/pcie_aer.c | 14 +-
include/hw/cxl/cxl_component.h | 26 +++
include/hw/cxl/cxl_device.h | 11 ++
include/hw/pci/pcie_aer.h | 1 +
include/hw/pci/pcie_regs.h | 3 +
qapi/cxl.json | 113 ++++++++++++
qapi/meson.build | 1 +
qapi/qapi-schema.json | 1 +
14 files changed, 551 insertions(+), 3 deletions(-)
create mode 100644 hw/mem/cxl_type3_stubs.c
create mode 100644 qapi/cxl.json
--
2.37.2
^ permalink raw reply [flat|nested] 21+ messages in thread
* [PATCH v2 0/7] hw/cxl: RAS error emulation and injection
@ 2023-01-20 14:24 ` Jonathan Cameron
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
v2: Thanks to Mike Maslenkin for review.
- Fix wrong parameter type to ct3d_qmp_cor_err_to_cxl()
- Rework use of CXLError local variable in ct3d_reg_write() to improve
code readability.
CXL error reporting is complex. This series only covers the protocol
related errors reported via PCIE AER - Ira Weiny has posted support for
Event log based injection and I will post an update of Poison list injection
shortly. My proposal is to upstream this one first, followed by Ira's Event
Log series, then finally the Poison List handling. That is based on likely
order of Linux kernel support (the support for this type of error reporting
went in during the recent merge window, the others are still under review).
Note we may propose other non error related features in between!
The current revisions of all the error injection can be found at:
https://gitlab.com/jic23/qemu/-/tree/cxl-2023-01-11
In order to test the kernel support for RAS error handling, I previously
provided this series via gitlab, enabling David Jiang's kernel patches
to be tested.
Now that Linux kernel support is upstream, this series is proposing the
support for upstream inclusion in QEMU. Note that support for Multiple
Header Recording has been added to QEMU the meantime and a kernel
patch to use that feature sent out.
https://lore.kernel.org/linux-cxl/20230113154058.16227-1-Jonathan.Cameron@huawei.com/T/#t
There are two generic PCI AER precursor feature additions.
1) The PCI_ERR_UCOR_MASK register has not been implemented until now
and is necessary for correct emulation.
2) The routing for AER errors, via existing AER error injection, only
covered one of two paths given in the PCIe base specification,
unfortunately not the one used by the Linux kernel CXL support.
The use of MSI for the CXL root ports, both makes sense from the point
of view of how it may well be implemented, and works around the documented
lack of PCI interrupt routing in i386/q35. I have a hack that lets
us correctly route those interrupts but don't currently plan to post it.
The actual CXL error injection uses a new QMP interface as documented
in the final patch description. The existing AER error injection
internals are reused though it's HMP interface is not.
Injection via QMP:
{ "execute": "qmp_capabilities" }
...
{ "execute": "cxl-inject-uncorrectable-errors",
"arguments": {
"path": "/machine/peripheral/cxl-pmem0",
"errors": [
{
"type": "cache-address-parity",
"header": [ 3, 4]
},
{
"type": "cache-data-parity",
"header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
},
{
"type": "internal",
"header": [ 1, 2, 4]
}
]
}}
...
{ "execute": "cxl-inject-correctable-error",
"arguments": {
"path": "/machine/peripheral/cxl-pmem0",
"type": "physical",
"header": [ 3, 4]
} }
Based on top of:
https://lore.kernel.org/all/20230112102644.27830-1-Jonathan.Cameron@huawei.com/
[PATCH v2 0/8] hw/cxl: CXL emulation cleanups and minor fixes for upstream
Jonathan Cameron (7):
hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register
hw/pci/aer: Add missing routing for AER errors
hw/pci-bridge/cxl_root_port: Wire up AER
hw/pci-bridge/cxl_root_port: Wire up MSI
hw/mem/cxl-type3: Add AER extended capability
hw/pci/aer: Make PCIE AER error injection facility available for other
emulation to use.
hw/mem/cxl_type3: Add CXL RAS Error Injection Support.
hw/cxl/cxl-component-utils.c | 4 +-
hw/mem/cxl_type3.c | 303 +++++++++++++++++++++++++++++++++
hw/mem/cxl_type3_stubs.c | 10 ++
hw/mem/meson.build | 2 +
hw/pci-bridge/cxl_root_port.c | 64 +++++++
hw/pci/pci-internal.h | 1 -
hw/pci/pcie_aer.c | 14 +-
include/hw/cxl/cxl_component.h | 26 +++
include/hw/cxl/cxl_device.h | 11 ++
include/hw/pci/pcie_aer.h | 1 +
include/hw/pci/pcie_regs.h | 3 +
qapi/cxl.json | 113 ++++++++++++
qapi/meson.build | 1 +
qapi/qapi-schema.json | 1 +
14 files changed, 551 insertions(+), 3 deletions(-)
create mode 100644 hw/mem/cxl_type3_stubs.c
create mode 100644 qapi/cxl.json
--
2.37.2
^ permalink raw reply [flat|nested] 21+ messages in thread
* [PATCH v2 1/7] hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register
2023-01-20 14:24 ` Jonathan Cameron
@ 2023-01-20 14:24 ` Jonathan Cameron via
-1 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
This register in AER should be both writeable and should
have a default value with a couple of the errors masked
including the Uncorrectable Internal Error used by CXL for
it's error reporting.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci/pcie_aer.c | 4 ++++
include/hw/pci/pcie_regs.h | 3 +++
2 files changed, 7 insertions(+)
diff --git a/hw/pci/pcie_aer.c b/hw/pci/pcie_aer.c
index 9a19be44ae..909e027d99 100644
--- a/hw/pci/pcie_aer.c
+++ b/hw/pci/pcie_aer.c
@@ -112,6 +112,10 @@ int pcie_aer_init(PCIDevice *dev, uint8_t cap_ver, uint16_t offset,
pci_set_long(dev->w1cmask + offset + PCI_ERR_UNCOR_STATUS,
PCI_ERR_UNC_SUPPORTED);
+ pci_set_long(dev->config + offset + PCI_ERR_UNCOR_MASK,
+ PCI_ERR_UNC_MASK_DEFAULT);
+ pci_set_long(dev->wmask + offset + PCI_ERR_UNCOR_MASK,
+ PCI_ERR_UNC_SUPPORTED);
pci_set_long(dev->config + offset + PCI_ERR_UNCOR_SEVER,
PCI_ERR_UNC_SEVERITY_DEFAULT);
diff --git a/include/hw/pci/pcie_regs.h b/include/hw/pci/pcie_regs.h
index 963dc2e170..6ec4785448 100644
--- a/include/hw/pci/pcie_regs.h
+++ b/include/hw/pci/pcie_regs.h
@@ -155,6 +155,9 @@ typedef enum PCIExpLinkWidth {
PCI_ERR_UNC_ATOP_EBLOCKED | \
PCI_ERR_UNC_TLP_PRF_BLOCKED)
+#define PCI_ERR_UNC_MASK_DEFAULT (PCI_ERR_UNC_INTN | \
+ PCI_ERR_UNC_TLP_PRF_BLOCKED)
+
#define PCI_ERR_UNC_SEVERITY_DEFAULT (PCI_ERR_UNC_DLP | \
PCI_ERR_UNC_SDN | \
PCI_ERR_UNC_FCP | \
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 1/7] hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register
@ 2023-01-20 14:24 ` Jonathan Cameron via
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
This register in AER should be both writeable and should
have a default value with a couple of the errors masked
including the Uncorrectable Internal Error used by CXL for
it's error reporting.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci/pcie_aer.c | 4 ++++
include/hw/pci/pcie_regs.h | 3 +++
2 files changed, 7 insertions(+)
diff --git a/hw/pci/pcie_aer.c b/hw/pci/pcie_aer.c
index 9a19be44ae..909e027d99 100644
--- a/hw/pci/pcie_aer.c
+++ b/hw/pci/pcie_aer.c
@@ -112,6 +112,10 @@ int pcie_aer_init(PCIDevice *dev, uint8_t cap_ver, uint16_t offset,
pci_set_long(dev->w1cmask + offset + PCI_ERR_UNCOR_STATUS,
PCI_ERR_UNC_SUPPORTED);
+ pci_set_long(dev->config + offset + PCI_ERR_UNCOR_MASK,
+ PCI_ERR_UNC_MASK_DEFAULT);
+ pci_set_long(dev->wmask + offset + PCI_ERR_UNCOR_MASK,
+ PCI_ERR_UNC_SUPPORTED);
pci_set_long(dev->config + offset + PCI_ERR_UNCOR_SEVER,
PCI_ERR_UNC_SEVERITY_DEFAULT);
diff --git a/include/hw/pci/pcie_regs.h b/include/hw/pci/pcie_regs.h
index 963dc2e170..6ec4785448 100644
--- a/include/hw/pci/pcie_regs.h
+++ b/include/hw/pci/pcie_regs.h
@@ -155,6 +155,9 @@ typedef enum PCIExpLinkWidth {
PCI_ERR_UNC_ATOP_EBLOCKED | \
PCI_ERR_UNC_TLP_PRF_BLOCKED)
+#define PCI_ERR_UNC_MASK_DEFAULT (PCI_ERR_UNC_INTN | \
+ PCI_ERR_UNC_TLP_PRF_BLOCKED)
+
#define PCI_ERR_UNC_SEVERITY_DEFAULT (PCI_ERR_UNC_DLP | \
PCI_ERR_UNC_SDN | \
PCI_ERR_UNC_FCP | \
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 2/7] hw/pci/aer: Add missing routing for AER errors
2023-01-20 14:24 ` Jonathan Cameron
@ 2023-01-20 14:24 ` Jonathan Cameron
-1 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
PCIe r6.0 Figure 6-3 "Pseudo Logic Diagram for Selected Error Message Control
and Status Bits" includes a right hand branch under "All PCI Express devices"
that allows for messages to be generated or sent onwards without SERR#
being set as long as the appropriate per error class bit in the PCIe
Device Control Register is set.
Implement that branch thus enabling routing of ERR_COR, ERR_NONFATAL
and ERR_FATAL under OSes that set these bits appropriately (e.g. Linux)
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci/pcie_aer.c | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/hw/pci/pcie_aer.c b/hw/pci/pcie_aer.c
index 909e027d99..103667c368 100644
--- a/hw/pci/pcie_aer.c
+++ b/hw/pci/pcie_aer.c
@@ -192,8 +192,16 @@ static void pcie_aer_update_uncor_status(PCIDevice *dev)
static bool
pcie_aer_msg_alldev(PCIDevice *dev, const PCIEAERMsg *msg)
{
+ uint16_t devctl = pci_get_word(dev->config + dev->exp.exp_cap +
+ PCI_EXP_DEVCTL);
if (!(pcie_aer_msg_is_uncor(msg) &&
- (pci_get_word(dev->config + PCI_COMMAND) & PCI_COMMAND_SERR))) {
+ (pci_get_word(dev->config + PCI_COMMAND) & PCI_COMMAND_SERR)) &&
+ !((msg->severity == PCI_ERR_ROOT_CMD_NONFATAL_EN) &&
+ (devctl & PCI_EXP_DEVCTL_NFERE)) &&
+ !((msg->severity == PCI_ERR_ROOT_CMD_COR_EN) &&
+ (devctl & PCI_EXP_DEVCTL_CERE)) &&
+ !((msg->severity == PCI_ERR_ROOT_CMD_FATAL_EN) &&
+ (devctl & PCI_EXP_DEVCTL_FERE))) {
return false;
}
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 2/7] hw/pci/aer: Add missing routing for AER errors
@ 2023-01-20 14:24 ` Jonathan Cameron
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
PCIe r6.0 Figure 6-3 "Pseudo Logic Diagram for Selected Error Message Control
and Status Bits" includes a right hand branch under "All PCI Express devices"
that allows for messages to be generated or sent onwards without SERR#
being set as long as the appropriate per error class bit in the PCIe
Device Control Register is set.
