On 02/09/2025 10:10, Francois Dugast wrote:
On Mon, Aug 18, 2025 at 10:39:10AM -0700, Matthew Brost wrote:
On Mon, Aug 18, 2025 at 04:21:54PM +0100, Matthew Auld wrote:
Handle the case where the hmm range partially covers a huge page (like
2M), otherwise we can potentially end up doing
struct (Matt B)
v3:
- Small kernel-doc fixes
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 67 +++-
drivers/gpu/drm/xe/xe_pt.c | 2 +-
drivers/gpu/drm/xe/xe_svm.c
sing conversion with CONFIG_DRM_XE_USERPTR_INVAL_INJECT
v6:
- Convert the new user in xe_vm_madise.
Signed-off-by: Matthew Auld
Cc: Himal Prasad Ghimiray
Cc: Thomas Hellström
Cc: Dafna Hirschfeld
Cc: Matthew Brost
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/xe/Kconfig | 2 +-
drivers
we lack DRM_GPUSVM, since we also need them for
userptr.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Himal Prasad Ghimiray
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/xe/xe_pt.c | 123 ---
drivers/gpu/drm/xe
ld be no functional change here.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/xe/Makefile | 1 +
drivers/gpu/drm/xe/xe_pt.c | 1 +
drivers/gpu/drm/xe/xe_userptr.c | 305 +++
driver
Refactor the core API of get/unmap/free pages to all operate on
drm_gpusvm_pages. In the next patch we want to export a simplified core
API without needing fully blown svm range etc.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
Export get/unmap/free pages API. We also need to tweak the SVM init to
allow skipping much of the unneeded parts.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 66
include
If we are only reading the memory then from the device pov the direction
can be DMA_TO_DEVICE. This aligns with the xe-userptr code. Using the
most restrictive data direction to represent the access is normally a
good idea.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
gpusvm, so needs alignment here.
v2:
- Add kernel-doc (Matt B)
- s/fls/ilog2/ (Thomas)
Reported-by: Thomas Hellström
Signed-off-by: Matthew Auld
Cc: Matthew Brost
Reviewed-by: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c | 33 +++--
1 file changed, 31
As a first step to moving userptr handling over to drm, replace the hmm
usage in xe over to gpusvm, which already offers similar functionality. As
some prep steps we also align on some of the missing pieces that were
already handled in xe_hmm.
v2:
- Rework the gpusvm API based on feedback.
- U
On 25/08/2025 12:31, Arunpravin Paneer Selvam wrote:
On 8/22/2025 5:32 PM, Matthew Auld wrote:
On 21/08/2025 16:06, Arunpravin Paneer Selvam wrote:
Maintain two separate RB trees per order - one for clear (zeroed) blocks
and another for dirty (uncleared) blocks. This separation improves
code
On 22/08/2025 09:37, Matthew Auld wrote:
On 21/08/2025 16:06, Arunpravin Paneer Selvam wrote:
Replace the freelist (O(n)) used for free block management with a
red-black tree, providing more efficient O(log n) search, insert,
and delete operations. This improves scalability and performance
when
tree macro.
- Use the existing dirty/free bit instead of new tree field.
- Make free_trees[] instead of clear_tree and dirty_tree for
more cleaner approach.
Signed-off-by: Arunpravin Paneer Selvam
Suggested-by: Matthew Auld
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4260
On 21/08/2025 16:06, Arunpravin Paneer Selvam wrote:
Replace the freelist (O(n)) used for free block management with a
red-black tree, providing more efficient O(log n) search, insert,
and delete operations. This improves scalability and performance
when managing large numbers of free blocks per
On 20/08/2025 08:56, Arunpravin Paneer Selvam wrote:
Hi Matthew,
On 8/14/2025 4:15 PM, Matthew Auld wrote:
On 24/07/2025 11:46, Arunpravin Paneer Selvam wrote:
Replace the freelist (O(n)) used for free block management with a
red-black tree, providing more efficient O(log n) search, insert
This should be enough to get scripts/kernel-doc passing for
drm_gpusvm.[ch], so we can then add them to our CI build infra.
v2:
- Also drop misplaced range_evict()
Link: https://gitlab.freedesktop.org/drm/xe/ci/-/merge_requests/86
Signed-off-by: Matthew Auld
Cc: Matthew Brost
---
drivers/gpu
This should be enough to get scripts/kernel-doc passing for
drm_gpusvm.[ch], so we can then add them to our CI build infra.
