- Linux-stable-mirror - lists.linaro.org

FAILED: patch "[PATCH] membarrier: Explicitly sync remote cores when SYNC_CORE is" failed to apply to 4.19-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.19-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 758c9373d84168dc7d039cf85a0e920046b17b41 Mon Sep 17 00:00:00 2001 From: Andy Lutomirski <luto(a)kernel.org> Date: Thu, 3 Dec 2020 21:07:05 -0800 Subject: [PATCH] membarrier: Explicitly sync remote cores when SYNC_CORE is requested membarrier() does not explicitly sync_core() remote CPUs; instead, it relies on the assumption that an IPI will result in a core sync. On x86, this may be true in practice, but it's not architecturally reliable. In particular, the SDM and APM do not appear to guarantee that interrupt delivery is serializing. While IRET does serialize, IPI return can schedule, thereby switching to another task in the same mm that was sleeping in a syscall. The new task could then SYSRET back to usermode without ever executing IRET. Make this more robust by explicitly calling sync_core_before_usermode() on remote cores. (This also helps people who search the kernel tree for instances of sync_core() and sync_core_before_usermode() -- one might be surprised that the core membarrier code doesn't currently show up in a such a search.) Fixes: 70216e18e519 ("membarrier: Provide core serializing command, *_SYNC_CORE") Signed-off-by: Andy Lutomirski <luto(a)kernel.org> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/r/776b448d5f7bd6b12690707f5ed67bcda7f1d427.16070583… diff --git a/kernel/sched/membarrier.c b/kernel/sched/membarrier.c index 7d98ef5d3bcd..1c278dff4f2d 100644 --- a/kernel/sched/membarrier.c +++ b/kernel/sched/membarrier.c @@ -38,6 +38,23 @@ static void ipi_mb(void *info) smp_mb(); /* IPIs should be serializing but paranoid. */ } +static void ipi_sync_core(void *info) +{ + /* + * The smp_mb() in membarrier after all the IPIs is supposed to + * ensure that memory on remote CPUs that occur before the IPI + * become visible to membarrier()'s caller -- see scenario B in + * the big comment at the top of this file. + * + * A sync_core() would provide this guarantee, but + * sync_core_before_usermode() might end up being deferred until + * after membarrier()'s smp_mb(). + */ + smp_mb(); /* IPIs should be serializing but paranoid. */ + + sync_core_before_usermode(); +} + static void ipi_rseq(void *info) { /* @@ -162,6 +179,7 @@ static int membarrier_private_expedited(int flags, int cpu_id) if (!(atomic_read(&mm->membarrier_state) & MEMBARRIER_STATE_PRIVATE_EXPEDITED_SYNC_CORE_READY)) return -EPERM; + ipi_func = ipi_sync_core; } else if (flags == MEMBARRIER_FLAG_RSEQ) { if (!IS_ENABLED(CONFIG_RSEQ)) return -EINVAL;

