From: Arvind Yadav <arvind.yadav@intel.com>
To: intel-xe@lists.freedesktop.org, dri-devel@lists.freedesktop.org
Cc: matthew.brost@intel.com, himal.prasad.ghimiray@intel.com,
thomas.hellstrom@linux.intel.com
Subject: [RFC v3 1/7] drm/xe/vm: Track CPU_AUTORESET state in xe_vma
Date: Thu, 4 Jun 2026 15:20:06 +0530 [thread overview]
Message-ID: <20260604095012.367231-2-arvind.yadav@intel.com> (raw)
In-Reply-To: <20260604095012.367231-1-arvind.yadav@intel.com>
Track MADVISE_AUTORESET CPU-only state in struct xe_vma with a dedicated
cpu_autoreset_active bool.
Add XE_VMA_CPU_AUTORESET_ACTIVE as a pipeline-only bit to carry this
state through MAP/REMAP operations. Runtime state lives in
xe_vma.cpu_autoreset_active rather than vma->gpuva.flags.
The state is set at bind time and cleared on the first successful GPU
fault.
v2:
- Move runtime state from gpuva.flags to xe_vma bool. (Matt)
- Keep XE_VMA_CPU_AUTORESET_ACTIVE pipeline-only. (Matt)
v3:
- Add kernel-doc and lockdep_assert_held to
xe_vma_has_cpu_autoreset_active(). (Matt)
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Signed-off-by: Arvind Yadav <arvind.yadav@intel.com>
---
drivers/gpu/drm/xe/xe_vm.h | 14 ++++++++++++++
drivers/gpu/drm/xe/xe_vm_types.h | 13 +++++++++++++
2 files changed, 27 insertions(+)
diff --git a/drivers/gpu/drm/xe/xe_vm.h b/drivers/gpu/drm/xe/xe_vm.h
index c5b900f38ded..0dd201cce968 100644
--- a/drivers/gpu/drm/xe/xe_vm.h
+++ b/drivers/gpu/drm/xe/xe_vm.h
@@ -181,6 +181,20 @@ static inline bool xe_vma_is_userptr(struct xe_vma *vma)
!xe_vma_is_cpu_addr_mirror(vma);
}
+/**
+ * xe_vma_has_cpu_autoreset_active() - Check if VMA is still CPU-only
+ * @vma: VMA to check
+ *
+ * The caller must hold vm->lock (read or write).
+ *
+ * Return: true if CPU-only autoreset is armed, false after the first GPU touch.
+ */
+static inline bool xe_vma_has_cpu_autoreset_active(struct xe_vma *vma)
+{
+ lockdep_assert_held(&xe_vma_vm(vma)->lock);
+ return vma->cpu_autoreset_active;
+}
+
struct xe_vma *xe_vm_find_vma_by_addr(struct xe_vm *vm, u64 page_addr);
int xe_vma_need_vram_for_atomic(struct xe_device *xe, struct xe_vma *vma, bool is_atomic);
diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h
index 635ed29b9a69..e6380458272e 100644
--- a/drivers/gpu/drm/xe/xe_vm_types.h
+++ b/drivers/gpu/drm/xe/xe_vm_types.h
@@ -52,6 +52,11 @@ struct xe_vm_pgtable_update_op;
#define XE_VMA_DUMPABLE (DRM_GPUVA_USERBITS << 8)
#define XE_VMA_SYSTEM_ALLOCATOR (DRM_GPUVA_USERBITS << 9)
#define XE_VMA_MADV_AUTORESET (DRM_GPUVA_USERBITS << 10)
+/*
+ * Pipeline-only bit used to carry cpu_autoreset_active through MAP/REMAP.
+ * It must not be stored persistently in vma->gpuva.flags.
+ */
+#define XE_VMA_CPU_AUTORESET_ACTIVE (DRM_GPUVA_USERBITS << 11)
/**
* struct xe_vma_mem_attr - memory attributes associated with vma
@@ -157,6 +162,14 @@ struct xe_vma {
/** @tile_staged: bind is staged for this VMA */
u8 tile_staged;
+ /**
+ * @cpu_autoreset_active: CPU mirror VMA has not been GPU-faulted yet.
+ *
+ * Set for MADVISE_AUTORESET CPU mirror VMAs and cleared on the first
+ * successful GPU fault. Protected by vm->lock.
+ */
+ bool cpu_autoreset_active;
+
/**
* @skip_invalidation: Used in madvise to avoid invalidation
* if mem attributes doesn't change
--
2.43.0
next prev parent reply other threads:[~2026-06-04 9:50 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-04 9:50 [RFC v3 0/7] drm/xe/svm: Add MMU notifier-based madvise autoreset on munmap Arvind Yadav
2026-06-04 9:50 ` Arvind Yadav [this message]
2026-06-04 9:50 ` [RFC v3 2/7] drm/xe/vm: Preserve cpu_autoreset_active across GPUVA operations Arvind Yadav
2026-06-04 10:07 ` sashiko-bot
2026-06-04 9:50 ` [RFC v3 3/7] drm/xe/svm: Clear CPU_AUTORESET_ACTIVE on first GPU fault Arvind Yadav
2026-06-04 9:50 ` [RFC v3 4/7] drm/xe/vm: Add madvise autoreset notifier worker Arvind Yadav
2026-06-04 10:14 ` sashiko-bot
2026-06-04 9:50 ` [RFC v3 5/7] drm/xe/vm: Disable madvise notifier on GPU touch Arvind Yadav
2026-06-04 10:03 ` sashiko-bot
2026-06-04 9:50 ` [RFC v3 6/7] drm/xe/vm: Wire MADVISE_AUTORESET notifiers into VM lifecycle Arvind Yadav
2026-06-04 10:05 ` sashiko-bot
2026-06-04 9:50 ` [RFC v3 7/7] drm/xe/svm: Correct memory attribute reset for partial unmap Arvind Yadav
2026-06-04 10:12 ` sashiko-bot
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=20260604095012.367231-2-arvind.yadav@intel.com \
--to=arvind.yadav@intel.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=himal.prasad.ghimiray@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=matthew.brost@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