Link: https://gitlab.freedesktop.org/drm/xe/ci/-/merge_requests/86
Signed-off-by: Matthew Auld
Cc: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 4 ++--
1 file changed
As a first step to moving userptr handling over to drm, replace the hmm
usage in xe over to gpusvm, which already offers similar functionality. As
some prep steps we also align on some of the missing pieces that were
already handled in xe_hmm.
v2:
- Rework the gpusvm API based on feedback.
- U
Signed-off-by: Matthew Auld
Cc: Himal Prasad Ghimiray
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/xe/xe_pt.c | 123 +++
drivers/gpu/drm/xe/xe_pt_types.h | 2 -
2 files changed, 44 insertions(+), 81 deletions(-)
diff --git a/drivers
full SVM, that way we can
use the same lock (Matt B & Himal)
v3:
- Re-use svm_init/fini for userptr.
v4:
- Allow building xe without userptr if we are missing DRM_GPUSVM
config. (Matt B)
- Always make .read_only match xe_vma_read_only() for the ctx. (Dafna)
Signed-off-by: Matthew
ld be no functional change here.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/xe/Makefile | 1 +
drivers/gpu/drm/xe/xe_pt.c | 1 +
drivers/gpu/drm/xe/xe_userptr.c | 305 +++
driver
Export get/unmap/free pages API. We also need to tweak the SVM init to
allow skipping much of the unneeded parts.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 66
include
struct (Matt B)
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 69 +++-
drivers/gpu/drm/xe/xe_pt.c | 2 +-
drivers/gpu/drm/xe/xe_svm.c | 16 -
drivers/gpu/drm
Refactor the core API of get/unmap/free pages to all operate on
drm_gpusvm_pages. In the next patch we want to export a simplified core
API without needing fully blown svm range etc.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
If we are only reading the memory then from the device pov the direction
can be DMA_TO_DEVICE. This aligns with the xe-userptr code. Using the
most restrictive data direction to represent the access is normally a
good idea.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
gpusvm, so needs alignment here.
v2:
- Add kernel-doc (Matt B)
- s/fls/ilog2/ (Thomas)
Reported-by: Thomas Hellström
Signed-off-by: Matthew Auld
Cc: Matthew Brost
Reviewed-by: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c | 33 +++--
1 file changed, 31
correctness and avoid regressions.
v2: Missed adding the suggested-by tag. Added it in v2.
Signed-off-by: Arunpravin Paneer Selvam
Suggested-by: Matthew Auld
---
drivers/gpu/drm/drm_buddy.c | 316 ++--
include/drm/drm_buddy.h | 15 +-
2 files changed, 204
On 24/07/2025 11:46, Arunpravin Paneer Selvam wrote:
Replace the freelist (O(n)) used for free block management with a
red-black tree, providing more efficient O(log n) search, insert,
and delete operations. This improves scalability and performance
when managing large numbers of free blocks per
On 06/08/2025 14:28, Christian König wrote:
On some old x86 systems we had the problem that changing the caching flags
of system memory requires changing the global MTRR/PAT tables.
But on any modern x86 system (CPUs introduced rughly after 2004) we
actually don't need that any more and can upda
On 10/07/2025 08:14, Arunpravin Paneer Selvam wrote:
On 7/8/2025 2:30 PM, Matthew Auld wrote:
On 08/07/2025 07:54, Arunpravin Paneer Selvam wrote:
- Added a handler in DRM buddy manager to reset the cleared
flag for the blocks in the freelist.
