From: Matthew Brost <matthew.brost@intel.com>
To: "Thomas Hellström" <thomas.hellstrom@linux.intel.com>
Cc: <intel-xe@lists.freedesktop.org>,
John Falkowski <john.falkowski@intel.com>,
Michal Mrozek <michal.mrozek@intel.com>
Subject: Re: [PATCH] drm/xe/uapi: Introduce a flag to disallow vm overcommit in fault mode
Date: Wed, 4 Feb 2026 10:56:24 -0800 [thread overview]
Message-ID: <aYOWWAg9O1KUhnpO@lstrano-desk.jf.intel.com> (raw)
In-Reply-To: <20260204153320.17989-1-thomas.hellstrom@linux.intel.com>
On Wed, Feb 04, 2026 at 04:33:20PM +0100, Thomas Hellström wrote:
> Some compute applications may try to allocate device memory to probe
> how much device memory is actually available, assuming that the
> application will be the only one running on the particular GPU.
>
> That strategy fails in fault mode since it allows VM overcommit.
>
> While this could be resolved in user-space it's further complicated
> by cgroups potentially restricting the amount of memory available
> to the application.
>
> Introduce a vm create flag, DRM_XE_VM_CREATE_NO_VM_OVERCOMMIT, that
> allows fault mode to mimic the behaviour of !fault mode WRT this. It
> blocks evicting same vm bos during VM_BIND processing. However,
> it does *not* block evicting same-vm bos during pagefault
> processing, preferring eviction rather than VM banning in
> OOM situations.
>
> Cc: John Falkowski <john.falkowski@intel.com>
> Cc: Michal Mrozek <michal.mrozek@intel.com>
> Cc: Matthew Brost <matthew.brost@intel.com>
> Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
> ---
> drivers/gpu/drm/xe/xe_vm.c | 11 +++++++++--
> drivers/gpu/drm/xe/xe_vm.h | 7 +++++++
> drivers/gpu/drm/xe/xe_vm_types.h | 1 +
> include/uapi/drm/xe_drm.h | 6 ++++++
> 4 files changed, 23 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
> index 8fe54a998385..cf92b6e13a16 100644
> --- a/drivers/gpu/drm/xe/xe_vm.c
> +++ b/drivers/gpu/drm/xe/xe_vm.c
> @@ -1938,7 +1938,8 @@ find_ufence_get(struct xe_sync_entry *syncs, u32 num_syncs)
>
> #define ALL_DRM_XE_VM_CREATE_FLAGS (DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE | \
> DRM_XE_VM_CREATE_FLAG_LR_MODE | \
> - DRM_XE_VM_CREATE_FLAG_FAULT_MODE)
> + DRM_XE_VM_CREATE_FLAG_FAULT_MODE | \
> + DRM_XE_VM_CREATE_FLAG_NO_VM_OVERCOMMIT)
>
> int xe_vm_create_ioctl(struct drm_device *dev, void *data,
> struct drm_file *file)
> @@ -1977,12 +1978,18 @@ int xe_vm_create_ioctl(struct drm_device *dev, void *data,
> args->flags & DRM_XE_VM_CREATE_FLAG_FAULT_MODE))
> return -EINVAL;
>
> + if (XE_IOCTL_DBG(xe, !(args->flags & DRM_XE_VM_CREATE_FLAG_FAULT_MODE) &&
> + args->flags & DRM_XE_VM_CREATE_FLAG_NO_VM_OVERCOMMIT))
> + return -EINVAL;
> +
> if (args->flags & DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE)
> flags |= XE_VM_FLAG_SCRATCH_PAGE;
> if (args->flags & DRM_XE_VM_CREATE_FLAG_LR_MODE)
> flags |= XE_VM_FLAG_LR_MODE;
> if (args->flags & DRM_XE_VM_CREATE_FLAG_FAULT_MODE)
> flags |= XE_VM_FLAG_FAULT_MODE;
> + if (args->flags & DRM_XE_VM_CREATE_FLAG_NO_VM_OVERCOMMIT)
> + flags |= XE_VM_FLAG_NO_VM_OVERCOMMIT;
>
> vm = xe_vm_create(xe, flags, xef);
> if (IS_ERR(vm))
> @@ -2903,7 +2910,7 @@ static int vma_lock_and_validate(struct drm_exec *exec, struct xe_vma *vma,
> err = drm_exec_lock_obj(exec, &bo->ttm.base);
> if (!err && validate)
> err = xe_bo_validate(bo, vm,
> - !xe_vm_in_preempt_fence_mode(vm) &&
> + xe_vm_allow_vm_eviction(vm) &&
One question. This is existing code but can you refresh my memory why we
allow overcommit on dma-fencing VMs? Wouldn't the next exec IOCTL
immediately fail?
