From: Michal Wajdeczko <michal.wajdeczko@intel.com>
To: Satyanarayana K V P <satyanarayana.k.v.p@intel.com>,
<intel-xe@lists.freedesktop.org>
Cc: Sashiko <sashiko-bot@kernel.org>,
Matthew Brost <matthew.brost@intel.com>,
Maarten Lankhorst <dev@lankhorst.se>
Subject: Re: [PATCH v10 05/10] drm/xe/ggtt: Avoid integer overflow when validating VF GGTT range
Date: Mon, 21 Sep 2026 16:03:15 +0200 [thread overview]
Message-ID: <4faf44b6-c321-4ce3-a88f-01fa442f58bd@intel.com> (raw)
In-Reply-To: <20260921092101.1243989-17-satyanarayana.k.v.p@intel.com>
On 9/21/2026 11:21 AM, Satyanarayana K V P wrote:
> Use check_add_overflow() for the VF range check, and rearrange the
> subsequent clamping into the subtraction form so the expression
> does not overflow.
>
> Fixes: e904c56ba6e0 ("drm/xe: Rewrite GGTT VF initialization")
> Reported-by: Sashiko <sashiko-bot@kernel.org>
> Closes: https://sashiko.dev/#/patchset/20260728095026.2197333-6-satyanarayana.k.v.p%40intel.com
> Signed-off-by: Satyanarayana K V P <satyanarayana.k.v.p@intel.com>
> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
> Cc: Matthew Brost <matthew.brost@intel.com>
> Cc: Maarten Lankhorst <dev@lankhorst.se>
> ---
> V9 -> V10:
> - Cleanup old stale code added in the previous revision.
>
> V8 -> V9:
> - Fixed review comments (Michal W).
> - Used check_add_overflow() for range checks (Michal W).
> ---
> drivers/gpu/drm/xe/xe_ggtt.c | 20 +++++++++++++-------
> 1 file changed, 13 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/gpu/drm/xe/xe_ggtt.c b/drivers/gpu/drm/xe/xe_ggtt.c
> index 7cac01cef1ee..2557e51bb7cc 100644
> --- a/drivers/gpu/drm/xe/xe_ggtt.c
> +++ b/drivers/gpu/drm/xe/xe_ggtt.c
> @@ -8,6 +8,7 @@
> #include <kunit/visibility.h>
> #include <linux/fault-inject.h>
> #include <linux/io-64-nonatomic-lo-hi.h>
> +#include <linux/overflow.h>
> #include <linux/sizes.h>
>
> #include <drm/drm_drv.h>
> @@ -393,8 +394,9 @@ int xe_ggtt_init_early(struct xe_ggtt *ggtt)
> {
> struct xe_device *xe = tile_to_xe(ggtt->tile);
> struct pci_dev *pdev = to_pci_dev(xe->drm.dev);
> + u64 ggtt_start, ggtt_size, ggtt_end;
> + u64 wopcm = xe_wopcm_size(xe);
> unsigned int gsm_size;
> - u64 ggtt_start, wopcm = xe_wopcm_size(xe), ggtt_size;
> int err;
>
> if (!IS_SRIOV_VF(xe)) {
> @@ -408,14 +410,21 @@ int xe_ggtt_init_early(struct xe_ggtt *ggtt)
> }
> ggtt_start = wopcm;
> ggtt_size = (gsm_size / 8) * (u64)XE_PAGE_SIZE - ggtt_start;
nit: if we are paranoid, shouldn't we use check_mul_overflow() too?
nit: one day, we should change this magic 8 into sizeof(u64)
nit: shouldn't we check/assert that GSM covers more than just WOPCM?
then:
if (check_mul_overflow(gsm_size / sizeof(u64),
XE_PAGE_SIZE, &ggtt_end) ||
ggtt_end > GUC_GGTT_TOP) {
ggtt_end = GUC_GGTT_TOP;
} else if (ggtt_end < wopcm) {
xe_tile_err(tile, "Hardware reported too small GGTT %llu\n", ..
return -ENOSPC;
}
ggtt_start = wopcm;
ggtt_size = ggtt_end - ggtt_start;
or
u64 gsm_to_ggtt_size(unsigned int gsm_size)
{
u64 ggtt_size;
if (WARN_ONCE(check_mul_overflow(gsm_size / sizeof(u64),
XE_PAGE_SIZE, &ggtt_size)))
return SZ_4G;
return ggtt_size;
}
ggtt_end = min(GUC_GGTT_TOP, gsm_to_ggtt_size(gsm_size));
if (ggtt_end < wopcm) {
xe_tile_err(tile, "Hardware reported too small GGTT %llu\n", ..
return -ENOSPC;
}
ggtt_start = wopcm;
ggtt_size = ggtt_end - ggtt_start;
> +
> + if (check_add_overflow(ggtt_start, ggtt_size, &ggtt_end) ||
> + ggtt_end > GUC_GGTT_TOP)
> + ggtt_size = GUC_GGTT_TOP - ggtt_start;
> +
> } else {
> ggtt_start = xe_tile_sriov_vf_ggtt_base(ggtt->tile);
> ggtt_size = xe_tile_sriov_vf_ggtt(ggtt->tile);
>
> if (ggtt_start < wopcm ||
> - ggtt_start + ggtt_size > GUC_GGTT_TOP) {
> - xe_tile_err(ggtt->tile, "Invalid GGTT configuration: %#llx-%#llx\n",
> - ggtt_start, ggtt_start + ggtt_size - 1);
> + check_add_overflow(ggtt_start, ggtt_size, &ggtt_end) ||
> + ggtt_end > GUC_GGTT_TOP) {
> + xe_tile_err(ggtt->tile,
> + "Invalid GGTT configuration: base %#llx, size %#llx\n",
> + ggtt_start, ggtt_size);
> return -ERANGE;
> }
> }
> @@ -424,9 +433,6 @@ int xe_ggtt_init_early(struct xe_ggtt *ggtt)
> if (xe_vram_needs_64k(xe))
> ggtt->flags |= XE_GGTT_FLAGS_64K;
>
> - if (ggtt_size + ggtt_start > GUC_GGTT_TOP)
> - ggtt_size = GUC_GGTT_TOP - ggtt_start;
> -
> if (GRAPHICS_VERx100(xe) >= 1270)
> ggtt->pt_ops =
> (ggtt->tile->media_gt && XE_GT_WA(ggtt->tile->media_gt, 22019338487)) ||
next prev parent reply other threads:[~2026-09-21 14:03 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-21 9:20 [PATCH v10 00/10] KUnit test for VF provisioning error handling Satyanarayana K V P
2026-09-21 9:21 ` [PATCH v10 01/10] drm/xe/guc: Allow to replace xe_guc_mmio_send_recv() with KUNIT stub Satyanarayana K V P
2026-09-21 9:21 ` [PATCH v10 02/10] drm/xe/vf: Split submission config query helpers Satyanarayana K V P
2026-09-21 9:21 ` [PATCH v10 03/10] drm/xe/vf: Add bounds checking for queried context and doorbell counts Satyanarayana K V P
2026-09-21 9:21 ` [PATCH v10 04/10] drm/xe: Introduce helpers for VRAM alignment Satyanarayana K V P
2026-09-21 9:21 ` [PATCH v10 05/10] drm/xe/ggtt: Avoid integer overflow when validating VF GGTT range Satyanarayana K V P
2026-09-21 14:03 ` Michal Wajdeczko [this message]
2026-09-21 9:21 ` [PATCH v10 06/10] drm/xe/sriov: Add helper for VF GGTT provisioning alignment Satyanarayana K V P
2026-09-21 13:26 ` Michal Wajdeczko
2026-09-21 9:21 ` [PATCH v10 07/10] drm/xe/vf: Add bounds checking for queried GGTT base and size Satyanarayana K V P
2026-09-21 14:36 ` Michal Wajdeczko
2026-09-21 9:21 ` [PATCH v10 08/10] drm/xe/vf: Add alignment check for queried VRAM size Satyanarayana K V P
2026-09-21 9:21 ` [PATCH v10 09/10] drm/xe/ggtt: Add KUnit stub for xe_ggtt_shift_nodes() Satyanarayana K V P
2026-09-21 14:45 ` Michal Wajdeczko
2026-09-21 9:21 ` [PATCH v10 10/10] drm/xe/tests: Add KUnit tests for VF provisioning error handling Satyanarayana K V P
2026-09-21 16:47 ` Michal Wajdeczko
2026-09-21 9:49 ` ✗ CI.checkpatch: warning for KUnit test for VF provisioning error handling (rev10) Patchwork
2026-09-21 9:51 ` ✓ CI.KUnit: success " Patchwork
2026-09-21 11:32 ` ✗ Xe.CI.BAT: failure " Patchwork
2026-09-21 14:10 ` ✗ 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=4faf44b6-c321-4ce3-a88f-01fa442f58bd@intel.com \
--to=michal.wajdeczko@intel.com \
--cc=dev@lankhorst.se \
--cc=intel-xe@lists.freedesktop.org \
--cc=matthew.brost@intel.com \
--cc=sashiko-bot@kernel.org \
--cc=satyanarayana.k.v.p@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