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] membarrier: Explicitly sync remote cores when SYNC_CORE is" failed to apply to 5.4-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 758c9373d84168dc7d039cf85a0e920046b17b41 Mon Sep 17 00:00:00 2001 From: Andy Lutomirski <luto(a)kernel.org> Date: Thu, 3 Dec 2020 21:07:05 -0800 Subject: [PATCH] membarrier: Explicitly sync remote cores when SYNC_CORE is requested membarrier() does not explicitly sync_core() remote CPUs; instead, it relies on the assumption that an IPI will result in a core sync. On x86, this may be true in practice, but it's not architecturally reliable. In particular, the SDM and APM do not appear to guarantee that interrupt delivery is serializing. While IRET does serialize, IPI return can schedule, thereby switching to another task in the same mm that was sleeping in a syscall. The new task could then SYSRET back to usermode without ever executing IRET. Make this more robust by explicitly calling sync_core_before_usermode() on remote cores. (This also helps people who search the kernel tree for instances of sync_core() and sync_core_before_usermode() -- one might be surprised that the core membarrier code doesn't currently show up in a such a search.) Fixes: 70216e18e519 ("membarrier: Provide core serializing command, *_SYNC_CORE") Signed-off-by: Andy Lutomirski <luto(a)kernel.org> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/r/776b448d5f7bd6b12690707f5ed67bcda7f1d427.16070583… diff --git a/kernel/sched/membarrier.c b/kernel/sched/membarrier.c index 7d98ef5d3bcd..1c278dff4f2d 100644 --- a/kernel/sched/membarrier.c +++ b/kernel/sched/membarrier.c @@ -38,6 +38,23 @@ static void ipi_mb(void *info) smp_mb(); /* IPIs should be serializing but paranoid. */ } +static void ipi_sync_core(void *info) +{ + /* + * The smp_mb() in membarrier after all the IPIs is supposed to + * ensure that memory on remote CPUs that occur before the IPI + * become visible to membarrier()'s caller -- see scenario B in + * the big comment at the top of this file. + * + * A sync_core() would provide this guarantee, but + * sync_core_before_usermode() might end up being deferred until + * after membarrier()'s smp_mb(). + */ + smp_mb(); /* IPIs should be serializing but paranoid. */ + + sync_core_before_usermode(); +} + static void ipi_rseq(void *info) { /* @@ -162,6 +179,7 @@ static int membarrier_private_expedited(int flags, int cpu_id) if (!(atomic_read(&mm->membarrier_state) & MEMBARRIER_STATE_PRIVATE_EXPEDITED_SYNC_CORE_READY)) return -EPERM; + ipi_func = ipi_sync_core; } else if (flags == MEMBARRIER_FLAG_RSEQ) { if (!IS_ENABLED(CONFIG_RSEQ)) return -EINVAL;

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] membarrier: Explicitly sync remote cores when SYNC_CORE is" failed to apply to 5.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 758c9373d84168dc7d039cf85a0e920046b17b41 Mon Sep 17 00:00:00 2001 From: Andy Lutomirski <luto(a)kernel.org> Date: Thu, 3 Dec 2020 21:07:05 -0800 Subject: [PATCH] membarrier: Explicitly sync remote cores when SYNC_CORE is requested membarrier() does not explicitly sync_core() remote CPUs; instead, it relies on the assumption that an IPI will result in a core sync. On x86, this may be true in practice, but it's not architecturally reliable. In particular, the SDM and APM do not appear to guarantee that interrupt delivery is serializing. While IRET does serialize, IPI return can schedule, thereby switching to another task in the same mm that was sleeping in a syscall. The new task could then SYSRET back to usermode without ever executing IRET. Make this more robust by explicitly calling sync_core_before_usermode() on remote cores. (This also helps people who search the kernel tree for instances of sync_core() and sync_core_before_usermode() -- one might be surprised that the core membarrier code doesn't currently show up in a such a search.) Fixes: 70216e18e519 ("membarrier: Provide core serializing command, *_SYNC_CORE") Signed-off-by: Andy Lutomirski <luto(a)kernel.org> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/r/776b448d5f7bd6b12690707f5ed67bcda7f1d427.16070583… diff --git a/kernel/sched/membarrier.c b/kernel/sched/membarrier.c index 7d98ef5d3bcd..1c278dff4f2d 100644 --- a/kernel/sched/membarrier.c +++ b/kernel/sched/membarrier.c @@ -38,6 +38,23 @@ static void ipi_mb(void *info) smp_mb(); /* IPIs should be serializing but paranoid. */ } +static void ipi_sync_core(void *info) +{ + /* + * The smp_mb() in membarrier after all the IPIs is supposed to + * ensure that memory on remote CPUs that occur before the IPI + * become visible to membarrier()'s caller -- see scenario B in + * the big comment at the top of this file. + * + * A sync_core() would provide this guarantee, but + * sync_core_before_usermode() might end up being deferred until + * after membarrier()'s smp_mb(). + */ + smp_mb(); /* IPIs should be serializing but paranoid. */ + + sync_core_before_usermode(); +} + static void ipi_rseq(void *info) { /* @@ -162,6 +179,7 @@ static int membarrier_private_expedited(int flags, int cpu_id) if (!(atomic_read(&mm->membarrier_state) & MEMBARRIER_STATE_PRIVATE_EXPEDITED_SYNC_CORE_READY)) return -EPERM; + ipi_func = ipi_sync_core; } else if (flags == MEMBARRIER_FLAG_RSEQ) { if (!IS_ENABLED(CONFIG_RSEQ)) return -EINVAL;

