From 32a59639c551c93dbec3a2ce91286e4613e82f9b Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 5 Oct 2023 17:29:50 +0200 Subject: [PATCH 1/8] rbd: move rbd_dev_refresh() definition commit 0b035401c57021fc6c300272cbb1c5a889d4fe45 upstream. Move rbd_dev_refresh() definition further down to avoid having to move struct parent_image_info definition in the next commit. This spares some forward declarations too. Signed-off-by: Ilya Dryomov Reviewed-by: Dongsheng Yang [idryomov@gmail.com: backport to 5.4: drop rbd_is_snap() assert, preserve rbd_exists_validate() call] Signed-off-by: Sasha Levin --- drivers/block/rbd.c | 76 ++++++++++++++++++++++----------------------- 1 file changed, 37 insertions(+), 39 deletions(-) diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index 9d21f90f93f0..e015b8610e27 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -627,8 +627,6 @@ static void rbd_dev_remove_parent(struct rbd_device *rbd_dev); static int rbd_dev_refresh(struct rbd_device *rbd_dev); static int rbd_dev_v2_header_onetime(struct rbd_device *rbd_dev); -static int rbd_dev_header_info(struct rbd_device *rbd_dev); -static int rbd_dev_v2_parent_info(struct rbd_device *rbd_dev); static const char *rbd_dev_v2_snap_name(struct rbd_device *rbd_dev, u64 snap_id); static int _rbd_dev_v2_snap_size(struct rbd_device *rbd_dev, u64 snap_id, @@ -5075,43 +5073,6 @@ static void rbd_dev_update_size(struct rbd_device *rbd_dev) } } -static int rbd_dev_refresh(struct rbd_device *rbd_dev) -{ - u64 mapping_size; - int ret; - - down_write(&rbd_dev->header_rwsem); - mapping_size = rbd_dev->mapping.size; - - ret = rbd_dev_header_info(rbd_dev); - if (ret) - goto out; - - /* - * If there is a parent, see if it has disappeared due to the - * mapped image getting flattened. - */ - if (rbd_dev->parent) { - ret = rbd_dev_v2_parent_info(rbd_dev); - if (ret) - goto out; - } - - if (rbd_dev->spec->snap_id == CEPH_NOSNAP) { - rbd_dev->mapping.size = rbd_dev->header.image_size; - } else { - /* validate mapped snapshot's EXISTS flag */ - rbd_exists_validate(rbd_dev); - } - -out: - up_write(&rbd_dev->header_rwsem); - if (!ret && mapping_size != rbd_dev->mapping.size) - rbd_dev_update_size(rbd_dev); - - return ret; -} - static int rbd_init_request(struct blk_mq_tag_set *set, struct request *rq, unsigned int hctx_idx, unsigned int numa_node) { @@ -7061,6 +7022,43 @@ err_out_format: return ret; } +static int rbd_dev_refresh(struct rbd_device *rbd_dev) +{ + u64 mapping_size; + int ret; + + down_write(&rbd_dev->header_rwsem); + mapping_size = rbd_dev->mapping.size; + + ret = rbd_dev_header_info(rbd_dev); + if (ret) + goto out; + + /* + * If there is a parent, see if it has disappeared due to the + * mapped image getting flattened. + */ + if (rbd_dev->parent) { + ret = rbd_dev_v2_parent_info(rbd_dev); + if (ret) + goto out; + } + + if (rbd_dev->spec->snap_id == CEPH_NOSNAP) { + rbd_dev->mapping.size = rbd_dev->header.image_size; + } else { + /* validate mapped snapshot's EXISTS flag */ + rbd_exists_validate(rbd_dev); + } + +out: + up_write(&rbd_dev->header_rwsem); + if (!ret && mapping_size != rbd_dev->mapping.size) + rbd_dev_update_size(rbd_dev); + + return ret; +} + static ssize_t do_rbd_add(struct bus_type *bus, const char *buf, size_t count) From 2e0114edeb47189c6f8c0c36de38ecb78d75375d Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 5 Oct 2023 17:29:51 +0200 Subject: [PATCH 2/8] rbd: decouple header read-in from updating rbd_dev->header commit 510a7330c82a7754d5df0117a8589e8a539067c7 upstream. Make rbd_dev_header_info() populate a passed struct rbd_image_header instead of rbd_dev->header and introduce rbd_dev_update_header() for updating mutable fields in rbd_dev->header upon refresh. The initial read-in of both mutable and immutable fields in rbd_dev_image_probe() passes in rbd_dev->header so no update step is required there. rbd_init_layout() is now called directly from rbd_dev_image_probe() instead of individually in format 1 and format 2 implementations. Signed-off-by: Ilya Dryomov Reviewed-by: Dongsheng Yang [idryomov@gmail.com: backport to 5.4: _rbd_dev_v2_snap_features() doesn't have read_only param] Signed-off-by: Sasha Levin --- drivers/block/rbd.c | 209 ++++++++++++++++++++++++-------------------- 1 file changed, 116 insertions(+), 93 deletions(-) diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index e015b8610e27..17d802effdc4 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -626,7 +626,8 @@ void rbd_warn(struct rbd_device *rbd_dev, const char *fmt, ...) static void rbd_dev_remove_parent(struct rbd_device *rbd_dev); static int rbd_dev_refresh(struct rbd_device *rbd_dev); -static int rbd_dev_v2_header_onetime(struct rbd_device *rbd_dev); +static int rbd_dev_v2_header_onetime(struct rbd_device *rbd_dev, + struct rbd_image_header *header); static const char *rbd_dev_v2_snap_name(struct rbd_device *rbd_dev, u64 snap_id); static int _rbd_dev_v2_snap_size(struct rbd_device *rbd_dev, u64 snap_id, @@ -1096,15 +1097,24 @@ static void rbd_init_layout(struct rbd_device *rbd_dev) RCU_INIT_POINTER(rbd_dev->layout.pool_ns, NULL); } +static void rbd_image_header_cleanup(struct rbd_image_header *header) +{ + kfree(header->object_prefix); + ceph_put_snap_context(header->snapc); + kfree(header->snap_sizes); + kfree(header->snap_names); + + memset(header, 0, sizeof(*header)); +} + /* * Fill an rbd image header with information from the given format 1 * on-disk header. */ -static int rbd_header_from_disk(struct rbd_device *rbd_dev, - struct rbd_image_header_ondisk *ondisk) +static int rbd_header_from_disk(struct rbd_image_header *header, + struct rbd_image_header_ondisk *ondisk, + bool first_time) { - struct rbd_image_header *header = &rbd_dev->header; - bool first_time = header->object_prefix == NULL; struct ceph_snap_context *snapc; char *object_prefix = NULL; char *snap_names = NULL; @@ -1171,11 +1181,6 @@ static int rbd_header_from_disk(struct rbd_device *rbd_dev, if (first_time) { header->object_prefix = object_prefix; header->obj_order = ondisk->options.order; - rbd_init_layout(rbd_dev); - } else { - ceph_put_snap_context(header->snapc); - kfree(header->snap_names); - kfree(header->snap_sizes); } /* The remaining fields always get updated (when we refresh) */ @@ -4981,7 +4986,9 @@ out_req: * return, the rbd_dev->header field will contain up-to-date * information about the image. */ -static int rbd_dev_v1_header_info(struct rbd_device *rbd_dev) +static int rbd_dev_v1_header_info(struct rbd_device *rbd_dev, + struct rbd_image_header *header, + bool first_time) { struct rbd_image_header_ondisk *ondisk = NULL; u32 snap_count = 0; @@ -5029,7 +5036,7 @@ static int rbd_dev_v1_header_info(struct rbd_device *rbd_dev) snap_count = le32_to_cpu(ondisk->snap_count); } while (snap_count != want_count); - ret = rbd_header_from_disk(rbd_dev, ondisk); + ret = rbd_header_from_disk(header, ondisk, first_time); out: kfree(ondisk); @@ -5642,17 +5649,12 @@ static int _rbd_dev_v2_snap_size(struct rbd_device *rbd_dev, u64 snap_id, return 0; } -static int rbd_dev_v2_image_size(struct rbd_device *rbd_dev) -{ - return _rbd_dev_v2_snap_size(rbd_dev, CEPH_NOSNAP, - &rbd_dev->header.obj_order, - &rbd_dev->header.image_size); -} - -static int rbd_dev_v2_object_prefix(struct rbd_device *rbd_dev) +static int rbd_dev_v2_object_prefix(struct rbd_device *rbd_dev, + char **pobject_prefix) { size_t size; void *reply_buf; + char *object_prefix; int ret; void *p; @@ -5670,16 +5672,16 @@ static int rbd_dev_v2_object_prefix(struct rbd_device *rbd_dev) goto out; p = reply_buf; - rbd_dev->header.object_prefix = ceph_extract_encoded_string(&p, - p + ret, NULL, GFP_NOIO); + object_prefix = ceph_extract_encoded_string(&p, p + ret, NULL, + GFP_NOIO); + if (IS_ERR(object_prefix)) { + ret = PTR_ERR(object_prefix); + goto out; + } ret = 0; - if (IS_ERR(rbd_dev->header.object_prefix)) { - ret = PTR_ERR(rbd_dev->header.object_prefix); - rbd_dev->header.object_prefix = NULL; - } else { - dout(" object_prefix = %s\n", rbd_dev->header.object_prefix); - } + *pobject_prefix = object_prefix; + dout(" object_prefix = %s\n", object_prefix); out: kfree(reply_buf); @@ -5724,12 +5726,6 @@ static int _rbd_dev_v2_snap_features(struct rbd_device *rbd_dev, u64 snap_id, return 0; } -static int rbd_dev_v2_features(struct rbd_device *rbd_dev) -{ - return _rbd_dev_v2_snap_features(rbd_dev, CEPH_NOSNAP, - &rbd_dev->header.features); -} - /* * These are generic image flags, but since they are used only for * object map, store them in rbd_dev->object_map_flags. @@ -6004,14 +6000,14 @@ out_err: return ret; } -static int rbd_dev_v2_striping_info(struct rbd_device *rbd_dev) +static int rbd_dev_v2_striping_info(struct rbd_device *rbd_dev, + u64 *stripe_unit, u64 *stripe_count) { struct { __le64 stripe_unit; __le64 stripe_count; } __attribute__ ((packed)) striping_info_buf = { 0 }; size_t size = sizeof (striping_info_buf); - void *p; int ret; ret = rbd_obj_method_sync(rbd_dev, &rbd_dev->header_oid, @@ -6023,27 +6019,33 @@ static int rbd_dev_v2_striping_info(struct rbd_device *rbd_dev) if (ret < size) return -ERANGE; - p = &striping_info_buf; - rbd_dev->header.stripe_unit = ceph_decode_64(&p); - rbd_dev->header.stripe_count = ceph_decode_64(&p); + *stripe_unit = le64_to_cpu(striping_info_buf.stripe_unit); + *stripe_count = le64_to_cpu(striping_info_buf.stripe_count); + dout(" stripe_unit = %llu stripe_count = %llu\n", *stripe_unit, + *stripe_count); + return 0; } -static int rbd_dev_v2_data_pool(struct rbd_device *rbd_dev) +static int rbd_dev_v2_data_pool(struct rbd_device *rbd_dev, s64 *data_pool_id) { - __le64 data_pool_id; + __le64 data_pool_buf; int ret; ret = rbd_obj_method_sync(rbd_dev, &rbd_dev->header_oid, &rbd_dev->header_oloc, "get_data_pool", - NULL, 0, &data_pool_id, sizeof(data_pool_id)); + NULL, 0, &data_pool_buf, + sizeof(data_pool_buf)); + dout("%s: rbd_obj_method_sync returned %d\n", __func__, ret); if (ret < 0) return ret; - if (ret < sizeof(data_pool_id)) + if (ret < sizeof(data_pool_buf)) return -EBADMSG; - rbd_dev->header.data_pool_id = le64_to_cpu(data_pool_id); - WARN_ON(rbd_dev->header.data_pool_id == CEPH_NOPOOL); + *data_pool_id = le64_to_cpu(data_pool_buf); + dout(" data_pool_id = %lld\n", *data_pool_id); + WARN_ON(*data_pool_id == CEPH_NOPOOL); + return 0; } @@ -6235,7 +6237,8 @@ out_err: return ret; } -static int rbd_dev_v2_snap_context(struct rbd_device *rbd_dev) +static int rbd_dev_v2_snap_context(struct rbd_device *rbd_dev, + struct ceph_snap_context **psnapc) { size_t size; int ret; @@ -6296,9 +6299,7 @@ static int rbd_dev_v2_snap_context(struct rbd_device *rbd_dev) for (i = 0; i < snap_count; i++) snapc->snaps[i] = ceph_decode_64(&p); - ceph_put_snap_context(rbd_dev->header.snapc); - rbd_dev->header.snapc = snapc; - + *psnapc = snapc; dout(" snap context seq = %llu, snap_count = %u\n", (unsigned long long)seq, (unsigned int)snap_count); out: @@ -6347,38 +6348,42 @@ out: return snap_name; } -static int rbd_dev_v2_header_info(struct rbd_device *rbd_dev) +static int rbd_dev_v2_header_info(struct rbd_device *rbd_dev, + struct rbd_image_header *header, + bool first_time) { - bool first_time = rbd_dev->header.object_prefix == NULL; int ret; - ret = rbd_dev_v2_image_size(rbd_dev); + ret = _rbd_dev_v2_snap_size(rbd_dev, CEPH_NOSNAP, + first_time ? &header->obj_order : NULL, + &header->image_size); if (ret) return ret; if (first_time) { - ret = rbd_dev_v2_header_onetime(rbd_dev); + ret = rbd_dev_v2_header_onetime(rbd_dev, header); if (ret) return ret; } - ret = rbd_dev_v2_snap_context(rbd_dev); - if (ret && first_time) { - kfree(rbd_dev->header.object_prefix); - rbd_dev->header.object_prefix = NULL; - } + ret = rbd_dev_v2_snap_context(rbd_dev, &header->snapc); + if (ret) + return ret; - return ret; + return 0; } -static int rbd_dev_header_info(struct rbd_device *rbd_dev) +static int rbd_dev_header_info(struct rbd_device *rbd_dev, + struct rbd_image_header *header, + bool first_time) { rbd_assert(rbd_image_format_valid(rbd_dev->image_format)); + rbd_assert(!header->object_prefix && !header->snapc); if (rbd_dev->image_format == 1) - return rbd_dev_v1_header_info(rbd_dev); + return rbd_dev_v1_header_info(rbd_dev, header, first_time); - return rbd_dev_v2_header_info(rbd_dev); + return rbd_dev_v2_header_info(rbd_dev, header, first_time); } /* @@ -6728,60 +6733,49 @@ out: */ static void rbd_dev_unprobe(struct rbd_device *rbd_dev) { - struct rbd_image_header *header; - rbd_dev_parent_put(rbd_dev); rbd_object_map_free(rbd_dev); rbd_dev_mapping_clear(rbd_dev); /* Free dynamic fields from the header, then zero it out */ - header = &rbd_dev->header; - ceph_put_snap_context(header->snapc); - kfree(header->snap_sizes); - kfree(header->snap_names); - kfree(header->object_prefix); - memset(header, 0, sizeof (*header)); + rbd_image_header_cleanup(&rbd_dev->header); } -static int rbd_dev_v2_header_onetime(struct rbd_device *rbd_dev) +static int rbd_dev_v2_header_onetime(struct rbd_device *rbd_dev, + struct rbd_image_header *header) { int ret; - ret = rbd_dev_v2_object_prefix(rbd_dev); + ret = rbd_dev_v2_object_prefix(rbd_dev, &header->object_prefix); if (ret) - goto out_err; + return ret; /* * Get the and check features for the image. Currently the * features are assumed to never change. */ - ret = rbd_dev_v2_features(rbd_dev); + ret = _rbd_dev_v2_snap_features(rbd_dev, CEPH_NOSNAP, + &header->features); if (ret) - goto out_err; + return ret; /* If the image supports fancy striping, get its