On Fri, 3 Oct 2025 at 05:54, Linus Torvalds
wrote:
>
> [ Miguel: Rust conflict resolution question at the end ]
>
> On Tue, 30 Sept 2025 at 21:06, Dave Airlie wrote:
> >
> > As usual, let me know if there are any problems.
>
> You are still corrupting indentati
drm/amd/display: Use mpc.preblend flag to indicate preblend
Dave Airlie (4):
Merge tag 'amd-drm-fixes-6.17-2025-09-24' of
https://gitlab.freedesktop.org/agd5f/linux into drm-fixes
Merge tag 'drm-intel-fixes-2025-09-25' of
https://gitlab.freedesktop.org/drm/i915/ker
omment"
Dan Carpenter (1):
drm/xe: Fix a NULL vs IS_ERR() in xe_vm_add_compute_exec_queue()
Daniele Ceraolo Spurio (2):
drm/xe: Fix error handling if PXP fails to start
drm/xe/guc: Set RCS/CCS yield policy
Dave Airlie (4):
Merge tag 'drm-intel-fixes-2025-09-17' of
Hi Inki,
I've already gotten the last drm-misc-next pull for 6.18 in my tree,
there are some patches being pulled in that might be for 6.19? or do
we want to get another drm-misc-next pull into 6.18?
Dave.
On Sun, 14 Sept 2025 at 13:15, Inki Dae wrote:
>
> Hi Dave and Daniel,
>
>Add DSIM br
> Questions for Community
> ===
>
> This issue raises several questions about DisplayID validation approach:
>
> 1. Is this strict validation intentional for all hardware? What are the
>security or stability reasons for treating checksum errors as fatal?
>
> 2. Are minor che
Krummrich (3):
gpu: nova-core: depend on CONFIG_64BIT
MAINTAINERS: Add drm-rust tree for Rust DRM drivers and infrastructure
MAINTAINERS: drm-misc: fix X: entries for nova/nouveau
Dave Airlie (6):
Merge tag 'drm-rust-fixes-2025-09-05' of
https://gitlab.freedesktop
On Thu, 4 Sept 2025 at 21:30, Christian König wrote:
>
> On 04.09.25 04:25, Dave Airlie wrote:
> > On Wed, 3 Sept 2025 at 00:23, Christian König
> > wrote:
> >>
> >> On 02.09.25 06:06, Dave Airlie wrote:
> >>> From: Dave Airlie
> >>>
ice remove
Alex Deucher (2):
drm/amdgpu/mes11: make MES_MISC_OP_CHANGE_CONFIG failure non-fatal
drm/amdgpu: drop hw access in non-DC audio fini
Colin Ian King (1):
drm/amd/amdgpu: Fix missing error return on kzalloc failure
Dave Airlie (4):
no
From: Dave Airlie
Before exporting a buffer, make sure it has been populated with
pages at least once.
Cc: Thomas Hellström
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/xe/xe_dma_buf.c | 12
1 file changed, 12 insertions(+)
diff --git a/drivers/gpu/drm/xe/xe_dma_buf.c b
On Wed, 3 Sept 2025 at 00:23, Christian König wrote:
>
> On 02.09.25 06:06, Dave Airlie wrote:
> > From: Dave Airlie
> >
> > This enables all the backend code to use the list lru in memcg mode,
> > and set the shrinker to be memcg aware.
> >
> > It adds
From: Dave Airlie
While discussing cgroups we noticed a problem where you could export
a BO to a dma-buf without having it ever being backed or accounted for.
This meant in low memory situations or eventually with cgroups, a
lower privledged process might cause the compositor to try and
From: Dave Airlie
Before exporting a buffer, make sure it has been populated with
pages at least once.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/nouveau/nouveau_prime.c | 12
1 file changed, 12 insertions(+)
diff --git a/drivers/gpu/drm/nouveau/nouveau_prime.c
b/drivers
On Wed, 3 Sept 2025 at 00:07, Christian König wrote:
>
>
>
> On 02.09.25 06:06, Dave Airlie wrote:
> > From: Dave Airlie
> >
> > Later memcg enablement needs the shrinker initialised before the list lru,
> > Just move it for now.