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] membarrier: Add an actual barrier before rseq_preempt()" failed to apply to 5.4-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 2ecedd7569080fd05c1a457e8af2165afecfa29f Mon Sep 17 00:00:00 2001 From: Andy Lutomirski <luto(a)kernel.org> Date: Thu, 3 Dec 2020 21:07:04 -0800 Subject: [PATCH] membarrier: Add an actual barrier before rseq_preempt() It seems that most RSEQ membarrier users will expect any stores done before the membarrier() syscall to be visible to the target task(s). While this is extremely likely to be true in practice, nothing actually guarantees it by a strict reading of the x86 manuals. Rather than providing this guarantee by accident and potentially causing a problem down the road, just add an explicit barrier. Fixes: 70216e18e519 ("membarrier: Provide core serializing command, *_SYNC_CORE") Signed-off-by: Andy Lutomirski <luto(a)kernel.org> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/r/d3e7197e034fa4852afcf370ca49c30496e58e40.16070583… diff --git a/kernel/sched/membarrier.c b/kernel/sched/membarrier.c index e23e74d52db5..7d98ef5d3bcd 100644 --- a/kernel/sched/membarrier.c +++ b/kernel/sched/membarrier.c @@ -40,6 +40,14 @@ static void ipi_mb(void *info) static void ipi_rseq(void *info) { + /* + * Ensure that all stores done by the calling thread are visible + * to the current task before the current task resumes. We could + * probably optimize this away on most architectures, but by the + * time we've already sent an IPI, the cost of the extra smp_mb() + * is negligible. + */ + smp_mb(); rseq_preempt(current); }

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] membarrier: Add an actual barrier before rseq_preempt()" failed to apply to 5.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 2ecedd7569080fd05c1a457e8af2165afecfa29f Mon Sep 17 00:00:00 2001 From: Andy Lutomirski <luto(a)kernel.org> Date: Thu, 3 Dec 2020 21:07:04 -0800 Subject: [PATCH] membarrier: Add an actual barrier before rseq_preempt() It seems that most RSEQ membarrier users will expect any stores done before the membarrier() syscall to be visible to the target task(s). While this is extremely likely to be true in practice, nothing actually guarantees it by a strict reading of the x86 manuals. Rather than providing this guarantee by accident and potentially causing a problem down the road, just add an explicit barrier. Fixes: 70216e18e519 ("membarrier: Provide core serializing command, *_SYNC_CORE") Signed-off-by: Andy Lutomirski <luto(a)kernel.org> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/r/d3e7197e034fa4852afcf370ca49c30496e58e40.16070583… diff --git a/kernel/sched/membarrier.c b/kernel/sched/membarrier.c index e23e74d52db5..7d98ef5d3bcd 100644 --- a/kernel/sched/membarrier.c +++ b/kernel/sched/membarrier.c @@ -40,6 +40,14 @@ static void ipi_mb(void *info) static void ipi_rseq(void *info) { + /* + * Ensure that all stores done by the calling thread are visible + * to the current task before the current task resumes. We could + * probably optimize this away on most architectures, but by the + * time we've already sent an IPI, the cost of the extra smp_mb() + * is negligible. + */ + smp_mb(); rseq_preempt(current); }

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] membarrier: Add an actual barrier before rseq_preempt()" failed to apply to 4.19-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.19-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 2ecedd7569080fd05c1a457e8af2165afecfa29f Mon Sep 17 00:00:00 2001 From: Andy Lutomirski <luto(a)kernel.org> Date: Thu, 3 Dec 2020 21:07:04 -0800 Subject: [PATCH] membarrier: Add an actual barrier before rseq_preempt() It seems that most RSEQ membarrier users will expect any stores done before the membarrier() syscall to be visible to the target task(s). While this is extremely likely to be true in practice, nothing actually guarantees it by a strict reading of the x86 manuals. Rather than providing this guarantee by accident and potentially causing a problem down the road, just add an explicit barrier. Fixes: 70216e18e519 ("membarrier: Provide core serializing command, *_SYNC_CORE") Signed-off-by: Andy Lutomirski <luto(a)kernel.org> Signed-off-by: Thomas Gleixner <tglx(a)linutronix.de> Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers(a)efficios.com> Cc: stable(a)vger.kernel.org Link: https://lore.kernel.org/r/d3e7197e034fa4852afcf370ca49c30496e58e40.16070583… diff --git a/kernel/sched/membarrier.c b/kernel/sched/membarrier.c index e23e74d52db5..7d98ef5d3bcd 100644 --- a/kernel/sched/membarrier.c +++ b/kernel/sched/membarrier.c @@ -40,6 +40,14 @@ static void ipi_mb(void *info) static void ipi_rseq(void *info) { + /* + * Ensure that all stores done by the calling thread are visible + * to the current task before the current task resumes. We could + * probably optimize this away on most architectures, but by the + * time we've already sent an IPI, the cost of the extra smp_mb() + * is negligible. + */ + smp_mb(); rseq_preempt(current); }

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] drm/amd/pm: update smu10.h WORKLOAD_PPLIB setting for raven" failed to apply to 5.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From c9918d1f63a3e77ec20997a77c997a6fa7282f2f Mon Sep 17 00:00:00 2001 From: Changfeng <Changfeng.Zhu(a)amd.com> Date: Mon, 7 Dec 2020 15:42:29 +0800 Subject: [PATCH] drm/amd/pm: update smu10.h WORKLOAD_PPLIB setting for raven When using old WORKLOAD_PPLIB setting in smu10.h, there is problem that it can't be able to switch to mak gpu clk during compute workload. It needs to update WORKLOAD_PPLIB setting to fix this issue. Signed-off-by: Changfeng <Changfeng.Zhu(a)amd.com> Reviewed-by: Huang Rui <ray.huang(a)amd.com> Acked-by: Alex Deucher <alexander.deucher(a)amd.com> Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Cc: stable(a)vger.kernel.org diff --git a/drivers/gpu/drm/amd/pm/inc/smu10.h b/drivers/gpu/drm/amd/pm/inc/smu10.h index b96520528240..9e837a5014c5 100644 --- a/drivers/gpu/drm/amd/pm/inc/smu10.h +++ b/drivers/gpu/drm/amd/pm/inc/smu10.h @@ -136,14 +136,12 @@ #define FEATURE_CORE_CSTATES_MASK (1 << FEATURE_CORE_CSTATES_BIT) /* Workload bits */ -#define WORKLOAD_DEFAULT_BIT 0 -#define WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT 1 -#define WORKLOAD_PPLIB_POWER_SAVING_BIT 2 -#define WORKLOAD_PPLIB_VIDEO_BIT 3 -#define WORKLOAD_PPLIB_VR_BIT 4 -#define WORKLOAD_PPLIB_COMPUTE_BIT 5 -#define WORKLOAD_PPLIB_CUSTOM_BIT 6 -#define WORKLOAD_PPLIB_COUNT 7 +#define WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT 0 +#define WORKLOAD_PPLIB_VIDEO_BIT 2 +#define WORKLOAD_PPLIB_VR_BIT 3 +#define WORKLOAD_PPLIB_COMPUTE_BIT 4 +#define WORKLOAD_PPLIB_CUSTOM_BIT 5 +#define WORKLOAD_PPLIB_COUNT 6 typedef struct { /* MP1_EXT_SCRATCH0 */ diff --git a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu10_hwmgr.c b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu10_hwmgr.c index cf60f3992303..e6f40ee9f313 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu10_hwmgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu10_hwmgr.c @@ -1297,15 +1297,9 @@ static int conv_power_profile_to_pplib_workload(int power_profile) int pplib_workload = 0; switch (power_profile) { - case PP_SMC_POWER_PROFILE_BOOTUP_DEFAULT: - pplib_workload = WORKLOAD_DEFAULT_BIT; - break; case PP_SMC_POWER_PROFILE_FULLSCREEN3D: pplib_workload = WORKLOAD_PPLIB_FULL_SCREEN_3D_BIT; break; - case PP_SMC_POWER_PROFILE_POWERSAVING: - pplib_workload = WORKLOAD_PPLIB_POWER_SAVING_BIT; - break; case PP_SMC_POWER_PROFILE_VIDEO: pplib_workload = WORKLOAD_PPLIB_VIDEO_BIT; break; @@ -1315,6 +1309,9 @@ static int conv_power_profile_to_pplib_workload(int power_profile) case PP_SMC_POWER_PROFILE_COMPUTE: pplib_workload = WORKLOAD_PPLIB_COMPUTE_BIT; break; + case PP_SMC_POWER_PROFILE_CUSTOM: + pplib_workload = WORKLOAD_PPLIB_CUSTOM_BIT; + break; } return pplib_workload;

