From nobody Sun Sep 14 20:39:57 2025 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 vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 1DC2AC77B73 for ; Sun, 16 Apr 2023 11:53:39 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230163AbjDPLxg (ORCPT ); Sun, 16 Apr 2023 07:53:36 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:59388 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230021AbjDPLx0 (ORCPT ); Sun, 16 Apr 2023 07:53:26 -0400 Received: from madras.collabora.co.uk (madras.collabora.co.uk [IPv6:2a00:1098:0:82:1000:25:2eeb:e5ab]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id BC0B2211B for ; Sun, 16 Apr 2023 04:53:24 -0700 (PDT) Received: from workpc.. (unknown [IPv6:2a00:1370:817e:4eb4:c5e6:4b85:1e3f:55e4]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) (Authenticated sender: dmitry.osipenko) by madras.collabora.co.uk (Postfix) with ESMTPSA id 5668B6603242; Sun, 16 Apr 2023 12:53:22 +0100 (BST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=collabora.com; s=mail; t=1681646003; bh=sqddcWUiE06YeLZPkiTfcRceF34Y3LsZ4KnYjbnMu0s=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=UP5xeBqJS4VQFtlc3ur/90p2DlVGY8uUUhnfanMKiH/ddpQ2Wnkap9Da+mu4s+ANz 9o2IOsTkg5zPnh4zen7K2adK3W/Wte4Or3lousanFKSTaesD60PezgWqu8yD4/WuKL eKeYwjzW7JRVRivG4EhJ08BFyxLvcwfWDxNmHNNTnDtj7hV4JrheI6HyEkCgh/BBaX C3iJ6fu3/NeIHP2tyUIICtDmCU+48R/TVJP7OCYgJ7V8FGmV3375W0SVUZGtrNr5h+ 5CAf+WC77jKUoyDz62cb0ESVDtkUCwB9ztxUHzS2al3TXZHNuOPyLZ0zI5pyax8ufj G6gORnanW/SDA== From: Dmitry Osipenko To: David Airlie , Gerd Hoffmann , Gurchetan Singh , Chia-I Wu , Daniel Vetter , Rob Clark , =?UTF-8?q?Marek=20Ol=C5=A1=C3=A1k?= , Pierre-Eric Pelloux-Prayer , Emil Velikov Cc: dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org, kernel@collabora.com, virtualization@lists.linux-foundation.org Subject: [PATCH v6 3/3] drm/virtio: Support sync objects Date: Sun, 16 Apr 2023 14:52:37 +0300 Message-Id: <20230416115237.798604-4-dmitry.osipenko@collabora.com> X-Mailer: git-send-email 2.39.2 In-Reply-To: <20230416115237.798604-1-dmitry.osipenko@collabora.com> References: <20230416115237.798604-1-dmitry.osipenko@collabora.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Type: text/plain; charset="utf-8" Add sync object DRM UAPI support to VirtIO-GPU driver. Sync objects support is needed by native context VirtIO-GPU Mesa drivers, it also will be used by Venus and Virgl contexts. Reviewed-by; Emil Velikov Signed-off-by: Dmitry Osipenko Reviewed-by; Emil Velikov Tested-by: Pierre-Eric Pelloux-Prayer --- drivers/gpu/drm/virtio/virtgpu_drv.c | 3 +- drivers/gpu/drm/virtio/virtgpu_submit.c | 219 ++++++++++++++++++++++++ include/uapi/drm/virtgpu_drm.h | 16 +- 3 files changed, 236 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.c b/drivers/gpu/drm/virtio/= virtgpu_drv.c index add075681e18..a22155577152 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.c +++ b/drivers/gpu/drm/virtio/virtgpu_drv.c @@ -176,7 +176,8 @@ static const struct drm_driver driver =3D { * If KMS is disabled DRIVER_MODESET and DRIVER_ATOMIC are masked * out via drm_device::driver_features: */ - .driver_features =3D DRIVER_MODESET | DRIVER_GEM | DRIVER_RENDER | DRIVER= _ATOMIC, + .driver_features =3D DRIVER_MODESET | DRIVER_GEM | DRIVER_RENDER | DRIVER= _ATOMIC | + DRIVER_SYNCOBJ | DRIVER_SYNCOBJ_TIMELINE, .open =3D virtio_gpu_driver_open, .postclose =3D virtio_gpu_driver_postclose, =20 diff --git a/drivers/gpu/drm/virtio/virtgpu_submit.c b/drivers/gpu/drm/virt= io/virtgpu_submit.c index cf3c04b16a7a..5a0f2526c1a0 100644 --- a/drivers/gpu/drm/virtio/virtgpu_submit.c +++ b/drivers/gpu/drm/virtio/virtgpu_submit.c @@ -14,11 +14,24 @@ #include =20 #include +#include #include =20 #include "virtgpu_drv.h" =20 +struct virtio_gpu_submit_post_dep { + struct drm_syncobj *syncobj; + struct dma_fence_chain *chain; + u64 point; +}; + struct virtio_gpu_submit { + struct virtio_gpu_submit_post_dep *post_deps; + unsigned int num_out_syncobjs; + + struct drm_syncobj **in_syncobjs; + unsigned int num_in_syncobjs; + struct virtio_gpu_object_array *buflist; struct drm_virtgpu_execbuffer *exbuf; struct virtio_gpu_fence *out_fence; @@ -59,6 +72,199 @@ static int virtio_gpu_dma_fence_wait(struct virtio_gpu_= submit *submit, return 0; } =20 +static void virtio_gpu_free_syncobjs(struct drm_syncobj **syncobjs, + u32 nr_syncobjs) +{ + u32 i =3D nr_syncobjs; + + while (i--) { + if (syncobjs[i]) + drm_syncobj_put(syncobjs[i]); + } + + kvfree(syncobjs); +} + +static int +virtio_gpu_parse_deps(struct virtio_gpu_submit *submit) +{ + struct drm_virtgpu_execbuffer *exbuf =3D submit->exbuf; + struct drm_virtgpu_execbuffer_syncobj syncobj_desc; + size_t syncobj_stride =3D exbuf->syncobj_stride; + u32 num_in_syncobjs =3D exbuf->num_in_syncobjs; + struct drm_syncobj **syncobjs; + int ret =3D 0, i; + + if (!num_in_syncobjs) + return 0; + + /* + * kvalloc at first tries to allocate memory using kmalloc and + * falls back to vmalloc only on failure. It also uses GFP_NOWARN + * internally for allocations larger than a page size, preventing + * storm of KMSG warnings. + */ + syncobjs =3D kvcalloc(num_in_syncobjs, sizeof(*syncobjs), GFP_KERNEL); + if (!syncobjs) + return -ENOMEM; + + for (i =3D 0; i < num_in_syncobjs; i++) { + u64 address =3D exbuf->in_syncobjs + i * syncobj_stride; + struct dma_fence *fence; + + memset(&syncobj_desc, 0, sizeof(syncobj_desc)); + + if (copy_from_user(&syncobj_desc, + u64_to_user_ptr(address), + min(syncobj_stride, sizeof(syncobj_desc)))) { + ret =3D -EFAULT; + break; + } + + if (syncobj_desc.flags & ~VIRTGPU_EXECBUF_SYNCOBJ_FLAGS) { + ret =3D -EINVAL; + break; + } + + ret =3D drm_syncobj_find_fence(submit->file, syncobj_desc.handle, + syncobj_desc.point, 0, &fence); + if (ret) + break; + + ret =3D virtio_gpu_dma_fence_wait(submit, fence); + + dma_fence_put(fence); + if (ret) + break; + + if (syncobj_desc.flags & VIRTGPU_EXECBUF_SYNCOBJ_RESET) { + syncobjs[i] =3D drm_syncobj_find(submit->file, + syncobj_desc.handle); + if (!syncobjs[i]) { + ret =3D -EINVAL; + break; + } + } + } + + if (ret) { + virtio_gpu_free_syncobjs(syncobjs, i); + return ret; + } + + submit->num_in_syncobjs =3D num_in_syncobjs; + submit->in_syncobjs =3D syncobjs; + + return ret; +} + +static void virtio_gpu_reset_syncobjs(struct drm_syncobj **syncobjs, + u32 nr_syncobjs) +{ + u32 i; + + for (i =3D 0; i < nr_syncobjs; i++) { + if (syncobjs[i]) + drm_syncobj_replace_fence(syncobjs[i], NULL); + } +} + +static void +virtio_gpu_free_post_deps(struct virtio_gpu_submit_post_dep *post_deps, + u32 nr_syncobjs) +{ + u32 i =3D nr_syncobjs; + + while (i--) { + kfree(post_deps[i].chain); + drm_syncobj_put(post_deps[i].syncobj); + } + + kvfree(post_deps); +} + +static int virtio_gpu_parse_post_deps(struct virtio_gpu_submit *submit) +{ + struct drm_virtgpu_execbuffer *exbuf =3D submit->exbuf; + struct drm_virtgpu_execbuffer_syncobj syncobj_desc; + struct virtio_gpu_submit_post_dep *post_deps; + u32 num_out_syncobjs =3D exbuf->num_out_syncobjs; + size_t syncobj_stride =3D exbuf->syncobj_stride; + int ret =3D 0, i; + + if (!num_out_syncobjs) + return 0; + + post_deps =3D kvcalloc(num_out_syncobjs, sizeof(*post_deps), GFP_KERNEL); + if (!post_deps) + return -ENOMEM; + + for (i =3D 0; i < num_out_syncobjs; i++) { + u64 address =3D exbuf->out_syncobjs + i * syncobj_stride; + + memset(&syncobj_desc, 