>
> Hui? That sh
On Wed, 3 Sept 2025 at 21:10, Maarten Lankhorst wrote:
>
> Hey,
>
> Den 2025-09-02 kl. 06:06, skrev Dave Airlie:
> > From: Dave Airlie
> >
> > This uses the newly introduced per-node gpu tracking stats,
> > to track GPU memory allocated via TTM and reclaima
From: Dave Airlie
Before exporting a buffer, make sure it has been populated with
pages at least once.
Reviewed-by: Christian Koenig
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c | 12
1 file changed, 12 insertions(+)
diff --git a/drivers/gpu/drm
From: Dave Airlie
This is an initial port of the TTM pools for
write combined and uncached pages to use the list_lru.
This makes the pool's more NUMA aware and avoids
needing separate NUMA pools (later commit enables this).
Cc: Christian Koenig
Cc: Johannes Weiner
Cc: Dave Chinner
S
From: Dave Airlie
Later memcg enablement needs the shrinker initialised before the list lru,
Just move it for now.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_pool.c | 22 +++---
1 file changed, 11 insertions(+), 11 deletions(-)
diff --git a/drivers/gpu/drm/ttm
From: Dave Airlie
This gets the memory sizes from the nodes and stores the limit
as 50% of those. I think eventually we should drop the limits
once we have memcg aware shrinking, but this should be more NUMA
friendly, and I think seems like what people would prefer to
happen on NUMA aware
From: Dave Airlie
This enable NUMA awareness for the shrinker on the
ttm pools.
Cc: Christian Koenig
Cc: Dave Chinner
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_pool.c | 38 +++---
1 file changed, 21 insertions(+), 17 deletions(-)
diff --git a
From: Dave Airlie
This uses the newly introduced per-node gpu tracking stats,
to track GPU memory allocated via TTM and reclaimable memory in
the TTM page pools.
These stats will be useful later for system information and
later when mem cgroups are integrated.
Cc: Christian Koenig
Cc: Matthew
From: Dave Airlie
While discussing cgroups we noticed a problem where you could export
a BO to a dma-buf without having it ever being backed or accounted for.
This meant in low memory situations or eventually with cgroups, a
lower privledged process might cause the compositor to try and
From: Dave Airlie
This enables all the backend code to use the list lru in memcg mode,
and set the shrinker to be memcg aware.
It adds the loop case for when pooled pages end up being reparented
to a higher memcg group, that newer memcg can search for them there
and take them back.
Signed-off
From: Dave Airlie
The list_lru will now handle numa for us, so need to keep
separate pool types for it. Just consoldiate into the global ones.
This adds a debugfs change to avoid dumping non-existant orders due
to this change.
Cc: Christian Koenig
Cc: Johannes Weiner
Signed-off-by: Dave
From: Dave Airlie
This adds a kconfig and a module option to turn off ttm memcg
integration completely.
When this is used, no object will ever end up using memcg aware
paths.
There is an existing workload that cgroup support might regress,
the systems are setup to allocate 1GB of uncached
From: Dave Airlie
This adds support for adding a obj cgroup to a buffer object,
and passing in the placement flags to make sure it's accounted
properly.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c| 2 ++
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
From: Dave Airlie
amdgpu wants to use the objcg api and not have to enable ifdef
around it, so just add a dummy function for the config off path.
Signed-off-by: Dave Airlie
---
include/linux/memcontrol.h | 5 +
1 file changed, 5 insertions(+)
diff --git a/include/linux/memcontrol.h b
From: Dave Airlie
This adds a placement flag that requests that any bo with this
placement flag set gets accounted for memcg if it's a system memory
allocation.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_bo.c | 2 +-
drivers/gpu/drm/ttm/ttm_bo_util.c | 6 +++---
driver
Hi all,
This is a repost with some fixes and cleanups.
I'd really like to land this into drm-next, Maarten posted xe support for this
and some other work
and I think we need to start moving this forward in tree as I'm not sure what
else I can really do
out of tree.
Differences since last posti
From: Dave Airlie
This just adds the obj cgroup pointer to the bo and tt structs,
and sets it between them.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_tt.c | 1 +
include/drm/ttm/ttm_bo.h | 6 ++
include/drm/ttm/ttm_tt.h | 2 ++
3 files changed, 9 insertions(+)
diff
From: Dave Airlie
This introduces 2 new statistics and 3 new memcontrol APIs for dealing
with GPU system memory allocations.