Implement that branch thus enabling routing of ERR_COR, ERR_NONFATAL
and ERR_FATAL under OSes that set these bits appropriately (e.g. Linux)
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci/pcie_aer.c | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/hw/pci/pcie_aer.c b/hw/pci/pcie_aer.c
index 909e027d99..103667c368 100644
--- a/hw/pci/pcie_aer.c
+++ b/hw/pci/pcie_aer.c
@@ -192,8 +192,16 @@ static void pcie_aer_update_uncor_status(PCIDevice *dev)
static bool
pcie_aer_msg_alldev(PCIDevice *dev, const PCIEAERMsg *msg)
{
+ uint16_t devctl = pci_get_word(dev->config + dev->exp.exp_cap +
+ PCI_EXP_DEVCTL);
if (!(pcie_aer_msg_is_uncor(msg) &&
- (pci_get_word(dev->config + PCI_COMMAND) & PCI_COMMAND_SERR))) {
+ (pci_get_word(dev->config + PCI_COMMAND) & PCI_COMMAND_SERR)) &&
+ !((msg->severity == PCI_ERR_ROOT_CMD_NONFATAL_EN) &&
+ (devctl & PCI_EXP_DEVCTL_NFERE)) &&
+ !((msg->severity == PCI_ERR_ROOT_CMD_COR_EN) &&
+ (devctl & PCI_EXP_DEVCTL_CERE)) &&
+ !((msg->severity == PCI_ERR_ROOT_CMD_FATAL_EN) &&
+ (devctl & PCI_EXP_DEVCTL_FERE))) {
return false;
}
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 3/7] hw/pci-bridge/cxl_root_port: Wire up AER
2023-01-20 14:24 ` Jonathan Cameron
@ 2023-01-20 14:24 ` Jonathan Cameron via
-1 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
We are missing necessary config write handling for AER emulation in
the CXL root port. Add it based on pcie_root_port.c
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci-bridge/cxl_root_port.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/hw/pci-bridge/cxl_root_port.c b/hw/pci-bridge/cxl_root_port.c
index 6664783974..00195257f7 100644
--- a/hw/pci-bridge/cxl_root_port.c
+++ b/hw/pci-bridge/cxl_root_port.c
@@ -187,12 +187,15 @@ static void cxl_rp_write_config(PCIDevice *d, uint32_t address, uint32_t val,
int len)
{
uint16_t slt_ctl, slt_sta;
+ uint32_t root_cmd =
+ pci_get_long(d->config + d->exp.aer_cap + PCI_ERR_ROOT_COMMAND);
pcie_cap_slot_get(d, &slt_ctl, &slt_sta);
pci_bridge_write_config(d, address, val, len);
pcie_cap_flr_write_config(d, address, val, len);
pcie_cap_slot_write_config(d, slt_ctl, slt_sta, address, val, len);
pcie_aer_write_config(d, address, val, len);
+ pcie_aer_root_write_config(d, address, val, len, root_cmd);
cxl_rp_dvsec_write_config(d, address, val, len);
}
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 3/7] hw/pci-bridge/cxl_root_port: Wire up AER
@ 2023-01-20 14:24 ` Jonathan Cameron via
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
We are missing necessary config write handling for AER emulation in
the CXL root port. Add it based on pcie_root_port.c
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci-bridge/cxl_root_port.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/hw/pci-bridge/cxl_root_port.c b/hw/pci-bridge/cxl_root_port.c
index 6664783974..00195257f7 100644
--- a/hw/pci-bridge/cxl_root_port.c
+++ b/hw/pci-bridge/cxl_root_port.c
@@ -187,12 +187,15 @@ static void cxl_rp_write_config(PCIDevice *d, uint32_t address, uint32_t val,
int len)
{
uint16_t slt_ctl, slt_sta;
+ uint32_t root_cmd =
+ pci_get_long(d->config + d->exp.aer_cap + PCI_ERR_ROOT_COMMAND);
pcie_cap_slot_get(d, &slt_ctl, &slt_sta);
pci_bridge_write_config(d, address, val, len);
pcie_cap_flr_write_config(d, address, val, len);
pcie_cap_slot_write_config(d, slt_ctl, slt_sta, address, val, len);
pcie_aer_write_config(d, address, val, len);
+ pcie_aer_root_write_config(d, address, val, len, root_cmd);
cxl_rp_dvsec_write_config(d, address, val, len);
}
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 4/7] hw/pci-bridge/cxl_root_port: Wire up MSI
2023-01-20 14:24 ` Jonathan Cameron
@ 2023-01-20 14:24 ` Jonathan Cameron
-1 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
Done to avoid fixing ACPI route description of traditional PCI interrupts on q35
and because we should probably move with the times anyway.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci-bridge/cxl_root_port.c | 61 +++++++++++++++++++++++++++++++++++
1 file changed, 61 insertions(+)
diff --git a/hw/pci-bridge/cxl_root_port.c b/hw/pci-bridge/cxl_root_port.c
index 00195257f7..7dfd20aa67 100644
--- a/hw/pci-bridge/cxl_root_port.c
+++ b/hw/pci-bridge/cxl_root_port.c
@@ -22,6 +22,7 @@
#include "qemu/range.h"
#include "hw/pci/pci_bridge.h"
#include "hw/pci/pcie_port.h"
+#include "hw/pci/msi.h"
#include "hw/qdev-properties.h"
#include "hw/sysbus.h"
#include "qapi/error.h"
@@ -29,6 +30,10 @@
#define CXL_ROOT_PORT_DID 0x7075
+#define CXL_RP_MSI_OFFSET 0x60
+#define CXL_RP_MSI_SUPPORTED_FLAGS PCI_MSI_FLAGS_MASKBIT
+#define CXL_RP_MSI_NR_VECTOR 2
+
/* Copied from the gen root port which we derive */
#define GEN_PCIE_ROOT_PORT_AER_OFFSET 0x100
#define GEN_PCIE_ROOT_PORT_ACS_OFFSET \
@@ -47,6 +52,49 @@ typedef struct CXLRootPort {
#define TYPE_CXL_ROOT_PORT "cxl-rp"
DECLARE_INSTANCE_CHECKER(CXLRootPort, CXL_ROOT_PORT, TYPE_CXL_ROOT_PORT)
+/*
+ * If two MSI vector are allocated, Advanced Error Interrupt Message Number
+ * is 1. otherwise 0.
+ * 17.12.5.10 RPERRSTS, 32:27 bit Advanced Error Interrupt Message Number.
+ */
+static uint8_t cxl_rp_aer_vector(const PCIDevice *d)
+{
+ switch (msi_nr_vectors_allocated(d)) {
+ case 1:
+ return 0;
+ case 2:
+ return 1;
+ case 4:
+ case 8:
+ case 16:
+ case 32:
+ default:
+ break;
+ }
+ abort();
+ return 0;
+}
+
+static int cxl_rp_interrupts_init(PCIDevice *d, Error **errp)
+{
+ int rc;
+
+ rc = msi_init(d, CXL_RP_MSI_OFFSET, CXL_RP_MSI_NR_VECTOR,
+ CXL_RP_MSI_SUPPORTED_FLAGS & PCI_MSI_FLAGS_64BIT,
+ CXL_RP_MSI_SUPPORTED_FLAGS & PCI_MSI_FLAGS_MASKBIT,
+ errp);
+ if (rc < 0) {
+ assert(rc == -ENOTSUP);
+ }
+
+ return rc;
+}
+
+static void cxl_rp_interrupts_uninit(PCIDevice *d)
+{
+ msi_uninit(d);
+}
+
static void latch_registers(CXLRootPort *crp)
{
uint32_t *reg_state = crp->cxl_cstate.crb.cache_mem_registers;
@@ -183,6 +231,15 @@ static void cxl_rp_dvsec_write_config(PCIDevice *dev, uint32_t addr,
}
}
+static void cxl_rp_aer_vector_update(PCIDevice *d)
+{
+ PCIERootPortClass *rpc = PCIE_ROOT_PORT_GET_CLASS(d);
+
+ if (rpc->aer_vector) {
+ pcie_aer_root_set_vector(d, rpc->aer_vector(d));
+ }
+}
+
static void cxl_rp_write_config(PCIDevice *d, uint32_t address, uint32_t val,
int len)
{
@@ -192,6 +249,7 @@ static void cxl_rp_write_config(PCIDevice *d, uint32_t address, uint32_t val,
pcie_cap_slot_get(d, &slt_ctl, &slt_sta);
pci_bridge_write_config(d, address, val, len);
+ cxl_rp_aer_vector_update(d);
pcie_cap_flr_write_config(d, address, val, len);
pcie_cap_slot_write_config(d, slt_ctl, slt_sta, address, val, len);
pcie_aer_write_config(d, address, val, len);
@@ -220,6 +278,9 @@ static void cxl_root_port_class_init(ObjectClass *oc, void *data)
rpc->aer_offset = GEN_PCIE_ROOT_PORT_AER_OFFSET;
rpc->acs_offset = GEN_PCIE_ROOT_PORT_ACS_OFFSET;
+ rpc->aer_vector = cxl_rp_aer_vector;
+ rpc->interrupts_init = cxl_rp_interrupts_init;
+ rpc->interrupts_uninit = cxl_rp_interrupts_uninit;
dc->hotpluggable = false;
}
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 4/7] hw/pci-bridge/cxl_root_port: Wire up MSI
@ 2023-01-20 14:24 ` Jonathan Cameron
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
Done to avoid fixing ACPI route description of traditional PCI interrupts on q35
and because we should probably move with the times anyway.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci-bridge/cxl_root_port.c | 61 +++++++++++++++++++++++++++++++++++
1 file changed, 61 insertions(+)
diff --git a/hw/pci-bridge/cxl_root_port.c b/hw/pci-bridge/cxl_root_port.c
index 00195257f7..7dfd20aa67 100644
--- a/hw/pci-bridge/cxl_root_port.c
+++ b/hw/pci-bridge/cxl_root_port.c
@@ -22,6 +22,7 @@
#include "qemu/range.h"
#include "hw/pci/pci_bridge.h"
#include "hw/pci/pcie_port.h"
+#include "hw/pci/msi.h"
#include "hw/qdev-properties.h"
#include "hw/sysbus.h"
#include "qapi/error.h"
@@ -29,6 +30,10 @@
#define CXL_ROOT_PORT_DID 0x7075
+#define CXL_RP_MSI_OFFSET 0x60
+#define CXL_RP_MSI_SUPPORTED_FLAGS PCI_MSI_FLAGS_MASKBIT
+#define CXL_RP_MSI_NR_VECTOR 2
+
/* Copied from the gen root port which we derive */
#define GEN_PCIE_ROOT_PORT_AER_OFFSET 0x100
#define GEN_PCIE_ROOT_PORT_ACS_OFFSET \
@@ -47,6 +52,49 @@ typedef struct CXLRootPort {
#define TYPE_CXL_ROOT_PORT "cxl-rp"
DECLARE_INSTANCE_CHECKER(CXLRootPort, CXL_ROOT_PORT, TYPE_CXL_ROOT_PORT)
+/*
+ * If two MSI vector are allocated, Advanced Error Interrupt Message Number
+ * is 1. otherwise 0.
+ * 17.12.5.10 RPERRSTS, 32:27 bit Advanced Error Interrupt Message Number.
+ */
+static uint8_t cxl_rp_aer_vector(const PCIDevice *d)
+{
+ switch (msi_nr_vectors_allocated(d)) {
+ case 1:
+ return 0;
+ case 2:
+ return 1;
+ case 4:
+ case 8:
+ case 16:
+ case 32:
+ default:
+ break;
+ }
+ abort();
+ return 0;
+}
+
+static int cxl_rp_interrupts_init(PCIDevice *d, Error **errp)
+{
+ int rc;
+
+ rc = msi_init(d, CXL_RP_MSI_OFFSET, CXL_RP_MSI_NR_VECTOR,
+ CXL_RP_MSI_SUPPORTED_FLAGS & PCI_MSI_FLAGS_64BIT,
+ CXL_RP_MSI_SUPPORTED_FLAGS & PCI_MSI_FLAGS_MASKBIT,
+ errp);
+ if (rc < 0) {
+ assert(rc == -ENOTSUP);
+ }
+
+ return rc;
+}
+
+static void cxl_rp_interrupts_uninit(PCIDevice *d)
+{
+ msi_uninit(d);
+}
+
static void latch_registers(CXLRootPort *crp)
{
uint32_t *reg_state = crp->cxl_cstate.crb.cache_mem_registers;
@@ -183,6 +231,15 @@ static void cxl_rp_dvsec_write_config(PCIDevice *dev, uint32_t addr,
}
}
+static void cxl_rp_aer_vector_update(PCIDevice *d)
+{
+ PCIERootPortClass *rpc = PCIE_ROOT_PORT_GET_CLASS(d);
+
+ if (rpc->aer_vector) {
+ pcie_aer_root_set_vector(d, rpc->aer_vector(d));
+ }
+}
+
static void cxl_rp_write_config(PCIDevice *d, uint32_t address, uint32_t val,
int len)
{
@@ -192,6 +249,7 @@ static void cxl_rp_write_config(PCIDevice *d, uint32_t address, uint32_t val,
pcie_cap_slot_get(d, &slt_ctl, &slt_sta);
pci_bridge_write_config(d, address, val, len);
+ cxl_rp_aer_vector_update(d);
pcie_cap_flr_write_config(d, address, val, len);
pcie_cap_slot_write_config(d, slt_ctl, slt_sta, address, val, len);
pcie_aer_write_config(d, address, val, len);
@@ -220,6 +278,9 @@ static void cxl_root_port_class_init(ObjectClass *oc, void *data)
rpc->aer_offset = GEN_PCIE_ROOT_PORT_AER_OFFSET;
rpc->acs_offset = GEN_PCIE_ROOT_PORT_ACS_OFFSET;
+ rpc->aer_vector = cxl_rp_aer_vector;
+ rpc->interrupts_init = cxl_rp_interrupts_init;
+ rpc->interrupts_uninit = cxl_rp_interrupts_uninit;
dc->hotpluggable = false;
}
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 5/7] hw/mem/cxl-type3: Add AER extended capability
2023-01-20 14:24 ` Jonathan Cameron
@ 2023-01-20 14:24 ` Jonathan Cameron
-1 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
This enables AER error injection to function as expected.
It is intended as a building block in enabling CXL RAS error injection
in the following patches.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/mem/cxl_type3.c | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/hw/mem/cxl_type3.c b/hw/mem/cxl_type3.c
index 217a5e639b..6cdd988d1d 100644
--- a/hw/mem/cxl_type3.c
+++ b/hw/mem/cxl_type3.c
@@ -250,6 +250,7 @@ static void ct3d_config_write(PCIDevice *pci_dev, uint32_t addr, uint32_t val,
pcie_doe_write_config(&ct3d->doe_cdat, addr, val, size);
pci_default_write_config(pci_dev, addr, val, size);
+ pcie_aer_write_config(pci_dev, addr, val, size);
}
/*
@@ -452,8 +453,19 @@ static void ct3_realize(PCIDevice *pci_dev, Error **errp)
cxl_cstate->cdat.free_cdat_table = ct3_free_cdat_table;
cxl_cstate->cdat.private = ct3d;
cxl_doe_cdat_init(cxl_cstate, errp);
+
+ pcie_cap_deverr_init(pci_dev);
+ /* Leave a bit of room for expansion */
+ rc = pcie_aer_init(pci_dev, PCI_ERR_VER, 0x200, PCI_ERR_SIZEOF, NULL);
+ if (rc) {
+ goto err_release_cdat;
+ }
+
return;
+err_release_cdat:
+ cxl_doe_cdat_release(cxl_cstate);
+ g_free(regs->special_ops);
err_address_space_free:
address_space_destroy(&ct3d->hostmem_as);
return;
@@ -465,6 +477,7 @@ static void ct3_exit(PCIDevice *pci_dev)
CXLComponentState *cxl_cstate = &ct3d->cxl_cstate;
ComponentRegisters *regs = &cxl_cstate->crb;
+ pcie_aer_exit(pci_dev);
cxl_doe_cdat_release(cxl_cstate);
g_free(regs->special_ops);
address_space_destroy(&ct3d->hostmem_as);
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 5/7] hw/mem/cxl-type3: Add AER extended capability
@ 2023-01-20 14:24 ` Jonathan Cameron
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
This enables AER error injection to function as expected.
It is intended as a building block in enabling CXL RAS error injection
in the following patches.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/mem/cxl_type3.c | 13 +++++++++++++
1 file changed, 13 insertions(+)
diff --git a/hw/mem/cxl_type3.c b/hw/mem/cxl_type3.c
index 217a5e639b..6cdd988d1d 100644
--- a/hw/mem/cxl_type3.c
+++ b/hw/mem/cxl_type3.c
@@ -250,6 +250,7 @@ static void ct3d_config_write(PCIDevice *pci_dev, uint32_t addr, uint32_t val,
pcie_doe_write_config(&ct3d->doe_cdat, addr, val, size);
pci_default_write_config(pci_dev, addr, val, size);
+ pcie_aer_write_config(pci_dev, addr, val, size);
}
/*
@@ -452,8 +453,19 @@ static void ct3_realize(PCIDevice *pci_dev, Error **errp)
cxl_cstate->cdat.free_cdat_table = ct3_free_cdat_table;
cxl_cstate->cdat.private = ct3d;
cxl_doe_cdat_init(cxl_cstate, errp);
+
+ pcie_cap_deverr_init(pci_dev);
+ /* Leave a bit of room for expansion */
+ rc = pcie_aer_init(pci_dev, PCI_ERR_VER, 0x200, PCI_ERR_SIZEOF, NULL);
+ if (rc) {
+ goto err_release_cdat;
+ }
+
return;
+err_release_cdat:
+ cxl_doe_cdat_release(cxl_cstate);
+ g_free(regs->special_ops);
err_address_space_free:
address_space_destroy(&ct3d->hostmem_as);
return;
@@ -465,6 +477,7 @@ static void ct3_exit(PCIDevice *pci_dev)
CXLComponentState *cxl_cstate = &ct3d->cxl_cstate;
ComponentRegisters *regs = &cxl_cstate->crb;
+ pcie_aer_exit(pci_dev);
cxl_doe_cdat_release(cxl_cstate);
g_free(regs->special_ops);
address_space_destroy(&ct3d->hostmem_as);
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 6/7] hw/pci/aer: Make PCIE AER error injection facility available for other emulation to use.