Matt
> res_evict, exec);
> }
>
> diff --git a/drivers/gpu/drm/xe/xe_vm.h b/drivers/gpu/drm/xe/xe_vm.h
> index 288115c7844a..f849e369432b 100644
> --- a/drivers/gpu/drm/xe/xe_vm.h
> +++ b/drivers/gpu/drm/xe/xe_vm.h
> @@ -220,6 +220,13 @@ static inline bool xe_vm_in_preempt_fence_mode(struct xe_vm *vm)
> return xe_vm_in_lr_mode(vm) && !xe_vm_in_fault_mode(vm);
> }
>
> +static inline bool xe_vm_allow_vm_eviction(struct xe_vm *vm)
> +{
> + return !xe_vm_in_lr_mode(vm) ||
> + (xe_vm_in_fault_mode(vm) &&
> + !(vm->flags & XE_VM_FLAG_NO_VM_OVERCOMMIT));
> +}
> +
> int xe_vm_add_compute_exec_queue(struct xe_vm *vm, struct xe_exec_queue *q);
> void xe_vm_remove_compute_exec_queue(struct xe_vm *vm, struct xe_exec_queue *q);
>
> diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h
> index 43203e90ee3e..1f6f7e30e751 100644
> --- a/drivers/gpu/drm/xe/xe_vm_types.h
> +++ b/drivers/gpu/drm/xe/xe_vm_types.h
> @@ -232,6 +232,7 @@ struct xe_vm {
> #define XE_VM_FLAG_TILE_ID(flags) FIELD_GET(GENMASK(7, 6), flags)
> #define XE_VM_FLAG_SET_TILE_ID(tile) FIELD_PREP(GENMASK(7, 6), (tile)->id)
> #define XE_VM_FLAG_GSC BIT(8)
> +#define XE_VM_FLAG_NO_VM_OVERCOMMIT BIT(9)
> unsigned long flags;
>
> /**
> diff --git a/include/uapi/drm/xe_drm.h b/include/uapi/drm/xe_drm.h
> index 077e66a682e2..e54f8e12acd9 100644
> --- a/include/uapi/drm/xe_drm.h
> +++ b/include/uapi/drm/xe_drm.h
> @@ -975,6 +975,11 @@ struct drm_xe_gem_mmap_offset {
> * demand when accessed, and also allows per-VM overcommit of memory.
> * The xe driver internally uses recoverable pagefaults to implement
> * this.
> + * - %DRM_XE_VM_CREATE_FLAG_NO_VM_OVERCOMMIT - Requires also
> + * DRM_XE_VM_CREATE_FLAG_FAULT_MODE. This disallows per-VM overcommit
> + * but only during a &DRM_IOCTL_XE_VM_BIND operation with the
> + * %DRM_XE_VM_BIND_FLAG_IMMEDIATE flag set. This may be useful for
> + * user-space naively probing the amount of available memory.
> */
> struct drm_xe_vm_create {
> /** @extensions: Pointer to the first extension struct, if any */
> @@ -983,6 +988,7 @@ struct drm_xe_vm_create {
> #define DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE (1 << 0)
> #define DRM_XE_VM_CREATE_FLAG_LR_MODE (1 << 1)
> #define DRM_XE_VM_CREATE_FLAG_FAULT_MODE (1 << 2)
> +#define DRM_XE_VM_CREATE_FLAG_NO_VM_OVERCOMMIT (1 << 3)
> /** @flags: Flags */
> __u32 flags;
>
> --
> 2.52.0
>
next prev parent reply other threads:[~2026-02-04 18:56 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-02-04 15:33 [PATCH] drm/xe/uapi: Introduce a flag to disallow vm overcommit in fault mode Thomas Hellström
2026-02-04 18:56 ` Matthew Brost [this message]
2026-02-05 8:06 ` Thomas Hellström
2026-02-05 17:58 ` Matthew Brost
2026-02-05 0:50 ` ✓ CI.KUnit: success for " Patchwork
2026-02-05 1:23 ` ✓ Xe.CI.BAT: " Patchwork
2026-02-05 16:14 ` ✓ Xe.CI.FULL: " Patchwork
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=aYOWWAg9O1KUhnpO@lstrano-desk.jf.intel.com \
--to=matthew.brost@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=john.falkowski@intel.com \
--cc=michal.mrozek@intel.com \
--cc=thomas.hellstrom@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox