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: media: meson: vdec: Fix concurrent STREAMON / STREAMOFF race conditions
Date: Thu, 04 Jun 2026 15:48:46 +1000	[thread overview]
Message-ID: <review-patch2-20260530094326.11892-3-linux.amoon@gmail.com> (raw)
In-Reply-To: <20260530094326.11892-3-linux.amoon@gmail.com>

Patch Review

**Good idea:** Adding mutex protection around `core->cur_sess` and `sess->status` transitions.

**Issue — `STATUS_INIT` set too early:** The patch sets `sess->status = STATUS_INIT` and `core->cur_sess = sess` inside the mutex *before* any hardware initialization (vififo allocation, poweron, codec start, kthread). If any of these fail, the error unwind correctly resets the state, but there's a window where `sess->status == STATUS_INIT` while the hardware isn't actually initialized. Another call to `vdec_stop_streaming()` during this window could try to stop a partially-initialized session, calling `kthread_stop()` on an uninitialized `recycle_thread`, `vdec_poweroff()` on hardware that was never powered on, etc. The original code set `STATUS_INIT` and `cur_sess` at the end, which was better in that regard (though also racy).

**Issue — `STATUS_INIT` early return bypasses mutex unlock properly:**
```c
if (sess->status == STATUS_INIT) {
    mutex_unlock(&core->lock);
    return 0;
}
```
This is fine structurally but means a second STREAMON on a session that's already STATUS_INIT but not yet STATUS_RUNNING returns success without completing initialization. This might already be the intent of the original code though.

**Issue — stop_streaming sets STATUS_STOPPED before hardware teardown:**
```c
mutex_lock(&core->lock);
old_status = sess->status;
if (core->cur_sess == sess)
    core->cur_sess = NULL;
sess->status = STATUS_STOPPED;
mutex_unlock(&core->lock);
```
Setting `STATUS_STOPPED` and clearing `cur_sess` before actually stopping the hardware means that during the hardware teardown (kthread_stop, vdec_poweroff, etc.), a concurrent `vdec_start_streaming()` could see `STATUS_STOPPED` / `cur_sess == NULL` and attempt to start a new session while the old one is still being torn down. The mutex only protects the state variables, not the actual hardware operations that follow.

---
Generated by Claude Code Patch Reviewer

  reply	other threads:[~2026-06-04  5:48 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-30  9:42 [PATCH v6 0/8] media: meson: Fix memory leak in error path in vdec Anand Moon
2026-05-30  9:42 ` [PATCH v6 1/8] media: meson: vdec: Fix memory leaks and lifetime of m2m device Anand Moon
2026-06-04  5:48   ` Claude review: " Claude Code Review Bot
2026-05-30  9:42 ` [PATCH v6 2/8] media: meson: vdec: Fix concurrent STREAMON / STREAMOFF race conditions Anand Moon
2026-06-04  5:48   ` Claude Code Review Bot [this message]
2026-05-30  9:42 ` [PATCH v6 3/8] media: meson: vdec: Handle kthread failure and free codec state Anand Moon
2026-06-04  5:48   ` Claude review: " Claude Code Review Bot
2026-05-30  9:42 ` [PATCH v6 4/8] media: meson: vdec: Condition buffer flushing on queue type in start_streaming Anand Moon
2026-06-04  5:48   ` Claude review: " Claude Code Review Bot
2026-05-30  9:42 ` [PATCH v6 5/8] media: meson: vdec: Cancel esparser work during teardown Anand Moon
2026-06-04  5:48   ` Claude review: " Claude Code Review Bot
2026-05-30  9:42 ` [PATCH v6 6/8] media: meson: vdec: Configure DMA mask and segment size in probe Anand Moon
2026-06-04  5:48   ` Claude review: " Claude Code Review Bot
2026-05-30  9:42 ` [PATCH v6 7/8] media: meson: vdec: Fix NULL pointer dereference in ISR handlers Anand Moon
2026-06-04  5:48   ` Claude review: " Claude Code Review Bot
2026-05-30  9:42 ` [PATCH v6 8/8] gpu: drm: meson: Fix DMA max segment size for DMABUF imports Anand Moon
2026-06-04  5:48   ` Claude review: " Claude Code Review Bot
2026-06-04  5:48 ` Claude review: media: meson: Fix memory leak in error path in vdec Claude Code Review 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=review-patch2-20260530094326.11892-3-linux.amoon@gmail.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