parameters */ - if (rbd_dev->header.features & RBD_FEATURE_STRIPINGV2) { - ret = rbd_dev_v2_striping_info(rbd_dev); - if (ret < 0) - goto out_err; - } - - if (rbd_dev->header.features & RBD_FEATURE_DATA_POOL) { - ret = rbd_dev_v2_data_pool(rbd_dev); + if (header->features & RBD_FEATURE_STRIPINGV2) { + ret = rbd_dev_v2_striping_info(rbd_dev, &header->stripe_unit, + &header->stripe_count); if (ret) - goto out_err; + return ret; } - rbd_init_layout(rbd_dev); - return 0; + if (header->features & RBD_FEATURE_DATA_POOL) { + ret = rbd_dev_v2_data_pool(rbd_dev, &header->data_pool_id); + if (ret) + return ret; + } -out_err: - rbd_dev->header.features = 0; - kfree(rbd_dev->header.object_prefix); - rbd_dev->header.object_prefix = NULL; - return ret; + return 0; } /* @@ -6959,10 +6953,12 @@ static int rbd_dev_image_probe(struct rbd_device *rbd_dev, int depth) if (!depth) down_write(&rbd_dev->header_rwsem); - ret = rbd_dev_header_info(rbd_dev); + ret = rbd_dev_header_info(rbd_dev, &rbd_dev->header, true); if (ret) goto err_out_probe; + rbd_init_layout(rbd_dev); + /* * If this image is the one being mapped, we have pool name and * id, image name and id, and snap name - need to fill snap id. @@ -7022,15 +7018,39 @@ err_out_format: return ret; } +static void rbd_dev_update_header(struct rbd_device *rbd_dev, + struct rbd_image_header *header) +{ + rbd_assert(rbd_image_format_valid(rbd_dev->image_format)); + rbd_assert(rbd_dev->header.object_prefix); /* !first_time */ + + rbd_dev->header.image_size = header->image_size; + + ceph_put_snap_context(rbd_dev->header.snapc); + rbd_dev->header.snapc = header->snapc; + header->snapc = NULL; + + if (rbd_dev->image_format == 1) { + kfree(rbd_dev->header.snap_names); + rbd_dev->header.snap_names = header->snap_names; + header->snap_names = NULL; + + kfree(rbd_dev->header.snap_sizes); + rbd_dev->header.snap_sizes = header->snap_sizes; + header->snap_sizes = NULL; + } +} + static int rbd_dev_refresh(struct rbd_device *rbd_dev) { + struct rbd_image_header header = { 0 }; u64 mapping_size; int ret; down_write(&rbd_dev->header_rwsem); mapping_size = rbd_dev->mapping.size; - ret = rbd_dev_header_info(rbd_dev); + ret = rbd_dev_header_info(rbd_dev, &header, false); if (ret) goto out; @@ -7044,6 +7064,8 @@ static int rbd_dev_refresh(struct rbd_device *rbd_dev) goto out; } + rbd_dev_update_header(rbd_dev, &header); + if (rbd_dev->spec->snap_id == CEPH_NOSNAP) { rbd_dev->mapping.size = rbd_dev->header.image_size; } else { @@ -7056,6 +7078,7 @@ out: if (!ret && mapping_size != rbd_dev->mapping.size) rbd_dev_update_size(rbd_dev); + rbd_image_header_cleanup(&header); return ret; } From bc2a3044015f90453dc885a4a072e38c20ce63b1 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 5 Oct 2023 17:29:52 +0200 Subject: [PATCH 3/8] rbd: decouple parent info read-in from updating rbd_dev commit c10311776f0a8ddea2276df96e255625b07045a8 upstream. Unlike header read-in, parent info read-in is already decoupled in get_parent_info(), but it's buried in rbd_dev_v2_parent_info() along with the processing logic. Separate the initial read-in and update read-in logic into rbd_dev_setup_parent() and rbd_dev_update_parent() respectively and have rbd_dev_v2_parent_info() just populate struct parent_image_info (i.e. what get_parent_info() did). Some existing QoI issues, like flatten of a standalone clone being