From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: drm/msm: Add sysprof accessors Date: Mon, 25 May 2026 18:21:40 +1000 Message-ID: In-Reply-To: <20260522173349.55491-8-robin.clark@oss.qualcomm.com> References: <20260522173349.55491-1-robin.clark@oss.qualcomm.com> <20260522173349.55491-8-robin.clark@oss.qualcomm.com> X-Mailer: Claude Code Patch Reviewer Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit MIME-Version: 1.0 Patch Review Good preparatory split. One observation: In `a8xx_gpu.c` hw_init: ```c - if (!ret && (refcount_read(&gpu->sysprof_active) > 1)) { + if (!ret && msm_gpu_sysprof_no_perfcntr_zap(gpu)) { ``` While `a6xx_gpu.c` hw_init uses `msm_gpu_sysprof_no_ifpc(gpu)`. At this point in the series the two functions are identical, but they diverge later (patch 10 adds perfcntr stream check to `_no_perfcntr_zap`, patch 16 removes it from `_no_ifpc`). For a8xx hw_init, this means IFPC will be blocked during hw_init if a perfcntr stream is active, which is a reasonable conservative approach during the initialization window before SEL regs are restored. The `a8xx_gpu_funcs` addition of `.sysprof_setup` looks correct -- this was previously missing from the a8xx funcs table. --- Generated by Claude Code Patch Reviewer