The stats corresponds to the same stats in the global vmstat,
for number of active GPU pages, and number of pages in pools that
can be reclaimed.
The first API charges a
From: Dave Airlie
This flag does nothing yet, but this just changes the APIs to accept
it in the future across all users.
This flag will eventually be filled out with when to account a tt
populate to a memcg.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
From: Dave Airlie
While discussing memcg intergration with gpu memory allocations,
it was pointed out that there was no numa/system counters for
GPU memory allocations.
With more integrated memory GPU server systems turning up, and
more requirements for memory tracking it seems we should start
On Tue, 2 Sept 2025 at 04:18, M Henning wrote:
>
> On Thu, Aug 28, 2025 at 10:17 PM Dave Airlie wrote:
> >
> > From: Dave Airlie
> >
> > Nouveau has code that when it gets an IRQ with no allowed handler
> > it disables it to avoid storms.
> >
>
From: Dave Airlie
While discussing cgroups we noticed a problem where you could export
a BO to a dma-buf without having it ever being backed or accounted for.
This meant in low memory situations or eventually with cgroups, a
lower privledged process might cause the compositor to try and
ode block
Carlos Llamas (1):
drm/xe: switch to local xbasename() helper
Chenyuan Yang (1):
drm/msm/dpu: Add a null ptr check for dpu_encoder_needs_modeset
Colin Ian King (1):
drm/msm: Fix dereference of pointer minor before null check
Dave Airlie (5):
Merge tag 'drm
From: Dave Airlie
Nouveau has code that when it gets an IRQ with no allowed handler
it disables it to avoid storms.
However with nonstall interrupts, we often disable them from
the drm driver, but still request their emission via the push submission.
Just don't disable nonstall irqs ev
d-off-by: Faith Ekstrand
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/nouveau/gv100_fence.c | 7 +-
.../drm/nouveau/include/nvhw/class/clc36f.h | 85 +++
2 files changed, 91 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/nouveau/gv100_fence.c
b/drivers/
> > +}
> > +
> > int
> > ga100_fifo_runl_ctor(struct nvkm_fifo *fifo)
> > {
> > @@ -599,6 +604,7 @@ ga100_fifo = {
> > .runl_ctor = ga100_fifo_runl_ctor,
> > .mmu_fault = &tu102_fifo_mmu_fault,
> > .nonstall_ctor = ga100_fifo_nonstall_ctor,
> > + .nonstall_dtor = ga
From: Dave Airlie
Nouveau has code that when it gets an IRQ with no allowed handler
it disables it to avoid storms.
However with nonstall interrupts, we often disable them from
the drm driver, but still request their emission via the push submission.
Just don't disable nonstall irqs ev
d-off-by: Faith Ekstrand
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/nouveau/gv100_fence.c | 7 +-
.../drm/nouveau/include/nvhw/class/clc36f.h | 85 +++
2 files changed, 91 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/nouveau/gv100_fence.c
b/drivers/
> Louis-Alexis Eyraud (2):
> drm/mediatek: dsi: Fix DSI host and panel bridge pre-enable order
> drm/mediatek: mtk_hdmi: Fix inverted parameters in some
> regmap_update_bits calls
Hey,
^^^ this patch is missing a S-o-b line from the committer, please fix.
Thanks,
Dave.
>
> Ma Ke (1
lo Krummrich (5):
MAINTAINERS: entry for DRM GPUVM
rust: alloc: replace aligned_size() with Kmalloc::aligned_layout()
rust: drm: ensure kmalloc() compatible Layout
rust: drm: remove pin annotations from drm::Device
rust: drm: don't pass the address of drm::Device to
ewly introduced drm_connector
parameter for drm_bridge_detect
Dave Airlie (4):
Merge tag 'drm-misc-next-fixes-2025-08-12' of
https://gitlab.freedesktop.org/drm/misc/kernel into drm-fixes
Merge tag 'amd-drm-fixes-6.17-2025-08-13' of
https://gitlab.freedesktop.org/agd5f/l
ery: fix fw based ip discovery
drm/amdgpu: add missing vram lost check for LEGACY RESET
Amber Lin (1):
drm/amdkfd: Destroy KFD debugfs after destroy KFD wq
Balasubramani Vivekanandan (1):
drm/xe/devcoredump: Defer devcoredump initialization during probe
Dave Airlie (3):
On Mon, 4 Aug 2025 at 19:22, Christian König wrote:
>
> Sorry for the delayed response, just back from vacation.
>
> On 22.07.25 01:16, David Airlie wrote:
> @@ -162,7 +164,10 @@ static struct page *ttm_pool_alloc_page(struct
> ttm_pool *pool, gfp_t gfp_flags,
> p = a
On Tue, 5 Aug 2025 at 20:58, Maarten Lankhorst
wrote:
>
> Hey,
>
> Den 2025-07-14 kl. 07:18, skrev Dave Airlie:
> > Hi all,
> >
> > This is a repost with some fixes and cleanups.
> >
> > Differences since last posting:
> > 1. Added patch 18: add
/amdgpu: track whether a queue is a kernel queue in amdgpu_mqd_prop
drm/amdgpu: update mmhub 4.1.0 client id mappings
Ashutosh Dixit (1):
drm/xe/oa: Fix static checker warning about null gt
Dan Carpenter (1):
drm/xe: Fix a NULL vs IS_ERR() bug in xe_i2c_register_adapter()
Dave A
On Thu, 31 Jul 2025 at 14:59, Linus Torvalds
wrote:
>
> On Wed, 30 Jul 2025 at 21:48, Linus Torvalds
> wrote:
> >
> > Well, it's one of these:
> >
> > 3f2b24a1ef35 drm/amd/display: Monitor patch to ignore EDID audio SAB check
> > aef3af22a456 drm/amd/display: Add definitions to support DID Ty
On Thu, 31 Jul 2025 at 14:32, Dave Airlie wrote:
>
> On Thu, 31 Jul 2025 at 14:27, Linus Torvalds
> wrote:
> >
> > On Wed, 30 Jul 2025 at 21:21, Dave Airlie wrote:
> > >
> > > Okay I don't have an rx580, but I have an rx480 which is pretty close,
&g
On Thu, 31 Jul 2025 at 14:27, Linus Torvalds
wrote:
>
> On Wed, 30 Jul 2025 at 21:21, Dave Airlie wrote:
> >
> > Okay I don't have an rx580, but I have an rx480 which is pretty close,
> > but it is booting fine with your tree at least, DP and HDMI connected,
&
On Thu, 31 Jul 2025 at 14:03, Linus Torvalds
wrote:
>
> On Wed, 30 Jul 2025 at 20:40, Linus Torvalds
> wrote:
> >
> > I'm very unhappy with the end result, because it just results in a
> > black screen at boot for me. No signal.
>
> It's not something in the merge, and it's not something in my tr
On Thu, 31 Jul 2025 at 13:41, Linus Torvalds
wrote:
>
> On Wed, 30 Jul 2025 at 20:05, Linus Torvalds
> wrote:
> >
> > Again: I'm not going to guarantee that I got it right. I *think* I did
> > - I'm not feeling particularly unhappy with my merge end result.
>
> I spoke too soon.
>
> I'm very unha
On Thu, 31 Jul 2025 at 13:05, Linus Torvalds
wrote:
>
> ,
>
> On Tue, 29 Jul 2025 at 14:06, Dave Airlie wrote:
> >
> > I've done a pass at merging mostly taking from drm-tip:
> > https://github.com/airlied/linux/tree/drm-next-6.17-rc1-merged
>
> Hmm. My
akash Deep Sarkar (1):
drm/i915/display: Fix dma_fence_wait_timeout() return value handling
Dave Airlie (2):
Merge tag 'drm-intel-fixes-2025-07-24' of
https://gitlab.freedesktop.org/drm/i915/kernel into drm-fixes
Merge tag 'drm-xe-fixes-2025-07-24' of
https://gitla
On Thu, 24 Jul 2025 at 11:59, Linus Torvalds
wrote:
>
> On Wed, 23 Jul 2025 at 17:40, Dave Airlie wrote:
> >
> > (this time for sure, plain text).
>
> I knew you could do it! Third time's the charm!
