September 2024 - Linux-stable-mirror

[PATCH 5.10] drm/vmwgfx: Remove rcu locks from user resources

by Huang Xiaojia

[ Upstream commit a309c7194e8a2f8bd4539b9449917913f6c2cd50 ] User resource lookups used rcu to avoid two extra atomics. Unfortunately the rcu paths were buggy and it was easy to make the driver crash by submitting command buffers from two different threads. Because the lookups never show up in performance profiles replace them with a regular spin lock which fixes the races in accesses to those shared resources. Fixes kernel oops'es in IGT's vmwgfx execution_buffer stress test and seen crashes with apps using shared resources. Fixes: e14c02e6b699 ("drm/vmwgfx: Look up objects without taking a reference") Signed-off-by: Huang Xiaojia <huangxiaojia2(a)huawei.com> --- drivers/gpu/drm/drm_hashtab.c | 5 +- drivers/gpu/drm/vmwgfx/ttm_object.c | 55 +++---- drivers/gpu/drm/vmwgfx/ttm_object.h | 14 -- drivers/gpu/drm/vmwgfx/vmwgfx_bo.c | 42 ------ drivers/gpu/drm/vmwgfx/vmwgfx_drv.h | 27 +--- drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c | 176 +++++++++++------------ drivers/gpu/drm/vmwgfx/vmwgfx_resource.c | 35 ----- include/drm/drm_hashtab.h | 2 + 8 files changed, 105 insertions(+), 251 deletions(-) diff --git a/drivers/gpu/drm/drm_hashtab.c b/drivers/gpu/drm/drm_hashtab.c index c50fa6f0709f..a61642835542 100644 --- a/drivers/gpu/drm/drm_hashtab.c +++ b/drivers/gpu/drm/drm_hashtab.c @@ -74,8 +74,8 @@ void drm_ht_verbose_list(struct drm_open_hash *ht, unsigned long key) DRM_DEBUG("count %d, key: 0x%08lx\n", count++, entry->key); } -static struct hlist_node *drm_ht_find_key(struct drm_open_hash *ht, - unsigned long key) +struct hlist_node *drm_ht_find_key(struct drm_open_hash *ht, + unsigned long key) { struct drm_hash_item *entry; struct hlist_head *h_list; @@ -91,6 +91,7 @@ static struct hlist_node *drm_ht_find_key(struct drm_open_hash *ht, } return NULL; } +EXPORT_SYMBOL(drm_ht_find_key); static struct hlist_node *drm_ht_find_key_rcu(struct drm_open_hash *ht, unsigned long key) diff --git a/drivers/gpu/drm/vmwgfx/ttm_object.c b/drivers/gpu/drm/vmwgfx/ttm_object.c index 16077785ad47..72a4aaeba0c7 100644 --- a/drivers/gpu/drm/vmwgfx/ttm_object.c +++ b/drivers/gpu/drm/vmwgfx/ttm_object.c @@ -137,6 +137,20 @@ ttm_object_file_ref(struct ttm_object_file *tfile) return tfile; } +static int ttm_tfile_find_ref(struct drm_open_hash *ht, + uint32_t key, + struct drm_hash_item **item) +{ + struct hlist_node *h_node; + + h_node = drm_ht_find_key(ht, key); + if (!h_node) + return -EINVAL; + + *item = hlist_entry(h_node, struct drm_hash_item, head); + return 0; +} + static void ttm_object_file_destroy(struct kref *kref) { struct ttm_object_file *tfile = @@ -225,41 +239,6 @@ void ttm_base_object_unref(struct ttm_base_object **p_base) kref_put(&base->refcount, ttm_release_base); } -/** - * ttm_base_object_noref_lookup - look up a base object without reference - * @tfile: The struct ttm_object_file the object is registered with. - * @key: The object handle. - * - * This function looks up a ttm base object and returns a pointer to it - * without refcounting the pointer. The returned pointer is only valid - * until ttm_base_object_noref_release() is called, and the object - * pointed to by the returned pointer may be doomed. Any persistent usage - * of the object requires a refcount to be taken using kref_get_unless_zero(). - * Iff this function returns successfully it needs to be paired with - * ttm_base_object_noref_release() and no sleeping- or scheduling functions - * may be called inbetween these function callse. - * - * Return: A pointer to the object if successful or NULL otherwise. - */ -struct ttm_base_object * -ttm_base_object_noref_lookup(struct ttm_object_file *tfile, uint32_t key) -{ - struct drm_hash_item *hash; - struct drm_open_hash *ht = &tfile->ref_hash[TTM_REF_USAGE]; - int ret; - - rcu_read_lock(); - ret = drm_ht_find_item_rcu(ht, key, &hash); - if (ret) { - rcu_read_unlock(); - return NULL; - } - - __release(RCU); - return drm_hash_entry(hash, struct ttm_ref_object, hash)->obj; -} -EXPORT_SYMBOL(ttm_base_object_noref_lookup); - struct ttm_base_object *ttm_base_object_lookup(struct ttm_object_file *tfile, uint32_t key) { @@ -268,15 +247,15 @@ struct ttm_base_object *ttm_base_object_lookup(struct ttm_object_file *tfile, struct drm_open_hash *ht = &tfile->ref_hash[TTM_REF_USAGE]; int ret; - rcu_read_lock(); - ret = drm_ht_find_item_rcu(ht, key, &hash); + spin_lock(&tfile->lock); + ret = ttm_tfile_find_ref(ht, key, &hash); if (likely(ret == 0)) { base = drm_hash_entry(hash, struct ttm_ref_object, hash)->obj; if (!kref_get_unless_zero(&base->refcount)) base = NULL; } - rcu_read_unlock(); + spin_unlock(&tfile->lock); return base; } diff --git a/drivers/gpu/drm/vmwgfx/ttm_object.h b/drivers/gpu/drm/vmwgfx/ttm_object.h index ede26df87c93..d16334e5d509 100644 --- a/drivers/gpu/drm/vmwgfx/ttm_object.h +++ b/drivers/gpu/drm/vmwgfx/ttm_object.h @@ -359,18 +359,4 @@ extern int ttm_prime_handle_to_fd(struct ttm_object_file *tfile, */ #define TTM_OBJ_EXTRA_SIZE 128 -struct ttm_base_object * -ttm_base_object_noref_lookup(struct ttm_object_file *tfile, uint32_t key); - -/** - * ttm_base_object_noref_release - release a base object pointer looked up - * without reference - * - * Releases a base object pointer looked up with ttm_base_object_noref_lookup(). - */ -static inline void ttm_base_object_noref_release(void) -{ - __acquire(RCU); - rcu_read_unlock(); -} #endif diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c b/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c index 9a66ba254326..eb76cca4b29c 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_bo.c @@ -976,48 +976,6 @@ int vmw_user_bo_lookup(struct ttm_object_file *tfile, return 0; } -/** - * vmw_user_bo_noref_lookup - Look up a vmw user buffer object without reference - * @tfile: The TTM object file the handle is registered with. - * @handle: The user buffer object handle. - * - * This function looks up a struct vmw_user_bo and returns a pointer to the - * struct vmw_buffer_object it derives from without refcounting the pointer. - * The returned pointer is only valid until vmw_user_bo_noref_release() is - * called, and the object pointed to by the returned pointer may be doomed. - * Any persistent usage of the object requires a refcount to be taken using - * ttm_bo_reference_unless_doomed(). Iff this function returns successfully it - * needs to be paired with vmw_user_bo_noref_release() and no sleeping- - * or scheduling functions may be called inbetween these function calls. - * - * Return: A struct vmw_buffer_object pointer if successful or negative - * error pointer on failure. - */ -struct vmw_buffer_object * -vmw_user_bo_noref_lookup(struct ttm_object_file *tfile, u32 handle) -{ - struct vmw_user_buffer_object *vmw_user_bo; - struct ttm_base_object *base; - - base = ttm_base_object_noref_lookup(tfile, handle); - if (!base) { - DRM_ERROR("Invalid buffer object handle 0x%08lx.