disregarded on refresh, remain. Signed-off-by: Ilya Dryomov Reviewed-by: Dongsheng Yang [idryomov@gmail.com: backport to 5.4: context] Signed-off-by: Sasha Levin --- drivers/block/rbd.c | 142 +++++++++++++++++++++++++------------------- 1 file changed, 80 insertions(+), 62 deletions(-) diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index 17d802effdc4..e86dca49fae7 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -5762,6 +5762,14 @@ struct parent_image_info { u64 overlap; }; +static void rbd_parent_info_cleanup(struct parent_image_info *pii) +{ + kfree(pii->pool_ns); + kfree(pii->image_id); + + memset(pii, 0, sizeof(*pii)); +} + /* * The caller is responsible for @pii. */ @@ -5831,6 +5839,9 @@ static int __get_parent_info(struct rbd_device *rbd_dev, if (pii->has_overlap) ceph_decode_64_safe(&p, end, pii->overlap, e_inval); + dout("%s pool_id %llu pool_ns %s image_id %s snap_id %llu has_overlap %d overlap %llu\n", + __func__, pii->pool_id, pii->pool_ns, pii->image_id, pii->snap_id, + pii->has_overlap, pii->overlap); return 0; e_inval: @@ -5869,14 +5880,17 @@ static int __get_parent_info_legacy(struct rbd_device *rbd_dev, pii->has_overlap = true; ceph_decode_64_safe(&p, end, pii->overlap, e_inval); + dout("%s pool_id %llu pool_ns %s image_id %s snap_id %llu has_overlap %d overlap %llu\n", + __func__, pii->pool_id, pii->pool_ns, pii->image_id, pii->snap_id, + pii->has_overlap, pii->overlap); return 0; e_inval: return -EINVAL; } -static int get_parent_info(struct rbd_device *rbd_dev, - struct parent_image_info *pii) +static int rbd_dev_v2_parent_info(struct rbd_device *rbd_dev, + struct parent_image_info *pii) { struct page *req_page, *reply_page; void *p; @@ -5904,7 +5918,7 @@ static int get_parent_info(struct rbd_device *rbd_dev, return ret; } -static int rbd_dev_v2_parent_info(struct rbd_device *rbd_dev) +static int rbd_dev_setup_parent(struct rbd_device *rbd_dev) { struct rbd_spec *parent_spec; struct parent_image_info pii = { 0 }; @@ -5914,37 +5928,12 @@ static int rbd_dev_v2_parent_info(struct rbd_device *rbd_dev) if (!parent_spec) return -ENOMEM; - ret = get_parent_info(rbd_dev, &pii); + ret = rbd_dev_v2_parent_info(rbd_dev, &pii); if (ret) goto out_err; - dout("%s pool_id %llu pool_ns %s image_id %s snap_id %llu has_overlap %d overlap %llu\n", - __func__, pii.pool_id, pii.pool_ns, pii.image_id, pii.snap_id, - pii.has_overlap, pii.overlap); - - if (pii.pool_id == CEPH_NOPOOL || !pii.has_overlap) { - /* - * Either the parent never existed, or we have - * record of it but the image got flattened so it no - * longer has a parent. When the parent of a - * layered image disappears we immediately set the - * overlap to 0. The effect of this is that all new - * requests will be treated as if the image had no - * parent. - * - * If !pii.has_overlap, the parent image spec is not - * applicable. It's there to avoid duplication in each - * snapshot record. - */ - if (rbd_dev->parent_overlap) { - rbd_dev->parent_overlap = 0; - rbd_dev_parent_put(rbd_dev); - pr_info("%s: clone image has been flattened\n", - rbd_dev->disk->disk_name); - } - + if (pii.pool_id == CEPH_NOPOOL || !pii.has_overlap) goto out; /* No parent? No problem. */ - } /* The ceph file layout needs to fit pool id in 32 bits */ @@ -5956,46 +5945,34 @@ static int rbd_dev_v2_parent_info(struct rbd_device *rbd_dev) } /* - * The parent won't