>
> I hope I don't need to worry about the branch co
uddy during resume
Ben Skeggs (1):
drm/nouveau/nvif: fix null ptr deref on pre-fermi boards
Dave Airlie (1):
Merge tag 'drm-misc-fixes-2025-07-23' of
https://gitlab.freedesktop.org/drm/misc/kernel into drm-fixes
Douglas Anderson (1):
drm/bridge: ti-sn65dsi86: Remo
Reset the clear flag in buddy during resume
Ben Skeggs (1):
drm/nouveau/nvif: fix null ptr deref on pre-fermi boards
Dave Airlie (1):
Merge tag 'drm-misc-fixes-2025-07-23' of
https://gitlab.freedesktop.org/drm/misc/kernel into drm-fixes
Douglas Anderson (1):
drm/bridge:
):
drm/nouveau/nvif: fix null ptr deref on pre-fermi boards
Dave Airlie (1):
Merge tag 'drm-misc-fixes-2025-07-23' of
https://gitlab.freedesktop.org/drm/misc/kernel into drm-fixes
Douglas Anderson (1):
drm/bridge: ti-sn65dsi86: Remove extra semicolon in
ti_sn_bridge_probe
From: Dave Airlie
This adds a kconfig and a module option to turn off ttm memcg
integration completely.
When this is used, no object will ever end up using memcg aware
paths.
There is an existing workload that cgroup support might regress,
the systems are setup to allocate 1GB of uncached
From: Dave Airlie
This adds support for adding a obj cgroup to a buffer object,
and passing in the placement flags to make sure it's accounted
properly.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c| 2 ++
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
From: Dave Airlie
amdgpu wants to use the objcg api and not have to enable ifdef
around it, so just add a dummy function for the config off path.
Signed-off-by: Dave Airlie
---
include/linux/memcontrol.h | 5 +
1 file changed, 5 insertions(+)
diff --git a/include/linux/memcontrol.h b
From: Dave Airlie
This adds a placement flag that requests that any bo with this
placement flag set gets accounted for memcg if it's a system memory
allocation.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_bo.c | 2 +-
drivers/gpu/drm/ttm/ttm_bo_util.c | 6 +++---
driver
From: Dave Airlie
This enables all the backend code to use the list lru in memcg mode,
and set the shrinker to be memcg aware.
It adds the loop case for when pooled pages end up being reparented
to a higher memcg group, that newer memcg can search for them there
and take them back.
Signed-off
From: Dave Airlie
This just adds the obj cgroup pointer to the bo and tt structs,
and sets it between them.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_tt.c | 1 +
include/drm/ttm/ttm_bo.h | 6 ++
include/drm/ttm/ttm_tt.h | 2 ++
3 files changed, 9 insertions(+)
diff
Hi all,
This is a 2nd repost with some fixes and cleanups. Original post is below.
https://lore.kernel.org/dri-devel/20250714052243.1149732-1-airl...@gmail.com/
is the 2nd post.
https://lore.kernel.org/dri-devel/20250630045005.1337339-1-airl...@gmail.com/
is the 1st post.
Differences since las
From: Dave Airlie
Later memcg enablement needs the shrinker initialised before the list lru,
Just move it for now.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_pool.c | 22 +++---
1 file changed, 11 insertions(+), 11 deletions(-)
diff --git a/drivers/gpu/drm/ttm
From: Dave Airlie
This introduces 2 new statistics and 3 new memcontrol APIs for dealing
with GPU system memory allocations.
The stats corresponds to the same stats in the global vmstat,
for number of active GPU pages, and number of pages in pools that
can be reclaimed.
The first API charges a
From: Dave Airlie
This flag does nothing yet, but this just changes the APIs to accept
it in the future across all users.
This flag will eventually be filled out with when to account a tt
populate to a memcg.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
From: Dave Airlie
This gets the memory sizes from the nodes and stores the limit
as 50% of those. I think eventually we should drop the limits
once we have memcg aware shrinking, but this should be more NUMA
friendly, and I think seems like what people would prefer to
happen on NUMA aware
From: Dave Airlie
This enable NUMA awareness for the shrinker on the
ttm pools.
