From patchwork Wed May 22 05:33:39 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 13670425 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 9F9F7C25B7E for ; Wed, 22 May 2024 05:33:27 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 1AEBE10E4C9; Wed, 22 May 2024 05:33:25 +0000 (UTC) Received: from mblankhorst.nl (lankhorst.se [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id CEC1D10E1E2; Wed, 22 May 2024 05:33:22 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Cc: intel-xe@lists.freedesktop.org, Maarten Lankhorst , Chaitanya Kumar Borah Subject: [v6 1/3] drm: Add drm_vblank_work_flush_all(). Date: Wed, 22 May 2024 07:33:39 +0200 Message-ID: <20240522053341.137592-2-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20240522053341.137592-1-maarten.lankhorst@linux.intel.com> References: <20240522053341.137592-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" In some cases we want to flush all vblank work, right before vblank_off for example. Add a simple function to make this possible. Check that both pending_work and running work are empty when flushing. Co-Developed-by: Chaitanya Kumar Borah Signed-off-by: Maarten Lankhorst Reviewed-by: Uma Shankar --- drivers/gpu/drm/drm_vblank_work.c | 22 ++++++++++++++++++++++ include/drm/drm_vblank_work.h | 2 ++ 2 files changed, 24 insertions(+) diff --git a/drivers/gpu/drm/drm_vblank_work.c b/drivers/gpu/drm/drm_vblank_work.c index 4fe9b1d3b00f..1752ffb44e1d 100644 --- a/drivers/gpu/drm/drm_vblank_work.c +++ b/drivers/gpu/drm/drm_vblank_work.c @@ -232,6 +232,28 @@ void drm_vblank_work_flush(struct drm_vblank_work *work) } EXPORT_SYMBOL(drm_vblank_work_flush); +/** + * drm_vblank_work_flush_all - flush all currently pending vblank work on crtc. + * @crtc: crtc for which vblank work to flush + * + * Wait until all currently queued vblank work on @crtc + * has finished executing once. + */ +void drm_vblank_work_flush_all(struct drm_crtc *crtc) +{ + struct drm_device *dev = crtc->dev; + struct drm_vblank_crtc *vblank = &dev->vblank[drm_crtc_index(crtc)]; + + spin_lock_irq(&dev->event_lock); + wait_event_lock_irq(vblank->work_wait_queue, + list_empty(&vblank->pending_work), + dev->event_lock); + spin_unlock_irq(&dev->event_lock); + + kthread_flush_worker(vblank->worker); +} +EXPORT_SYMBOL(drm_vblank_work_flush_all); + /** * drm_vblank_work_init - initialize a vblank work item * @work: vblank work item diff --git a/include/drm/drm_vblank_work.h b/include/drm/drm_vblank_work.h index eb41d0810c4f..e04d436b7297 100644 --- a/include/drm/drm_vblank_work.h +++ b/include/drm/drm_vblank_work.h @@ -17,6 +17,7 @@ struct drm_crtc; * drm_vblank_work_init() * drm_vblank_work_cancel_sync() * drm_vblank_work_flush() + * drm_vblank_work_flush_all() */ struct drm_vblank_work { /** @@ -67,5 +68,6 @@ void drm_vblank_work_init(struct drm_vblank_work *work, struct drm_crtc *crtc, void (*func)(struct kthread_work *work)); bool drm_vblank_work_cancel_sync(struct drm_vblank_work *work); void drm_vblank_work_flush(struct drm_vblank_work *work); +void drm_vblank_work_flush_all(struct drm_crtc *crtc); #endif /* !_DRM_VBLANK_WORK_H_ */ From patchwork Wed May 22 05:33:40 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 13670426 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 C31E2C25B77 for ; Wed, 22 May 2024 05:33:34 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 1FB4C10E201; Wed, 22 May 2024 05:33:34 +0000 (UTC) Received: from mblankhorst.nl (lankhorst.se [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id AD4CC10E12D; Wed, 22 May 2024 05:33:23 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Cc: intel-xe@lists.freedesktop.org, =?utf-8?b?VmlsbGUgU3lyasOkbMOk?= , Maarten Lankhorst Subject: [v6 2/3] drm/i915: Use vblank worker to unpin old legacy cursor fb safely Date: Wed, 22 May 2024 07:33:40 +0200 Message-ID: <20240522053341.137592-3-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20240522053341.137592-1-maarten.lankhorst@linux.intel.com> References: <20240522053341.137592-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" From: Ville Syrjälä The cursor hardware only does sync updates, and thus the hardware will be scanning out from the old fb until the next start of vblank. So in order to make the legacy cursor fastpath actually safe we should not unpin the old fb until we're sure the hardware has ceased accessing it. The simplest approach is to just use a vblank work here to do the delayed unpin. Not 100% sure it's a good idea to put this onto the same high priority vblank worker as eg. our timing critical gamma updates. But let's keep it simple for now, and it we later discover that this is causing problems we can think about adding a lower priority worker for such things. This patch is slightly reworked by Maarten Cc: Maarten Lankhorst Signed-off-by: Ville Syrjälä Signed-off-by: Maarten Lankhorst Reviewed-by: Uma Shankar --- drivers/gpu/drm/i915/display/intel_cursor.c | 26 +++++++++++++++++-- drivers/gpu/drm/i915/display/intel_display.c | 3 +++ .../drm/i915/display/intel_display_types.h | 3 +++ 3 files changed, 30 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_cursor.c b/drivers/gpu/drm/i915/display/intel_cursor.c index c780ce146131..36e2dcbe3614 100644 --- a/drivers/gpu/drm/i915/display/intel_cursor.c +++ b/drivers/gpu/drm/i915/display/intel_cursor.c @@ -710,6 +710,17 @@ static bool intel_cursor_format_mod_supported(struct drm_plane *_plane, return format == DRM_FORMAT_ARGB8888; } +static void intel_cursor_unpin_work(struct kthread_work *base) +{ + struct drm_vblank_work *work = to_drm_vblank_work(base); + struct intel_plane_state *plane_state = + container_of(work, typeof(*plane_state), unpin_work); + struct intel_plane *plane = to_intel_plane(plane_state->uapi.plane); + + intel_plane_unpin_fb(plane_state); + intel_plane_destroy_state(&plane->base, &plane_state->uapi); +} + static int intel_legacy_cursor_update(struct drm_plane *_plane, struct drm_crtc *_crtc, @@ -853,14 +864,25 @@ intel_legacy_cursor_update(struct drm_plane *_plane, intel_psr_unlock(crtc_state); - intel_plane_unpin_fb(old_plane_state); + if (old_plane_state->ggtt_vma != new_plane_state->ggtt_vma) { + drm_vblank_work_init(&old_plane_state->unpin_work, &crtc->base, + intel_cursor_unpin_work); + + drm_vblank_work_schedule(&old_plane_state->unpin_work, + drm_crtc_accurate_vblank_count(&crtc->base) + 1, + false); + + old_plane_state = NULL; + } else { + intel_plane_unpin_fb(old_plane_state); + } out_free: if (new_crtc_state) intel_crtc_destroy_state(&crtc->base, &new_crtc_state->uapi); if (ret) intel_plane_destroy_state(&plane->base, &new_plane_state->uapi); - else + else if (old_plane_state) intel_plane_destroy_state(&plane->base, &old_plane_state->uapi); return ret; diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index cce1420fb541..715672c142b4 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -66,6 +66,7 @@ #include "intel_crtc.h" #include "intel_crtc_state_dump.h" #include "intel_cursor_regs.h" +#include "intel_cursor.h" #include "intel_ddi.h" #include "intel_de.h" #include "intel_display_driver.h" @@ -6925,6 +6926,8 @@ static void intel_commit_modeset_disables(struct intel_atomic_state *state) continue; intel_crtc_disable_planes(state, crtc); + + drm_vblank_work_flush_all(&crtc->base); } /* Only disable port sync and MST slaves */ diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h index 9678c2b157f6..51fa73a2a161 100644 --- a/drivers/gpu/drm/i915/display/intel_display_types.h +++ b/drivers/gpu/drm/i915/display/intel_display_types.h @@ -735,6 +735,9 @@ struct intel_plane_state { struct intel_fb_view view; u32 phys_dma_addr; /* for cursor_needs_physical */ + /* for legacy cursor fb unpin */ + struct drm_vblank_work unpin_work; + /* Plane pxp decryption state */ bool decrypt; From patchwork Wed May 22 05:33:41 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 13670427 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 1A863C25B78 for ; Wed, 22 May 2024 05:33:35 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 5AA4E10E9AC; Wed, 22 May 2024 05:33:34 +0000 (UTC) Received: from mblankhorst.nl (lankhorst.se [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 9B8B110E201; Wed, 22 May 2024 05:33:24 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Cc: intel-xe@lists.freedesktop.org, Maarten Lankhorst Subject: [v6 3/3] drm/i915: Use the same vblank worker for atomic unpin Date: Wed, 22 May 2024 07:33:41 +0200 Message-ID: <20240522053341.137592-4-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.43.0 In-Reply-To: <20240522053341.137592-1-maarten.lankhorst@linux.intel.com> References: <20240522053341.137592-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" In case of legacy cursor update, the cursor VMA needs to be unpinned only after vblank. This exceeds the lifetime of the whole atomic commit. Any trick I attempted to keep the atomic commit alive didn't work, as drm_atomic_helper_setup_commit() force throttles on any old commit that wasn't cleaned up. The only option remaining is to remove the plane from the atomic commit, and use the same path as the legacy cursor update to clean the state after vblank. Changes since previous version: - Call the memset for plane state immediately when scheduling vblank, this prevents a use-after-free in cursor cleanup. Signed-off-by: Maarten Lankhorst Reviewed-by: Uma Shankar --- .../gpu/drm/i915/display/intel_atomic_plane.c | 13 +++++++- .../gpu/drm/i915/display/intel_atomic_plane.h | 2 ++ drivers/gpu/drm/i915/display/intel_crtc.c | 31 +++++++++++++++++++ drivers/gpu/drm/i915/display/intel_cursor.c | 2 +- drivers/gpu/drm/i915/display/intel_cursor.h | 3 ++ 5 files changed, 49 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_atomic_plane.c b/drivers/gpu/drm/i915/display/intel_atomic_plane.c index 27224ecdc94c..a06c676c9bb3 100644 --- a/drivers/gpu/drm/i915/display/intel_atomic_plane.c +++ b/drivers/gpu/drm/i915/display/intel_atomic_plane.c @@ -42,6 +42,7 @@ #include "i915_reg.h" #include "intel_atomic_plane.h" #include "intel_cdclk.h" +#include "intel_cursor.h" #include "intel_display_rps.h" #include "intel_display_trace.h" #include "intel_display_types.h" @@ -1188,7 +1189,6 @@ intel_cleanup_plane_fb(struct drm_plane *plane, intel_display_rps_mark_interactive(dev_priv, state, false); - /* Should only be called after a successful intel_prepare_plane_fb()! */ intel_plane_unpin_fb(old_plane_state); } @@ -1201,3 +1201,14 @@ void intel_plane_helper_add(struct intel_plane *plane) { drm_plane_helper_add(&plane->base, &intel_plane_helper_funcs); } + +void intel_plane_init_cursor_vblank_work(struct intel_plane_state *old_plane_state, + struct intel_plane_state *new_plane_state) +{ + if (!old_plane_state->ggtt_vma || + old_plane_state->ggtt_vma == new_plane_state->ggtt_vma) + return; + + drm_vblank_work_init(&old_plane_state->unpin_work, old_plane_state->uapi.crtc, + intel_cursor_unpin_work); +} diff --git a/drivers/gpu/drm/i915/display/intel_atomic_plane.h b/drivers/gpu/drm/i915/display/intel_atomic_plane.h index e7a0699f17c8..4c2031fc3504 100644 --- a/drivers/gpu/drm/i915/display/intel_atomic_plane.h +++ b/drivers/gpu/drm/i915/display/intel_atomic_plane.h @@ -67,5 +67,7 @@ void intel_plane_set_invisible(struct intel_crtc_state *crtc_state, struct intel_plane_state *plane_state); void intel_plane_helper_add(struct intel_plane *plane); bool intel_plane_needs_physical(struct intel_plane *plane); +void intel_plane_init_cursor_vblank_work(struct intel_plane_state *old_plane_state, + struct intel_plane_state *new_plane_state); #endif /* __INTEL_ATOMIC_PLANE_H__ */ diff --git a/drivers/gpu/drm/i915/display/intel_crtc.c b/drivers/gpu/drm/i915/display/intel_crtc.c index 339010384b86..283106558b2a 100644 --- a/drivers/gpu/drm/i915/display/intel_crtc.c +++ b/drivers/gpu/drm/i915/display/intel_crtc.c @@ -499,6 +499,19 @@ void intel_pipe_update_start(struct intel_atomic_state *state, if (intel_crtc_needs_vblank_work(new_crtc_state)) intel_crtc_vblank_work_init(new_crtc_state); + if (state->base.legacy_cursor_update) { + struct intel_plane *plane; + struct intel_plane_state *old_plane_state, *new_plane_state; + int i; + + for_each_oldnew_intel_plane_in_state(state, plane, old_plane_state, + new_plane_state, i) { + if (old_plane_state->uapi.crtc == &crtc->base) + intel_plane_init_cursor_vblank_work(old_plane_state, + new_plane_state); + } + } + intel_vblank_evade_init(old_crtc_state, new_crtc_state, &evade); if (drm_WARN_ON(&dev_priv->drm, drm_crtc_vblank_get(&crtc->base))) @@ -615,6 +628,24 @@ void intel_pipe_update_end(struct intel_atomic_state *state, new_crtc_state->uapi.event = NULL; } + if (state->base.legacy_cursor_update) { + struct intel_plane *plane; + struct intel_plane_state *old_plane_state; + int i; + + for_each_old_intel_plane_in_state(state, plane, old_plane_state, i) { + if (old_plane_state->uapi.crtc == &crtc->base && + old_plane_state->unpin_work.vblank) { + drm_vblank_work_schedule(&old_plane_state->unpin_work, + drm_crtc_accurate_vblank_count(&crtc->base) + 1, + false); + + /* Remove plane from atomic state, cleanup/free is done from vblank worker. */ + memset(&state->base.planes[i], 0, sizeof(state->base.planes[i])); + } + } + } + /* * Send VRR Push to terminate Vblank. If we are already in vblank * this has to be done _after_ sampling the frame counter, as diff --git a/drivers/gpu/drm/i915/display/intel_cursor.c b/drivers/gpu/drm/i915/display/intel_cursor.c index 36e2dcbe3614..a6dcc4d95ff2 100644 --- a/drivers/gpu/drm/i915/display/intel_cursor.c +++ b/drivers/gpu/drm/i915/display/intel_cursor.c @@ -710,7 +710,7 @@ static bool intel_cursor_format_mod_supported(struct drm_plane *_plane, return format == DRM_FORMAT_ARGB8888; } -static void intel_cursor_unpin_work(struct kthread_work *base) +void intel_cursor_unpin_work(struct kthread_work *base) { struct drm_vblank_work *work = to_drm_vblank_work(base); struct intel_plane_state *plane_state = diff --git a/drivers/gpu/drm/i915/display/intel_cursor.h b/drivers/gpu/drm/i915/display/intel_cursor.h index ce333bf4c2d5..e2d9ec710a86 100644 --- a/drivers/gpu/drm/i915/display/intel_cursor.h +++ b/drivers/gpu/drm/i915/display/intel_cursor.h @@ -9,9 +9,12 @@ enum pipe; struct drm_i915_private; struct intel_plane; +struct kthread_work; struct intel_plane * intel_cursor_plane_create(struct drm_i915_private *dev_priv, enum pipe pipe); +void intel_cursor_unpin_work(struct kthread_work *base); + #endif