2023-01-20 14:24 ` Jonathan Cameron
@ 2023-01-20 14:24 ` Jonathan Cameron via
-1 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
This infrastructure will be reused for CXL RAS error injection
in patches that follow.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci/pci-internal.h | 1 -
include/hw/pci/pcie_aer.h | 1 +
2 files changed, 1 insertion(+), 1 deletion(-)
diff --git a/hw/pci/pci-internal.h b/hw/pci/pci-internal.h
index 2ea356bdf5..a7d6d8a732 100644
--- a/hw/pci/pci-internal.h
+++ b/hw/pci/pci-internal.h
@@ -20,6 +20,5 @@ void pcibus_dev_print(Monitor *mon, DeviceState *dev, int indent);
int pcie_aer_parse_error_string(const char *error_name,
uint32_t *status, bool *correctable);
-int pcie_aer_inject_error(PCIDevice *dev, const PCIEAERErr *err);
#endif
diff --git a/include/hw/pci/pcie_aer.h b/include/hw/pci/pcie_aer.h
index 65e71d98fe..1234fdc4e2 100644
--- a/include/hw/pci/pcie_aer.h
+++ b/include/hw/pci/pcie_aer.h
@@ -100,4 +100,5 @@ void pcie_aer_root_write_config(PCIDevice *dev,
uint32_t addr, uint32_t val, int len,
uint32_t root_cmd_prev);
+int pcie_aer_inject_error(PCIDevice *dev, const PCIEAERErr *err);
#endif /* QEMU_PCIE_AER_H */
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 6/7] hw/pci/aer: Make PCIE AER error injection facility available for other emulation to use.
@ 2023-01-20 14:24 ` Jonathan Cameron via
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
This infrastructure will be reused for CXL RAS error injection
in patches that follow.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/pci/pci-internal.h | 1 -
include/hw/pci/pcie_aer.h | 1 +
2 files changed, 1 insertion(+), 1 deletion(-)
diff --git a/hw/pci/pci-internal.h b/hw/pci/pci-internal.h
index 2ea356bdf5..a7d6d8a732 100644
--- a/hw/pci/pci-internal.h
+++ b/hw/pci/pci-internal.h
@@ -20,6 +20,5 @@ void pcibus_dev_print(Monitor *mon, DeviceState *dev, int indent);
int pcie_aer_parse_error_string(const char *error_name,
uint32_t *status, bool *correctable);
-int pcie_aer_inject_error(PCIDevice *dev, const PCIEAERErr *err);
#endif
diff --git a/include/hw/pci/pcie_aer.h b/include/hw/pci/pcie_aer.h
index 65e71d98fe..1234fdc4e2 100644
--- a/include/hw/pci/pcie_aer.h
+++ b/include/hw/pci/pcie_aer.h
@@ -100,4 +100,5 @@ void pcie_aer_root_write_config(PCIDevice *dev,
uint32_t addr, uint32_t val, int len,
uint32_t root_cmd_prev);
+int pcie_aer_inject_error(PCIDevice *dev, const PCIEAERErr *err);
#endif /* QEMU_PCIE_AER_H */
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 7/7] hw/mem/cxl_type3: Add CXL RAS Error Injection Support
2023-01-20 14:24 ` Jonathan Cameron
@ 2023-01-20 14:24 ` Jonathan Cameron via
-1 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
CXL uses PCI AER Internal errors to signal to the host that an error has
occurred. The host can then read more detailed status from the CXL RAS
capability.
For uncorrectable errors: support multiple injection in one operation
as this is needed to reliably test multiple header logging support in an
OS. The equivalent feature doesn't exist for correctable errors, so only
one error need be injected at a time.
Note:
- Header content needs to be manually specified in a fashion that
matches the specification for what can be in the header for each
error type.
Injection via QMP:
{ "execute": "qmp_capabilities" }
...
{ "execute": "cxl-inject-uncorrectable-errors",
"arguments": {
"path": "/machine/peripheral/cxl-pmem0",
"errors": [
{
"type": "cache-address-parity",
"header": [ 3, 4]
},
{
"type": "cache-data-parity",
"header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
},
{
"type": "internal",
"header": [ 1, 2, 4]
}
]
}}
...
{ "execute": "cxl-inject-correctable-error",
"arguments": {
"path": "/machine/peripheral/cxl-pmem0",
"type": "physical",
"header": [ 3, 4]
} }
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/cxl/cxl-component-utils.c | 4 +-
hw/mem/cxl_type3.c | 290 +++++++++++++++++++++++++++++++++
hw/mem/cxl_type3_stubs.c | 10 ++
hw/mem/meson.build | 2 +
include/hw/cxl/cxl_component.h | 26 +++
include/hw/cxl/cxl_device.h | 11 ++
qapi/cxl.json | 113 +++++++++++++
qapi/meson.build | 1 +
qapi/qapi-schema.json | 1 +
9 files changed, 457 insertions(+), 1 deletion(-)
diff --git a/hw/cxl/cxl-component-utils.c b/hw/cxl/cxl-component-utils.c
index 3edd303a33..02fb6c17b9 100644
--- a/hw/cxl/cxl-component-utils.c
+++ b/hw/cxl/cxl-component-utils.c
@@ -142,16 +142,18 @@ static void ras_init_common(uint32_t *reg_state, uint32_t *write_msk)
* be handled as RO.
*/
reg_state[R_CXL_RAS_UNC_ERR_STATUS] = 0;
+ write_msk[R_CXL_RAS_UNC_ERR_STATUS] = 0x1cfff;
/* Bits 12-13 and 17-31 reserved in CXL 2.0 */
reg_state[R_CXL_RAS_UNC_ERR_MASK] = 0x1cfff;
write_msk[R_CXL_RAS_UNC_ERR_MASK] = 0x1cfff;
reg_state[R_CXL_RAS_UNC_ERR_SEVERITY] = 0x1cfff;
write_msk[R_CXL_RAS_UNC_ERR_SEVERITY] = 0x1cfff;
reg_state[R_CXL_RAS_COR_ERR_STATUS] = 0;
+ write_msk[R_CXL_RAS_COR_ERR_STATUS] = 0x7f;
reg_state[R_CXL_RAS_COR_ERR_MASK] = 0x7f;
write_msk[R_CXL_RAS_COR_ERR_MASK] = 0x7f;
/* CXL switches and devices must set */
- reg_state[R_CXL_RAS_ERR_CAP_CTRL] = 0x00;
+ reg_state[R_CXL_RAS_ERR_CAP_CTRL] = 0x200;
}
static void hdm_init_common(uint32_t *reg_state, uint32_t *write_msk,
diff --git a/hw/mem/cxl_type3.c b/hw/mem/cxl_type3.c
index 6cdd988d1d..41ae19bf66 100644
--- a/hw/mem/cxl_type3.c
+++ b/hw/mem/cxl_type3.c
@@ -1,6 +1,7 @@
#include "qemu/osdep.h"
#include "qemu/units.h"
#include "qemu/error-report.h"
+#include "qapi/qapi-commands-cxl.h"
#include "hw/mem/memory-device.h"
#include "hw/mem/pc-dimm.h"
#include "hw/pci/pci.h"
@@ -323,6 +324,66 @@ static void hdm_decoder_commit(CXLType3Dev *ct3d, int which)
ARRAY_FIELD_DP32(cache_mem, CXL_HDM_DECODER0_CTRL, COMMITTED, 1);
}
+static int ct3d_qmp_uncor_err_to_cxl(CxlUncorErrorType qmp_err)
+{
+ switch (qmp_err) {
+ case CXL_UNCOR_ERROR_TYPE_CACHE_DATA_PARITY:
+ return CXL_RAS_UNC_ERR_CACHE_DATA_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_CACHE_ADDRESS_PARITY:
+ return CXL_RAS_UNC_ERR_CACHE_ADDRESS_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_CACHE_BE_PARITY:
+ return CXL_RAS_UNC_ERR_CACHE_BE_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_CACHE_DATA_ECC:
+ return CXL_RAS_UNC_ERR_CACHE_DATA_ECC;
+ case CXL_UNCOR_ERROR_TYPE_MEM_DATA_PARITY:
+ return CXL_RAS_UNC_ERR_MEM_DATA_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_MEM_ADDRESS_PARITY:
+ return CXL_RAS_UNC_ERR_MEM_ADDRESS_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_MEM_BE_PARITY:
+ return CXL_RAS_UNC_ERR_MEM_BE_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_MEM_DATA_ECC:
+ return CXL_RAS_UNC_ERR_MEM_DATA_ECC;
+ case CXL_UNCOR_ERROR_TYPE_REINIT_THRESHOLD:
+ return CXL_RAS_UNC_ERR_REINIT_THRESHOLD;
+ case CXL_UNCOR_ERROR_TYPE_RSVD_ENCODING:
+ return CXL_RAS_UNC_ERR_RSVD_ENCODING;
+ case CXL_UNCOR_ERROR_TYPE_POISON_RECEIVED:
+ return CXL_RAS_UNC_ERR_POISON_RECEIVED;
+ case CXL_UNCOR_ERROR_TYPE_RECEIVER_OVERFLOW:
+ return CXL_RAS_UNC_ERR_RECEIVER_OVERFLOW;
+ case CXL_UNCOR_ERROR_TYPE_INTERNAL:
+ return CXL_RAS_UNC_ERR_INTERNAL;
+ case CXL_UNCOR_ERROR_TYPE_CXL_IDE_TX:
+ return CXL_RAS_UNC_ERR_CXL_IDE_TX;
+ case CXL_UNCOR_ERROR_TYPE_CXL_IDE_RX:
+ return CXL_RAS_UNC_ERR_CXL_IDE_RX;
+ default:
+ return -EINVAL;
+ }
+}
+
+static int ct3d_qmp_cor_err_to_cxl(CxlCorErrorType qmp_err)
+{
+ switch (qmp_err) {
+ case CXL_COR_ERROR_TYPE_CACHE_DATA_ECC:
+ return CXL_RAS_COR_ERR_CACHE_DATA_ECC;
+ case CXL_COR_ERROR_TYPE_MEM_DATA_ECC:
+ return CXL_RAS_COR_ERR_MEM_DATA_ECC;
+ case CXL_COR_ERROR_TYPE_CRC_THRESHOLD:
+ return CXL_RAS_COR_ERR_CRC_THRESHOLD;
+ case CXL_COR_ERROR_TYPE_RETRY_THRESHOLD:
+ return CXL_RAS_COR_ERR_RETRY_THRESHOLD;
+ case CXL_COR_ERROR_TYPE_CACHE_POISON_RECEIVED:
+ return CXL_RAS_COR_ERR_CACHE_POISON_RECEIVED;
+ case CXL_COR_ERROR_TYPE_MEM_POISON_RECEIVED:
+ return CXL_RAS_COR_ERR_MEM_POISON_RECEIVED;
+ case CXL_COR_ERROR_TYPE_PHYSICAL:
+ return CXL_RAS_COR_ERR_PHYSICAL;
+ default:
+ return -EINVAL;
+ }
+}
+
static void ct3d_reg_write(void *opaque, hwaddr offset, uint64_t value,
unsigned size)
{
@@ -341,6 +402,84 @@ static void ct3d_reg_write(void *opaque, hwaddr offset, uint64_t value,
should_commit = FIELD_EX32(value, CXL_HDM_DECODER0_CTRL, COMMIT);
which_hdm = 0;
break;
+ case A_CXL_RAS_UNC_ERR_STATUS:
+ {
+ uint32_t capctrl = ldl_le_p(cache_mem + R_CXL_RAS_ERR_CAP_CTRL);
+ uint32_t fe = FIELD_EX32(capctrl, CXL_RAS_ERR_CAP_CTRL, FIRST_ERROR_POINTER);
+ CXLError *cxl_err;
+ uint32_t unc_err;
+
+ /*
+ * If single bit written that corresponds to the first error
+ * pointer being cleared, update the status and header log.
+ */
+ if (!QTAILQ_EMPTY(&ct3d->error_list)) {
+ uint32_t *header_log = &cache_mem[R_CXL_RAS_ERR_HEADER0];
+
+ if ((1 << fe) ^ value) {
+ CXLError *cxl_next;
+ /*
+ * Software is using wrong flow for multiple header recording
+ * Following behaviour in PCIe r6.0 and assuming multiple
+ * header support. Imdef choice to clear all matching records
+ * if more than one bit set - which corresponds closest to
+ * behavior of hardware not capable of multiple header
+ * recording.
+ */
+ QTAILQ_FOREACH_SAFE(cxl_err, &ct3d->error_list, node, cxl_next) {
+ if ((1 << cxl_err->type) & value) {
+ QTAILQ_REMOVE(&ct3d->error_list, cxl_err, node);
+ g_free(cxl_err);
+ }
+ }
+ } else {
+ /* Done with previous FE, so drop from list */
+ cxl_err = QTAILQ_FIRST(&ct3d->error_list);
+ QTAILQ_REMOVE(&ct3d->error_list, cxl_err, node);
+ g_free(cxl_err);
+ }
+
+ /*
+ * If there is another FE, then put that in place and update
+ * the header log
+ */
+ if (!QTAILQ_EMPTY(&ct3d->error_list)) {
+ int i;
+
+ cxl_err = QTAILQ_FIRST(&ct3d->error_list);
+ for (i = 0; i < CXL_RAS_ERR_HEADER_NUM; i++) {
+ stl_le_p(header_log + i, cxl_err->header[i]);
+ }
+ capctrl = FIELD_DP32(capctrl, CXL_RAS_ERR_CAP_CTRL,
+ FIRST_ERROR_POINTER, cxl_err->type);
+ } else {
+ /*
+ * If no more errors, then follow recomendation of PCI spec
+ * r6.0 6.2.4.2 to set the first error pointer to a status
+ * bit that will never be used.
+ */
+ capctrl = FIELD_DP32(capctrl, CXL_RAS_ERR_CAP_CTRL,
+ FIRST_ERROR_POINTER,
+ CXL_RAS_UNC_ERR_CXL_UNUSED);
+ }
+ stl_le_p((uint8_t *)cache_mem + A_CXL_RAS_ERR_CAP_CTRL, capctrl);
+ }
+ unc_err = 0;
+ QTAILQ_FOREACH(cxl_err, &ct3d->error_list, node) {
+ unc_err |= 1 << cxl_err->type;
+ }
+ stl_le_p((uint8_t *)cache_mem + offset, unc_err);
+
+ return;
+ }
+ case A_CXL_RAS_COR_ERR_STATUS:
+ {
+ uint32_t rw1c = value;
+ uint32_t temp = ldl_le_p((uint8_t *)cache_mem + offset);
+ temp &= ~rw1c;
+ stl_le_p((uint8_t *)cache_mem + offset, temp);
+ return;
+ }
default:
break;
}
@@ -404,6 +543,8 @@ static void ct3_realize(PCIDevice *pci_dev, Error **errp)
unsigned short msix_num = 1;
int i, rc;
+ QTAILQ_INIT(&ct3d->error_list);
+
if (!cxl_setup_memory(ct3d, errp)) {
return;
}
@@ -631,6 +772,155 @@ static void set_lsa(CXLType3Dev *ct3d, const void *buf, uint64_t size,
*/
}
+/* For uncorrectable errors include support for multiple header recording */
+void qmp_cxl_inject_uncorrectable_errors(const char *path,
+ CXLUncorErrorRecordList *errors,
+ Error **errp)
+{
+ Object *obj = object_resolve_path(path, NULL);
+ static PCIEAERErr err = {};
+ CXLType3Dev *ct3d;
+ CXLError *cxl_err;
+ uint32_t *reg_state;
+ uint32_t unc_err;
+ bool first;
+
+ if (!obj) {
+ error_setg(errp, "Unable to resolve path");
+ return;
+ }
+
+ if (!object_dynamic_cast(obj, TYPE_CXL_TYPE3)) {
+ error_setg(errp, "Path does not point to a CXL type 3 device");
+ return;
+ }
+
+ err.status = PCI_ERR_UNC_INTN;
+ err.source_id = pci_requester_id(PCI_DEVICE(obj));
+ err.flags = 0;
+
+ ct3d = CXL_TYPE3(obj);
+
+ first = QTAILQ_EMPTY(&ct3d->error_list);
+ reg_state = ct3d->cxl_cstate.crb.cache_mem_registers;
+ while (errors) {
+ uint32List *header = errors->value->header;
+ uint8_t header_count = 0;
+ int cxl_err_code;
+
+ cxl_err_code = ct3d_qmp_uncor_err_to_cxl(errors->value->type);
+ if (cxl_err_code < 0) {
+ error_setg(errp, "Unknown error code");
+ return;
+ }
+
+ /* If the error is masked, nothting to do here */
+ if (!((1 << cxl_err_code) & ~reg_state[R_CXL_RAS_UNC_ERR_MASK])) {
+ errors = errors->next;
+ continue;
+ }
+
+ cxl_err = g_malloc0(sizeof(*cxl_err));
+ if (!cxl_err) {
+ return;
+ }
+
+ cxl_err->type = cxl_err_code;
+ while (header && header_count < 32) {
+ cxl_err->header[header_count++] = header->value;
+ header = header->next;
+ }
+ if (header_count > 32) {
+ error_setg(errp, "Header must be 32 DWORD or less");
+ return;
+ }
+ QTAILQ_INSERT_TAIL(&ct3d->error_list, cxl_err, node);
+
+ errors = errors->next;
+ }
+
+ if (first && !QTAILQ_EMPTY(&ct3d->error_list)) {
+ uint32_t *cache_mem = ct3d->cxl_cstate.crb.cache_mem_registers;
+ uint32_t capctrl = ldl_le_p((uint8_t *)cache_mem + A_CXL_RAS_ERR_CAP_CTRL);
+ uint32_t *header_log = &cache_mem[R_CXL_RAS_ERR_HEADER0];
+ int i;
+
+ cxl_err = QTAILQ_FIRST(&ct3d->error_list);
+ for (i = 0; i < CXL_RAS_ERR_HEADER_NUM; i++) {
+ stl_le_p(header_log + i, cxl_err->header[i]);
+ }
+
+ capctrl = FIELD_DP32(capctrl, CXL_RAS_ERR_CAP_CTRL,
+ FIRST_ERROR_POINTER, cxl_err->type);
+ stl_le_p(cache_mem + R_CXL_RAS_ERR_CAP_CTRL, capctrl);
+ }
+
+ unc_err = 0;
+ QTAILQ_FOREACH(cxl_err, &ct3d->error_list, node) {
+ unc_err |= (1 << cxl_err->type);
+ }
+ if (!unc_err) {
+ return;
+ }
+
+ reg_state[R_CXL_RAS_UNC_ERR_STATUS] = unc_err;
+ pcie_aer_inject_error(PCI_DEVICE(obj), &err);
+
+ return;
+}
+
+void qmp_cxl_inject_correctable_error(const char *path, CxlCorErrorType type,
+ uint32List *header, Error **errp)
+{
+ static PCIEAERErr err = {};
+ Object *obj = object_resolve_path(path, NULL);
+ CXLType3Dev *ct3d;
+ uint32_t *reg_state;
+ uint8_t header_count = 0;
+ uint32_t cor_err;
+ int cxl_err_type;
+
+ if (!obj) {
+ error_setg(errp, "Unable to resolve path");
+ return;
+ }
+ if (!object_dynamic_cast(obj, TYPE_CXL_TYPE3)) {
+ error_setg(errp, "Path does not point to a CXL type 3 device");
+ return;
+ }
+
+ err.status = PCI_ERR_COR_INTERNAL;
+ err.source_id = pci_requester_id(PCI_DEVICE(obj));
+ err.flags = PCIE_AER_ERR_IS_CORRECTABLE;
+
+ ct3d = CXL_TYPE3(obj);
+ reg_state = ct3d->cxl_cstate.crb.cache_mem_registers;
+ cor_err = reg_state[R_CXL_RAS_COR_ERR_STATUS];
+
+ cxl_err_type = ct3d_qmp_cor_err_to_cxl(type);
+ if (cxl_err_type < 0) {
+ error_setg(errp, "Invalid COR error");
+ return;
+ }
+ /* If the error is masked, nothting to do here */
+ if (!((1 << cxl_err_type) & ~reg_state[R_CXL_RAS_COR_ERR_MASK])) {
+ return;
+ }
+
+ cor_err |= (1 << cxl_err_type);
+ reg_state[R_CXL_RAS_COR_ERR_STATUS] = cor_err;
+ while (header && header_count < 32) {
+ reg_state[R_CXL_RAS_ERR_HEADER0 + header_count++] = header->value;
+ header = header->next;
+ }
+ if (header_count > 32) {
+ error_setg(errp, "Header must be 32 DWORD or less");
+ return;
+ }
+
+ pcie_aer_inject_error(PCI_DEVICE(obj), &err);
+}
+
static void ct3_class_init(ObjectClass *oc, void *data)
{
DeviceClass *dc = DEVICE_CLASS(oc);
diff --git a/hw/mem/cxl_type3_stubs.c b/hw/mem/cxl_type3_stubs.c
new file mode 100644
index 0000000000..df7f8dfa60
--- /dev/null
+++ b/hw/mem/cxl_type3_stubs.c
@@ -0,0 +1,10 @@
+
+#include "qemu/osdep.h"
+#include "qapi/qapi-commands-cxl.h"
+
+void qmp_cxl_inject_uncorrectable_errors(const char *path,
+ CXLUncorErrorRecordList *errors,
+ Error **errp) {}
+
+void qmp_cxl_inject_correctable_error(const char *path, CxlCorErrorType type,
+ uint32List *header, Error **errp) {}
diff --git a/hw/mem/meson.build b/hw/mem/meson.build
index 609b2b36fc..56c2618b84 100644
--- a/hw/mem/meson.build
+++ b/hw/mem/meson.build
@@ -4,6 +4,8 @@ mem_ss.add(when: 'CONFIG_DIMM', if_true: files('pc-dimm.c'))
mem_ss.add(when: 'CONFIG_NPCM7XX', if_true: files('npcm7xx_mc.c'))
mem_ss.add(when: 'CONFIG_NVDIMM', if_true: files('nvdimm.c'))
mem_ss.add(when: 'CONFIG_CXL_MEM_DEVICE', if_true: files('cxl_type3.c'))
+softmmu_ss.add(when: 'CONFIG_CXL_MEM_DEVICE', if_false: files('cxl_type3_stubs.c'))
+softmmu_ss.add(when: 'CONFIG_ALL', if_true: files('cxl_type3_stubs.c'))
softmmu_ss.add_all(when: 'CONFIG_MEM_DEVICE', if_true: mem_ss)
diff --git a/include/hw/cxl/cxl_component.h b/include/hw/cxl/cxl_component.h
index 5dca21e95b..8752171f70 100644
--- a/include/hw/cxl/cxl_component.h
+++ b/include/hw/cxl/cxl_component.h
@@ -67,11 +67,37 @@ CXLx_CAPABILITY_HEADER(SNOOP, 0x14)
#define CXL_RAS_REGISTERS_OFFSET 0x80
#define CXL_RAS_REGISTERS_SIZE 0x58
REG32(CXL_RAS_UNC_ERR_STATUS, CXL_RAS_REGISTERS_OFFSET)
+#define CXL_RAS_UNC_ERR_CACHE_DATA_PARITY 0
+#define CXL_RAS_UNC_ERR_CACHE_ADDRESS_PARITY 1
+#define CXL_RAS_UNC_ERR_CACHE_BE_PARITY 2
+#define CXL_RAS_UNC_ERR_CACHE_DATA_ECC 3
+#define CXL_RAS_UNC_ERR_MEM_DATA_PARITY 4
+#define CXL_RAS_UNC_ERR_MEM_ADDRESS_PARITY 5
+#define CXL_RAS_UNC_ERR_MEM_BE_PARITY 6
+#define CXL_RAS_UNC_ERR_MEM_DATA_ECC 7
+#define CXL_RAS_UNC_ERR_REINIT_THRESHOLD 8
+#define CXL_RAS_UNC_ERR_RSVD_ENCODING 9
+#define CXL_RAS_UNC_ERR_POISON_RECEIVED 10
+#define CXL_RAS_UNC_ERR_RECEIVER_OVERFLOW 11
+#define CXL_RAS_UNC_ERR_INTERNAL 14
+#define CXL_RAS_UNC_ERR_CXL_IDE_TX 15
+#define CXL_RAS_UNC_ERR_CXL_IDE_RX 16
+#define CXL_RAS_UNC_ERR_CXL_UNUSED 63 /* Magic value */
REG32(CXL_RAS_UNC_ERR_MASK, CXL_RAS_REGISTERS_OFFSET + 0x4)
REG32(CXL_RAS_UNC_ERR_SEVERITY, CXL_RAS_REGISTERS_OFFSET + 0x8)
REG32(CXL_RAS_COR_ERR_STATUS, CXL_RAS_REGISTERS_OFFSET + 0xc)
+#define CXL_RAS_COR_ERR_CACHE_DATA_ECC 0
+#define CXL_RAS_COR_ERR_MEM_DATA_ECC 1
+#define CXL_RAS_COR_ERR_CRC_THRESHOLD 2
+#define CXL_RAS_COR_ERR_RETRY_THRESHOLD 3
+#define CXL_RAS_COR_ERR_CACHE_POISON_RECEIVED 4
+#define CXL_RAS_COR_ERR_MEM_POISON_RECEIVED 5
+#define CXL_RAS_COR_ERR_PHYSICAL 6
REG32(CXL_RAS_COR_ERR_MASK, CXL_RAS_REGISTERS_OFFSET + 0x10)
REG32(CXL_RAS_ERR_CAP_CTRL, CXL_RAS_REGISTERS_OFFSET + 0x14)
+ FIELD(CXL_RAS_ERR_CAP_CTRL, FIRST_ERROR_POINTER, 0, 6)
+REG32(CXL_RAS_ERR_HEADER0, CXL_RAS_REGISTERS_OFFSET + 0x18)
+#define CXL_RAS_ERR_HEADER_NUM 32
/* Offset 0x18 - 0x58 reserved for RAS logs */
/* 8.2.5.10 - CXL Security Capability Structure */
diff --git a/include/hw/cxl/cxl_device.h b/include/hw/cxl/cxl_device.h
index 7e5ad65c1d..d589f78202 100644
--- a/include/hw/cxl/cxl_device.h
+++ b/include/hw/cxl/cxl_device.h
@@ -232,6 +232,14 @@ REG64(CXL_MEM_DEV_STS, 0)
FIELD(CXL_MEM_DEV_STS, MBOX_READY, 4, 1)
FIELD(CXL_MEM_DEV_STS, RESET_NEEDED, 5, 3)
+typedef struct CXLError {
+ QTAILQ_ENTRY(CXLError) node;
+ int type; /* Error code as per FE definition */
+ uint32_t header[32];
+} CXLError;
+
+typedef QTAILQ_HEAD(, CXLError) CXLErrorList;
+
struct CXLType3Dev {
/* Private */
PCIDevice parent_obj;
@@ -248,6 +256,9 @@ struct CXLType3Dev {
/* DOE */
DOECap doe_cdat;
+
+ /* Error injection */
+ CXLErrorList error_list;
};
#define TYPE_CXL_TYPE3 "cxl-type3"
diff --git a/qapi/cxl.json b/qapi/cxl.json
new file mode 100644
index 0000000000..3631bdba8f
--- /dev/null
+++ b/qapi/cxl.json
@@ -0,0 +1,113 @@
+# -*- Mode: Python -*-
+# vim: filetype=python
+
+##
+# = CXL devices
+##
+
+##
+# @CxlUncorErrorType:
+#
+# Type of CXL error to inject
+#
+# @cache-data-parity: Data error such as data parity or data ECC error CXL.cache
+# @cache-address-parity: Address parity or other errors associated with the
+# address field on CXL.cache
+# @cache-be-parity: Byte enable parity or other byte enable errors on CXL.cache
+# @cache-data-ecc: ECC error on CXL.cache
+# @mem-data-parity: Data error such as data parity or data ECC error on CXL.mem
+# @mem-address-parity: Address parity or other errors associated with the
+# address field on CXL.mem
+# @mem-be-parity: Byte enable parity or other byte enable errors on CXL.mem.
+# @mem-data-ecc: Data ECC error on CXL.mem.
+# @reinit-threshold: REINIT threshold hit.
+# @rsvd-encoding: Received unrecognized encoding.
+# @poison-received: Received poison from the peer.
+# @receiver-overflow: Buffer overflows (first 3 bits of header log indicate which)
+# @internal: Component specific error
+# @cxl-ide-tx: Integrity and data encryption tx error.
+# @cxl-ide-rx: Integrity and data encryption rx error.
+##
+
+{ 'enum': 'CxlUncorErrorType',
+ 'data': ['cache-data-parity',
+ 'cache-address-parity',
+ 'cache-be-parity',
+ 'cache-data-ecc',
+ 'mem-data-parity',
+ 'mem-address-parity',
+ 'mem-be-parity',
+ 'mem-data-ecc',
+ 'reinit-threshold',
+ 'rsvd-encoding',
+ 'poison-received',
+ 'receiver-overflow',
+ 'internal',
+ 'cxl-ide-tx',
+ 'cxl-ide-rx'
+ ]
+ }
+
+##
+# @CXLUncorErrorRecord:
+#
+# @type: Type of error
+# @header: 16 DWORD of header.
+##
+{ 'struct': 'CXLUncorErrorRecord',
+ 'data': {
+ 'type': 'CxlUncorErrorType',
+ 'header': [ 'uint32' ]
+ }
+}
+
+##
+# @cxl-inject-uncorrectable-errors:
+#
+# Inject mulitiple uncorrectable errors in one go.
+#
+# @path: CXL Type 3 device canonical QOM path
+#
+# @errors: Errors to inject
+##
+{ 'command': 'cxl-inject-uncorrectable-errors',
+ 'data': { 'path': 'str',
+ 'errors': [ 'CXLUncorErrorRecord' ] }}
+
+##
+# @CxlCorErrorType:
+#
+# Type of CXL correctable error to inject
+#
+# @cache-data-ecc: Data ECC error on CXL.cache
+# @mem-data-ecc: Data ECC error on CXL.mem
+# @crc-threshold: Component specific and applicable to 68 byte Flit mode only.
+# @cache-poison-received: Received poison from a peer on CXL.cache.
+# @mem-poison-received: Received poison from a peer on CXL.mem
+# @physical: Received error indication from the physical layer.
+##
+{ 'enum': 'CxlCorErrorType',
+ 'data': ['cache-data-ecc',
+ 'mem-data-ecc',
+ 'crc-threshold',
+ 'retry-threshold',
+ 'cache-poison-received',
+ 'mem-poison-received',
+ 'physical']
+}
+
+##
+# @cxl-inject-correctable-error:
+#
+# @path: CXL Type 3 device canonical QOM path
+#
+# @type: Type of error.
+#
+# @header : 16 DWORD of header.
+##
+{ 'command': 'cxl-inject-correctable-error',
+ 'data': { 'path': 'str',
+ 'type': 'CxlCorErrorType',
+ 'header': [ 'uint32' ]
+ }
+}
diff --git a/qapi/meson.build b/qapi/meson.build
index fbdb442fdf..73c3c8c31a 100644
--- a/qapi/meson.build
+++ b/qapi/meson.build
@@ -31,6 +31,7 @@ qapi_all_modules = [
'compat',
'control',
'crypto',
+ 'cxl',
'dump',
'error',
'introspect',
diff --git a/qapi/qapi-schema.json b/qapi/qapi-schema.json
index f000b90744..079f2a402a 100644
--- a/qapi/qapi-schema.json
+++ b/qapi/qapi-schema.json
@@ -95,3 +95,4 @@
{ 'include': 'pci.json' }
{ 'include': 'stats.json' }
{ 'include': 'virtio.json' }
+{ 'include': 'cxl.json' }
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* [PATCH v2 7/7] hw/mem/cxl_type3: Add CXL RAS Error Injection Support
@ 2023-01-20 14:24 ` Jonathan Cameron via
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-20 14:24 UTC (permalink / raw)
To: qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
CXL uses PCI AER Internal errors to signal to the host that an error has
occurred. The host can then read more detailed status from the CXL RAS
capability.
For uncorrectable errors: support multiple injection in one operation
as this is needed to reliably test multiple header logging support in an
OS. The equivalent feature doesn't exist for correctable errors, so only
one error need be injected at a time.
Note:
- Header content needs to be manually specified in a fashion that
matches the specification for what can be in the header for each
error type.
Injection via QMP:
{ "execute": "qmp_capabilities" }
...
{ "execute": "cxl-inject-uncorrectable-errors",
"arguments": {
"path": "/machine/peripheral/cxl-pmem0",
"errors": [
{
"type": "cache-address-parity",
"header": [ 3, 4]
},
{
"type": "cache-data-parity",
"header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
},
{
"type": "internal",
"header": [ 1, 2, 4]
}
]
}}
...
{ "execute": "cxl-inject-correctable-error",
"arguments": {
"path": "/machine/peripheral/cxl-pmem0",
"type": "physical",
"header": [ 3, 4]
} }
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
hw/cxl/cxl-component-utils.c | 4 +-
hw/mem/cxl_type3.c | 290 +++++++++++++++++++++++++++++++++
hw/mem/cxl_type3_stubs.c | 10 ++
hw/mem/meson.build | 2 +
include/hw/cxl/cxl_component.h | 26 +++
include/hw/cxl/cxl_device.h | 11 ++
qapi/cxl.json | 113 +++++++++++++
qapi/meson.build | 1 +
qapi/qapi-schema.json | 1 +
9 files changed, 457 insertions(+), 1 deletion(-)
diff --git a/hw/cxl/cxl-component-utils.c b/hw/cxl/cxl-component-utils.c
index 3edd303a33..02fb6c17b9 100644
--- a/hw/cxl/cxl-component-utils.c
+++ b/hw/cxl/cxl-component-utils.c
@@ -142,16 +142,18 @@ static void ras_init_common(uint32_t *reg_state, uint32_t *write_msk)
* be handled as RO.
*/
reg_state[R_CXL_RAS_UNC_ERR_STATUS] = 0;
+ write_msk[R_CXL_RAS_UNC_ERR_STATUS] = 0x1cfff;
/* Bits 12-13 and 17-31 reserved in CXL 2.0 */
reg_state[R_CXL_RAS_UNC_ERR_MASK] = 0x1cfff;
write_msk[R_CXL_RAS_UNC_ERR_MASK] = 0x1cfff;
reg_state[R_CXL_RAS_UNC_ERR_SEVERITY] = 0x1cfff;
write_msk[R_CXL_RAS_UNC_ERR_SEVERITY] = 0x1cfff;
reg_state[R_CXL_RAS_COR_ERR_STATUS] = 0;
+ write_msk[R_CXL_RAS_COR_ERR_STATUS] = 0x7f;
reg_state[R_CXL_RAS_COR_ERR_MASK] = 0x7f;
write_msk[R_CXL_RAS_COR_ERR_MASK] = 0x7f;
/* CXL switches and devices must set */
- reg_state[R_CXL_RAS_ERR_CAP_CTRL] = 0x00;
+ reg_state[R_CXL_RAS_ERR_CAP_CTRL] = 0x200;
}
static void hdm_init_common(uint32_t *reg_state, uint32_t *write_msk,
diff --git a/hw/mem/cxl_type3.c b/hw/mem/cxl_type3.c
index 6cdd988d1d..41ae19bf66 100644
--- a/hw/mem/cxl_type3.c
+++ b/hw/mem/cxl_type3.c
@@ -1,6 +1,7 @@
#include "qemu/osdep.h"
#include "qemu/units.h"
#include "qemu/error-report.h"
+#include "qapi/qapi-commands-cxl.h"
#include "hw/mem/memory-device.h"
#include "hw/mem/pc-dimm.h"
#include "hw/pci/pci.h"
@@ -323,6 +324,66 @@ static void hdm_decoder_commit(CXLType3Dev *ct3d, int which)
ARRAY_FIELD_DP32(cache_mem, CXL_HDM_DECODER0_CTRL, COMMITTED, 1);
}
+static int ct3d_qmp_uncor_err_to_cxl(CxlUncorErrorType qmp_err)
+{
+ switch (qmp_err) {
+ case CXL_UNCOR_ERROR_TYPE_CACHE_DATA_PARITY:
+ return CXL_RAS_UNC_ERR_CACHE_DATA_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_CACHE_ADDRESS_PARITY:
+ return CXL_RAS_UNC_ERR_CACHE_ADDRESS_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_CACHE_BE_PARITY:
+ return CXL_RAS_UNC_ERR_CACHE_BE_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_CACHE_DATA_ECC:
+ return CXL_RAS_UNC_ERR_CACHE_DATA_ECC;
+ case CXL_UNCOR_ERROR_TYPE_MEM_DATA_PARITY:
+ return CXL_RAS_UNC_ERR_MEM_DATA_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_MEM_ADDRESS_PARITY:
+ return CXL_RAS_UNC_ERR_MEM_ADDRESS_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_MEM_BE_PARITY:
+ return CXL_RAS_UNC_ERR_MEM_BE_PARITY;
+ case CXL_UNCOR_ERROR_TYPE_MEM_DATA_ECC:
+ return CXL_RAS_UNC_ERR_MEM_DATA_ECC;
+ case CXL_UNCOR_ERROR_TYPE_REINIT_THRESHOLD:
+ return CXL_RAS_UNC_ERR_REINIT_THRESHOLD;
+ case CXL_UNCOR_ERROR_TYPE_RSVD_ENCODING:
+ return CXL_RAS_UNC_ERR_RSVD_ENCODING;
+ case CXL_UNCOR_ERROR_TYPE_POISON_RECEIVED:
+ return CXL_RAS_UNC_ERR_POISON_RECEIVED;
+ case CXL_UNCOR_ERROR_TYPE_RECEIVER_OVERFLOW:
+ return CXL_RAS_UNC_ERR_RECEIVER_OVERFLOW;
+ case CXL_UNCOR_ERROR_TYPE_INTERNAL:
+ return CXL_RAS_UNC_ERR_INTERNAL;
+ case CXL_UNCOR_ERROR_TYPE_CXL_IDE_TX:
+ return CXL_RAS_UNC_ERR_CXL_IDE_TX;
+ case CXL_UNCOR_ERROR_TYPE_CXL_IDE_RX:
+ return CXL_RAS_UNC_ERR_CXL_IDE_RX;
+ default:
+ return -EINVAL;
+ }
+}
+
+static int ct3d_qmp_cor_err_to_cxl(CxlCorErrorType qmp_err)
+{
+ switch (qmp_err) {
+ case CXL_COR_ERROR_TYPE_CACHE_DATA_ECC:
+ return CXL_RAS_COR_ERR_CACHE_DATA_ECC;
+ case CXL_COR_ERROR_TYPE_MEM_DATA_ECC:
+ return CXL_RAS_COR_ERR_MEM_DATA_ECC;
+ case CXL_COR_ERROR_TYPE_CRC_THRESHOLD:
+ return CXL_RAS_COR_ERR_CRC_THRESHOLD;
+ case CXL_COR_ERROR_TYPE_RETRY_THRESHOLD:
+ return CXL_RAS_COR_ERR_RETRY_THRESHOLD;
+ case CXL_COR_ERROR_TYPE_CACHE_POISON_RECEIVED:
+ return CXL_RAS_COR_ERR_CACHE_POISON_RECEIVED;
+ case CXL_COR_ERROR_TYPE_MEM_POISON_RECEIVED:
+ return CXL_RAS_COR_ERR_MEM_POISON_RECEIVED;
+ case CXL_COR_ERROR_TYPE_PHYSICAL:
+ return CXL_RAS_COR_ERR_PHYSICAL;
+ default:
+ return -EINVAL;
+ }
+}
+
static void ct3d_reg_write(void *opaque, hwaddr offset, uint64_t value,
unsigned size)
{
@@ -341,6 +402,84 @@ static void ct3d_reg_write(void *opaque, hwaddr offset, uint64_t value,
should_commit = FIELD_EX32(value, CXL_HDM_DECODER0_CTRL, COMMIT);
which_hdm = 0;
break;
+ case A_CXL_RAS_UNC_ERR_STATUS:
+ {
+ uint32_t capctrl = ldl_le_p(cache_mem + R_CXL_RAS_ERR_CAP_CTRL);
+ uint32_t fe = FIELD_EX32(capctrl, CXL_RAS_ERR_CAP_CTRL, FIRST_ERROR_POINTER);
+ CXLError *cxl_err;
+ uint32_t unc_err;
+
+ /*
+ * If single bit written that corresponds to the first error
+ * pointer being cleared, update the status and header log.
+ */
+ if (!QTAILQ_EMPTY(&ct3d->error_list)) {
+ uint32_t *header_log = &cache_mem[R_CXL_RAS_ERR_HEADER0];
+
+ if ((1 << fe) ^ value) {
+ CXLError *cxl_next;
+ /*
+ * Software is using wrong flow for multiple header recording
+ * Following behaviour in PCIe r6.0 and assuming multiple
+ * header support. Imdef choice to clear all matching records
+ * if more than one bit set - which corresponds closest to
+ * behavior of hardware not capable of multiple header
+ * recording.
+ */
+ QTAILQ_FOREACH_SAFE(cxl_err, &ct3d->error_list, node, cxl_next) {
+ if ((1 << cxl_err->type) & value) {
+ QTAILQ_REMOVE(&ct3d->error_list, cxl_err, node);
+ g_free(cxl_err);
+ }
+ }
+ } else {
+ /* Done with previous FE, so drop from list */
+ cxl_err = QTAILQ_FIRST(&ct3d->error_list);
+ QTAILQ_REMOVE(&ct3d->error_list, cxl_err, node);
+ g_free(cxl_err);
+ }
+
+ /*
+ * If there is another FE, then put that in place and update
+ * the header log
+ */
+ if (!QTAILQ_EMPTY(&ct3d->error_list)) {
+ int i;
+
+ cxl_err = QTAILQ_FIRST(&ct3d->error_list);
+ for (i = 0; i < CXL_RAS_ERR_HEADER_NUM; i++) {
+ stl_le_p(header_log + i, cxl_err->header[i]);
+ }
+ capctrl = FIELD_DP32(capctrl, CXL_RAS_ERR_CAP_CTRL,
+ FIRST_ERROR_POINTER, cxl_err->type);
+ } else {
+ /*
+ * If no more errors, then follow recomendation of PCI spec
+ * r6.0 6.2.4.2 to set the first error pointer to a status
+ * bit that will never be used.
+ */
+ capctrl = FIELD_DP32(capctrl, CXL_RAS_ERR_CAP_CTRL,
+ FIRST_ERROR_POINTER,
+ CXL_RAS_UNC_ERR_CXL_UNUSED);
+ }
+ stl_le_p((uint8_t *)cache_mem + A_CXL_RAS_ERR_CAP_CTRL, capctrl);
+ }
+ unc_err = 0;
+ QTAILQ_FOREACH(cxl_err, &ct3d->error_list, node) {
+ unc_err |= 1 << cxl_err->type;
+ }
+ stl_le_p((uint8_t *)cache_mem + offset, unc_err);
+
+ return;
+ }
+ case A_CXL_RAS_COR_ERR_STATUS:
+ {
+ uint32_t rw1c = value;
+ uint32_t temp = ldl_le_p((uint8_t *)cache_mem + offset);
+ temp &= ~rw1c;
+ stl_le_p((uint8_t *)cache_mem + offset, temp);
+ return;
+ }
default:
break;
}
@@ -404,6 +543,8 @@ static void ct3_realize(PCIDevice *pci_dev, Error **errp)
unsigned short msix_num = 1;
int i, rc;
+ QTAILQ_INIT(&ct3d->error_list);
+
if (!cxl_setup_memory(ct3d, errp)) {
return;
}
@@ -631,6 +772,155 @@ static void set_lsa(CXLType3Dev *ct3d, const void *buf, uint64_t size,
*/
}
+/* For uncorrectable errors include support for multiple header recording */
+void qmp_cxl_inject_uncorrectable_errors(const char *path,
+ CXLUncorErrorRecordList *errors,
+ Error **errp)
+{
+ Object *obj = object_resolve_path(path, NULL);
+ static PCIEAERErr err = {};
+ CXLType3Dev *ct3d;
+ CXLError *cxl_err;
+ uint32_t *reg_state;
+ uint32_t unc_err;
+ bool first;
+
+ if (!obj) {
+ error_setg(errp, "Unable to resolve path");
+ return;
+ }
+
+ if (!object_dynamic_cast(obj, TYPE_CXL_TYPE3)) {
+ error_setg(errp, "Path does not point to a CXL type 3 device");
+ return;
+ }
+
+ err.status = PCI_ERR_UNC_INTN;
+ err.source_id = pci_requester_id(PCI_DEVICE(obj));
+ err.flags = 0;
+
+ ct3d = CXL_TYPE3(obj);
+
+ first = QTAILQ_EMPTY(&ct3d->error_list);
+ reg_state = ct3d->cxl_cstate.crb.cache_mem_registers;
+ while (errors) {
+ uint32List *header = errors->value->header;
+ uint8_t header_count = 0;
+ int cxl_err_code;
+
+ cxl_err_code = ct3d_qmp_uncor_err_to_cxl(errors->value->type);
+ if (cxl_err_code < 0) {
+ error_setg(errp, "Unknown error code");
+ return;
+ }
+
+ /* If the error is masked, nothting to do here */
+ if (!((1 << cxl_err_code) & ~reg_state[R_CXL_RAS_UNC_ERR_MASK])) {
+ errors = errors->next;
+ continue;
+ }
+
+ cxl_err = g_malloc0(sizeof(*cxl_err));
+ if (!cxl_err) {
+ return;
+ }
+
+ cxl_err->type = cxl_err_code;
+ while (header && header_count < 32) {
+ cxl_err->header[header_count++] = header->value;
+ header = header->next;
+ }
+ if (header_count > 32) {
+ error_setg(errp, "Header must be 32 DWORD or less");
+ return;
+ }
+ QTAILQ_INSERT_TAIL(&ct3d->error_list, cxl_err, node);
+
+ errors = errors->next;
+ }
+
+ if (first && !QTAILQ_EMPTY(&ct3d->error_list)) {
+ uint32_t *cache_mem = ct3d->cxl_cstate.crb.cache_mem_registers;
+ uint32_t capctrl = ldl_le_p((uint8_t *)cache_mem + A_CXL_RAS_ERR_CAP_CTRL);
+ uint32_t *header_log = &cache_mem[R_CXL_RAS_ERR_HEADER0];
+ int i;
+
+ cxl_err = QTAILQ_FIRST(&ct3d->error_list);
+ for (i = 0; i < CXL_RAS_ERR_HEADER_NUM; i++) {
+ stl_le_p(header_log + i, cxl_err->header[i]);
+ }
+
+ capctrl = FIELD_DP32(capctrl, CXL_RAS_ERR_CAP_CTRL,
+ FIRST_ERROR_POINTER, cxl_err->type);
+ stl_le_p(cache_mem + R_CXL_RAS_ERR_CAP_CTRL, capctrl);
+ }
+
+ unc_err = 0;
+ QTAILQ_FOREACH(cxl_err, &ct3d->error_list, node) {
+ unc_err |= (1 << cxl_err->type);
+ }
+ if (!unc_err) {
+ return;
+ }
+
+ reg_state[R_CXL_RAS_UNC_ERR_STATUS] = unc_err;
+ pcie_aer_inject_error(PCI_DEVICE(obj), &err);
+
+ return;
+}
+
+void qmp_cxl_inject_correctable_error(const char *path, CxlCorErrorType type,
+ uint32List *header, Error **errp)
+{
+ static PCIEAERErr err = {};
+ Object *obj = object_resolve_path(path, NULL);
+ CXLType3Dev *ct3d;
+ uint32_t *reg_state;
+ uint8_t header_count = 0;
+ uint32_t cor_err;
+ int cxl_err_type;
+
+ if (!obj) {
+ error_setg(errp, "Unable to resolve path");
+ return;
+ }
+ if (!object_dynamic_cast(obj, TYPE_CXL_TYPE3)) {
+ error_setg(errp, "Path does not point to a CXL type 3 device");
+ return;
+ }
+
+ err.status = PCI_ERR_COR_INTERNAL;
+ err.source_id = pci_requester_id(PCI_DEVICE(obj));
+ err.flags = PCIE_AER_ERR_IS_CORRECTABLE;
+
+ ct3d = CXL_TYPE3(obj);
+ reg_state = ct3d->cxl_cstate.crb.cache_mem_registers;
+ cor_err = reg_state[R_CXL_RAS_COR_ERR_STATUS];
+
+ cxl_err_type = ct3d_qmp_cor_err_to_cxl(type);
+ if (cxl_err_type < 0) {
+ error_setg(errp, "Invalid COR error");
+ return;
+ }
+ /* If the error is masked, nothting to do here */
+ if (!((1 << cxl_err_type) & ~reg_state[R_CXL_RAS_COR_ERR_MASK])) {
+ return;
+ }
+
+ cor_err |= (1 << cxl_err_type);
+ reg_state[R_CXL_RAS_COR_ERR_STATUS] = cor_err;
+ while (header && header_count < 32) {
+ reg_state[R_CXL_RAS_ERR_HEADER0 + header_count++] = header->value;
+ header = header->next;
+ }
+ if (header_count > 32) {
+ error_setg(errp, "Header must be 32 DWORD or less");
+ return;
+ }
+
+ pcie_aer_inject_error(PCI_DEVICE(obj), &err);
+}
+
static void ct3_class_init(ObjectClass *oc, void *data)
{
DeviceClass *dc = DEVICE_CLASS(oc);
diff --git a/hw/mem/cxl_type3_stubs.c b/hw/mem/cxl_type3_stubs.c
new file mode 100644
index 0000000000..df7f8dfa60
--- /dev/null
+++ b/hw/mem/cxl_type3_stubs.c
@@ -0,0 +1,10 @@
+
+#include "qemu/osdep.h"
+#include "qapi/qapi-commands-cxl.h"
+
+void qmp_cxl_inject_uncorrectable_errors(const char *path,
+ CXLUncorErrorRecordList *errors,
+ Error **errp) {}
+
+void qmp_cxl_inject_correctable_error(const char *path, CxlCorErrorType type,
+ uint32List *header, Error **errp) {}
diff --git a/hw/mem/meson.build b/hw/mem/meson.build
index 609b2b36fc..56c2618b84 100644
--- a/hw/mem/meson.build
+++ b/hw/mem/meson.build
@@ -4,6 +4,8 @@ mem_ss.add(when: 'CONFIG_DIMM', if_true: files('pc-dimm.c'))
mem_ss.add(when: 'CONFIG_NPCM7XX', if_true: files('npcm7xx_mc.c'))
mem_ss.add(when: 'CONFIG_NVDIMM', if_true: files('nvdimm.c'))
mem_ss.add(when: 'CONFIG_CXL_MEM_DEVICE', if_true: files('cxl_type3.c'))
+softmmu_ss.add(when: 'CONFIG_CXL_MEM_DEVICE', if_false: files('cxl_type3_stubs.c'))
+softmmu_ss.add(when: 'CONFIG_ALL', if_true: files('cxl_type3_stubs.c'))
softmmu_ss.add_all(when: 'CONFIG_MEM_DEVICE', if_true: mem_ss)
diff --git a/include/hw/cxl/cxl_component.h b/include/hw/cxl/cxl_component.h
index 5dca21e95b..8752171f70 100644
--- a/include/hw/cxl/cxl_component.h
+++ b/include/hw/cxl/cxl_component.h
@@ -67,11 +67,37 @@ CXLx_CAPABILITY_HEADER(SNOOP, 0x14)
#define CXL_RAS_REGISTERS_OFFSET 0x80
#define CXL_RAS_REGISTERS_SIZE 0x58
REG32(CXL_RAS_UNC_ERR_STATUS, CXL_RAS_REGISTERS_OFFSET)
+#define CXL_RAS_UNC_ERR_CACHE_DATA_PARITY 0
+#define CXL_RAS_UNC_ERR_CACHE_ADDRESS_PARITY 1
+#define CXL_RAS_UNC_ERR_CACHE_BE_PARITY 2
+#define CXL_RAS_UNC_ERR_CACHE_DATA_ECC 3
+#define CXL_RAS_UNC_ERR_MEM_DATA_PARITY 4
+#define CXL_RAS_UNC_ERR_MEM_ADDRESS_PARITY 5
+#define CXL_RAS_UNC_ERR_MEM_BE_PARITY 6
+#define CXL_RAS_UNC_ERR_MEM_DATA_ECC 7
+#define CXL_RAS_UNC_ERR_REINIT_THRESHOLD 8
+#define CXL_RAS_UNC_ERR_RSVD_ENCODING 9
+#define CXL_RAS_UNC_ERR_POISON_RECEIVED 10
+#define CXL_RAS_UNC_ERR_RECEIVER_OVERFLOW 11
+#define CXL_RAS_UNC_ERR_INTERNAL 14
+#define CXL_RAS_UNC_ERR_CXL_IDE_TX 15
+#define CXL_RAS_UNC_ERR_CXL_IDE_RX 16
+#define CXL_RAS_UNC_ERR_CXL_UNUSED 63 /* Magic value */
REG32(CXL_RAS_UNC_ERR_MASK, CXL_RAS_REGISTERS_OFFSET + 0x4)
REG32(CXL_RAS_UNC_ERR_SEVERITY, CXL_RAS_REGISTERS_OFFSET + 0x8)
REG32(CXL_RAS_COR_ERR_STATUS, CXL_RAS_REGISTERS_OFFSET + 0xc)
+#define CXL_RAS_COR_ERR_CACHE_DATA_ECC 0
+#define CXL_RAS_COR_ERR_MEM_DATA_ECC 1
+#define CXL_RAS_COR_ERR_CRC_THRESHOLD 2
+#define CXL_RAS_COR_ERR_RETRY_THRESHOLD 3
+#define CXL_RAS_COR_ERR_CACHE_POISON_RECEIVED 4
+#define CXL_RAS_COR_ERR_MEM_POISON_RECEIVED 5
+#define CXL_RAS_COR_ERR_PHYSICAL 6
REG32(CXL_RAS_COR_ERR_MASK, CXL_RAS_REGISTERS_OFFSET + 0x10)
REG32(CXL_RAS_ERR_CAP_CTRL, CXL_RAS_REGISTERS_OFFSET + 0x14)
+ FIELD(CXL_RAS_ERR_CAP_CTRL, FIRST_ERROR_POINTER, 0, 6)
+REG32(CXL_RAS_ERR_HEADER0, CXL_RAS_REGISTERS_OFFSET + 0x18)
+#define CXL_RAS_ERR_HEADER_NUM 32
/* Offset 0x18 - 0x58 reserved for RAS logs */
/* 8.2.5.10 - CXL Security Capability Structure */
diff --git a/include/hw/cxl/cxl_device.h b/include/hw/cxl/cxl_device.h
index 7e5ad65c1d..d589f78202 100644
--- a/include/hw/cxl/cxl_device.h
+++ b/include/hw/cxl/cxl_device.h
@@ -232,6 +232,14 @@ REG64(CXL_MEM_DEV_STS, 0)
FIELD(CXL_MEM_DEV_STS, MBOX_READY, 4, 1)
FIELD(CXL_MEM_DEV_STS, RESET_NEEDED, 5, 3)
+typedef struct CXLError {
+ QTAILQ_ENTRY(CXLError) node;
+ int type; /* Error code as per FE definition */
+ uint32_t header[32];
+} CXLError;
+
+typedef QTAILQ_HEAD(, CXLError) CXLErrorList;
+
struct CXLType3Dev {
/* Private */
PCIDevice parent_obj;
@@ -248,6 +256,9 @@ struct CXLType3Dev {
/* DOE */
DOECap doe_cdat;
+
+ /* Error injection */
+ CXLErrorList error_list;
};
#define TYPE_CXL_TYPE3 "cxl-type3"
diff --git a/qapi/cxl.json b/qapi/cxl.json
new file mode 100644
index 0000000000..3631bdba8f
--- /dev/null
+++ b/qapi/cxl.json
@@ -0,0 +1,113 @@
+# -*- Mode: Python -*-
+# vim: filetype=python
+
+##
+# = CXL devices
+##
+
+##
+# @CxlUncorErrorType:
+#
+# Type of CXL error to inject
+#
+# @cache-data-parity: Data error such as data parity or data ECC error CXL.cache
+# @cache-address-parity: Address parity or other errors associated with the
+# address field on CXL.cache
+# @cache-be-parity: Byte enable parity or other byte enable errors on CXL.cache
+# @cache-data-ecc: ECC error on CXL.cache
+# @mem-data-parity: Data error such as data parity or data ECC error on CXL.mem
+# @mem-address-parity: Address parity or other errors associated with the
+# address field on CXL.mem
+# @mem-be-parity: Byte enable parity or other byte enable errors on CXL.mem.
+# @mem-data-ecc: Data ECC error on CXL.mem.
+# @reinit-threshold: REINIT threshold hit.
+# @rsvd-encoding: Received unrecognized encoding.
+# @poison-received: Received poison from the peer.
+# @receiver-overflow: Buffer overflows (first 3 bits of header log indicate which)
+# @internal: Component specific error
+# @cxl-ide-tx: Integrity and data encryption tx error.
+# @cxl-ide-rx: Integrity and data encryption rx error.
+##
+
+{ 'enum': 'CxlUncorErrorType',
+ 'data': ['cache-data-parity',
+ 'cache-address-parity',
+ 'cache-be-parity',
+ 'cache-data-ecc',
+ 'mem-data-parity',
+ 'mem-address-parity',
+ 'mem-be-parity',
+ 'mem-data-ecc',
+ 'reinit-threshold',
+ 'rsvd-encoding',
+ 'poison-received',
+ 'receiver-overflow',
+ 'internal',
+ 'cxl-ide-tx',
+ 'cxl-ide-rx'
+ ]
+ }
+
+##
+# @CXLUncorErrorRecord:
+#
+# @type: Type of error
+# @header: 16 DWORD of header.
+##
+{ 'struct': 'CXLUncorErrorRecord',
+ 'data': {
+ 'type': 'CxlUncorErrorType',
+ 'header': [ 'uint32' ]
+ }
+}
+
+##
+# @cxl-inject-uncorrectable-errors:
+#
+# Inject mulitiple uncorrectable errors in one go.
+#
+# @path: CXL Type 3 device canonical QOM path
+#
+# @errors: Errors to inject
+##
+{ 'command': 'cxl-inject-uncorrectable-errors',
+ 'data': { 'path': 'str',
+ 'errors': [ 'CXLUncorErrorRecord' ] }}
+
+##
+# @CxlCorErrorType:
+#
+# Type of CXL correctable error to inject
+#
+# @cache-data-ecc: Data ECC error on CXL.cache
+# @mem-data-ecc: Data ECC error on CXL.mem
+# @crc-threshold: Component specific and applicable to 68 byte Flit mode only.
+# @cache-poison-received: Received poison from a peer on CXL.cache.
+# @mem-poison-received: Received poison from a peer on CXL.mem
+# @physical: Received error indication from the physical layer.
+##
+{ 'enum': 'CxlCorErrorType',
+ 'data': ['cache-data-ecc',
+ 'mem-data-ecc',
+ 'crc-threshold',
+ 'retry-threshold',
+ 'cache-poison-received',
+ 'mem-poison-received',
+ 'physical']
+}
+
+##
+# @cxl-inject-correctable-error:
+#
+# @path: CXL Type 3 device canonical QOM path
+#
+# @type: Type of error.
+#
+# @header : 16 DWORD of header.
+##
+{ 'command': 'cxl-inject-correctable-error',
+ 'data': { 'path': 'str',
+ 'type': 'CxlCorErrorType',
+ 'header': [ 'uint32' ]
+ }
+}
diff --git a/qapi/meson.build b/qapi/meson.build
index fbdb442fdf..73c3c8c31a 100644
--- a/qapi/meson.build
+++ b/qapi/meson.build
@@ -31,6 +31,7 @@ qapi_all_modules = [
'compat',
'control',
'crypto',
+ 'cxl',
'dump',
'error',
'introspect',
diff --git a/qapi/qapi-schema.json b/qapi/qapi-schema.json
index f000b90744..079f2a402a 100644
--- a/qapi/qapi-schema.json
+++ b/qapi/qapi-schema.json
@@ -95,3 +95,4 @@
{ 'include': 'pci.json' }
{ 'include': 'stats.json' }
{ 'include': 'virtio.json' }
+{ 'include': 'cxl.json' }
--
2.37.2
^ permalink raw reply related [flat|nested] 21+ messages in thread
* Re: [PATCH v2 0/7] hw/cxl: RAS error emulation and injection
2023-01-20 14:24 ` Jonathan Cameron
` (7 preceding siblings ...)
(?)
@ 2023-01-26 5:42 ` Ira Weiny
2023-01-26 10:00 ` Jonathan Cameron via
-1 siblings, 1 reply; 21+ messages in thread
From: Ira Weiny @ 2023-01-26 5:42 UTC (permalink / raw)
To: Jonathan Cameron, qemu-devel, Michael Tsirkin
Cc: Ben Widawsky, linux-cxl, linuxarm, Ira Weiny, Dave Jiang,
alison.schofield, Mike Maslenkin
Jonathan Cameron wrote:
> v2: Thanks to Mike Maslenkin for review.
> - Fix wrong parameter type to ct3d_qmp_cor_err_to_cxl()
> - Rework use of CXLError local variable in ct3d_reg_write() to improve
> code readability.
>
> CXL error reporting is complex. This series only covers the protocol
> related errors reported via PCIE AER - Ira Weiny has posted support for
> Event log based injection and I will post an update of Poison list injection
> shortly. My proposal is to upstream this one first, followed by Ira's Event
> Log series, then finally the Poison List handling. That is based on likely
> order of Linux kernel support (the support for this type of error reporting
> went in during the recent merge window, the others are still under review).
> Note we may propose other non error related features in between!
> The current revisions of all the error injection can be found at:
> https://gitlab.com/jic23/qemu/-/tree/cxl-2023-01-11
Thanks!
I see all of the patches for the event log stuff has landed in this
tree.
I see the following:
1) I have cleanup patches for[*]
a) The timestamp change
b) the g_new0() allocation
2) [PATCH v2 7/8] bswap: Add the ability to store to an unaligned 24 bit field
Was left alone. I'm good with that. But did you said you
wanted to move it into the CXL specific code. Did you
change your mind?
3) Thank you so much for fixing the optional variable stuff! :-D
4) And thanks for the CXLRetCode fix. Thanks!
5) In the latest code from 1/20 I see you fixed the static const
UUID, Thanks!
For the event stuff I have tested what is on this branch with the cleanup
patches.
I was not sure if you wanted me to re-roll them or just send fixes
patches. But I'd like to move forward with the fixes submitted if that is
ok. Those are all minor issues which don't affect the behavior much at
this point.
[*] https://lore.kernel.org/all/20230125-ira-cxl-events-fixups-2023-01-11-v1-0-1931378515f5@intel.com/
Thank you,
Ira
>
> In order to test the kernel support for RAS error handling, I previously
> provided this series via gitlab, enabling David Jiang's kernel patches
> to be tested.
>
> Now that Linux kernel support is upstream, this series is proposing the
> support for upstream inclusion in QEMU. Note that support for Multiple
> Header Recording has been added to QEMU the meantime and a kernel
> patch to use that feature sent out.
>
> https://lore.kernel.org/linux-cxl/20230113154058.16227-1-Jonathan.Cameron@huawei.com/T/#t
>
> There are two generic PCI AER precursor feature additions.
> 1) The PCI_ERR_UCOR_MASK register has not been implemented until now
> and is necessary for correct emulation.
> 2) The routing for AER errors, via existing AER error injection, only
> covered one of two paths given in the PCIe base specification,
> unfortunately not the one used by the Linux kernel CXL support.
>
> The use of MSI for the CXL root ports, both makes sense from the point
> of view of how it may well be implemented, and works around the documented
> lack of PCI interrupt routing in i386/q35. I have a hack that lets
> us correctly route those interrupts but don't currently plan to post it.
>
> The actual CXL error injection uses a new QMP interface as documented
> in the final patch description. The existing AER error injection
> internals are reused though it's HMP interface is not.
>
> Injection via QMP:
> { "execute": "qmp_capabilities" }
> ...
> { "execute": "cxl-inject-uncorrectable-errors",
> "arguments": {
> "path": "/machine/peripheral/cxl-pmem0",
> "errors": [
> {
> "type": "cache-address-parity",
> "header": [ 3, 4]
> },
> {
> "type": "cache-data-parity",
> "header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
> },
> {
> "type": "internal",
> "header": [ 1, 2, 4]
> }
> ]
> }}
> ...
> { "execute": "cxl-inject-correctable-error",
> "arguments": {
> "path": "/machine/peripheral/cxl-pmem0",
> "type": "physical",
> "header": [ 3, 4]
> } }
>
> Based on top of:
> https://lore.kernel.org/all/20230112102644.27830-1-Jonathan.Cameron@huawei.com/
> [PATCH v2 0/8] hw/cxl: CXL emulation cleanups and minor fixes for upstream
>
> Jonathan Cameron (7):
> hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register
> hw/pci/aer: Add missing routing for AER errors
> hw/pci-bridge/cxl_root_port: Wire up AER
> hw/pci-bridge/cxl_root_port: Wire up MSI
> hw/mem/cxl-type3: Add AER extended capability
> hw/pci/aer: Make PCIE AER error injection facility available for other
> emulation to use.
> hw/mem/cxl_type3: Add CXL RAS Error Injection Support.
>
> hw/cxl/cxl-component-utils.c | 4 +-
> hw/mem/cxl_type3.c | 303 +++++++++++++++++++++++++++++++++
> hw/mem/cxl_type3_stubs.c | 10 ++
> hw/mem/meson.build | 2 +
> hw/pci-bridge/cxl_root_port.c | 64 +++++++
> hw/pci/pci-internal.h | 1 -
> hw/pci/pcie_aer.c | 14 +-
> include/hw/cxl/cxl_component.h | 26 +++
> include/hw/cxl/cxl_device.h | 11 ++
> include/hw/pci/pcie_aer.h | 1 +
> include/hw/pci/pcie_regs.h | 3 +
> qapi/cxl.json | 113 ++++++++++++
> qapi/meson.build | 1 +
> qapi/qapi-schema.json | 1 +
> 14 files changed, 551 insertions(+), 3 deletions(-)
> create mode 100644 hw/mem/cxl_type3_stubs.c
> create mode 100644 qapi/cxl.json
>
> --
> 2.37.2
>
^ permalink raw reply [flat|nested] 21+ messages in thread
* Re: [PATCH v2 0/7] hw/cxl: RAS error emulation and injection
2023-01-26 5:42 ` [PATCH v2 0/7] hw/cxl: RAS error emulation and injection Ira Weiny
@ 2023-01-26 10:00 ` Jonathan Cameron via
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-26 10:00 UTC (permalink / raw)
To: Ira Weiny
Cc: qemu-devel, Michael Tsirkin, Ben Widawsky, linux-cxl, linuxarm,
Dave Jiang, alison.schofield, Mike Maslenkin
On Wed, 25 Jan 2023 21:42:04 -0800
Ira Weiny <ira.weiny@intel.com> wrote:
> Jonathan Cameron wrote:
> > v2: Thanks to Mike Maslenkin for review.
> > - Fix wrong parameter type to ct3d_qmp_cor_err_to_cxl()
> > - Rework use of CXLError local variable in ct3d_reg_write() to improve
> > code readability.
> >
> > CXL error reporting is complex. This series only covers the protocol
> > related errors reported via PCIE AER - Ira Weiny has posted support for
> > Event log based injection and I will post an update of Poison list injection
> > shortly. My proposal is to upstream this one first, followed by Ira's Event
> > Log series, then finally the Poison List handling. That is based on likely
> > order of Linux kernel support (the support for this type of error reporting
> > went in during the recent merge window, the others are still under review).
> > Note we may propose other non error related features in between!
> > The current revisions of all the error injection can be found at:
> > https://gitlab.com/jic23/qemu/-/tree/cxl-2023-01-11
>
> Thanks!
>
> I see all of the patches for the event log stuff has landed in this
> tree.
>
> I see the following:
>
> 1) I have cleanup patches for[*]
> a) The timestamp change
> b) the g_new0() allocation
>
> 2) [PATCH v2 7/8] bswap: Add the ability to store to an unaligned 24 bit field
> Was left alone. I'm good with that. But did you said you
> wanted to move it into the CXL specific code. Did you
> change your mind?
Let's proposing as a general function and see what feedback we get.
Easy to move later if we need to.
>
> 3) Thank you so much for fixing the optional variable stuff! :-D
>
> 4) And thanks for the CXLRetCode fix. Thanks!
>
> 5) In the latest code from 1/20 I see you fixed the static const
> UUID, Thanks!
>
> For the event stuff I have tested what is on this branch with the cleanup
> patches.
>
> I was not sure if you wanted me to re-roll them or just send fixes
> patches. But I'd like to move forward with the fixes submitted if that is
> ok. Those are all minor issues which don't affect the behavior much at
> this point.
I've shuffled the tree again as wasn't sure on ordering for posting for upstream,
but it should make minimal difference. Fixes are fine, I'll just squash them into
the relevant patches. Thanks!
>
> [*] https://lore.kernel.org/all/20230125-ira-cxl-events-fixups-2023-01-11-v1-0-1931378515f5@intel.com/
>
> Thank you,
> Ira
>
> >
> > In order to test the kernel support for RAS error handling, I previously
> > provided this series via gitlab, enabling David Jiang's kernel patches
> > to be tested.
> >
> > Now that Linux kernel support is upstream, this series is proposing the
> > support for upstream inclusion in QEMU. Note that support for Multiple
> > Header Recording has been added to QEMU the meantime and a kernel
> > patch to use that feature sent out.
> >
> > https://lore.kernel.org/linux-cxl/20230113154058.16227-1-Jonathan.Cameron@huawei.com/T/#t
> >
> > There are two generic PCI AER precursor feature additions.
> > 1) The PCI_ERR_UCOR_MASK register has not been implemented until now
> > and is necessary for correct emulation.
> > 2) The routing for AER errors, via existing AER error injection, only
> > covered one of two paths given in the PCIe base specification,
> > unfortunately not the one used by the Linux kernel CXL support.
> >
> > The use of MSI for the CXL root ports, both makes sense from the point
> > of view of how it may well be implemented, and works around the documented
> > lack of PCI interrupt routing in i386/q35. I have a hack that lets
> > us correctly route those interrupts but don't currently plan to post it.
> >
> > The actual CXL error injection uses a new QMP interface as documented
> > in the final patch description. The existing AER error injection
> > internals are reused though it's HMP interface is not.
> >
> > Injection via QMP:
> > { "execute": "qmp_capabilities" }
> > ...
> > { "execute": "cxl-inject-uncorrectable-errors",
> > "arguments": {
> > "path": "/machine/peripheral/cxl-pmem0",
> > "errors": [
> > {
> > "type": "cache-address-parity",
> > "header": [ 3, 4]
> > },
> > {
> > "type": "cache-data-parity",
> > "header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
> > },
> > {
> > "type": "internal",
> > "header": [ 1, 2, 4]
> > }
> > ]
> > }}
> > ...
> > { "execute": "cxl-inject-correctable-error",
> > "arguments": {
> > "path": "/machine/peripheral/cxl-pmem0",
> > "type": "physical",
> > "header": [ 3, 4]
> > } }
> >
> > Based on top of:
> > https://lore.kernel.org/all/20230112102644.27830-1-Jonathan.Cameron@huawei.com/
> > [PATCH v2 0/8] hw/cxl: CXL emulation cleanups and minor fixes for upstream
> >
> > Jonathan Cameron (7):
> > hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register
> > hw/pci/aer: Add missing routing for AER errors
> > hw/pci-bridge/cxl_root_port: Wire up AER
> > hw/pci-bridge/cxl_root_port: Wire up MSI
> > hw/mem/cxl-type3: Add AER extended capability
> > hw/pci/aer: Make PCIE AER error injection facility available for other
> > emulation to use.
