From: Stanley.Yang <Stanley.Yang@amd.com>
To: <amd-gfx@lists.freedesktop.org>
Cc: <Hawking.Zhang@amd.com>, <Tao.Zhou1@amd.com>,
<YiPeng.Chai@amd.com>, <Candice.Li@amd.com>,
Stanley.Yang <Stanley.Yang@amd.com>
Subject: [PATCH 4/5] drm/amd/ras: snapshot remote cmd header to fix double-fetch
Date: Mon, 18 May 2026 17:40:20 +0800 [thread overview]
Message-ID: <20260518094021.280968-4-Stanley.Yang@amd.com> (raw)
In-Reply-To: <20260518094021.280968-1-Stanley.Yang@amd.com>
The response header lives in PF-controlled shared memory. Copy it
into a local struct once, then read cmd_res and output_size from the
snapshot so the PF cannot flip cmd_res or grow output_size between
checks.
Signed-off-by: Stanley.Yang <Stanley.Yang@amd.com>
---
.../drm/amd/ras/ras_mgr/amdgpu_virt_ras_cmd.c | 29 +++++++++++++++----
1 file changed, 23 insertions(+), 6 deletions(-)
diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_virt_ras_cmd.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_virt_ras_cmd.c
index 838eb91aef39..ebbf92a2bd94 100644
--- a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_virt_ras_cmd.c
+++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_virt_ras_cmd.c
@@ -92,6 +92,7 @@ static int amdgpu_virt_ras_remote_ioctl_cmd(struct ras_core_context *ras_core,
struct amdgpu_virt_ras_cmd *virt_ras = ras_mgr->virt_ras_cmd;
uint32_t mem_len = ALIGN(sizeof(*cmd) + output_size, AMDGPU_GPU_PAGE_SIZE);
struct ras_cmd_ctx *rcmd;
+ struct ras_cmd_ctx hdr_snap;
struct amdgpu_virt_shared_mem shared_mem = {0};
int ret = 0;
@@ -108,15 +109,31 @@ static int amdgpu_virt_ras_remote_ioctl_cmd(struct ras_core_context *ras_core,
ret = amdgpu_virt_send_remote_ras_cmd(ras_core->dev,
shared_mem.gpa, mem_len);
if (!ret) {
- if (rcmd->cmd_res) {
- ret = rcmd->cmd_res;
+ /*
+ * rcmd lives in shared memory the PF can mutate at any time.
+ * Snapshot the entire fixed-size response header into a local
+ * struct in one shot so every subsequent decision (cmd_res,
+ * output_size, version, etc.) operates on a stable copy. This
+ * defeats double-fetch / TOCTOU attacks where a malicious or
+ * buggy PF could flip cmd_res from SUCCESS to an error after
+ * our success branch, or enlarge output_size between the
+ * bounds check and the memcpy below to corrupt the caller's
+ * local output buffer.
+ */
+ memcpy(&hdr_snap, rcmd, sizeof(hdr_snap));
+ barrier();
+
+ if (hdr_snap.cmd_res) {
+ ret = hdr_snap.cmd_res;
goto out;
}
- cmd->cmd_res = rcmd->cmd_res;
- cmd->output_size = rcmd->output_size;
- if (rcmd->output_size && (rcmd->output_size <= output_size) && output_data)
- memcpy(output_data, rcmd->output_buff_raw, rcmd->output_size);
+ cmd->cmd_res = hdr_snap.cmd_res;
+ cmd->output_size = hdr_snap.output_size;
+
+ if (hdr_snap.output_size && output_data &&
+ hdr_snap.output_size <= output_size)
+ memcpy(output_data, rcmd->output_buff_raw, hdr_snap.output_size);
}
out:
--
2.43.0
next prev parent reply other threads:[~2026-05-18 9:41 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-05-18 9:40 [PATCH 1/5] drm/amdgpu: fix potential overflow in fs_info.debugfs_name Stanley.Yang
2026-05-18 9:40 ` [PATCH 2/5] drm/amdgpu: init locals in umc_v12_0_convert_error_address Stanley.Yang
2026-05-18 9:40 ` [PATCH 3/5] drm/amd/ras: cap pending_ecc_list size Stanley.Yang
2026-05-18 9:40 ` Stanley.Yang [this message]
2026-05-18 9:40 ` [PATCH 5/5] drm/amdgpu: harden FRU PIA parsing with bounded helpers Stanley.Yang
2026-05-19 7:00 ` Zhou1, Tao
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=20260518094021.280968-4-Stanley.Yang@amd.com \
--to=stanley.yang@amd.com \
--cc=Candice.Li@amd.com \
--cc=Hawking.Zhang@amd.com \
--cc=Tao.Zhou1@amd.com \
--cc=YiPeng.Chai@amd.com \
--cc=amd-gfx@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;
as well as URLs for NNTP newsgroup(s).