- This is necessary because, upon resuming
everything has been cleared.
v2:
- Add lock before accessing drm_buddy_clear_reset_blocks()(Matthew Auld)
- Force merge the two dirty blocks.(Matthew Auld)
- Add a new unit test case for this issue.(Matthew Auld)
- Having this function being able to flip the state either way would be
everything has been cleared.
v2:
- Add lock before accessing drm_buddy_clear_reset_blocks()(Matthew Auld)
- Force merge the two dirty blocks.(Matthew Auld)
- Add a new unit test case for this issue.(Matthew Auld)
- Having this function being able to flip the state either way would be
+Matt B who is adding clear-on-free support in xe. I'm not sure if we
might also see something like this.
On 23/06/2025 06:52, Arunpravin Paneer Selvam wrote:
- Added a handler in DRM buddy manager to reset the cleared
flag for the blocks in the freelist.
- This is necessary because, upon r
full SVM, that way we can
use the same lock (Matt B & Himal)
v3:
- Re-use svm_init/fini for userptr.
v4:
- Allow building xe without userptr if we are missing DRM_GPUSVM
config. (Matt B)
- Always make .read_only match xe_vma_read_only() for the ctx. (Dafna)
Signed-off-by: Matthew
Signed-off-by: Matthew Auld
Cc: Himal Prasad Ghimiray
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/xe/xe_pt.c | 90 ++--
drivers/gpu/drm/xe/xe_pt_types.h | 2 -
2 files changed, 29 insertions(+), 63 deletions(-)
diff --git a/drivers
ld be no functional change here.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
---
drivers/gpu/drm/xe/Makefile | 1 +
drivers/gpu/drm/xe/xe_pt.c | 1 +
drivers/gpu/drm/xe/xe_userptr.c | 303 +++
drivers/gpu/drm/xe/xe_userptr.h
Export get/unmap/free pages API. We also need to tweak the SVM init to
allow skipping much of the unneeded parts.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 66
include
Refactor the core API of get/unmap/free pages to all operate on
drm_gpusvm_pages. In the next patch we want to export a simplified core
API without needing fully blown svm range etc.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
struct (Matt B)
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 66
drivers/gpu/drm/xe/xe_pt.c | 2 +-
drivers/gpu/drm/xe/xe_svm.c | 8 ++---
drivers/gpu/drm/xe
If we are only reading the memory then from the device pov the direction
can be DMA_TO_DEVICE. This aligns with the xe-userptr code. Using the
most restrictive data direction to represent the access is normally a
good idea.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
gpusvm, so needs alignment here.
v2:
- Add kernel-doc (Matt B)
- s/fls/ilog2/ (Thomas)
Reported-by: Thomas Hellström
Signed-off-by: Matthew Auld
Cc: Matthew Brost
Reviewed-by: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c | 33 +++--
1 file changed, 31
As a first step to moving userptr handling over to drm, replace the hmm
usage in xe over to gpusvm, which already offers similar functionality. As
some prep steps we also align on some of the missing pieces that were
already handled in xe_hmm.
v2:
- Rework the gpusvm API based on feedback.
- U
struct (Matt B)
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Reviewed-by: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 66
drivers/gpu/drm/xe/xe_pt.c | 2 +-
drivers/gpu/drm/xe/xe_svm.c | 8 ++---
drivers/gpu/drm/xe
Signed-off-by: Matthew Auld
Cc: Himal Prasad Ghimiray
Cc: Thomas Hellström
---
drivers/gpu/drm/xe/xe_pt.c | 90 ++--
drivers/gpu/drm/xe/xe_pt_types.h | 2 -
drivers/gpu/drm/xe/xe_svm.h | 19 +++
3 files changed, 39 insertions(+), 72 deletions
Refactor the core API of get/unmap/free pages to all operate on
drm_gpusvm_pages. In the next patch we want to export a simplified core
API without needing fully blown svm range etc.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c
If we are only reading the memory then from the device pov the direction
can be DMA_TO_DEVICE. This aligns with the xe-userptr code. Using the
most restrictive data direction to represent the access is normally a
good idea.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
As a first step to moving userptr handling over to drm, replace the hmm
usage in xe over to gpusvm, which already offers similar functionality. As
some prep steps we also align on some of the missing pieces that were
already handled in xe_hmm.
v2:
- Rework the gpusvm API based on feedback.
- U
full SVM, that way we can
use the same lock (Matt B & Himal)
v3:
- Re-use svm_init/fini for userptr.
