From: Ilija Hadzic <ihadzic@research.bell-labs.com> To: dri-devel@lists.freedesktop.org Subject: [PATCH 19/19] drm: hold mutex in critical sections of render-node code Date: Thu, 12 Apr 2012 14:19:44 -0400 [thread overview] Message-ID: <1334254784-3200-20-git-send-email-ihadzic@research.bell-labs.com> (raw) In-Reply-To: <1334254784-3200-1-git-send-email-ihadzic@research.bell-labs.com> Critical sections are parts of the code where we claim or release resources (we don't want two render-node create or remove ioctl called in the context of different processes to claim part of requested resources because of the race). Another critical section is manipulating the render node list. We can use dev->mode_config.mutex for both. v2: - Use dev->mode_config.mutex instead of dev->struct_mutex because we are also racing against drm_mode_getresources() and drm_mode_getplane_res() functions. Signed-off-by: Ilija Hadzic <ihadzic@research.bell-labs.com> --- drivers/gpu/drm/drm_stub.c | 15 ++++++++++++++- 1 files changed, 14 insertions(+), 1 deletions(-) diff --git a/drivers/gpu/drm/drm_stub.c b/drivers/gpu/drm/drm_stub.c index 13ff4c8..a5fd905 100644 --- a/drivers/gpu/drm/drm_stub.c +++ b/drivers/gpu/drm/drm_stub.c @@ -486,6 +486,7 @@ static void drm_release_render_node_resources(struct drm_device *dev, int *render_node_owner; int s, e, i, j; + /* no lock, assume we were called with mode_config mutex grabbed */ for (e = 0, j = 0; j < DRM_RN_NUM_EXP_TYPES; j++) { s = e; e += resource_count[j]; @@ -509,6 +510,7 @@ static int drm_claim_render_node_resources(struct drm_device *dev, int s, e, i, j; int ret = 0; + mutex_lock(&dev->mode_config.mutex); for (e = 0, j = 0; j < DRM_RN_NUM_EXP_TYPES; j++) { s = e; e += resource_count[j]; @@ -529,10 +531,12 @@ static int drm_claim_render_node_resources(struct drm_device *dev, *render_node_owner = minor; } } + mutex_unlock(&dev->mode_config.mutex); return ret; out_release: drm_release_render_node_resources(dev, id_list, resource_count, minor); + mutex_unlock(&dev->mode_config.mutex); return ret; } @@ -553,7 +557,9 @@ int drm_create_render_node(struct drm_device *dev, struct drm_minor **minor_p) } render_node->minor = minor; *minor_p = minor; + mutex_lock(&dev->mode_config.mutex); list_add_tail(&render_node->list, &dev->render_node_list); + mutex_unlock(&dev->mode_config.mutex); return 0; } @@ -561,13 +567,16 @@ int drm_destroy_render_node(struct drm_device *dev, int index) { struct drm_render_node *node, *tmp; + mutex_lock(&dev->mode_config.mutex); list_for_each_entry_safe(node, tmp, &dev->render_node_list, list) { if (node->minor->index == index) { struct drm_mode_group *group; int resource_count[DRM_RN_NUM_EXP_TYPES]; - if (node->minor->open_count) + if (node->minor->open_count) { + mutex_unlock(&dev->mode_config.mutex); return -EBUSY; + } group = &node->minor->mode_group; list_del(&node->list); resource_count[0] = group->num_crtcs; @@ -577,12 +586,14 @@ int drm_destroy_render_node(struct drm_device *dev, int index) drm_release_render_node_resources(dev, group->id_list, resource_count, node->minor->index); + mutex_unlock(&dev->mode_config.mutex); drm_put_minor(&node->minor); drm_mode_group_fini(group); kfree(node); return 0; } } + mutex_unlock(&dev->mode_config.mutex); return -ENODEV; } @@ -590,6 +601,7 @@ void drm_destroy_all_render_nodes(struct drm_device *dev) { struct drm_render_node *node, *tmp; + mutex_lock(&dev->mode_config.mutex); list_for_each_entry_safe(node, tmp, &dev->render_node_list, list) { struct drm_mode_group *group; int resource_count[DRM_RN_NUM_EXP_TYPES]; @@ -607,6 +619,7 @@ void drm_destroy_all_render_nodes(struct drm_device *dev) drm_mode_group_fini(group); kfree(node); } + mutex_unlock(&dev->mode_config.mutex); } /** -- 1.7.8.5
next prev parent reply other threads:[~2012-04-12 18:22 UTC|newest] Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top 2012-04-12 18:19 [RFC v2] Revive the work on render-nodes branch Ilija Hadzic 2012-04-12 18:19 ` [PATCH 01/19] drm: simplify dereferencing of node type Ilija Hadzic 2012-04-12 18:19 ` [PATCH 02/19] drm: track planes in drm_mode_group structure Ilija Hadzic 2012-04-12 18:19 ` [PATCH 03/19] drm: use drm_mode_group in drm_mode_getplane_res Ilija Hadzic 2012-04-12 18:19 ` [PATCH 04/19] drm: do not push inode down into drm_open_helper Ilija Hadzic 2012-04-12 18:19 ` [PATCH 05/19] drm: move dev_mapping to the minor node Ilija Hadzic 2012-04-20 10:04 ` Dave Airlie 2012-04-30 14:48 ` Ilija Hadzic 2012-04-30 16:39 ` Dave Airlie 2012-04-30 16:52 ` Ilija Hadzic 2012-04-30 17:53 ` Dave Airlie 2012-04-30 18:04 ` Dave Airlie 2012-05-15 20:48 ` Ilija Hadzic 2012-04-12 18:19 ` [PATCH 06/19] drm: add support for render nodes Ilija Hadzic 2012-04-12 18:19 ` [PATCH 07/19] drm: initial multiple nodes ioctl work Ilija Hadzic 2012-04-20 10:15 ` Dave Airlie 2012-04-12 18:19 ` [PATCH 08/19] drm: separate render node descriptor from minor Ilija Hadzic 2012-04-12 18:19 ` [PATCH 09/19] drm: cleanup render node ioctls Ilija Hadzic 2012-04-12 18:19 ` [PATCH 10/19] drm: only allow render node ioctls through control node Ilija Hadzic 2012-04-12 18:19 ` [PATCH 11/19] drm: do not remove a render node in use Ilija Hadzic 2012-04-12 18:19 ` [PATCH 12/19] drm: allocate correct id_list size for a render node Ilija Hadzic 2012-04-12 18:19 ` [PATCH 13/19] drm: add drm_mode_group_fini function Ilija Hadzic 2012-04-12 18:19 ` [PATCH 14/19] drm: properly free id_list when a render node is removed Ilija Hadzic 2012-04-12 18:19 ` [PATCH 15/19] drm: call drm_mode_group_fini on primary node Ilija Hadzic 2012-04-12 18:19 ` [PATCH 16/19] drm: more elaborate check for resource count Ilija Hadzic 2012-04-12 18:19 ` [PATCH 17/19] drm: validate id list when creating a render node Ilija Hadzic 2012-04-12 18:19 ` [PATCH 18/19] drm: keep track of which node holds which resource Ilija Hadzic 2012-04-12 18:19 ` Ilija Hadzic [this message] 2012-04-12 18:55 ` [RFC v2] Revive the work on render-nodes branch Ville Syrjälä 2012-04-12 19:09 ` Ilija Hadzic 2012-04-20 10:20 ` Dave Airlie 2012-04-20 13:46 ` Daniel Vetter 2012-04-30 15:16 ` Ilija Hadzic 2012-04-30 19:01 ` Kristian Høgsberg
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=1334254784-3200-20-git-send-email-ihadzic@research.bell-labs.com \ --to=ihadzic@research.bell-labs.com \ --cc=dri-devel@lists.freedesktop.org \ --subject='Re: [PATCH 19/19] drm: hold mutex in critical sections of render-node code' \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: link
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.