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] drm/amd/pm: typo fix (CUSTOM -> COMPUTE)" failed to apply to 5.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From c5b58c8c860db330c0b8b891b69014ee9d470dab Mon Sep 17 00:00:00 2001 From: Evan Quan <evan.quan(a)amd.com> Date: Wed, 9 Dec 2020 16:34:22 +0800 Subject: [PATCH] drm/amd/pm: typo fix (CUSTOM -> COMPUTE) The "COMPUTE" was wrongly spelled as "CUSTOM". Signed-off-by: Evan Quan <evan.quan(a)amd.com> Reviewed-by: Alex Deucher <alexander.deucher(a)amd.com> Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Cc: stable(a)vger.kernel.org # 5.9.x diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c index 895d89bea7fa..cf7c4f0e0a0b 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c @@ -217,7 +217,7 @@ static struct cmn2asic_mapping sienna_cichlid_workload_map[PP_SMC_POWER_PROFILE_ WORKLOAD_MAP(PP_SMC_POWER_PROFILE_POWERSAVING, WORKLOAD_PPLIB_POWER_SAVING_BIT), WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VIDEO, WORKLOAD_PPLIB_VIDEO_BIT), WORKLOAD_MAP(PP_SMC_POWER_PROFILE_VR, WORKLOAD_PPLIB_VR_BIT), - WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE, WORKLOAD_PPLIB_CUSTOM_BIT), + WORKLOAD_MAP(PP_SMC_POWER_PROFILE_COMPUTE, WORKLOAD_PPLIB_COMPUTE_BIT), WORKLOAD_MAP(PP_SMC_POWER_PROFILE_CUSTOM, WORKLOAD_PPLIB_CUSTOM_BIT), };

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] drm/amdgpu: fix size calculation with stolen vga memory" failed to apply to 5.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 157fe68d74c2ad2db438c91af9ed3d3a51de4ed7 Mon Sep 17 00:00:00 2001 From: Alex Deucher <alexander.deucher(a)amd.com> Date: Mon, 7 Dec 2020 13:12:29 -0500 Subject: [PATCH] drm/amdgpu: fix size calculation with stolen vga memory MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit If we need to keep the stolen vga memory, make sure it is at least as big as the legacy vga size. Acked-by: Christian König <christian.koenig(a)amd.com> Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Cc: stable(a)vger.kernel.org diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 36604d751d62..3e4892b7b7d3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -499,6 +499,9 @@ void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev) else size = amdgpu_gmc_get_vbios_fb_size(adev); + if (adev->mman.keep_stolen_vga_memory) + size = max(size, (unsigned)AMDGPU_VBIOS_VGA_ALLOCATION); + /* set to 0 if the pre-OS buffer uses up most of vram */ if ((adev->gmc.real_vram_size - size) < (8 * 1024 * 1024)) size = 0;

