From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: accel/rocket: fix ignored return value in rocket_ioctl_submit Date: Sat, 16 May 2026 16:08:43 +1000 Message-ID: In-Reply-To: <20260510200047.6117-1-contact.kartikn@gmail.com> References: <20260510200047.6117-1-contact.kartikn@gmail.com> X-Mailer: Claude Code Patch Reviewer Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Overall Series Review Subject: accel/rocket: fix ignored return value in rocket_ioctl_submit Author: Kartik Nair Patches: 1 Reviewed: 2026-05-16T16:08:43.322827 --- This is a single-patch fix for the `accel/rocket` driver where `rocket_ioct= l_submit()` was ignoring the return value of `rocket_ioctl_submit_job()`. T= he bug is real =E2=80=94 without this fix, a failed job submission is silen= tly swallowed and userspace receives a success return code despite the fail= ure. The fix is correct, minimal, and follows the existing error-handling p= attern already used in the same function (the `copy_from_user` loop above u= ses the identical `goto exit` pattern). **Verdict: The patch looks good.** One minor observation worth noting, but not necessarily a blocker: when a j= ob fails partway through a multi-job submission, the jobs that were already= successfully submitted (pushed to the scheduler) will still execute. This = is arguably the correct behavior =E2=80=94 `rocket_ioctl_submit_job()` full= y owns each job's lifecycle (it cleans up on error, and pushes to the sched= uler on success) =E2=80=94 but the caller has no way to cancel previously s= ubmitted jobs. Whether this is a problem depends on the userspace API contr= act. The original code had the same issue (it submitted all jobs regardless= of failures), so this patch is strictly an improvement. --- Generated by Claude Code Patch Reviewer