Signed-off-by: Matthew Auld
Cc: Himal Prasad Ghimiray
Cc: Thomas Hellström
Cc: Matthew Brost
---
drivers/gpu/drm/xe/Kconfig | 3 +-
drivers/gpu/drm/xe/Makefile |
Export get/unmap/free pages API. We also need to tweak the SVM init to
allow skipping much of the unneeded parts.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 66
include/drm/drm_gpusvm.h | 16
gpusvm, so needs alignment here.
v2:
- Add kernel-doc (Matt B)
- s/fls/ilog2/ (Thomas)
Reported-by: Thomas Hellström
Signed-off-by: Matthew Auld
Cc: Matthew Brost
Reviewed-by: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c | 33 +++--
1 file changed, 31
On 17/04/2025 10:32, Boris Brezillon wrote:
Right now the DRM_PANTHOR_BO_NO_MMAP flag is ignored by
panthor_ioctl_bo_mmap_offset(), meaning BOs with this flag set can
have a file offset but can't be mapped anyway, because
panthor_gem_mmap() will filter them out.
If we error out at mmap_offset cr
On 14/04/2025 22:40, Jonathan Cavitt wrote:
Add additional information to each VM so they can report up to the first
50 seen faults. Only pagefaults are saved this way currently, though in
the future, all faults should be tracked by the VM for future reporting.
Additionally, of the pagefaults r
On 04/04/2025 18:02, Matthew Brost wrote:
On Fri, Apr 04, 2025 at 09:19:34AM +0100, Matthew Auld wrote:
On 03/04/2025 22:25, Matthew Brost wrote:
On Fri, Mar 28, 2025 at 06:10:36PM +, Matthew Auld wrote:
We now use the same notifier lock for SVM and userptr, with that we can
combine
this change makes sense.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 138 +--
include/drm/drm_gpusvm.h | 23 ++
2 files changed, 137 insertions(+), 24 deletions(-)
diff --git a/drivers/gpu/drm
like get_pages(), but not all the other more complex
svm stuff.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c | 95 +++-
drivers/gpu/drm/xe/xe_pt.c | 2 +-
drivers/gpu/drm/xe/xe_svm.c | 6
Goal here is cut over to gpusvm and remove xe_hmm, relying instead on
common code. The core facilities we need are get_pages(), unmap_pages()
and free_pages() for a given useptr range, plus a vm level notifier
lock, which is now provided by gpusvm.
Signed-off-by: Matthew Auld
Cc: Thomas
If we are only reading the memory then from the device pov the direction
can be DMA_TO_DEVICE. This aligns with the xe-userptr code. Using the
most restrictive data direction to represent the access is normally a
good idea.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
On 20/03/2025 20:43, Matthew Brost wrote:
On Thu, Mar 20, 2025 at 05:30:01PM +, Matthew Auld wrote:
Pull the pages stuff from the svm range into its own substructure, with
the idea of having the main pages related routines, like get_pages(),
unmap_pages() and free_pages() all operating on
On 03/04/2025 22:25, Matthew Brost wrote:
On Fri, Mar 28, 2025 at 06:10:36PM +, Matthew Auld wrote:
We now use the same notifier lock for SVM and userptr, with that we can
combine xe_pt_userptr_pre_commit and xe_pt_svm_pre_commit.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
On 03/04/2025 22:23, Matthew Brost wrote:
On Fri, Mar 28, 2025 at 06:10:36PM +, Matthew Auld wrote:
We now use the same notifier lock for SVM and userptr, with that we can
combine xe_pt_userptr_pre_commit and xe_pt_svm_pre_commit.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Hi,
On 02/04/2025 11:53, Thomas Zimmermann wrote:
Hi
Am 02.04.25 um 12:44 schrieb Matthew Auld:
Build fails with:
error: multiple unsequenced modifications to 'sbuf32'
[-Werror,-Wunsequenced]
264 | le32_to_c
version")
Fixes: 3f31a017ddbc ("drm/format-helper: Optimize 32-to-16-bpp conversion")
Fixes: 65931bbc5177 ("drm/format-helper: Optimize 32-to-8-bpp conversion")
Signed-off-by: Matthew Auld
Cc: Thomas Zimmermann
Cc: Jocelyn Falempe
---
drivers/gpu/drm/drm_format_helper
Refactor the core API of get/unmap/free pages to all operate on
drm_gpusvm_pages. In the next patch we want to export a simplified core
API without needing fully blown svm range etc.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c
struct (Matt B)
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c | 66
drivers/gpu/drm/xe/xe_pt.c | 2 +-
drivers/gpu/drm/xe/xe_svm.c | 8 ++---
drivers/gpu/drm/xe/xe_svm.h | 6 ++--
include
Export get/unmap/free pages API. We also need to tweak the SVM init to
allow skipping much of the unneeded parts.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 66
include/drm/drm_gpusvm.h | 16
full SVM, that way we can
use the same lock (Matt B & Himal)
Signed-off-by: Matthew Auld
Cc: Himal Prasad Ghimiray
Cc: Thomas Hellström
Cc: Matthew Brost
---
drivers/gpu/drm/xe/Kconfig | 3 +-
drivers/gpu/drm/xe/Makefile | 1 -
drivers/gpu/drm/xe/xe_exec.c |
We now use the same notifier lock for SVM and userptr, with that we can
combine xe_pt_userptr_pre_commit and xe_pt_svm_pre_commit.
Suggested-by: Matthew Brost
Signed-off-by: Matthew Auld
Cc: Himal Prasad Ghimiray
Cc: Thomas Hellström
---
drivers/gpu/drm/xe/xe_pt.c | 95
gpusvm, so needs alignment here.
v2:
- Add kernel-doc (Matt B)
- s/fls/ilog2/ (Thomas)
Reported-by: Thomas Hellström
Signed-off-by: Matthew Auld
Cc: Matthew Brost
Reviewed-by: Thomas Hellström
---
drivers/gpu/drm/drm_gpusvm.c | 33 +++--
1 file changed, 31
As a first step to moving userptr handling over to drm, replace the hmm
usage in xe over to gpusvm, which already offers similar functionality. As
some prep steps we also align on some of the missing pieces that were
already handled in xe_hmm.
v2:
- Rework the gpusvm API based on feedback.
- U
As a first step to moving userptr handling over to drm, replace the hmm
usage in xe over to gpusvm, which already offers similar functionality. As
some prep steps we also align on some of the missing pieces that were
already handled in xe_hmm.
--
2.48.1
On 20/03/2025 20:40, Matthew Brost wrote:
On Thu, Mar 20, 2025 at 05:29:58PM +, Matthew Auld wrote:
Handle the case where the hmm range partially covers a huge page (like
2M), otherwise we can potentially end up doing something nasty like
mapping memory which potentially is outside the
On 20/03/2025 19:33, Matthew Brost wrote:
On Thu, Mar 20, 2025 at 08:29:42PM +0100, Thomas Hellström wrote:
On Thu, 2025-03-20 at 17:30 +, Matthew Auld wrote:
If the memory is going to be accessed by the device, make sure we
mark
the pages accordingly such that the kernel knows this. This
If the memory is going to be accessed by the device, make sure we mark
the pages accordingly such that the kernel knows this. This aligns with
the xe-userptr code.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 9 +
1 file changed
directly use gpusvm, so needs alignment here.
Reported-by: Thomas Hellström
Signed-off-by: Matthew Auld
Cc: Matthew Brost
---
drivers/gpu/drm/drm_gpusvm.c | 25 +++--
1 file changed, 23 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/drm_gpusvm.c b/drivers/gpu/drm
Lower get/unmap pages to facilitate operating on the lowest level
pieces, without needing a full drm_gpusvm_range structure. In the next
patch we want to extract get/unmap/free to operate on a different range
type.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
---
drivers
If we are only reading the memory then from the device pov the direction
can be DMA_TO_DEVICE. This aligns with the xe-userptr code. Using the
most restrictive data direction to represent the access is normally a
good idea.
Signed-off-by: Matthew Auld
Cc: Thomas Hellström
Cc: Matthew Brost
On 05/03/2025 18:44, Matthew Brost wrote:
On Wed, Feb 26, 2025 at 06:05:55PM +, Matthew Auld wrote:
On 26/02/2025 15:12, Matthew Brost wrote:
On Wed, Feb 26, 2025 at 10:38:40AM +, Matthew Auld wrote:
On 26/02/2025 04:18, Matthew Brost wrote:
On Tue, Feb 25, 2025 at 09:13:09PM -0600
On 26/02/2025 15:12, Matthew Brost wrote:
On Wed, Feb 26, 2025 at 10:38:40AM +, Matthew Auld wrote:
On 26/02/2025 04:18, Matthew Brost wrote:
On Tue, Feb 25, 2025 at 09:13:09PM -0600, Lucas De Marchi wrote:
On Wed, Feb 26, 2025 at 10:00:18AM +0800, Mingcong Bai via B4 Relay wrote:
From
On 26/02/2025 04:18, Matthew Brost wrote:
On Tue, Feb 25, 2025 at 09:13:09PM -0600, Lucas De Marchi wrote:
On Wed, Feb 26, 2025 at 10:00:18AM +0800, Mingcong Bai via B4 Relay wrote:
From: Mingcong Bai
The bo/ttm interfaces with kernel memory mapping from dedicated GPU
memory. It is not correc
On 25/02/2025 04:42, Matthew Brost wrote:
Clear root PT entry and invalidate entire VM's address space when
closing the VM. Will prevent the GPU from accessing any of the VM's
memory after closing.
v2:
- s/vma/vm in kernel doc (CI)
- Don't nuke migration VM as this occur at driver unload (CI
On 25/02/2025 04:42, Matthew Brost wrote:
This patch introduces support for GPU Shared Virtual Memory (SVM) in the
Direct Rendering Manager (DRM) subsystem. SVM allows for seamless
sharing of memory between the CPU and GPU, enhancing performance and
flexibility in GPU computing tasks.
The patch
On 25/02/2025 04:42, Matthew Brost wrote:
From: Thomas Hellström
Introduce drm_pagemap ops to map and unmap dma to VRAM resources. In the
local memory case it's a matter of merely providing an offset into the
device's physical address. For future p2p the map and unmap functions may
encode as ne
On 20/02/2025 19:55, Matthew Brost wrote:
On Thu, Feb 20, 2025 at 04:59:29PM +0100, Thomas Hellström wrote:
On Thu, 2025-02-20 at 15:53 +, Matthew Auld wrote:
On 13/02/2025 02:11, Matthew Brost wrote:
Migration is implemented with range granularity, with VRAM backing
being
a VM private
On 13/02/2025 02:11, Matthew Brost wrote:
Migration is implemented with range granularity, with VRAM backing being
a VM private TTM BO (i.e., shares dma-resv with VM). The lifetime of the
TTM BO is limited to when the SVM range is in VRAM (i.e., when a VRAM
SVM range is migrated to SRAM, the TTM
m"
Cc: Rodrigo Vivi
Reviewed-by: Matthew Auld
---
drivers/gpu/drm/xe/xe_bo.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c
index 78d09c5ed26d..b34f446ad57d 100644
--- a/drivers/gpu/drm/xe/xe_bo.c
+++
On 06/02/2025 10:35, Thomas Hellström wrote:
Add Matthew Auld and Matthew Brost as TTM reviewers
Cc: Matthew Auld
Cc: Matthew Brost
Cc: Christian König
Cc: Dave Airlie
Cc: Simona Vetter
Cc: dri-devel@lists.freedesktop.org
Cc: linux-ker...@vger.kernel.org
Signed-off-by: Thomas Hellström
On 03/02/2025 12:12, Boris Brezillon wrote:
On Mon, 3 Feb 2025 11:23:53 +
Liviu Dudau wrote:
On Mon, Feb 03, 2025 at 10:21:53AM +0100, Boris Brezillon wrote:
+Akash with whom we've been discussing adding a 'REPEAT' mode to
drm_gpuvm/panthor.
On Sun, 2 Feb 2025 19:53:47 +0100
Danilo Krumm
On 30/01/2025 16:32, Matthew Brost wrote:
On Thu, Jan 30, 2025 at 02:22:55PM +, Matthew Auld wrote:
On 29/01/2025 19:52, Matthew Brost wrote:
Migration is implemented with range granularity, with VRAM backing being
a VM private TTM BO (i.e., shares dma-resv with VM). The lifetime of the
On 29/01/2025 19:52, Matthew Brost wrote:
Migration is implemented with range granularity, with VRAM backing being
a VM private TTM BO (i.e., shares dma-resv with VM). The lifetime of the
TTM BO is limited to when the SVM range is in VRAM (i.e., when a VRAM
SVM range is migrated to SRAM, the TTM
On 29/01/2025 19:51, Matthew Brost wrote:
This patch introduces support for GPU Shared Virtual Memory (SVM) in the
Direct Rendering Manager (DRM) subsystem. SVM allows for seamless
sharing of memory between the CPU and GPU, enhancing performance and
flexibility in GPU computing tasks.
The patch
On 29/01/2025 19:52, Matthew Brost wrote:
From: Thomas Hellström
Add support for mapping device pages to Xe SVM by attaching drm_pagemap
to a memory region, which is then linked to a GPU SVM devmem allocation.
This enables GPU SVM to derive the device page address.
v3:
- Better commit messag
On 29/01/2025 19:51, Matthew Brost wrote:
Clear root PT entry and invalidate entire VM's address space when
closing the VM. Will prevent the GPU from accessing any of the VM's
memory after closing.
v2:
- s/vma/vm in kernel doc (CI)
- Don't nuke migration VM as this occur at driver unload (CI
: Add a testcase to verify the multiroot fini")
Signed-off-by: Matthew Auld
Cc: Arunpravin Paneer Selvam
Cc: Christian König
Reviewed-by: Thomas Hellström
---
drivers/gpu/drm/tests/drm_buddy_test.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/drivers/gpu/drm/tests/drm_buddy_test.c
);
Reported-by: Jani Nikula
Reported-by: Thomas Hellström
Fixes: 8cb3a1e2b350 ("drm/buddy: Add a testcase to verify the multiroot fini")
Signed-off-by: Matthew Auld
Cc: Arunpravin Paneer Selvam
Cc: Christian König
---
drivers/gpu/drm/tests/drm_buddy_test.c | 1 -
1 file changed,
On 09/01/2025 16:26, Thomas Zimmermann wrote:
Hi
Am 09.01.25 um 17:05 schrieb Matthew Auld:
On 09/01/2025 14:57, Thomas Zimmermann wrote:
Call drm_mode_size_dumb() to compute dumb-buffer scanline pitch
and buffer size. Align the pitch to a multiple of 8. Align the
buffer size according to
On 09/01/2025 14:57, Thomas Zimmermann wrote:
Call drm_mode_size_dumb() to compute dumb-buffer scanline pitch
and buffer size. Align the pitch to a multiple of 8. Align the
buffer size according to hardware requirements.
Signed-off-by: Thomas Zimmermann
Cc: Lucas De Marchi
Cc: "Thomas Hellströ
Hey,
On 06/01/2025 12:37, Paneer Selvam, Arunpravin wrote:
Hi Matthew,
Ping?
Just got back from time off. I already slapped an r-b here:
https://lore.kernel.org/dri-devel/aa27455c-04b8-494c-a3b5-c4385a774...@intel.com/
Regards,
Arun.
On 12/26/2024 12:31 PM, Arunpravin Paneer Selvam wrot
: Arunpravin Paneer Selvam
Signed-off-by: Lin.Cao
Reviewed-by: Matthew Auld
could fix this issue.
Signed-off-by: Lin.Cao
Reviewed-by: Matthew Auld
On 16/12/2024 13:07, Arunpravin Paneer Selvam wrote:
- Added a testcase to verify the multiroot force merge fini.
- Added a new field in_use to track the mm freed status.
Signed-off-by: Arunpravin Paneer Selvam
Signed-off-by: Lin.Cao
---
drivers/gpu/drm/drm_buddy.c| 20 ++
1 - 100 of 1447 matches
Mail list logo