\n", - (unsigned long)handle); - return ERR_PTR(-ESRCH); - } - - if (unlikely(ttm_base_object_type(base) != ttm_buffer_type)) { - ttm_base_object_noref_release(); - DRM_ERROR("Invalid buffer object handle 0x%08lx.\n", - (unsigned long)handle); - return ERR_PTR(-EINVAL); - } - - vmw_user_bo = container_of(base, struct vmw_user_buffer_object, - prime.base); - return &vmw_user_bo->vbo; -} - /** * vmw_user_bo_reference - Open a handle to a vmw user buffer object. * diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h index fa285c20d6da..0f7a0f9169f8 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_drv.h @@ -768,12 +768,7 @@ extern int vmw_user_resource_lookup_handle( uint32_t handle, const struct vmw_user_resource_conv *converter, struct vmw_resource **p_res); -extern struct vmw_resource * -vmw_user_resource_noref_lookup_handle(struct vmw_private *dev_priv, - struct ttm_object_file *tfile, - uint32_t handle, - const struct vmw_user_resource_conv * - converter); + extern int vmw_stream_claim_ioctl(struct drm_device *dev, void *data, struct drm_file *file_priv); extern int vmw_stream_unref_ioctl(struct drm_device *dev, void *data, @@ -811,15 +806,6 @@ static inline bool vmw_resource_mob_attached(const struct vmw_resource *res) return !RB_EMPTY_NODE(&res->mob_node); } -/** - * vmw_user_resource_noref_release - release a user resource pointer looked up - * without reference - */ -static inline void vmw_user_resource_noref_release(void) -{ - ttm_base_object_noref_release(); -} - /** * Buffer object helper functions - vmwgfx_bo.c */ @@ -880,17 +866,6 @@ extern void vmw_bo_unmap(struct vmw_buffer_object *vbo); extern void vmw_bo_move_notify(struct ttm_buffer_object *bo, struct ttm_resource *mem); extern void vmw_bo_swap_notify(struct ttm_buffer_object *bo); -extern struct vmw_buffer_object * -vmw_user_bo_noref_lookup(struct ttm_object_file *tfile, u32 handle); - -/** - * vmw_user_bo_noref_release - release a buffer object pointer looked up - * without reference - */ -static inline void vmw_user_bo_noref_release(void) -{ - ttm_base_object_noref_release(); -} /** * vmw_bo_adjust_prio - Adjust the buffer object eviction priority diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c index 616f6cb62278..c1aab5ee2a35 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_execbuf.c @@ -285,20 +285,26 @@ static void vmw_execbuf_rcache_update(struct vmw_res_cache_entry *rcache, rcache->valid_handle = 0; } +enum vmw_val_add_flags { + vmw_val_add_flag_none = 0, + vmw_val_add_flag_noctx = 1 << 0, +}; + /** - * vmw_execbuf_res_noref_val_add - Add a resource described by an unreferenced - * rcu-protected pointer to the validation list. + * vmw_execbuf_res_val_add - Add a resource to the validation list. * * @sw_context: Pointer to the software context. * @res: Unreferenced rcu-protected pointer to the resource. * @dirty: Whether to change dirty status. + * @flags: specifies whether to use the context or not * * Returns: 0 on success. Negative error code on failure. Typical error codes * are %-EINVAL on inconsistency and %-ESRCH if the resource was doomed. */ -static int vmw_execbuf_res_noref_val_add(struct vmw_sw_context *sw_context, - struct vmw_resource *res, - u32 dirty) +static int vmw_execbuf_res_val_add(struct vmw_sw_context *sw_context, + struct vmw_resource *res, + u32 dirty, + u32 flags) { struct vmw_private *dev_priv = res->dev_priv; int ret; @@ -313,24 +319,30 @@ static int vmw_execbuf_res_noref_val_add(struct vmw_sw_context *sw_context, if (dirty) vmw_validation_res_set_dirty(sw_context->ctx, rcache->private, dirty); - vmw_user_resource_noref_release(); return 0; } - priv_size = vmw_execbuf_res_size(dev_priv, res_type); - ret = vmw_validation_add_resource(sw_context->ctx, res, priv_size, - dirty, (void **)&ctx_info, - &first_usage); - vmw_user_resource_noref_release(); - if (ret) - return ret; + if ((flags & vmw_val_add_flag_noctx) != 0) { + ret = vmw_validation_add_resource(sw_context->ctx, res, 0, dirty, + (void **)&ctx_info, NULL); + if (ret) + return ret; - if (priv_size && first_usage) { - ret = vmw_cmd_ctx_first_setup(dev_priv, sw_context, res, - ctx_info); - if (ret) { - VMW_DEBUG_USER("Failed first usage context setup.\n"); + } else { + priv_size = vmw_execbuf_res_size(dev_priv, res_type); + ret = vmw_validation_add_resource(sw_context->ctx, res, priv_size, + dirty, (void **)&ctx_info, + &first_usage); + if (ret) return ret; + + if (priv_size && first_usage) { + ret = vmw_cmd_ctx_first_setup(dev_priv, sw_context, res, + ctx_info); + if (ret) { + VMW_DEBUG_USER("Failed first usage context setup.\n"); + return ret; + } } } @@ -338,43 +350,6 @@ static int vmw_execbuf_res_noref_val_add(struct vmw_sw_context *sw_context, return 0; } -/** - * vmw_execbuf_res_noctx_val_add - Add a non-context resource to the resource - * validation list if it's not already on it - * - * @sw_context: Pointer to the software context. - * @res: Pointer to the resource. - * @dirty: Whether to change dirty status. - * - * Returns: Zero on success. Negative error code on failure. - */ -static int vmw_execbuf_res_noctx_val_add(struct vmw_sw_context *sw_context, - struct vmw_resource *res, - u32 dirty) -{ - struct vmw_res_cache_entry *rcache; - enum vmw_res_type res_type = vmw_res_type(res); - void *ptr; - int ret; - - rcache = &sw_context->res_cache[res_type]; - if (likely(rcache->valid && rcache->res == res)) { - if (dirty) - vmw_validation_res_set_dirty(sw_context->ctx, - rcache->private, dirty); - return 0; - } - - ret = vmw_validation_add_resource(sw_context->ctx, res, 0, dirty, - &ptr, NULL); - if (ret) - return ret; - - vmw_execbuf_rcache_update(rcache, res, ptr); - - return 0; -} - /** * vmw_view_res_val_add - Add a view and the surface it's pointing to to the * validation list @@ -393,13 +368,13 @@ static int vmw_view_res_val_add(struct vmw_sw_context *sw_context, * First add the resource the view is pointing to, otherwise it may be * swapped out when the view is validated. */ - ret = vmw_execbuf_res_noctx_val_add(sw_context, vmw_view_srf(view), - vmw_view_dirtying(view)); + ret = vmw_execbuf_res_val_add(sw_context, vmw_view_srf(view), + vmw_view_dirtying(view), vmw_val_add_flag_noctx); if (ret) return ret; - return vmw_execbuf_res_noctx_val_add(sw_context, view, - VMW_RES_DIRTY_NONE); + return vmw_execbuf_res_val_add(sw_context, view, VMW_RES_DIRTY_NONE, + vmw_val_add_flag_noctx); } /** @@ -470,8 +445,9 @@ static int vmw_resource_context_res_add(struct vmw_private *dev_priv, if (IS_ERR_OR_NULL(res)) continue; - ret = vmw_execbuf_res_noctx_val_add(sw_context, res, - VMW_RES_DIRTY_SET); + ret = vmw_execbuf_res_val_add(sw_context, res, + VMW_RES_DIRTY_SET, + vmw_val_add_flag_noctx); if (unlikely(ret != 0)) return ret; } @@ -485,9 +461,9 @@ static int vmw_resource_context_res_add(struct vmw_private *dev_priv, if (vmw_res_type(entry->res) == vmw_res_view) ret = vmw_view_res_val_add(sw_context, entry->res); else - ret = vmw_execbuf_res_noctx_val_add - (sw_context, entry->res, - vmw_binding_dirtying(entry->bt)); + ret = vmw_execbuf_res_val_add(sw_context, entry->res, + vmw_binding_dirtying(entry->bt), + vmw_val_add_flag_noctx); if (unlikely(ret != 0)) break; } @@ -653,7 +629,8 @@ vmw_cmd_res_check(struct vmw_private *dev_priv, { struct vmw_res_cache_entry *rcache = &sw_context->res_cache[res_type]; struct vmw_resource *res; - int ret; + int ret = 0; + bool needs_unref = false; if (p_res) *p_res = NULL; @@ -678,17 +655,18 @@ vmw_cmd_res_check(struct vmw_private *dev_priv, if (ret) return ret; - res = vmw_user_resource_noref_lookup_handle - (dev_priv, sw_context->fp->tfile, *id_loc, converter); - if (IS_ERR(res)) { + ret = vmw_user_resource_lookup_handle + (dev_priv, sw_context->fp->tfile, *id_loc, converter, &res); + if (ret != 0) { VMW_DEBUG_USER("Could not find/use resource 0x%08x.