Cc: Christian Koenig
Cc: Dave Chinner
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_pool.c | 38 +++---
1 file changed, 21 insertions(+), 17 deletions(-)
diff --git a
From: Dave Airlie
The list_lru will now handle numa for us, so need to keep
separate pool types for it. Just consoldiate into the global ones.
This adds a debugfs change to avoid dumping non-existant orders due
to this change.
Cc: Christian Koenig
Cc: Johannes Weiner
Signed-off-by: Dave
From: Dave Airlie
This is an initial port of the TTM pools for
write combined and uncached pages to use the list_lru.
This makes the pool's more NUMA aware and avoids
needing separate NUMA pools (later commit enables this).
Cc: Christian Koenig
Cc: Johannes Weiner
Cc: Dave Chinner
S
From: Dave Airlie
This uses the newly introduced per-node gpu tracking stats,
to track GPU memory allocated via TTM and reclaimable memory in
the TTM page pools.
These stats will be useful later for system information and
later when mem cgroups are integrated.
Cc: Christian Koenig
Cc: Matthew
From: Dave Airlie
While discussing memcg intergration with gpu memory allocations,
it was pointed out that there was no numa/system counters for
GPU memory allocations.
With more integrated memory GPU server systems turning up, and
more requirements for memory tracking it seems we should start
MT8195/88
Arnd Bergmann (1):
drm/nouveau: check ioctl command codes better
Balasubramani Vivekanandan (1):
drm/xe/mocs: Initialize MOCS index early
Clayton King (1):
drm/amd/display: Free memory allocation
Dave
From: Dave Airlie
There is an existing workload that cgroup support might regress,
the systems are setup to allocate 1GB of uncached pages at system
startup to prime the pool, then any further users will take them
from the pool. The current cgroup code might handle that, but
it also may regress
From: Dave Airlie
This adds support for adding a obj cgroup to a buffer object,
and passing in the placement flags to make sure it's accounted
properly.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c| 2 ++
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
From: Dave Airlie
This is needed to use get_obj_cgroup_from_current from a module.
Signed-off-by: Dave Airlie
---
mm/memcontrol.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 4c8ded9501c6..4c041c5b3a15 100644
--- a/mm/memcontrol.c
+++ b/mm
From: Dave Airlie
amdgpu wants to use the objcg api and not have to enable ifdef
around it, so just add a dummy function for the config off path.
Signed-off-by: Dave Airlie
---
include/linux/memcontrol.h | 5 +
1 file changed, 5 insertions(+)
diff --git a/include/linux/memcontrol.h b
From: Dave Airlie
This adds a placement flag that requests that any bo with this
placement flag set gets accounted for memcg if it's a system memory
allocation.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_bo.c | 2 +-
drivers/gpu/drm/ttm/ttm_bo_util.c | 6 +++---
driver
From: Dave Airlie
This enables all the backend code to use the list lru in memcg mode,
and set the shrinker to be memcg aware.
It adds the loop case for when pooled pages end up being reparented
to a higher memcg group, that newer memcg can search for them there
and take them back.
Signed-off
From: Dave Airlie
This just adds the obj cgroup pointer to the bo and tt structs,
and sets it between them.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_tt.c | 1 +
include/drm/ttm/ttm_bo.h | 6 ++
include/drm/ttm/ttm_tt.h | 2 ++
3 files changed, 9 insertions(+)
diff
From: Dave Airlie
Later memcg enablement needs the shrinker initialised before the list lru,
Just move it for now.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_pool.c | 22 +++---
1 file changed, 11 insertions(+), 11 deletions(-)
diff --git a/drivers/gpu/drm/ttm
From: Dave Airlie
This is need to use list lru with memcg from a module. drm/ttm
wants to use this interface.
Signed-off-by: Dave Airlie
---
mm/list_lru.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/mm/list_lru.c b/mm/list_lru.c
index 315362e3df3d..2892c1d945dd 100644
--- a/mm
From: Dave Airlie
This flag does nothing yet, but this just changes the APIs to accept
it in the future across all users.
This flag will eventually be filled out with when to account a tt
populate to a memcg.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
From: Dave Airlie
This introduces 2 new statistics and 3 new memcontrol APIs for dealing
with GPU system memory allocations.
The stats corresponds to the same stats in the global vmstat,
for number of active GPU pages, and number of pages in pools that
can be reclaimed.
The first API charges a
From: Dave Airlie
This gets the memory sizes from the nodes and stores the limit
as 50% of those. I think eventually we should drop the limits
once we have memcg aware shrinking, but this should be more NUMA
friendly, and I think seems like what people would prefer to
happen on NUMA aware
From: Dave Airlie
This enable NUMA awareness for the shrinker on the
ttm pools.
Cc: Christian Koenig
Cc: Dave Chinner
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/ttm/ttm_pool.c | 38 +++---
1 file changed, 21 insertions(+), 17 deletions(-)
diff --git a
From: Dave Airlie
The list_lru will now handle numa for us, so need to keep
separate pool types for it. Just consoldiate into the global ones.
This adds a debugfs change to avoid dumping non-existant orders due
to this change.
Cc: Christian Koenig
Cc: Johannes Weiner
Signed-off-by: Dave
From: Dave Airlie
This is an initial port of the TTM pools for
write combined and uncached pages to use the list_lru.
This makes the pool's more NUMA aware and avoids
needing separate NUMA pools (later commit enables this).
Cc: Christian Koenig
Cc: Johannes Weiner
Cc: Dave Chinner
S
From: Dave Airlie
DRM/TTM wants to use this for it's page pool
LRU tracking.
This effective is a revert of
78c0ed09131b772f062b986a2fcca6600daa6285
Author: Kairui Song
Date: Tue Nov 5 01:52:53 2024 +0800
mm/list_lru: don't export list_lru_add
Cc: Kairui Song
Cc: Johannes
From: Dave Airlie
This uses the newly introduced per-node gpu tracking stats,
to track GPU memory allocated via TTM and reclaimable memory in
the TTM page pools.
These stats will be useful later for system information and
later when mem cgroups are integrated.
Cc: Christian Koenig
Cc: Matthew
From: Dave Airlie
While discussing memcg intergration with gpu memory allocations,
it was pointed out that there was no numa/system counters for
GPU memory allocations.
With more integrated memory GPU server systems turning up, and
more requirements for memory tracking it seems we should start
Hi all,
This is a repost with some fixes and cleanups.
Differences since last posting:
1. Added patch 18: add a module option to allow pooled pages to not be stored
in the lru per-memcg
(Requested by Christian Konig)
2. Converged the naming and stats between vmstat and memcg (Suggested by
Sh
for gfx7/8
Christian König (2):
dma-buf: fix timeout handling in dma_resv_wait_timeout v2
drm/ttm: fix error handling in ttm_buffer_object_transfer
Dan Carpenter (1):
drm/i915/selftests: Change mock_request() to return error pointers
Daniele Ceraolo Spurio (1):
drm/xe: Exte
In this commit :
a72002cb181f350734108228b24c5d10d358f95a
Author: André Almeida
Date: Tue Jun 17 09:49:49 2025 -0300
drm/amdgpu: Make use of drm_wedge_task_info
@@ -164,13 +165,15 @@ static enum drm_gpu_sched_stat
amdgpu_job_timedout(struct drm_sched_job *s_job)
i
From: Dave Airlie
This fixes a bunch of command hangs after runtime suspend/resume.
This fixes a regression caused by code movement in the commit below,
the commit seems to just change timings enough to cause this to happen
now, and adding the sleep seems to avoid it.
I've spent some
From: Dave Airlie
This adds support for adding a obj cgroup to a buffer object,
and passing in the placement flags to make sure it's accounted
properly.
Signed-off-by: Dave Airlie
---
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c| 2 ++
drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
From: Dave Airlie
This is needed to use get_obj_cgroup_from_current from a module.
Signed-off-by: Dave Airlie
---
mm/memcontrol.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 87d75963a9ed..1e52e43cc239 100644
--- a/mm/memcontrol.c
+++ b/mm
From: Dave Airlie
amdgpu wants to use the objcg api and not have to enable ifdef
around it, so just add a dummy function for the config off path.
Signed-off-by: Dave Airlie
---
include/linux/memcontrol.h | 5 +
1 file changed, 5 insertions(+)
diff --git a/include/linux/memcontrol.h b
1 - 100 of 1445 matches
Mail list logo