4 years, 9 months

1
0
0 0

FAILED: patch "[PATCH] drm/amdgpu/powerplay: parse fan table for CI asics" failed to apply to 5.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 0e830d2872cf6e75ef6619edd23050ddf3673358 Mon Sep 17 00:00:00 2001 From: Alex Deucher <alexander.deucher(a)amd.com> Date: Tue, 1 Dec 2020 17:44:58 -0500 Subject: [PATCH] drm/amdgpu/powerplay: parse fan table for CI asics Set up all the parameters required for SMU fan control if supported. Bug: https://bugzilla.kernel.org/show_bug.cgi?id=201539 Acked-by: Evan Quan <evan.quan(a)amd.com> Signed-off-by: Alex Deucher <alexander.deucher(a)amd.com> Cc: stable(a)vger.kernel.org diff --git a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/processpptables.c b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/processpptables.c index 719597c5d27d..6606511891e3 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/processpptables.c +++ b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/processpptables.c @@ -24,6 +24,8 @@ #include <linux/types.h> #include <linux/kernel.h> #include <linux/slab.h> +#include <linux/pci.h> + #include <drm/amdgpu_drm.h> #include "processpptables.h" #include <atom-types.h> @@ -984,6 +986,8 @@ static int init_thermal_controller( struct pp_hwmgr *hwmgr, const ATOM_PPLIB_POWERPLAYTABLE *powerplay_table) { + struct amdgpu_device *adev = hwmgr->adev; + hwmgr->thermal_controller.ucType = powerplay_table->sThermalController.ucType; hwmgr->thermal_controller.ucI2cLine = @@ -1008,7 +1012,104 @@ static int init_thermal_controller( ATOM_PP_THERMALCONTROLLER_NONE != hwmgr->thermal_controller.ucType, PHM_PlatformCaps_ThermalController); - hwmgr->thermal_controller.use_hw_fan_control = 1; + if (powerplay_table->usTableSize >= sizeof(ATOM_PPLIB_POWERPLAYTABLE3)) { + const ATOM_PPLIB_POWERPLAYTABLE3 *powerplay_table3 = + (const ATOM_PPLIB_POWERPLAYTABLE3 *)powerplay_table; + + if (0 == le16_to_cpu(powerplay_table3->usFanTableOffset)) { + hwmgr->thermal_controller.use_hw_fan_control = 1; + return 0; + } else { + const ATOM_PPLIB_FANTABLE *fan_table = + (const ATOM_PPLIB_FANTABLE *)(((unsigned long)powerplay_table) + + le16_to_cpu(powerplay_table3->usFanTableOffset)); + + if (1 <= fan_table->ucFanTableFormat) { + hwmgr->thermal_controller.advanceFanControlParameters.ucTHyst = + fan_table->ucTHyst; + hwmgr->thermal_controller.advanceFanControlParameters.usTMin = + le16_to_cpu(fan_table->usTMin); + hwmgr->thermal_controller.advanceFanControlParameters.usTMed = + le16_to_cpu(fan_table->usTMed); + hwmgr->thermal_controller.advanceFanControlParameters.usTHigh = + le16_to_cpu(fan_table->usTHigh); + hwmgr->thermal_controller.advanceFanControlParameters.usPWMMin = + le16_to_cpu(fan_table->usPWMMin); + hwmgr->thermal_controller.advanceFanControlParameters.usPWMMed = + le16_to_cpu(fan_table->usPWMMed); + hwmgr->thermal_controller.advanceFanControlParameters.usPWMHigh = + le16_to_cpu(fan_table->usPWMHigh); + hwmgr->thermal_controller.advanceFanControlParameters.usTMax = 10900; + hwmgr->thermal_controller.advanceFanControlParameters.ulCycleDelay = 100000; + + phm_cap_set(hwmgr->platform_descriptor.platformCaps, + PHM_PlatformCaps_MicrocodeFanControl); + } + + if (2 <= fan_table->ucFanTableFormat) { + const ATOM_PPLIB_FANTABLE2 *fan_table2 = + (const ATOM_PPLIB_FANTABLE2 *)(((unsigned long)powerplay_table) + + le16_to_cpu(powerplay_table3->usFanTableOffset)); + hwmgr->thermal_controller.advanceFanControlParameters.usTMax = + le16_to_cpu(fan_table2->usTMax); + } + + if (3 <= fan_table->ucFanTableFormat) { + const ATOM_PPLIB_FANTABLE3 *fan_table3 = + (const ATOM_PPLIB_FANTABLE3 *) (((unsigned long)powerplay_table) + + le16_to_cpu(powerplay_table3->usFanTableOffset)); + + hwmgr->thermal_controller.advanceFanControlParameters.ucFanControlMode = + fan_table3->ucFanControlMode; + + if ((3 == fan_table->ucFanTableFormat) && + (0x67B1 == adev->pdev->device)) + hwmgr->thermal_controller.advanceFanControlParameters.usDefaultMaxFanPWM = + 47; + else + hwmgr->thermal_controller.advanceFanControlParameters.usDefaultMaxFanPWM = + le16_to_cpu(fan_table3->usFanPWMMax); + + hwmgr->thermal_controller.advanceFanControlParameters.usDefaultFanOutputSensitivity = + 4836; + hwmgr->thermal_controller.advanceFanControlParameters.usFanOutputSensitivity = + le16_to_cpu(fan_table3->usFanOutputSensitivity); + } + + if (6 <= fan_table->ucFanTableFormat) { + const ATOM_PPLIB_FANTABLE4 *fan_table4 = + (const ATOM_PPLIB_FANTABLE4 *)(((unsigned long)powerplay_table) + + le16_to_cpu(powerplay_table3->usFanTableOffset)); + + phm_cap_set(hwmgr->platform_descriptor.platformCaps, + PHM_PlatformCaps_FanSpeedInTableIsRPM); + + hwmgr->thermal_controller.advanceFanControlParameters.usDefaultMaxFanRPM = + le16_to_cpu(fan_table4->usFanRPMMax); + } + + if (7 <= fan_table->ucFanTableFormat) { + const ATOM_PPLIB_FANTABLE5 *fan_table5 = + (const ATOM_PPLIB_FANTABLE5 *)(((unsigned long)powerplay_table) + + le16_to_cpu(powerplay_table3->usFanTableOffset)); + + if (0x67A2 == adev->pdev->device || + 0x67A9 == adev->pdev->device || + 0x67B9 == adev->pdev->device) { + phm_cap_set(hwmgr->platform_descriptor.platformCaps, + PHM_PlatformCaps_GeminiRegulatorFanControlSupport); + hwmgr->thermal_controller.advanceFanControlParameters.usFanCurrentLow = + le16_to_cpu(fan_table5->usFanCurrentLow); + hwmgr->thermal_controller.advanceFanControlParameters.usFanCurrentHigh = + le16_to_cpu(fan_table5->usFanCurrentHigh); + hwmgr->thermal_controller.advanceFanControlParameters.usFanRPMLow = + le16_to_cpu(fan_table5->usFanRPMLow); + hwmgr->thermal_controller.advanceFanControlParameters.usFanRPMHigh = + le16_to_cpu(fan_table5->usFanRPMHigh); + } + } + } + } return 0; }

4 years, 9 months

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror