[PATCH 0/6] kexec fixes

[PATCH 0/6] kexec fixes

[Alex Deucher]

Fixes for kexec.  The GPU needs to be reset before the driver
can be loaded again reliably.  Needed for kexec for booting some
ppc platforms.  See:
https://bugs.freedesktop.org/show_bug.cgi?id=108585

Alex Deucher (6):
  drm/amdgpu: add need_reset_on_init asic callback (v2)
  drm/amdgpu/si: add need_reset_on_init asic callback for SI (v2)
  drm/amdgpu/cik: add need_reset_on_init asic callback for CIK (v2)
  drm/amdgpu/vi: add need_reset_on_init asic callback for VI (v2)
  drm/amdgpu/soc15: add need_reset_on_init asic callback for SOC15 (v2)
  drm/amdgpu: check if we need to reset at init time (v2)

 drivers/gpu/drm/amd/amdgpu/amdgpu.h        |  3 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 11 +++++++++++
 drivers/gpu/drm/amd/amdgpu/cik.c           | 18 ++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/si.c            |  6 ++++++
 drivers/gpu/drm/amd/amdgpu/soc15.c         | 20 ++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/vi.c            | 18 ++++++++++++++++++
 6 files changed, 76 insertions(+)

-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 1/6] drm/amdgpu: add need_reset_on_init asic callback (v2)

[Alex Deucher]

Used to determine if we need to reset the asic on init due
to the driver having been previously loaded or not shutdown
cleanly.  E.g., kexec or VM passthrough.

v2: rebase

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu.h | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index bcef6ea4bcf9..d27fa2ffb2f0 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -542,6 +542,8 @@ struct amdgpu_asic_funcs {
 	bool (*need_full_reset)(struct amdgpu_device *adev);
 	/* initialize doorbell layout for specific asic*/
 	void (*init_doorbell_index)(struct amdgpu_device *adev);
+	/* do we need to reset the asic at init time (e.g., kexec) */
+	bool (*need_reset_on_init)(struct amdgpu_device *adev);
 };
 
 /*
@@ -1042,6 +1044,7 @@ int emu_soc_asic_init(struct amdgpu_device *adev);
 #define amdgpu_asic_invalidate_hdp(adev, r) (adev)->asic_funcs->invalidate_hdp((adev), (r))
 #define amdgpu_asic_need_full_reset(adev) (adev)->asic_funcs->need_full_reset((adev))
 #define amdgpu_asic_init_doorbell_index(adev) (adev)->asic_funcs->init_doorbell_index((adev))
+#define amdgpu_asic_need_reset_on_init(adev) (adev)->asic_funcs->need_reset_on_init((adev))
 
 /* Common functions */
 bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev);
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 2/6] drm/amdgpu/si: add need_reset_on_init asic callback for SI (v2)

[Alex Deucher]

SI chips don't require a reset on reload due to the nature of
the SMU on them.

v2: rebase

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/si.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c
index f8408f88cd37..51adfe986da3 100644
--- a/drivers/gpu/drm/amd/amdgpu/si.c
+++ b/drivers/gpu/drm/amd/amdgpu/si.c
@@ -1258,6 +1258,11 @@ static bool si_need_full_reset(struct amdgpu_device *adev)
 	return true;
 }
 
+static bool si_need_reset_on_init(struct amdgpu_device *adev)
+{
+	return false;
+}
+
 static int si_get_pcie_lanes(struct amdgpu_device *adev)
 {
 	u32 link_width_cntl;
@@ -1339,6 +1344,7 @@ static const struct amdgpu_asic_funcs si_asic_funcs =
 	.flush_hdp = &si_flush_hdp,
 	.invalidate_hdp = &si_invalidate_hdp,
 	.need_full_reset = &si_need_full_reset,
+	.need_reset_on_init = &si_need_reset_on_init,
 };
 
 static uint32_t si_get_rev_id(struct amdgpu_device *adev)
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 3/6] drm/amdgpu/cik: add need_reset_on_init asic callback for CIK (v2)

[Alex Deucher]

CIK chips require a reset if the driver was previously loaded
because the SMU can only be loaded once between each reset.

v2: rebase

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/cik.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c
index 71c50d8900e3..09dfaff63502 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik.c
@@ -1741,6 +1741,23 @@ static bool cik_need_full_reset(struct amdgpu_device *adev)
 	return true;
 }
 
+static bool cik_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 clock_cntl, pc;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* check if the SMC is already running */
+	clock_cntl = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	pc = RREG32_SMC(ixSMC_PC_C);
+	if ((0 == REG_GET_FIELD(clock_cntl, SMC_SYSCON_CLOCK_CNTL_0, ck_disable)) &&
+	    (0x20100 <= pc))
+		return true;
+
+	return false;
+}
+
 static const struct amdgpu_asic_funcs cik_asic_funcs =
 {
 	.read_disabled_bios = &cik_read_disabled_bios,
@@ -1756,6 +1773,7 @@ static const struct amdgpu_asic_funcs cik_asic_funcs =
 	.invalidate_hdp = &cik_invalidate_hdp,
 	.need_full_reset = &cik_need_full_reset,
 	.init_doorbell_index = &legacy_doorbell_index_init,
+	.need_reset_on_init = &cik_need_reset_on_init,
 };
 
 static int cik_common_early_init(void *handle)
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 4/6] drm/amdgpu/vi: add need_reset_on_init asic callback for VI (v2)

[Alex Deucher]

VI chips require a reset if the driver was previously loaded
because the SMU can only be loaded once between each reset.

v2: rebase

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/vi.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c
index 03e7be595a0d..60ce2c1dac5c 100644
--- a/drivers/gpu/drm/amd/amdgpu/vi.c
+++ b/drivers/gpu/drm/amd/amdgpu/vi.c
@@ -941,6 +941,23 @@ static bool vi_need_full_reset(struct amdgpu_device *adev)
 	}
 }
 
+static bool vi_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 clock_cntl, pc;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* check if the SMC is already running */
+	clock_cntl = RREG32_SMC(ixSMC_SYSCON_CLOCK_CNTL_0);
+	pc = RREG32_SMC(ixSMC_PC_C);
+	if ((0 == REG_GET_FIELD(clock_cntl, SMC_SYSCON_CLOCK_CNTL_0, ck_disable)) &&
+	    (0x20100 <= pc))
+		return true;
+
+	return false;
+}
+
 static const struct amdgpu_asic_funcs vi_asic_funcs =
 {
 	.read_disabled_bios = &vi_read_disabled_bios,
@@ -956,6 +973,7 @@ static const struct amdgpu_asic_funcs vi_asic_funcs =
 	.invalidate_hdp = &vi_invalidate_hdp,
 	.need_full_reset = &vi_need_full_reset,
 	.init_doorbell_index = &legacy_doorbell_index_init,
+	.need_reset_on_init = &vi_need_reset_on_init,
 };
 
 #define CZ_REV_BRISTOL(rev)	 \
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 5/6] drm/amdgpu/soc15: add need_reset_on_init asic callback for SOC15 (v2)

[Alex Deucher]

SOC15 chips require a reset if the driver was previously loaded
because the PSP can only be loaded once between each reset.

v2: rebase, handle multiple asic funcs

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/soc15.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index 8849b74078d6..fb01a6cda262 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -43,6 +43,7 @@
 #include "hdp/hdp_4_0_sh_mask.h"
 #include "smuio/smuio_9_0_offset.h"
 #include "smuio/smuio_9_0_sh_mask.h"
+#include "mp/mp_9_0_offset.h"
 
 #include "soc15.h"
 #include "soc15_common.h"
@@ -602,6 +603,23 @@ static bool soc15_need_full_reset(struct amdgpu_device *adev)
 	return true;
 }
 
+static bool soc15_need_reset_on_init(struct amdgpu_device *adev)
+{
+	u32 sol_reg;
+
+	if (adev->flags & AMD_IS_APU)
+		return false;
+
+	/* Check sOS sign of life register to confirm sys driver and sOS
+	 * are already been loaded.
+	 */
+	sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
+	if (sol_reg)
+		return true;
+
+	return false;
+}
+
 static const struct amdgpu_asic_funcs soc15_asic_funcs =
 {
 	.read_disabled_bios = &soc15_read_disabled_bios,
@@ -617,6 +635,7 @@ static const struct amdgpu_asic_funcs soc15_asic_funcs =
 	.invalidate_hdp = &soc15_invalidate_hdp,
 	.need_full_reset = &soc15_need_full_reset,
 	.init_doorbell_index = &vega10_doorbell_index_init,
+	.need_reset_on_init = &soc15_need_reset_on_init,
 };
 
 static const struct amdgpu_asic_funcs vega20_asic_funcs =
@@ -634,6 +653,7 @@ static const struct amdgpu_asic_funcs vega20_asic_funcs =
 	.invalidate_hdp = &soc15_invalidate_hdp,
 	.need_full_reset = &soc15_need_full_reset,
 	.init_doorbell_index = &vega20_doorbell_index_init,
+	.need_reset_on_init = &soc15_need_reset_on_init,
 };
 
 static int soc15_common_early_init(void *handle)
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

[PATCH 6/6] drm/amdgpu: check if we need to reset at init time (v2)

[Alex Deucher]

To deal with situations like kexec or GPU VM passthrough
where the device may have been used previously without a
proper GPU reset between.

v2: rebase

bug: https://bugs.freedesktop.org/show_bug.cgi?id=108585
bug: https://bugs.freedesktop.org/show_bug.cgi?id=108754
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 1a558dc41ba6..8a61764e64cf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2550,6 +2550,17 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	/* detect if we are with an SRIOV vbios */
 	amdgpu_device_detect_sriov_bios(adev);
 
+	/* check if we need to reset the asic
+	 *  E.g., driver was not cleanly unloaded previously, etc.
+	 */
+	if (amdgpu_asic_need_reset_on_init(adev)) {
+		r = amdgpu_asic_reset(adev);
+		if (r) {
+			dev_err(adev->dev, "asic reset on init failed\n");
+			goto failed;
+		}
+	}
+
 	/* Post card if necessary */
 	if (amdgpu_device_need_post(adev)) {
 		if (!adev->bios) {
-- 
2.20.1

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH 0/6] kexec fixes

[Kuehling, Felix]

Thanks for this. I trust that your method for detecting that SMU or PSP
are running is correct. With that caveat the series is Reviewed-by:
Felix Kuehling <Felix.Kuehling@amd.com>

Regards,
  Felix

On 2019-01-08 5:11 p.m., Alex Deucher wrote:
> Fixes for kexec.  The GPU needs to be reset before the driver
> can be loaded again reliably.  Needed for kexec for booting some
> ppc platforms.  See:
> https://bugs.freedesktop.org/show_bug.cgi?id=108585
>
> Alex Deucher (6):
>   drm/amdgpu: add need_reset_on_init asic callback (v2)
>   drm/amdgpu/si: add need_reset_on_init asic callback for SI (v2)
>   drm/amdgpu/cik: add need_reset_on_init asic callback for CIK (v2)
>   drm/amdgpu/vi: add need_reset_on_init asic callback for VI (v2)
>   drm/amdgpu/soc15: add need_reset_on_init asic callback for SOC15 (v2)
>   drm/amdgpu: check if we need to reset at init time (v2)
>
>  drivers/gpu/drm/amd/amdgpu/amdgpu.h        |  3 +++
>  drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 11 +++++++++++
>  drivers/gpu/drm/amd/amdgpu/cik.c           | 18 ++++++++++++++++++
>  drivers/gpu/drm/amd/amdgpu/si.c            |  6 ++++++
>  drivers/gpu/drm/amd/amdgpu/soc15.c         | 20 ++++++++++++++++++++
>  drivers/gpu/drm/amd/amdgpu/vi.c            | 18 ++++++++++++++++++
>  6 files changed, 76 insertions(+)
>
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Re: [PATCH 0/6] kexec fixes

[Deucher, Alexander]

[-- Attachment #1.1: Type: text/plain, Size: 1667 bytes --]

Yes, the checks are taken from the existing SMU and PSP code.


Alex

________________________________
From: Kuehling, Felix
Sent: Tuesday, January 8, 2019 6:22:48 PM
To: Alex Deucher; amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org
Cc: Deucher, Alexander
Subject: Re: [PATCH 0/6] kexec fixes

Thanks for this. I trust that your method for detecting that SMU or PSP
are running is correct. With that caveat the series is Reviewed-by:
Felix Kuehling <Felix.Kuehling-5C7GfCeVMHo@public.gmane.org>

Regards,
  Felix

On 2019-01-08 5:11 p.m., Alex Deucher wrote:
> Fixes for kexec.  The GPU needs to be reset before the driver
> can be loaded again reliably.  Needed for kexec for booting some
> ppc platforms.  See:
> https://bugs.freedesktop.org/show_bug.cgi?id=108585
>
> Alex Deucher (6):
>   drm/amdgpu: add need_reset_on_init asic callback (v2)
>   drm/amdgpu/si: add need_reset_on_init asic callback for SI (v2)
>   drm/amdgpu/cik: add need_reset_on_init asic callback for CIK (v2)
>   drm/amdgpu/vi: add need_reset_on_init asic callback for VI (v2)
>   drm/amdgpu/soc15: add need_reset_on_init asic callback for SOC15 (v2)
>   drm/amdgpu: check if we need to reset at init time (v2)
>
>  drivers/gpu/drm/amd/amdgpu/amdgpu.h        |  3 +++
>  drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 11 +++++++++++
>  drivers/gpu/drm/amd/amdgpu/cik.c           | 18 ++++++++++++++++++
>  drivers/gpu/drm/amd/amdgpu/si.c            |  6 ++++++
>  drivers/gpu/drm/amd/amdgpu/soc15.c         | 20 ++++++++++++++++++++
>  drivers/gpu/drm/amd/amdgpu/vi.c            | 18 ++++++++++++++++++
>  6 files changed, 76 insertions(+)
>

[-- Attachment #1.2: Type: text/html, Size: 3426 bytes --]

[-- Attachment #2: Type: text/plain, Size: 154 bytes --]

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

RE: [PATCH 6/6] drm/amdgpu: check if we need to reset at init time (v2)

[Quan, Evan]

Series is Reviewed-by: Evan Quan <evan.quan@amd.com>

> -----Original Message-----
> From: amd-gfx <amd-gfx-bounces@lists.freedesktop.org> On Behalf Of Alex
> Deucher
> Sent: 2019年1月9日 6:12
> To: amd-gfx@lists.freedesktop.org
> Cc: Deucher, Alexander <Alexander.Deucher@amd.com>
> Subject: [PATCH 6/6] drm/amdgpu: check if we need to reset at init time (v2)
> 
> To deal with situations like kexec or GPU VM passthrough where the device
> may have been used previously without a proper GPU reset between.
> 
> v2: rebase
> 
> bug: https://bugs.freedesktop.org/show_bug.cgi?id=108585
> bug: https://bugs.freedesktop.org/show_bug.cgi?id=108754
> Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 1a558dc41ba6..8a61764e64cf 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -2550,6 +2550,17 @@ int amdgpu_device_init(struct amdgpu_device
> *adev,
>  	/* detect if we are with an SRIOV vbios */
>  	amdgpu_device_detect_sriov_bios(adev);
> 
> +	/* check if we need to reset the asic
> +	 *  E.g., driver was not cleanly unloaded previously, etc.
> +	 */
> +	if (amdgpu_asic_need_reset_on_init(adev)) {
> +		r = amdgpu_asic_reset(adev);
> +		if (r) {
> +			dev_err(adev->dev, "asic reset on init failed\n");
> +			goto failed;
> +		}
> +	}
> +
>  	/* Post card if necessary */
>  	if (amdgpu_device_need_post(adev)) {
>  		if (!adev->bios) {
> --
> 2.20.1
> 
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx