From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id BD2D4C61DBE for ; Wed, 26 Aug 2026 21:35:33 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id DE4CA10E6BF; Wed, 26 Aug 2026 21:35:32 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=kernel.org header.i=@kernel.org header.b="Oi2wvBIL"; dkim-atps=neutral Received: from tor.source.kernel.org (tor.source.kernel.org [172.105.4.254]) by gabe.freedesktop.org (Postfix) with ESMTPS id 51BB010ECD1 for ; Wed, 26 Aug 2026 21:35:32 +0000 (UTC) Received: from smtp.kernel.org (quasi.space.kernel.org [100.103.45.18]) by tor.source.kernel.org (Postfix) with ESMTP id 8DF4C6020E; Wed, 26 Aug 2026 21:35:31 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 2CDFD1F000E9; Wed, 26 Aug 2026 21:35:31 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1787780131; bh=QBeIZhuh5fpMy2E0EEJ987/+M7qbUynbI/25X0DGK9A=; h=From:Subject:Reply-To:To:Cc:In-Reply-To:References:Date; b=Oi2wvBILSynfOM077XwpPZdTMxaWs93RVlTRzgIAW9TpSLpdn7W6n6RGicRy1tdIc XXc/alfnI+DQ0V/8LBUBAoXpXcg4uKtzULpQKM5amGuYp2NqhRoQzTK65cB6xCvq6Q eNJLyz+h9yxJHZUTBxlA6cUFIR7pRs+nUu4qLG6Dujl0qGliTheOq8soJ2cDXwNXqe lxhsba1d0Z/PnIpbTWtKf2IyL1RuskKaBEE8e5e0rn6MYZNOP8W1QZAJf0N/dAgEox pt+Yipnz3qYVgXoAtlKHoXIpwHwVYVhPHptIoWTxr8ZS1ggYqNflKP+Od1otO2NAMb q9+UnadWM8GOA== From: sashiko-bot@kernel.org Subject: Re: [PATCH v6 15/16] drm/panfrost: Fix races between perfcnt and reset sequence To: =?utf-8?b?QWRyacOhbiBMYXJ1bWJl?= Cc: dri-devel@lists.freedesktop.org In-Reply-To: <20260826-claude-fixes-v6-15-78119af409e7@collabora.com> References: <20260826-claude-fixes-v6-0-78119af409e7@collabora.com> <20260826-claude-fixes-v6-15-78119af409e7@collabora.com> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Date: Wed, 26 Aug 2026 21:35:30 +0000 Message-Id: <20260826213531.2CDFD1F000E9@smtp.kernel.org> X-BeenThere: dri-devel@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Direct Rendering Infrastructure - Development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Reply-To: sashiko-reviews@lists.linux.dev Errors-To: dri-devel-bounces@lists.freedesktop.org Sender: "dri-devel" Thank you for your contribution! Sashiko AI review found 3 potential issue(= s) to consider: - [High] Hardware initialization is bypassed on new perfcnt sessions, break= ing performance counters. - [High] Unconditional MMU AS reference put causes reference count underflo= w. - [Medium] Mixing goto-based error handling with scoped_guard violates the = kernel cleanup API. -- commit 036d049b8b6f60dc9ba18742b07e01b404ef96c3 Author: Adri=C3=A1n Larumbe drm/panfrost: Fix races between perfcnt and reset sequence Formerly, the reset sequence would race with panfrost_mmu_as_put() when tearing down a perfcnt session. On top of that, poking GPU registers to program a perfcnt session or obtaining a dump might lead to undefined behaviour when done at the same time a reset was ongoing. > diff --git a/drivers/gpu/drm/panfrost/panfrost_perfcnt.c b/drivers/gpu/dr= m/panfrost/panfrost_perfcnt.c > index b3f71d7fd82a9..7dcf27add7ae7 100644 > --- a/drivers/gpu/drm/panfrost/panfrost_perfcnt.c > +++ b/drivers/gpu/drm/panfrost/panfrost_perfcnt.c [ ... ] > @@ -132,44 +206,20 @@ static int panfrost_perfcnt_enable_locked(struct pa= nfrost_device *pfdev, > gpu_write(pfdev, GPU_INT_CLEAR, GPU_IRQ_PERFCNT_SAMPLE_COMPLETED); > gpu_write(pfdev, GPU_CMD, GPU_CMD_PERFCNT_CLEAR); > =20 > - ret =3D panfrost_mmu_as_get(pfdev, perfcnt->mapping->mmu); > - if (ret < 0) > - goto err_vunmap; > - [ ... ] > + scoped_guard(rwsem_read, &pfdev->reset.lock) { > + if (perfcnt->reset_failed) { > + ret =3D panfrost_perfcnt_hw_enable(pfdev); > + if (ret) > + goto err_vunmap; > + } [Severity: High] Does this code inadvertently skip hardware initialization for new perfcnt sessions? Earlier in panfrost_perfcnt_enable_locked(), the function returns early if perfcnt->user is not NULL. Because reset_failed is only set during a reset while a session is already active, reset_failed is guaranteed to be false when creating a new session. Consequently, panfrost_perfcnt_hw_enable() is never called here, the MMU address space is not acquired, and hardware performance counters are never enabled. [Severity: Medium] Is it safe to mix goto-based error handling with scoped_guard() blocks here? According to the kernel cleanup subsystem guidelines, the usage of "goto" a= nd cleanup helpers should never be mixed in the same function to prevent confusing ownership semantics and potential resource leak bugs. The jump to err_vunmap directly exits the scoped_guard block, breaking this rule. > + perfcnt->reset_happened =3D false; > + perfcnt->reset_failed =3D false; > + perfcnt->user =3D user; > + } > =20 > /* The BO ref is retained by the mapping. */ > drm_gem_object_put(&bo->base); [ ... ] > @@ -195,13 +245,15 @@ static int panfrost_perfcnt_disable_locked(struct p= anfrost_device *pfdev, > if (user !=3D perfcnt->user) > return -EINVAL; > =20 > - panfrost_perfcnt_hw_disable(pfdev); > + scoped_guard(rwsem_read, &pfdev->reset.lock) { > + panfrost_perfcnt_hw_disable(pfdev); > + panfrost_mmu_as_put(pfdev, perfcnt->mapping->mmu); [Severity: High] Can this unconditional call to panfrost_mmu_as_put() cause a reference count underflow regression? If panfrost_perfcnt_hw_enable() was skipped during initialization as noted above, or if it failed to re-acquire the AS during reset recovery, the perf= cnt session does not actually hold an MMU AS reference. When a userspace client closes the perfcnt session, panfrost_mmu_as_put() is called unconditionally, which will drive the mmu->as_count reference count = to -1 and trigger a kernel warning. > + perfcnt->user =3D NULL; > + } > =20 > - perfcnt->user =3D NULL; > drm_gem_vunmap(&perfcnt->mapping->obj->base.base, &map); > perfcnt->buf =3D NULL; > panfrost_gem_close(&perfcnt->mapping->obj->base.base, file_priv); > - panfrost_mmu_as_put(pfdev, perfcnt->mapping->mmu); > panfrost_gem_mapping_put(perfcnt->mapping); > perfcnt->mapping =3D NULL; > pm_runtime_put_autosuspend(pfdev->base.dev); --=20 Sashiko AI review =C2=B7 https://sashiko.dev/#/patchset/20260826-claude-fix= es-v6-0-78119af409e7@collabora.com?part=3D15