From: Dave Airlie <airlied@gmail.com>
To: dri-devel@lists.freedesktop.org
Cc: nouveau@lists.freedesktop.org
Subject: [PATCH] nouveau/vmm: start tracking if the LPT PTE is valid. (v3)
Date: Thu, 29 Jan 2026 13:07:10 +1000 [thread overview]
Message-ID: <20260129030710.1574879-1-airlied@gmail.com> (raw)
From: Dave Airlie <airlied@redhat.com>
When NVK enabled large pages userspace tests were seeing fault
reports at a valid address.
There was a case where an address moving from 64k page to 4k pages
could expose a race between unmapping the 4k page, mapping the 64k
page and unref the 4k pages.
Unref 4k pages would cause the dual-page table handling to always
set the LPTE entry to SPARSE or INVALID, but if we'd mapped a valid
LPTE in the meantime, it would get trashed. Keep track of when
a valid LPTE has been referenced, and don't reset in that case.
This increase the tracking to 32-bit, because it turns out if
unref can get delayed, you can get a lot of these outstanding
and this can cause strange behaviours.
Cc: stable@vger.kernel.org
Link: https://gitlab.freedesktop.org/mesa/mesa/-/issues/14610
Signed-off-by: Dave Airlie <airlied@redhat.com>
--
v2: move to 32-bit from 8-bit tracker
fix some more flag changes.
v3: missed one BIG_PTE unset
---
drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c | 44 ++++++++++++++-----
drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.h | 9 ++--
2 files changed, 37 insertions(+), 16 deletions(-)
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c
index f95c58b67633..970e5f94a6e3 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.c
@@ -53,7 +53,7 @@ nvkm_vmm_pt_new(const struct nvkm_vmm_desc *desc, bool sparse,
}
}
- if (!(pgt = kzalloc(sizeof(*pgt) + lpte, GFP_KERNEL)))
+ if (!(pgt = kzalloc(sizeof(*pgt) + (sizeof(pgt->pte[0]) * lpte), GFP_KERNEL)))
return NULL;
pgt->page = page ? page->shift : 0;
pgt->sparse = sparse;
@@ -242,14 +242,17 @@ nvkm_vmm_unref_sptes(struct nvkm_vmm_iter *it, struct nvkm_vmm_pt *pgt,
if (pgt->pte[pteb] & NVKM_VMM_PTE_SPARSE) {
TRA(it, "LPTE %05x: U -> S %d PTEs", pteb, ptes);
pair->func->sparse(vmm, pgt->pt[0], pteb, ptes);
- } else
- if (pair->func->invalid) {
- /* If the MMU supports it, restore the LPTE to the
- * INVALID state to tell the MMU there is no point
- * trying to fetch the corresponding SPTEs.
- */
- TRA(it, "LPTE %05x: U -> I %d PTEs", pteb, ptes);
- pair->func->invalid(vmm, pgt->pt[0], pteb, ptes);
+ } else if (!(pgt->pte[pteb] & NVKM_VMM_PTE_BIG_VALID)) {
+ if (pair->func->invalid) {
+ /* If the MMU supports it, restore the LPTE to the
+ * INVALID state to tell the MMU there is no point
+ * trying to fetch the corresponding SPTEs.
+ */
+ TRA(it, "LPTE %05x: U -> I %d PTEs", pteb, ptes);
+ pair->func->invalid(vmm, pgt->pt[0], pteb, ptes);
+ }
+ } else {
+ TRA(it, "LPTE %05x: V %d PTEs", pteb, ptes);
}
}
}
@@ -280,6 +283,13 @@ nvkm_vmm_unref_ptes(struct nvkm_vmm_iter *it, bool pfn, u32 ptei, u32 ptes)
if (desc->type == SPT && (pgt->refs[0] || pgt->refs[1]))
nvkm_vmm_unref_sptes(it, pgt, desc, ptei, ptes);
+ if (desc->type == LPT && (pgt->refs[0] || pgt->refs[1])) {
+ for (u32 lpti = ptei; ptes; lpti++) {
+ pgt->pte[lpti] &= ~NVKM_VMM_PTE_BIG_VALID;
+ ptes--;
+ }
+ }
+
/* PT no longer needed? Destroy it. */
if (!pgt->refs[type]) {
it->lvl++;
@@ -332,10 +342,12 @@ nvkm_vmm_ref_sptes(struct nvkm_vmm_iter *it, struct nvkm_vmm_pt *pgt,
* Determine how many LPTEs need to transition state.
*/
pgt->pte[ptei] |= NVKM_VMM_PTE_VALID;
+ pgt->pte[ptei] &= ~NVKM_VMM_PTE_BIG_VALID;
for (ptes = 1, ptei++; ptei < lpti; ptes++, ptei++) {
if (pgt->pte[ptei] & NVKM_VMM_PTE_VALID)
break;
pgt->pte[ptei] |= NVKM_VMM_PTE_VALID;
+ pgt->pte[ptei] &= ~NVKM_VMM_PTE_BIG_VALID;
}
if (pgt->pte[pteb] & NVKM_VMM_PTE_SPARSE) {
@@ -374,6 +386,14 @@ nvkm_vmm_ref_ptes(struct nvkm_vmm_iter *it, bool pfn, u32 ptei, u32 ptes)
if (desc->type == SPT)
nvkm_vmm_ref_sptes(it, pgt, desc, ptei, ptes);
+ if (desc->type == LPT) {
+ for (u32 lpti = ptei; ptes; lpti++) {
+ pgt->pte[lpti] &= ~NVKM_VMM_PTE_VALID;
+ pgt->pte[lpti] |= NVKM_VMM_PTE_BIG_VALID;
+ ptes--;
+ }
+ }
+
return true;
}
@@ -386,7 +406,7 @@ nvkm_vmm_sparse_ptes(const struct nvkm_vmm_desc *desc,
pgt->pde[ptei++] = NVKM_VMM_PDE_SPARSE;
} else
if (desc->type == LPT) {
- memset(&pgt->pte[ptei], NVKM_VMM_PTE_SPARSE, ptes);
+ memset32(&pgt->pte[ptei], NVKM_VMM_PTE_SPARSE, ptes);
}
}
@@ -398,7 +418,7 @@ nvkm_vmm_sparse_unref_ptes(struct nvkm_vmm_iter *it, bool pfn, u32 ptei, u32 pte
memset(&pt->pde[ptei], 0x00, sizeof(pt->pde[0]) * ptes);
else
if (it->desc->type == LPT)
- memset(&pt->pte[ptei], 0x00, sizeof(pt->pte[0]) * ptes);
+ memset32(&pt->pte[ptei], 0x00, ptes);
return nvkm_vmm_unref_ptes(it, pfn, ptei, ptes);
}
@@ -457,7 +477,7 @@ nvkm_vmm_ref_hwpt(struct nvkm_vmm_iter *it, struct nvkm_vmm_pt *pgd, u32 pdei)
desc->func->sparse(vmm, pt, pteb, ptes);
else
desc->func->invalid(vmm, pt, pteb, ptes);
- memset(&pgt->pte[pteb], 0x00, ptes);
+ memset32(&pgt->pte[pteb], 0x00, ptes);
} else {
desc->func->unmap(vmm, pt, pteb, ptes);
while (ptes--)
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.h b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.h
index 4586a425dbe4..3720579f4bf7 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.h
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/vmm.h
@@ -44,10 +44,11 @@ struct nvkm_vmm_pt {
*
* This information is used to manage LPTE state transitions.
*/
-#define NVKM_VMM_PTE_SPARSE 0x80
-#define NVKM_VMM_PTE_VALID 0x40
-#define NVKM_VMM_PTE_SPTES 0x3f
- u8 pte[];
+#define NVKM_VMM_PTE_SPARSE 0x80000000
+#define NVKM_VMM_PTE_VALID 0x40000000
+#define NVKM_VMM_PTE_BIG_VALID 0x20000000
+#define NVKM_VMM_PTE_SPTES 0x1fffffff
+ u32 pte[];
};
typedef void (*nvkm_vmm_pxe_func)(struct nvkm_vmm *,
--
2.52.0
reply other threads:[~2026-01-29 3:07 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
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=20260129030710.1574879-1-airlied@gmail.com \
--to=airlied@gmail.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=nouveau@lists.freedesktop.org \
/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