\n", (unsigned int) *id_loc); - return PTR_ERR(res); + return ret; } + needs_unref = true; - ret = vmw_execbuf_res_noref_val_add(sw_context, res, dirty); + ret = vmw_execbuf_res_val_add(sw_context, res, dirty, vmw_val_add_flag_none); if (unlikely(ret != 0)) - return ret; + goto res_check_done; if (rcache->valid && rcache->res == res) { rcache->valid_handle = true; @@ -703,7 +681,11 @@ vmw_cmd_res_check(struct vmw_private *dev_priv, if (p_res) *p_res = res; - return 0; +res_check_done: + if (needs_unref) + vmw_resource_unreference(&res); + + return ret; } /** @@ -1166,14 +1148,14 @@ static int vmw_translate_mob_ptr(struct vmw_private *dev_priv, int ret; vmw_validation_preload_bo(sw_context->ctx); - vmw_bo = vmw_user_bo_noref_lookup(sw_context->fp->tfile, handle); - if (IS_ERR(vmw_bo)) { + ret = vmw_user_bo_lookup(sw_context->fp->tfile, handle, &vmw_bo, NULL); + if (ret != 0) { VMW_DEBUG_USER("Could not find or use MOB buffer.\n"); return PTR_ERR(vmw_bo); } ret = vmw_validation_add_bo(sw_context->ctx, vmw_bo, true, false); - vmw_user_bo_noref_release(); + ttm_bo_put(&vmw_bo->base); if (unlikely(ret != 0)) return ret; @@ -1221,14 +1203,14 @@ static int vmw_translate_guest_ptr(struct vmw_private *dev_priv, int ret; vmw_validation_preload_bo(sw_context->ctx); - vmw_bo = vmw_user_bo_noref_lookup(sw_context->fp->tfile, handle); - if (IS_ERR(vmw_bo)) { + ret = vmw_user_bo_lookup(sw_context->fp->tfile, handle, &vmw_bo, NULL); + if (ret != 0) { VMW_DEBUG_USER("Could not find or use GMR region.\n"); return PTR_ERR(vmw_bo); } ret = vmw_validation_add_bo(sw_context->ctx, vmw_bo, false, false); - vmw_user_bo_noref_release(); + ttm_bo_put(&vmw_bo->base); if (unlikely(ret != 0)) return ret; @@ -2024,8 +2006,9 @@ static int vmw_cmd_set_shader(struct vmw_private *dev_priv, res = vmw_shader_lookup(vmw_context_res_man(ctx), cmd->body.shid, cmd->body.type); if (!IS_ERR(res)) { - ret = vmw_execbuf_res_noctx_val_add(sw_context, res, - VMW_RES_DIRTY_NONE); + ret = vmw_execbuf_res_val_add(sw_context, res, + VMW_RES_DIRTY_NONE, + vmw_val_add_flag_noctx); if (unlikely(ret != 0)) return ret; @@ -2227,8 +2210,9 @@ static int vmw_cmd_dx_set_shader(struct vmw_private *dev_priv, return PTR_ERR(res); } - ret = vmw_execbuf_res_noctx_val_add(sw_context, res, - VMW_RES_DIRTY_NONE); + ret = vmw_execbuf_res_val_add(sw_context, res, + VMW_RES_DIRTY_NONE, + vmw_val_add_flag_noctx); if (ret) return ret; } @@ -2735,8 +2719,8 @@ static int vmw_cmd_dx_bind_shader(struct vmw_private *dev_priv, return PTR_ERR(res); } - ret = vmw_execbuf_res_noctx_val_add(sw_context, res, - VMW_RES_DIRTY_NONE); + ret = vmw_execbuf_res_val_add(sw_context, res, VMW_RES_DIRTY_NONE, + vmw_val_add_flag_noctx); if (ret) { VMW_DEBUG_USER("Error creating resource validation node.\n"); return ret; @@ -3058,8 +3042,8 @@ static int vmw_cmd_dx_bind_streamoutput(struct vmw_private *dev_priv, vmw_dx_streamoutput_set_size(res, cmd->body.sizeInBytes); - ret = vmw_execbuf_res_noctx_val_add(sw_context, res, - VMW_RES_DIRTY_NONE); + ret = vmw_execbuf_res_val_add(sw_context, res, VMW_RES_DIRTY_NONE, + vmw_val_add_flag_noctx); if (ret) { DRM_ERROR("Error creating resource validation node.\n"); return ret; @@ -3108,8 +3092,8 @@ static int vmw_cmd_dx_set_streamoutput(struct vmw_private *dev_priv, return 0; } - ret = vmw_execbuf_res_noctx_val_add(sw_context, res, - VMW_RES_DIRTY_NONE); + ret = vmw_execbuf_res_val_add(sw_context, res, VMW_RES_DIRTY_NONE, + vmw_val_add_flag_noctx); if (ret) { DRM_ERROR("Error creating resource validation node.\n"); return ret; @@ -4008,22 +3992,26 @@ static int vmw_execbuf_tie_context(struct vmw_private *dev_priv, if (ret) return ret; - res = vmw_user_resource_noref_lookup_handle + ret = vmw_user_resource_lookup_handle (dev_priv, sw_context->fp->tfile, handle, - user_context_converter); - if (IS_ERR(res)) { + user_context_converter, &res); + if (ret != 0) { VMW_DEBUG_USER("Could not find or user DX context 0x%08x.\n", (unsigned int) handle); - return PTR_ERR(res); + return ret; } - ret = vmw_execbuf_res_noref_val_add(sw_context, res, VMW_RES_DIRTY_SET); - if (unlikely(ret != 0)) + ret = vmw_execbuf_res_val_add(sw_context, res, VMW_RES_DIRTY_SET, + vmw_val_add_flag_none); + if (unlikely(ret != 0)) { + vmw_resource_unreference(&res); return ret; + } sw_context->dx_ctx_node = vmw_execbuf_info_from_res(sw_context, res); sw_context->man = vmw_context_res_man(res); + vmw_resource_unreference(&res); return 0; } diff --git a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c index 26f88d64879f..64f9f407752b 100644 --- a/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c +++ b/drivers/gpu/drm/vmwgfx/vmwgfx_resource.c @@ -282,41 +282,6 @@ int vmw_user_resource_lookup_handle(struct vmw_private *dev_priv, return ret; } -/** - * vmw_user_resource_lookup_handle - lookup a struct resource from a - * TTM user-space handle and perform basic type checks - * - * @dev_priv: Pointer to a device private struct - * @tfile: Pointer to a struct ttm_object_file identifying the caller - * @handle: The TTM user-space handle - * @converter: Pointer to an object describing the resource type - * @p_res: On successful return the location pointed to will contain - * a pointer to a refcounted struct vmw_resource. - * - * If the handle can't be found or is associated with an incorrect resource - * type, -EINVAL will be returned. - */ -struct vmw_resource * -vmw_user_resource_noref_lookup_handle(struct vmw_private *dev_priv, - struct ttm_object_file *tfile, - uint32_t handle, - const struct vmw_user_resource_conv - *converter) -{ - struct ttm_base_object *base; - - base = ttm_base_object_noref_lookup(tfile, handle); - if (!base) - return ERR_PTR(-ESRCH); - - if (unlikely(ttm_base_object_type(base) != converter->object_type)) { - ttm_base_object_noref_release(); - return ERR_PTR(-EINVAL); - } - - return converter->base_obj_to_res(base); -} - /** * Helper function that looks either a surface or bo. * diff --git a/include/drm/drm_hashtab.h b/include/drm/drm_hashtab.h index bb95ff011baf..fe5ad4793a15 100644 --- a/include/drm/drm_hashtab.h +++ b/include/drm/drm_hashtab.h @@ -49,6 +49,8 @@ struct drm_open_hash { u8 order; }; +struct hlist_node *drm_ht_find_key(struct drm_open_hash *ht, unsigned long key); + int drm_ht_create(struct drm_open_hash *ht, unsigned int order); int drm_ht_insert_item(struct drm_open_hash *ht, struct drm_hash_item *item); int drm_ht_just_insert_please(struct drm_open_hash *ht, struct drm_hash_item *item, -- 2.34.1

1 month, 1 week

2
1
0 0

[PATCH v2] Revert "vrf: Remove unnecessary RCU-bh critical section"

by greearb＠candelatech.com

From: Ben Greear <greearb(a)candelatech.com> This reverts commit 504fc6f4f7f681d2a03aa5f68aad549d90eab853. dev_queue_xmit_nit needs to run with bh locking, otherwise it conflicts with packets coming in from a nic in softirq context and packets being transmitted from user context. ================================ WARNING: inconsistent lock state 6.11.0 #1 Tainted: G W -------------------------------- inconsistent {IN-SOFTIRQ-W} -> {SOFTIRQ-ON-W} usage. btserver/134819 [HC0[0]:SC0[0]:HE1:SE1] takes: ffff8882da30c118 (rlock-AF_PACKET){+.?.}-{2:2}, at: tpacket_rcv+0x863/0x3b30 {IN-SOFTIRQ-W} state was registered at: lock_acquire+0x19a/0x4f0 _raw_spin_lock+0x27/0x40 packet_rcv+0xa33/0x1320 __netif_receive_skb_core.constprop.0+0xcb0/0x3a90 __netif_receive_skb_list_core+0x2c9/0x890 netif_receive_skb_list_internal+0x610/0xcc0 napi_complete_done+0x1c0/0x7c0 igb_poll+0x1dbb/0x57e0 [igb] __napi_poll.constprop.0+0x99/0x430 net_rx_action+0x8e7/0xe10 handle_softirqs+0x1b7/0x800 __irq_exit_rcu+0x91/0xc0 irq_exit_rcu+0x5/0x10 [snip] other info that might help us debug this: Possible unsafe locking scenario: CPU0 ---- lock(rlock-AF_PACKET); <Interrupt> lock(rlock-AF_PACKET); *** DEADLOCK *** Call Trace: <TASK> dump_stack_lvl+0x73/0xa0 mark_lock+0x102e/0x16b0 __lock_acquire+0x9ae/0x6170 lock_acquire+0x19a/0x4f0 _raw_spin_lock+0x27/0x40 tpacket_rcv+0x863/0x3b30 dev_queue_xmit_nit+0x709/0xa40 vrf_finish_direct+0x26e/0x340 [vrf] vrf_l3_out+0x5f4/0xe80 [vrf] __ip_local_out+0x51e/0x7a0 [snip] Fixes: 504fc6f4f7f6 ("vrf: Remove unnecessary RCU-bh critical section") Link: https://lore.kernel.org/netdev/05765015-f727-2f30-58da-2ad6fa7ea99f@candela… Signed-off-by: Ben Greear <greearb(a)candelatech.com> --- v2: Edit patch description. drivers/net/vrf.c | 2 ++ net/core/dev.c | 1 + 2 files changed, 3 insertions(+) diff --git a/drivers/net/vrf.c b/drivers/net/vrf.c index 4d8ccaf9a2b4..4087f72f0d2b 100644 --- a/drivers/net/vrf.c +++ b/drivers/net/vrf.c @@ -608,7 +608,9 @@ static void vrf_finish_direct(struct sk_buff *skb) eth_zero_addr(eth->h_dest); eth->h_proto = skb->protocol; + rcu_read_lock_bh(); dev_queue_xmit_nit(skb, vrf_dev); + rcu_read_unlock_bh(); skb_pull(skb, ETH_HLEN); } diff --git a/net/core/dev.c b/net/core/dev.c index cd479f5f22f6..566e69a38eed 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -2285,6 +2285,7 @@ EXPORT_SYMBOL_GPL(dev_nit_active); /* * Support routine. Sends outgoing frames to any network * taps currently in use. + * BH must be disabled before calling this. */ void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev) -- 2.42.0

1 month, 1 week

4
5
0 0

[PATCH] virtio: console: Fix atomicity violation in fill_readbuf()

by Qiu-ji Chen

The atomicity violation issue is due to the invalidation of the function port_has_data()'s check caused by concurrency. Imagine a scenario where a port that contains data passes the validity check but is simultaneously assigned a value with no data. This could result in an empty port passing the validity check, potentially leading to a null pointer dereference error later in the program, which is inconsistent. To address this issue, we believe there is a problem with the original logic. Since the validity check and the read operation were separated, it could lead to inconsistencies between the data that passes the check and the data being read. Therefore, we moved the main logic of the port_has_data function into this function and placed the read operation within a lock, ensuring that the validity check and read operation are not separated, thus resolving the problem. This possible bug is found by an experimental static analysis tool developed by our team. This tool analyzes the locking APIs to extract function pairs that can be concurrently executed, and then analyzes the instructions in the paired functions to identify possible concurrency bugs including data races and atomicity violations. Fixes: 203baab8ba31 ("virtio: console: Introduce function to hand off data from host to readers") Cc: stable(a)vger.kernel.org Signed-off-by: Qiu-ji Chen <chenqiuji666(a)gmail.com> --- drivers/char/virtio_console.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c index de7d720d99fa..5aaf07f71a4c 100644 --- a/drivers/char/virtio_console.c +++ b/drivers/char/virtio_console.c @@ -656,10 +656,14 @@ static ssize_t fill_readbuf(struct port *port, u8 __user *out_buf, struct port_buffer *buf; unsigned long flags; - if (!out_count || !port_has_data(port)) + if (!out_count) return 0; - buf = port->inbuf; + spin_lock_irqsave(&port->inbuf_lock, flags); + buf = port->inbuf = get_inbuf(port); + spin_unlock_irqrestore(&port->inbuf_lock, flags); + if (!buf) + return 0; out_count = min(out_count, buf->len - buf->offset); if (to_user) { -- 2.34.1

1 month, 1 week

1
0
0 0

[PATCH] drivers:media:radio: Fix atomicity violation in fmc_send_cmd()

by Qiu-ji Chen

Atomicity violation occurs when the fmc_send_cmd() function is executed simultaneously with the modification of the fmdev->resp_skb value. Consider a scenario where, after passing the validity check within the function, a non-null fmdev->resp_skb variable is assigned a null value. This results in an invalid fmdev->resp_skb variable passing the validity check. As seen in the later part of the function, skb = fmdev->resp_skb; when the invalid fmdev->resp_skb passes the check, a null pointer dereference error may occur at line 478, evt_hdr = (void *)skb->data; To address this issue, it is recommended to include the validity check of fmdev->resp_skb within the locked section of the function. This modification ensures that the value of fmdev->resp_skb does not change during the validation process, thereby maintaining its validity. This possible bug is found by an experimental static analysis tool developed by our team. This tool analyzes the locking APIs to extract function pairs that can be concurrently executed, and then analyzes the instructions in the paired functions to identify possible concurrency bugs including data races and atomicity violations. Fixes: e8454ff7b9a4 ("[media] drivers:media:radio: wl128x: FM Driver Common sources") Cc: stable(a)vger.kernel.org Signed-off-by: Qiu-ji Chen <chenqiuji666(a)gmail.com> --- drivers/media/radio/wl128x/fmdrv_common.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/media/radio/wl128x/fmdrv_common.c b/drivers/media/radio/wl128x/fmdrv_common.c index 3d36f323a8f8..4d032436691c 100644 --- a/drivers/media/radio/wl128x/fmdrv_common.c +++ b/drivers/media/radio/wl128x/fmdrv_common.c @@ -466,11 +466,12 @@ int fmc_send_cmd(struct fmdev *fmdev, u8 fm_op, u16 type, void *payload, jiffies_to_msecs(FM_DRV_TX_TIMEOUT) / 1000); return -ETIMEDOUT; } + spin_lock_irqsave(&fmdev->resp_skb_lock, flags); if (!fmdev->resp_skb) { + spin_unlock_irqrestore(&fmdev->resp_skb_lock, flags); fmerr("Response SKB is missing\n"); return -EFAULT; } - spin_lock_irqsave(&fmdev->resp_skb_lock, flags); skb = fmdev->resp_skb; fmdev->resp_skb = NULL; spin_unlock_irqrestore(&fmdev->resp_skb_lock, flags); -- 2.34.1

1 month, 1 week

1
0
0 0

[PATCH 6.1.y] btrfs: calculate the right space for delayed refs when updating global reserve

by Diogo Jahchan Koike

From: Filipe Manana <fdmanana(a)suse.com> commit f8f210dc84709804c9f952297f2bfafa6ea6b4bd upstream. When updating the global block reserve, we account for the 6 items needed by an unlink operation and the 6 delayed references for each one of those items. However the calculation for the delayed references is not correct in case we have the free space tree enabled, as in that case we need to touch the free space tree as well and therefore need twice the number of bytes. So use the btrfs_calc_delayed_ref_bytes() helper to calculate the number of bytes need for the delayed references at btrfs_update_global_block_rsv(). Reviewed-by: Josef Bacik <josef(a)toxicpanda.com> Signed-off-by: Filipe Manana <fdmanana(a)suse.com> Reviewed-by: David Sterba <dsterba(a)suse.com> Signed-off-by: David Sterba <dsterba(a)suse.com> [Diogo: this patch has been cherry-picked from the original commit; conflicts included lack of a define (picked from commit 5630e2bcfe223) and lack of btrfs_calc_delayed_ref_bytes (picked from commit 0e55a54502b97) - changed const struct -> struct for compatibility.] Signed-off-by: Diogo Jahchan Koike <djahchankoike(a)gmail.com> --- Fixes WARNING in btrfs_chunk_alloc (2) [0] [0]: https://syzkaller.appspot.com/bug?extid=57de2b05959bc1e659af --- fs/btrfs/block-rsv.c | 16 +++++++++------- fs/btrfs/block-rsv.h | 12 ++++++++++++ fs/btrfs/delayed-ref.h | 21 +++++++++++++++++++++ 3 files changed, 42 insertions(+), 7 deletions(-) diff --git a/fs/btrfs/block-rsv.c b/fs/btrfs/block-rsv.c index ec96285357e0..f7e3d6c2e302 100644 --- a/fs/btrfs/block-rsv.c +++ b/fs/btrfs/block-rsv.c @@ -378,17 +378,19 @@ void btrfs_update_global_block_rsv(struct btrfs_fs_info *fs_info) /* * But we also want to reserve enough space so we can do the fallback - * global reserve for an unlink, which is an additional 5 items (see the - * comment in __unlink_start_trans for what we're modifying.) + * global reserve for an unlink, which is an additional + * BTRFS_UNLINK_METADATA_UNITS items. * * But we also need space for the delayed ref updates from the unlink, - * so its 10, 5 for the actual operation, and 5 for the delayed ref - * updates. + * so add BTRFS_UNLINK_METADATA_UNITS units for delayed refs, one for + * each unlink metadata item. */ - min_items += 10; - + min_items += BTRFS_UNLINK_METADATA_UNITS; + num_bytes = max_t(u64, num_bytes, - btrfs_calc_insert_metadata_size(fs_info, min_items)); + btrfs_calc_insert_metadata_size(fs_info, min_items) + + btrfs_calc_delayed_ref_bytes(fs_info, + BTRFS_UNLINK_METADATA_UNITS)); spin_lock(&sinfo->lock); spin_lock(&block_rsv->lock); diff --git a/fs/btrfs/block-rsv.h b/fs/btrfs/block-rsv.h index 578c3497a455..662c52b4bd44 100644 --- a/fs/btrfs/block-rsv.h +++ b/fs/btrfs/block-rsv.h @@ -50,6 +50,18 @@ struct btrfs_block_rsv { u64 qgroup_rsv_reserved; }; +/* + * Number of metadata items necessary for an unlink operation: + * + * 1 for the possible orphan item + * 1 for the dir item + * 1 for the dir index + * 1 for the inode ref + * 1 for the inode + * 1 for the parent inode + */ +#define BTRFS_UNLINK_METADATA_UNITS 6 + void btrfs_init_block_rsv(struct btrfs_block_rsv *rsv, enum btrfs_rsv_type type); void btrfs_init_root_block_rsv(struct btrfs_root *root); struct btrfs_block_rsv *btrfs_alloc_block_rsv(struct btrfs_fs_info *fs_info, diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h index d6304b690ec4..5c6bb23d45a5 100644 --- a/fs/btrfs/delayed-ref.h +++ b/fs/btrfs/delayed-ref.h @@ -253,6 +253,27 @@ extern struct kmem_cache *btrfs_delayed_extent_op_cachep; int __init btrfs_delayed_ref_init(void); void __cold btrfs_delayed_ref_exit(void); +static inline u64 btrfs_calc_delayed_ref_bytes(struct btrfs_fs_info *fs_info, + int num_delayed_refs) +{ + u64 num_bytes; + + num_bytes = btrfs_calc_insert_metadata_size(fs_info, num_delayed_refs); + + /* + * We have to check the mount option here because we could be enabling + * the free space tree for the first time and don't have the compat_ro + * option set yet. + * + * We need extra reservations if we have the free space tree because + * we'll have to modify that tree as well. + */ + if (btrfs_test_opt(fs_info, FREE_SPACE_TREE)) + num_bytes *= 2; + + return num_bytes; +} + static inline void btrfs_init_generic_ref(struct btrfs_ref *generic_ref, int action, u64 bytenr, u64 len, u64 parent) { -- 2.43.0

1 month, 1 week

2
1
0 0

[PATCH 6.1] wifi: mac80211: Avoid address calculations via out of bounds array indexing

by Xiangyu Chen

From: Kenton Groombridge <concord(a)gentoo.org> [ Upstream commit 2663d0462eb32ae7c9b035300ab6b1523886c718 ] req->n_channels must be set before req->channels[] can be used. This patch fixes one of the issues encountered in [1]. [ 83.964255] UBSAN: array-index-out-of-bounds in net/mac80211/scan.c:364:4 [ 83.964258] index 0 is out of range for type 'struct ieee80211_channel *[]' [...] [ 83.964264] Call Trace: [ 83.964267] <TASK> [ 83.964269] dump_stack_lvl+0x3f/0xc0 [ 83.964274] __ubsan_handle_out_of_bounds+0xec/0x110 [ 83.964278] ieee80211_prep_hw_scan+0x2db/0x4b0 [ 83.964281] __ieee80211_start_scan+0x601/0x990 [ 83.964291] nl80211_trigger_scan+0x874/0x980 [ 83.964295] genl_family_rcv_msg_doit+0xe8/0x160 [ 83.964298] genl_rcv_msg+0x240/0x270 [...] [1] https://bugzilla.kernel.org/show_bug.cgi?id=218810 Co-authored-by: Kees Cook <keescook(a)chromium.org> Signed-off-by: Kees Cook <kees(a)kernel.org> Signed-off-by: Kenton Groombridge <concord(a)gentoo.org> Link: https://msgid.link/20240605152218.236061-1-concord@gentoo.org Signed-off-by: Johannes Berg <johannes.berg(a)intel.com> Signed-off-by: Xiangyu Chen <xiangyu.chen(a)windriver.com> --- net/mac80211/scan.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) diff --git a/net/mac80211/scan.c b/net/mac80211/scan.c index 62c22ff329ad..d81b49fb6458 100644 --- a/net/mac80211/scan.c +++ b/net/mac80211/scan.c @@ -357,7 +357,8 @@ static bool ieee80211_prep_hw_scan(struct ieee80211_sub_if_data *sdata) struct cfg80211_scan_request *req; struct cfg80211_chan_def chandef; u8 bands_used = 0; - int i, ielen, n_chans; + int i, ielen; + u32 *n_chans; u32 flags = 0; req = rcu_dereference_protected(local->scan_req, @@ -367,34 +368,34 @@ static bool ieee80211_prep_hw_scan(struct ieee80211_sub_if_data *sdata) return false; if (ieee80211_hw_check(&local->hw, SINGLE_SCAN_ON_ALL_BANDS)) { + local->hw_scan_req->req.n_channels = req->n_channels; + for (i = 0; i < req->n_channels; i++) { local->hw_scan_req->req.channels[i] = req->channels[i]; bands_used |= BIT(req->channels[i]->band); } - - n_chans = req->n_channels; } else { do { if (local->hw_scan_band == NUM_NL80211_BANDS) return false; - n_chans = 0; + n_chans = &local->hw_scan_req->req.n_channels; + *n_chans = 0; for (i = 0; i < req->n_channels; i++) { if (req->channels[i]->band != local->hw_scan_band) continue; - local->hw_scan_req->req.channels[n_chans] = + local->hw_scan_req->req.channels[(*n_chans)++] = req->channels[i]; - n_chans++; + bands_used |= BIT(req->channels[i]->band); } local->hw_scan_band++; - } while (!n_chans); + } while (!*n_chans); } - local->hw_scan_req->req.n_channels = n_chans; ieee80211_prepare_scan_chandef(&chandef, req->scan_width); if (req->flags & NL80211_SCAN_FLAG_MIN_PREQ_CONTENT) -- 2.43.0

1 month, 1 week

2
1
0 0

FAILED: patch "[PATCH] mptcp: pm: Fix uaf in __timer_delete_sync" failed to apply to 5.10-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.10-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.10.y git checkout FETCH_HEAD git cherry-pick -x b4cd80b0338945a94972ac3ed54f8338d2da2076 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2024091330-reps-craftsman-ab67@gregkh' --subject-prefix 'PATCH 5.10.y' HEAD^.. Possible dependencies: b4cd80b03389 ("mptcp: pm: Fix uaf in __timer_delete_sync") d58300c3185b ("mptcp: validate 'id' when stopping the ADD_ADDR retransmit timer") f7dafee18538 ("mptcp: use mptcp_addr_info in mptcp_options_received") dc87efdb1a5c ("mptcp: add mptcp reset option support") 557963c383e8 ("mptcp: move to next addr when subflow creation fail") d88c476f4a7d ("mptcp: export lookup_anno_list_by_saddr") efd13b71a3fa ("Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net") thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From b4cd80b0338945a94972ac3ed54f8338d2da2076 Mon Sep 17 00:00:00 2001 From: Edward Adam Davis <eadavis(a)qq.com> Date: Tue, 10 Sep 2024 17:58:56 +0800 Subject: [PATCH] mptcp: pm: Fix uaf in __timer_delete_sync There are two paths to access mptcp_pm_del_add_timer, result in a race condition: CPU1 CPU2 ==== ==== net_rx_action napi_poll netlink_sendmsg __napi_poll netlink_unicast process_backlog netlink_unicast_kernel __netif_receive_skb genl_rcv __netif_receive_skb_one_core netlink_rcv_skb NF_HOOK genl_rcv_msg ip_local_deliver_finish genl_family_rcv_msg ip_protocol_deliver_rcu genl_family_rcv_msg_doit tcp_v4_rcv mptcp_pm_nl_flush_addrs_doit tcp_v4_do_rcv mptcp_nl_remove_addrs_list tcp_rcv_established mptcp_pm_remove_addrs_and_subflows tcp_data_queue remove_anno_list_by_saddr mptcp_incoming_options mptcp_pm_del_add_timer mptcp_pm_del_add_timer kfree(entry) In remove_anno_list_by_saddr(running on CPU2), after leaving the critical zone protected by "pm.lock", the entry will be released, which leads to the occurrence of uaf in the mptcp_pm_del_add_timer(running on CPU1). Keeping a reference to add_timer inside the lock, and calling sk_stop_timer_sync() with this reference, instead of "entry->add_timer". Move list_del(&entry->list) to mptcp_pm_del_add_timer and inside the pm lock, do not directly access any members of the entry outside the pm lock, which can avoid similar "entry->x" uaf. Fixes: 00cfd77b9063 ("mptcp: retransmit ADD_ADDR when timeout") Cc: stable(a)vger.kernel.org Reported-and-tested-by: syzbot+f3a31fb909db9b2a5c4d(a)syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=f3a31fb909db9b2a5c4d Signed-off-by: Matthieu Baerts (NGI0) <matttbe(a)kernel.org> Signed-off-by: Edward Adam Davis <eadavis(a)qq.com> Acked-by: Paolo Abeni <pabeni(a)redhat.com> Link: https://patch.msgid.link/tencent_7142963A37944B4A74EF76CD66EA3C253609@qq.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c index f891bc714668..ad935d34c973 100644 --- a/net/mptcp/pm_netlink.c +++ b/net/mptcp/pm_netlink.c @@ -334,15 +334,21 @@ mptcp_pm_del_add_timer(struct mptcp_sock *msk, { struct mptcp_pm_add_entry *entry; struct sock *sk = (struct sock *)msk; + struct timer_list *add_timer = NULL; spin_lock_bh(&msk->pm.lock); entry = mptcp_lookup_anno_list_by_saddr(msk, addr); - if (entry && (!check_id || entry->addr.id == addr->id)) + if (entry && (!check_id || entry->addr.id == addr->id)) { entry->retrans_times = ADD_ADDR_RETRANS_MAX; + add_timer = &entry->add_timer; + } + if (!check_id && entry) + list_del(&entry->list); spin_unlock_bh(&msk->pm.lock); - if (entry && (!check_id || entry->addr.id == addr->id)) - sk_stop_timer_sync(sk, &entry->add_timer); + /* no lock, because sk_stop_timer_sync() is calling del_timer_sync() */ + if (add_timer) + sk_stop_timer_sync(sk, add_timer); return entry; } @@ -1462,7 +1468,6 @@ static bool remove_anno_list_by_saddr(struct mptcp_sock *msk, entry = mptcp_pm_del_add_timer(msk, addr, false); if (entry) { - list_del(&entry->list); kfree(entry); return true; }

