public inbox for drm-ai-reviews@public-inbox.freedesktop.org
 help / color / mirror / Atom feed
From: Claude Code Review Bot <claude-review@example.com>
To: dri-devel-reviews@example.com
Subject: Claude review: drm/amdgpu/userq: fix memory leak in MQD creation error paths
Date: Mon, 16 Mar 2026 11:58:18 +1000	[thread overview]
Message-ID: <review-patch1-SYBPR01MB7881A279A361F81B670CDEEAAF42A@SYBPR01MB7881.ausprd01.prod.outlook.com> (raw)
In-Reply-To: <SYBPR01MB7881A279A361F81B670CDEEAAF42A@SYBPR01MB7881.ausprd01.prod.outlook.com>

Patch Review

**Bug analysis is correct.** Looking at the current tree (`mes_userqueue.c:309-398`), each branch does:
1. `memdup_user()` → allocates a struct (e.g., `compute_mqd`)
2. `amdgpu_userq_input_va_validate()` → may fail
3. `kfree()` → only reached on the success path

The `free_mqd` label does not free these IP-specific allocations, so the patch correctly adds `kfree()` before each `goto free_mqd`.

**All four error paths are addressed:**
- COMPUTE branch: `kfree(compute_mqd)` at line 192
- GFX branch: `kfree(mqd_gfx_v11)` at lines 204 and 211 (two validation calls)
- SDMA branch: `kfree(mqd_sdma_v11)` at line 223

**Minor observations:**

1. **Correctness concern in COMPUTE branch:** In the current tree at line 334, `userq_props->tmz_queue` and other properties are set *after* the validation call. The patch correctly places the `kfree()` before any of those assignments, so this is fine.

2. **Correctness concern in GFX branch:** At lines 361-364 in the tree, `userq_props->shadow_addr` and `userq_props->csa_addr` are assigned from `mqd_gfx_v11` *before* the validation calls. This means on the error path, `userq_props` contains stale VA values from user input. This isn't a problem introduced by this patch (and `userq_props` itself is freed by `free_mqd`), but it's worth noting.

3. **Style suggestion:** Rather than sprinkling `kfree()` at each validation failure, a cleaner approach would be to use a local `goto` label per branch (e.g., `free_compute_mqd`) or restructure so the `kfree()` is always done before any `goto`. However, this fix is minimal and correct, which is appropriate for a stable-targeted bugfix.

4. **Fixes tag:** The referenced commit `9e46b8bb0539` ("drm/amdgpu: validate userq buffer virtual address and size") is the commit that introduced the validation calls (and thus the leak), so the Fixes tag is correct.

**Overall: The patch is correct and ready to apply.** It fixes a genuine memory leak with minimal, targeted changes. The `Cc: stable` tag is appropriate since this is a straightforward resource leak fix.

---
Generated by Claude Code Patch Reviewer

      parent reply	other threads:[~2026-03-16  1:58 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-14 15:33 [PATCH] drm/amdgpu/userq: fix memory leak in MQD creation error paths Junrui Luo
2026-03-14 20:04 ` Markus Elfring
2026-03-15  5:25   ` Junrui Luo
2026-03-15  9:50     ` Markus Elfring
2026-03-16  1:58 ` Claude review: " Claude Code Review Bot
2026-03-16  1:58 ` Claude Code Review Bot [this message]

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=review-patch1-SYBPR01MB7881A279A361F81B670CDEEAAF42A@SYBPR01MB7881.ausprd01.prod.outlook.com \
    --to=claude-review@example.com \
    --cc=dri-devel-reviews@example.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