From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: crypto: tegra - Don't touch bo refcount in host1x bo pin/unpin Date: Sat, 16 May 2026 10:12:28 +1000 Message-ID: In-Reply-To: <20260515-host1x-bocache-leak-v1-2-a0375f68aeab@nvidia.com> References: <20260515-host1x-bocache-leak-v1-0-a0375f68aeab@nvidia.com> <20260515-host1x-bocache-leak-v1-2-a0375f68aeab@nvidia.com> X-Mailer: Claude Code Patch Reviewer Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Patch Review Straightforward follow-up aligning the crypto driver with the new contract: ```c - map->bo =3D host1x_bo_get(bo); + map->bo =3D bo; ``` ```c - host1x_bo_put(map->bo); ``` The crypto driver (tegra-se) never uses caching (the `host1x_bo_pin` calls = pass `NULL` for cache), so no `host1x_bo_clear_cached_mappings` call is nee= ded. The cmdbuf manages its own lifetime via `cmdbuf->ref`, independent of = the host1x_bo refcount. The cover letter's claim that this is safe to merge independently is correc= t: without Patch 1, the old `host1x_bo_pin`/`host1x_bo_unpin` don't touch t= he BO refcount either (it was all in the callbacks), so removing the ref ma= nipulation from both pin and unpin is a no-op from a lifecycle perspective = =E2=80=94 the crypto driver's own kref on the cmdbuf ensures the BO outlive= s the mapping. **No issues with this patch.** --- Generated by Claude Code Patch Reviewer