1 month, 1 week

3
5
0 0

[PATCH 6.6.y] bnxt_en: Cap the size of HWRM_PORT_PHY_QCFG forwarded response

by Samasth Norway Ananda

From: Michael Chan <michael.chan(a)broadcom.com> commit 7d9df38c9c037ab84502ce7eeae9f1e1e7e72603 upstream. Firmware interface 1.10.2.118 has increased the size of HWRM_PORT_PHY_QCFG response beyond the maximum size that can be forwarded. When the VF's link state is not the default auto state, the PF will need to forward the response back to the VF to indicate the forced state. This regression may cause the VF to fail to initialize. Fix it by capping the HWRM_PORT_PHY_QCFG response to the maximum 96 bytes. The SPEEDS2_SUPPORTED flag needs to be cleared because the new speeds2 fields are beyond the legacy structure. Also modify bnxt_hwrm_fwd_resp() to print a warning if the message size exceeds 96 bytes to make this failure more obvious. Fixes: 84a911db8305 ("bnxt_en: Update firmware interface to 1.10.2.118") Reviewed-by: Somnath Kotur <somnath.kotur(a)broadcom.com> Reviewed-by: Pavan Chebbi <pavan.chebbi(a)broadcom.com> Signed-off-by: Michael Chan <michael.chan(a)broadcom.com> Link: https://lore.kernel.org/r/20240612231736.57823-1-michael.chan@broadcom.com Signed-off-by: Jakub Kicinski <kuba(a)kernel.org> [Samasth: backport to 6.6.y] Signed-off-by: Samasth Norway Ananda <samasth.norway.ananda(a)oracle.com> --- drivers/net/ethernet/broadcom/bnxt/bnxt.h | 51 +++++++++++++++++++ .../net/ethernet/broadcom/bnxt/bnxt_sriov.c | 12 ++++- 2 files changed, 61 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.h b/drivers/net/ethernet/broadcom/bnxt/bnxt.h index 0116f67593e3a..f323c77aaea0e 100644 --- a/drivers/net/ethernet/broadcom/bnxt/bnxt.h +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.h @@ -1195,6 +1195,57 @@ struct bnxt_ntuple_filter { #define BNXT_FLTR_UPDATE 1 }; +/* Compat version of hwrm_port_phy_qcfg_output capped at 96 bytes. The + * first 95 bytes are identical to hwrm_port_phy_qcfg_output in bnxt_hsi.h. + * The last valid byte in the compat version is different. + */ +struct hwrm_port_phy_qcfg_output_compat { + __le16 error_code; + __le16 req_type; + __le16 seq_id; + __le16 resp_len; + u8 link; + u8 active_fec_signal_mode; + __le16 link_speed; + u8 duplex_cfg; + u8 pause; + __le16 support_speeds; + __le16 force_link_speed; + u8 auto_mode; + u8 auto_pause; + __le16 auto_link_speed; + __le16 auto_link_speed_mask; + u8 wirespeed; + u8 lpbk; + u8 force_pause; + u8 module_status; + __le32 preemphasis; + u8 phy_maj; + u8 phy_min; + u8 phy_bld; + u8 phy_type; + u8 media_type; + u8 xcvr_pkg_type; + u8 eee_config_phy_addr; + u8 parallel_detect; + __le16 link_partner_adv_speeds; + u8 link_partner_adv_auto_mode; + u8 link_partner_adv_pause; + __le16 adv_eee_link_speed_mask; + __le16 link_partner_adv_eee_link_speed_mask; + __le32 xcvr_identifier_type_tx_lpi_timer; + __le16 fec_cfg; + u8 duplex_state; + u8 option_flags; + char phy_vendor_name[16]; + char phy_vendor_partnumber[16]; + __le16 support_pam4_speeds; + __le16 force_pam4_link_speed; + __le16 auto_pam4_link_speed_mask; + u8 link_partner_pam4_adv_speeds; + u8 valid; +}; + struct bnxt_link_info { u8 phy_type; u8 media_type; diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c index dde327f2c57ee..ac9a0c039b776 100644 --- a/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c @@ -942,8 +942,11 @@ static int bnxt_hwrm_fwd_resp(struct bnxt *bp, struct bnxt_vf_info *vf, struct hwrm_fwd_resp_input *req; int rc; - if (BNXT_FWD_RESP_SIZE_ERR(msg_size)) + if (BNXT_FWD_RESP_SIZE_ERR(msg_size)) { + netdev_warn_once(bp->dev, "HWRM fwd response too big (%d bytes)\n", + msg_size); return -EINVAL; + } rc = hwrm_req_init(bp, req, HWRM_FWD_RESP); if (!rc) { @@ -1077,7 +1080,7 @@ static int bnxt_vf_set_link(struct bnxt *bp, struct bnxt_vf_info *vf) rc = bnxt_hwrm_exec_fwd_resp( bp, vf, sizeof(struct hwrm_port_phy_qcfg_input)); } else { - struct hwrm_port_phy_qcfg_output phy_qcfg_resp = {0}; + struct hwrm_port_phy_qcfg_output_compat phy_qcfg_resp = {}; struct hwrm_port_phy_qcfg_input *phy_qcfg_req; phy_qcfg_req = @@ -1088,6 +1091,11 @@ static int bnxt_vf_set_link(struct bnxt *bp, struct bnxt_vf_info *vf) mutex_unlock(&bp->link_lock); phy_qcfg_resp.resp_len = cpu_to_le16(sizeof(phy_qcfg_resp)); phy_qcfg_resp.seq_id = phy_qcfg_req->seq_id; + /* New SPEEDS2 fields are beyond the legacy structure, so + * clear the SPEEDS2_SUPPORTED flag. + */ + phy_qcfg_resp.option_flags &= + ~PORT_PHY_QCAPS_RESP_FLAGS2_SPEEDS2_SUPPORTED; phy_qcfg_resp.valid = 1; if (vf->flags & BNXT_VF_LINK_UP) { -- 2.45.2

1 month, 1 week

2
1
0 0

[PATCH 5.10] bpf: Fix mismatch memory accounting for devmap maps

by Pu Lehui

From: Pu Lehui <pulehui(a)huawei.com> Commit 70294d8bc31f ("bpf: Eliminate rlimit-based memory accounting for devmap maps") relies on the v5.11+ basic mechanism of memcg-based memory accounting [0]. The commit cannot be independently backported to the 5.10 stable branch, otherwise the related memory when creating devmap will be unrestricted and the associated bpf selftest map_ptr will fail. Let's roll back to rlimit-based memory accounting mode for devmap and re-adapt the commit 225da02acdc9 ("bpf: Fix DEVMAP_HASH overflow check on 32-bit arches") to the 5.10 stable branch. Link: https://lore.kernel.org/bpf/20201201215900.3569844-1-guro@fb.com [0] Fixes: 225da02acdc9 ("bpf: Fix DEVMAP_HASH overflow check on 32-bit arches") Fixes: 70294d8bc31f ("bpf: Eliminate rlimit-based memory accounting for devmap maps") Signed-off-by: Pu Lehui <pulehui(a)huawei.com> --- kernel/bpf/devmap.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-) diff --git a/kernel/bpf/devmap.c b/kernel/bpf/devmap.c index 07b5edb2c70f..7eb1282edc8e 100644 --- a/kernel/bpf/devmap.c +++ b/kernel/bpf/devmap.c @@ -109,6 +109,8 @@ static inline struct hlist_head *dev_map_index_hash(struct bpf_dtab *dtab, static int dev_map_init_map(struct bpf_dtab *dtab, union bpf_attr *attr) { u32 valsize = attr->value_size; + u64 cost = 0; + int err; /* check sanity of attributes. 2 value sizes supported: * 4 bytes: ifindex @@ -136,11 +138,21 @@ static int dev_map_init_map(struct bpf_dtab *dtab, union bpf_attr *attr) return -EINVAL; dtab->n_buckets = roundup_pow_of_two(dtab->map.max_entries); + cost += (u64) sizeof(struct hlist_head) * dtab->n_buckets; + } else { + cost += (u64) dtab->map.max_entries * sizeof(struct bpf_dtab_netdev *); + } + /* if map size is larger than memlock limit, reject it */ + err = bpf_map_charge_init(&dtab->map.memory, cost); + if (err) + return -EINVAL; + + if (attr->map_type == BPF_MAP_TYPE_DEVMAP_HASH) { dtab->dev_index_head = dev_map_create_hash(dtab->n_buckets, dtab->map.numa_node); if (!dtab->dev_index_head) - return -ENOMEM; + goto free_charge; spin_lock_init(&dtab->index_lock); } else { @@ -148,10 +160,14 @@ static int dev_map_init_map(struct bpf_dtab *dtab, union bpf_attr *attr) sizeof(struct bpf_dtab_netdev *), dtab->map.numa_node); if (!dtab->netdev_map) - return -ENOMEM; + goto free_charge; } return 0; + +free_charge: + bpf_map_charge_finish(&dtab->map.memory); + return -ENOMEM; } static struct bpf_map *dev_map_alloc(union bpf_attr *attr) -- 2.34.1

1 month, 1 week

4
6
0 0

mcp251xfd: please add to the stable trees

by Marc Kleine-Budde

Hello stable-team, Francesco Dolcini reported a regression on v6.6.52 [1], it turns out since v6.6.51~34 a.k.a. 5ea24ddc26a7 ("can: mcp251xfd: rx: add workaround for erratum DS80000789E 6 of mcp2518fd") the mcp25xfd driver is broken. Cherry picking the following commits fixes the problem: 51b2a7216122 ("can: mcp251xfd: properly indent labels") a7801540f325 ("can: mcp251xfd: move mcp251xfd_timestamp_start()/stop() into mcp251xfd_chip_start/stop()") Please pick these patches for - 6.10.x - 6.6.x - 6.1.x [1] https://lore.kernel.org/all/20240923115310.GA138774@francesco-nb regards, Marc -- Pengutronix e.K. | Marc Kleine-Budde | Embedded Linux | https://www.pengutronix.de | Vertretung Nürnberg | Phone: +49-5121-206917-129 | Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-9 |

1 month, 1 week

2
1
0 0

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror September 2024