> > hw/mem/cxl_type3: Add CXL RAS Error Injection Support.
> >
> > hw/cxl/cxl-component-utils.c | 4 +-
> > hw/mem/cxl_type3.c | 303 +++++++++++++++++++++++++++++++++
> > hw/mem/cxl_type3_stubs.c | 10 ++
> > hw/mem/meson.build | 2 +
> > hw/pci-bridge/cxl_root_port.c | 64 +++++++
> > hw/pci/pci-internal.h | 1 -
> > hw/pci/pcie_aer.c | 14 +-
> > include/hw/cxl/cxl_component.h | 26 +++
> > include/hw/cxl/cxl_device.h | 11 ++
> > include/hw/pci/pcie_aer.h | 1 +
> > include/hw/pci/pcie_regs.h | 3 +
> > qapi/cxl.json | 113 ++++++++++++
> > qapi/meson.build | 1 +
> > qapi/qapi-schema.json | 1 +
> > 14 files changed, 551 insertions(+), 3 deletions(-)
> > create mode 100644 hw/mem/cxl_type3_stubs.c
> > create mode 100644 qapi/cxl.json
> >
> > --
> > 2.37.2
> >
>
>
^ permalink raw reply [flat|nested] 21+ messages in thread
* Re: [PATCH v2 0/7] hw/cxl: RAS error emulation and injection
@ 2023-01-26 10:00 ` Jonathan Cameron via
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-26 10:00 UTC (permalink / raw)
To: Ira Weiny
Cc: qemu-devel, Michael Tsirkin, Ben Widawsky, linux-cxl, linuxarm,
Dave Jiang, alison.schofield, Mike Maslenkin
On Wed, 25 Jan 2023 21:42:04 -0800
Ira Weiny <ira.weiny@intel.com> wrote:
> Jonathan Cameron wrote:
> > v2: Thanks to Mike Maslenkin for review.
> > - Fix wrong parameter type to ct3d_qmp_cor_err_to_cxl()
> > - Rework use of CXLError local variable in ct3d_reg_write() to improve
> > code readability.
> >
> > CXL error reporting is complex. This series only covers the protocol
> > related errors reported via PCIE AER - Ira Weiny has posted support for
> > Event log based injection and I will post an update of Poison list injection
> > shortly. My proposal is to upstream this one first, followed by Ira's Event
> > Log series, then finally the Poison List handling. That is based on likely
> > order of Linux kernel support (the support for this type of error reporting
> > went in during the recent merge window, the others are still under review).
> > Note we may propose other non error related features in between!
> > The current revisions of all the error injection can be found at:
> > https://gitlab.com/jic23/qemu/-/tree/cxl-2023-01-11
>
> Thanks!
>
> I see all of the patches for the event log stuff has landed in this
> tree.
>
> I see the following:
>
> 1) I have cleanup patches for[*]
> a) The timestamp change
> b) the g_new0() allocation
>
> 2) [PATCH v2 7/8] bswap: Add the ability to store to an unaligned 24 bit field
> Was left alone. I'm good with that. But did you said you
> wanted to move it into the CXL specific code. Did you
> change your mind?
Let's proposing as a general function and see what feedback we get.
Easy to move later if we need to.
>
> 3) Thank you so much for fixing the optional variable stuff! :-D
>
> 4) And thanks for the CXLRetCode fix. Thanks!
>
> 5) In the latest code from 1/20 I see you fixed the static const
> UUID, Thanks!
>
> For the event stuff I have tested what is on this branch with the cleanup
> patches.
>
> I was not sure if you wanted me to re-roll them or just send fixes
> patches. But I'd like to move forward with the fixes submitted if that is
> ok. Those are all minor issues which don't affect the behavior much at
> this point.
I've shuffled the tree again as wasn't sure on ordering for posting for upstream,
but it should make minimal difference. Fixes are fine, I'll just squash them into
the relevant patches. Thanks!
>
> [*] https://lore.kernel.org/all/20230125-ira-cxl-events-fixups-2023-01-11-v1-0-1931378515f5@intel.com/
>
> Thank you,
> Ira
>
> >
> > In order to test the kernel support for RAS error handling, I previously
> > provided this series via gitlab, enabling David Jiang's kernel patches
> > to be tested.
> >
> > Now that Linux kernel support is upstream, this series is proposing the
> > support for upstream inclusion in QEMU. Note that support for Multiple
> > Header Recording has been added to QEMU the meantime and a kernel
> > patch to use that feature sent out.
> >
> > https://lore.kernel.org/linux-cxl/20230113154058.16227-1-Jonathan.Cameron@huawei.com/T/#t
> >
> > There are two generic PCI AER precursor feature additions.
> > 1) The PCI_ERR_UCOR_MASK register has not been implemented until now
> > and is necessary for correct emulation.
> > 2) The routing for AER errors, via existing AER error injection, only
> > covered one of two paths given in the PCIe base specification,
> > unfortunately not the one used by the Linux kernel CXL support.
> >
> > The use of MSI for the CXL root ports, both makes sense from the point
> > of view of how it may well be implemented, and works around the documented
> > lack of PCI interrupt routing in i386/q35. I have a hack that lets
> > us correctly route those interrupts but don't currently plan to post it.
> >
> > The actual CXL error injection uses a new QMP interface as documented
> > in the final patch description. The existing AER error injection
> > internals are reused though it's HMP interface is not.
> >
> > Injection via QMP:
> > { "execute": "qmp_capabilities" }
> > ...
> > { "execute": "cxl-inject-uncorrectable-errors",
> > "arguments": {
> > "path": "/machine/peripheral/cxl-pmem0",
> > "errors": [
> > {
> > "type": "cache-address-parity",
> > "header": [ 3, 4]
> > },
> > {
> > "type": "cache-data-parity",
> > "header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
> > },
> > {
> > "type": "internal",
> > "header": [ 1, 2, 4]
> > }
> > ]
> > }}
> > ...
> > { "execute": "cxl-inject-correctable-error",
> > "arguments": {
> > "path": "/machine/peripheral/cxl-pmem0",
> > "type": "physical",
> > "header": [ 3, 4]
> > } }
> >
> > Based on top of:
> > https://lore.kernel.org/all/20230112102644.27830-1-Jonathan.Cameron@huawei.com/
> > [PATCH v2 0/8] hw/cxl: CXL emulation cleanups and minor fixes for upstream
> >
> > Jonathan Cameron (7):
> > hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register
> > hw/pci/aer: Add missing routing for AER errors
> > hw/pci-bridge/cxl_root_port: Wire up AER
> > hw/pci-bridge/cxl_root_port: Wire up MSI
> > hw/mem/cxl-type3: Add AER extended capability
> > hw/pci/aer: Make PCIE AER error injection facility available for other
> > emulation to use.
> > hw/mem/cxl_type3: Add CXL RAS Error Injection Support.
> >
> > hw/cxl/cxl-component-utils.c | 4 +-
> > hw/mem/cxl_type3.c | 303 +++++++++++++++++++++++++++++++++
> > hw/mem/cxl_type3_stubs.c | 10 ++
> > hw/mem/meson.build | 2 +
> > hw/pci-bridge/cxl_root_port.c | 64 +++++++
> > hw/pci/pci-internal.h | 1 -
> > hw/pci/pcie_aer.c | 14 +-
> > include/hw/cxl/cxl_component.h | 26 +++
> > include/hw/cxl/cxl_device.h | 11 ++
> > include/hw/pci/pcie_aer.h | 1 +
> > include/hw/pci/pcie_regs.h | 3 +
> > qapi/cxl.json | 113 ++++++++++++
> > qapi/meson.build | 1 +
> > qapi/qapi-schema.json | 1 +
> > 14 files changed, 551 insertions(+), 3 deletions(-)
> > create mode 100644 hw/mem/cxl_type3_stubs.c
> > create mode 100644 qapi/cxl.json
> >
> > --
> > 2.37.2
> >
>
>
^ permalink raw reply [flat|nested] 21+ messages in thread
* Re: [PATCH v2 7/7] hw/mem/cxl_type3: Add CXL RAS Error Injection Support
2023-01-20 14:24 ` Jonathan Cameron via
@ 2023-01-26 17:35 ` Jonathan Cameron via
-1 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron @ 2023-01-26 17:35 UTC (permalink / raw)
To: Jonathan Cameron via
Cc: Jonathan Cameron, Michael Tsirkin, Ben Widawsky, linux-cxl,
linuxarm, Ira Weiny, Dave Jiang, alison.schofield,
Mike Maslenkin
On Fri, 20 Jan 2023 14:24:50 +0000
Jonathan Cameron via <qemu-devel@nongnu.org> wrote:
> CXL uses PCI AER Internal errors to signal to the host that an error has
> occurred. The host can then read more detailed status from the CXL RAS
> capability.
>
> For uncorrectable errors: support multiple injection in one operation
> as this is needed to reliably test multiple header logging support in an
> OS. The equivalent feature doesn't exist for correctable errors, so only
> one error need be injected at a time.
>
> Note:
> - Header content needs to be manually specified in a fashion that
> matches the specification for what can be in the header for each
> error type.
>
> Injection via QMP:
> { "execute": "qmp_capabilities" }
> ...
> { "execute": "cxl-inject-uncorrectable-errors",
> "arguments": {
> "path": "/machine/peripheral/cxl-pmem0",
> "errors": [
> {
> "type": "cache-address-parity",
> "header": [ 3, 4]
> },
> {
> "type": "cache-data-parity",
> "header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
> },
> {
> "type": "internal",
> "header": [ 1, 2, 4]
> }
> ]
> }}
> ...
> { "execute": "cxl-inject-correctable-error",
> "arguments": {
> "path": "/machine/peripheral/cxl-pmem0",
> "type": "physical",
> "header": [ 3, 4]
Correctable errors don't have header logging (I misread the spec a long time
back and somehow didn't notice this was still here) I'll fix this up for v3
along with cleaning up the messy endian handling that was pointed out off list.
I don't intent (at this point) to deal with big endian support in general,
but it would be good not to make the situation worse than it already is
even if no one is shouting for big endian yet.
Thanks,
Jonathan
^ permalink raw reply [flat|nested] 21+ messages in thread
* Re: [PATCH v2 7/7] hw/mem/cxl_type3: Add CXL RAS Error Injection Support
@ 2023-01-26 17:35 ` Jonathan Cameron via
0 siblings, 0 replies; 21+ messages in thread
From: Jonathan Cameron via @ 2023-01-26 17:35 UTC (permalink / raw)
To: Jonathan Cameron via
Cc: Jonathan Cameron, Michael Tsirkin, Ben Widawsky, linux-cxl,
linuxarm, Ira Weiny, Dave Jiang, alison.schofield,
Mike Maslenkin
On Fri, 20 Jan 2023 14:24:50 +0000
Jonathan Cameron via <qemu-devel@nongnu.org> wrote:
> CXL uses PCI AER Internal errors to signal to the host that an error has
> occurred. The host can then read more detailed status from the CXL RAS
> capability.
>
> For uncorrectable errors: support multiple injection in one operation
> as this is needed to reliably test multiple header logging support in an
> OS. The equivalent feature doesn't exist for correctable errors, so only
> one error need be injected at a time.
>
> Note:
> - Header content needs to be manually specified in a fashion that
> matches the specification for what can be in the header for each
> error type.
>
> Injection via QMP:
> { "execute": "qmp_capabilities" }
> ...
> { "execute": "cxl-inject-uncorrectable-errors",
> "arguments": {
> "path": "/machine/peripheral/cxl-pmem0",
> "errors": [
> {
> "type": "cache-address-parity",
> "header": [ 3, 4]
> },
> {
> "type": "cache-data-parity",
> "header": [0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]
> },
> {
> "type": "internal",
> "header": [ 1, 2, 4]
> }
> ]
> }}
> ...
> { "execute": "cxl-inject-correctable-error",
> "arguments": {
> "path": "/machine/peripheral/cxl-pmem0",
> "type": "physical",
> "header": [ 3, 4]
Correctable errors don't have header logging (I misread the spec a long time
back and somehow didn't notice this was still here) I'll fix this up for v3
along with cleaning up the messy endian handling that was pointed out off list.
I don't intent (at this point) to deal with big endian support in general,
but it would be good not to make the situation worse than it already is
even if no one is shouting for big endian yet.
Thanks,
Jonathan
^ permalink raw reply [flat|nested] 21+ messages in thread
end of thread, other threads:[~2023-01-26 17:36 UTC | newest]
Thread overview: 21+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-01-20 14:24 [PATCH v2 0/7] hw/cxl: RAS error emulation and injection Jonathan Cameron via
2023-01-20 14:24 ` Jonathan Cameron
2023-01-20 14:24 ` [PATCH v2 1/7] hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register Jonathan Cameron
2023-01-20 14:24 ` Jonathan Cameron via
2023-01-20 14:24 ` [PATCH v2 2/7] hw/pci/aer: Add missing routing for AER errors Jonathan Cameron via
2023-01-20 14:24 ` Jonathan Cameron
2023-01-20 14:24 ` [PATCH v2 3/7] hw/pci-bridge/cxl_root_port: Wire up AER Jonathan Cameron
2023-01-20 14:24 ` Jonathan Cameron via
2023-01-20 14:24 ` [PATCH v2 4/7] hw/pci-bridge/cxl_root_port: Wire up MSI Jonathan Cameron via
2023-01-20 14:24 ` Jonathan Cameron
2023-01-20 14:24 ` [PATCH v2 5/7] hw/mem/cxl-type3: Add AER extended capability Jonathan Cameron via
2023-01-20 14:24 ` Jonathan Cameron
2023-01-20 14:24 ` [PATCH v2 6/7] hw/pci/aer: Make PCIE AER error injection facility available for other emulation to use Jonathan Cameron
2023-01-20 14:24 ` Jonathan Cameron via
2023-01-20 14:24 ` [PATCH v2 7/7] hw/mem/cxl_type3: Add CXL RAS Error Injection Support Jonathan Cameron
2023-01-20 14:24 ` Jonathan Cameron via
2023-01-26 17:35 ` Jonathan Cameron
2023-01-26 17:35 ` Jonathan Cameron via
2023-01-26 5:42 ` [PATCH v2 0/7] hw/cxl: RAS error emulation and injection Ira Weiny
2023-01-26 10:00 ` Jonathan Cameron
2023-01-26 10:00 ` Jonathan Cameron via
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.