0, sizeof(syncobj_desc)); + + if (copy_from_user(&syncobj_desc, + u64_to_user_ptr(address), + min(syncobj_stride, sizeof(syncobj_desc)))) { + ret =3D -EFAULT; + break; + } + + post_deps[i].point =3D syncobj_desc.point; + + if (syncobj_desc.flags) { + ret =3D -EINVAL; + break; + } + + if (syncobj_desc.point) { + post_deps[i].chain =3D dma_fence_chain_alloc(); + if (!post_deps[i].chain) { + ret =3D -ENOMEM; + break; + } + } + + post_deps[i].syncobj =3D drm_syncobj_find(submit->file, + syncobj_desc.handle); + if (!post_deps[i].syncobj) { + kfree(post_deps[i].chain); + ret =3D -EINVAL; + break; + } + } + + if (ret) { + virtio_gpu_free_post_deps(post_deps, i); + return ret; + } + + submit->num_out_syncobjs =3D num_out_syncobjs; + submit->post_deps =3D post_deps; + + return 0; +} + +static void +virtio_gpu_process_post_deps(struct virtio_gpu_submit *submit) +{ + struct virtio_gpu_submit_post_dep *post_deps =3D submit->post_deps; + struct dma_fence *fence =3D &submit->out_fence->f; + u32 i; + + for (i =3D 0; i < submit->num_out_syncobjs; i++) { + if (post_deps[i].chain) { + drm_syncobj_add_point(post_deps[i].syncobj, + post_deps[i].chain, + fence, post_deps[i].point); + post_deps[i].chain =3D NULL; + } else { + drm_syncobj_replace_fence(post_deps[i].syncobj, fence); + } + } +} + static int virtio_gpu_fence_event_create(struct drm_device *dev, struct drm_file *file, struct virtio_gpu_fence *fence, @@ -122,6 +328,10 @@ static int virtio_gpu_init_submit_buflist(struct virti= o_gpu_submit *submit) =20 static void virtio_gpu_cleanup_submit(struct virtio_gpu_submit *submit) { + virtio_gpu_reset_syncobjs(submit->in_syncobjs, submit->num_in_syncobjs); + virtio_gpu_free_syncobjs(submit->in_syncobjs, submit->num_in_syncobjs); + virtio_gpu_free_post_deps(submit->post_deps, submit->num_out_syncobjs); + if (!IS_ERR(submit->buf)) kvfree(submit->buf); =20 @@ -288,6 +498,14 @@ int virtio_gpu_execbuffer_ioctl(struct drm_device *dev= , void *data, * optimize the path by proceeding directly to the submission * to virtio after the waits. */ + ret =3D virtio_gpu_parse_post_deps(&submit); + if (ret) + goto cleanup; + + ret =3D virtio_gpu_parse_deps(&submit); + if (ret) + goto cleanup; + ret =3D virtio_gpu_wait_in_fence(&submit); if (ret) goto cleanup; @@ -303,6 +521,7 @@ int virtio_gpu_execbuffer_ioctl(struct drm_device *dev,= void *data, * the job submission path. */ virtio_gpu_install_out_fence_fd(&submit); + virtio_gpu_process_post_deps(&submit); virtio_gpu_complete_submit(&submit); cleanup: virtio_gpu_cleanup_submit(&submit); diff --git a/include/uapi/drm/virtgpu_drm.h b/include/uapi/drm/virtgpu_drm.h index 7b158fcb02b4..b1d0e56565bc 100644 --- a/include/uapi/drm/virtgpu_drm.h +++ b/include/uapi/drm/virtgpu_drm.h @@ -64,6 +64,16 @@ struct drm_virtgpu_map { __u32 pad; }; =20 +#define VIRTGPU_EXECBUF_SYNCOBJ_RESET 0x01 +#define VIRTGPU_EXECBUF_SYNCOBJ_FLAGS ( \ + VIRTGPU_EXECBUF_SYNCOBJ_RESET | \ + 0) +struct drm_virtgpu_execbuffer_syncobj { + __u32 handle; + __u32 flags; + __u64 point; +}; + /* fence_fd is modified on success if VIRTGPU_EXECBUF_FENCE_FD_OUT flag is= set. */ struct drm_virtgpu_execbuffer { __u32 flags; @@ -73,7 +83,11 @@ struct drm_virtgpu_execbuffer { __u32 num_bo_handles; __s32 fence_fd; /* in/out fence fd (see VIRTGPU_EXECBUF_FENCE_FD_IN/OUT) = */ __u32 ring_idx; /* command ring index (see VIRTGPU_EXECBUF_RING_IDX) */ - __u32 pad; + __u32 syncobj_stride; /* size of @drm_virtgpu_execbuffer_syncobj */ + __u32 num_in_syncobjs; + __u32 num_out_syncobjs; + __u64 in_syncobjs; + __u64 out_syncobjs; }; =20 #define VIRTGPU_PARAM_3D_FEATURES 1 /* do we have 3D features in the hw */ --=20 2.39.2