change (except when the clone is - * flattened, already handled that). So we only need to - * record the parent spec we have not already done so. + * The parent won't change except when the clone is flattened, + * so we only need to record the parent image spec once. */ - if (!rbd_dev->parent_spec) { - parent_spec->pool_id = pii.pool_id; - if (pii.pool_ns && *pii.pool_ns) { - parent_spec->pool_ns = pii.pool_ns; - pii.pool_ns = NULL; - } - parent_spec->image_id = pii.image_id; - pii.image_id = NULL; - parent_spec->snap_id = pii.snap_id; - - rbd_dev->parent_spec = parent_spec; - parent_spec = NULL; /* rbd_dev now owns this */ + parent_spec->pool_id = pii.pool_id; + if (pii.pool_ns && *pii.pool_ns) { + parent_spec->pool_ns = pii.pool_ns; + pii.pool_ns = NULL; } + parent_spec->image_id = pii.image_id; + pii.image_id = NULL; + parent_spec->snap_id = pii.snap_id; + + rbd_assert(!rbd_dev->parent_spec); + rbd_dev->parent_spec = parent_spec; + parent_spec = NULL; /* rbd_dev now owns this */ /* - * We always update the parent overlap. If it's zero we issue - * a warning, as we will proceed as if there was no parent. + * Record the parent overlap. If it's zero, issue a warning as + * we will proceed as if there is no parent. */ - if (!pii.overlap) { - if (parent_spec) { - /* refresh, careful to warn just once */ - if (rbd_dev->parent_overlap) - rbd_warn(rbd_dev, - "clone now standalone (overlap became 0)"); - } else { - /* initial probe */ - rbd_warn(rbd_dev, "clone is standalone (overlap 0)"); - } - } + if (!pii.overlap) + rbd_warn(rbd_dev, "clone is standalone (overlap 0)"); rbd_dev->parent_overlap = pii.overlap; out: ret = 0; out_err: - kfree(pii.pool_ns); - kfree(pii.image_id); + rbd_parent_info_cleanup(&pii); rbd_spec_put(parent_spec); return ret; } @@ -6992,7 +6969,7 @@ static int rbd_dev_image_probe(struct rbd_device *rbd_dev, int depth) } if (rbd_dev->header.features & RBD_FEATURE_LAYERING) { - ret = rbd_dev_v2_parent_info(rbd_dev); + ret = rbd_dev_setup_parent(rbd_dev); if (ret) goto err_out_probe; } @@ -7041,9 +7018,47 @@ static void rbd_dev_update_header(struct rbd_device *rbd_dev, } } +static void rbd_dev_update_parent(struct rbd_device *rbd_dev, + struct parent_image_info *pii) +{ + if (pii->pool_id == CEPH_NOPOOL || !pii->has_overlap) { + /* + * Either the parent never existed, or we have + * record of it but the image got flattened so it no + * longer has a parent. When the parent of a + * layered image disappears we immediately set the + * overlap to 0. The effect of this is that all new + * requests will be treated as if the image had no + * parent. + * + * If !pii.has_overlap, the parent image spec is not + * applicable. It's there to avoid duplication in each + * snapshot record. + */ + if (rbd_dev->parent_overlap) { + rbd_dev->parent_overlap = 0; + rbd_dev_parent_put(rbd_dev); + pr_info("%s: clone has been flattened\n", + rbd_dev->disk->disk_name); + } + } else { + rbd_assert(rbd_dev->parent_spec); + + /* + * Update the parent overlap. If it became zero, issue + * a warning as we will proceed as if there is no parent. + */ + if (!pii->overlap && rbd_dev->parent_overlap) + rbd_warn(rbd_dev, + "clone has become standalone (overlap 0)"); + rbd_dev->parent_overlap = pii->overlap; + } +} + static int rbd_dev_refresh(struct rbd_device *rbd_dev) { struct rbd_image_header header = { 0 }; + struct parent_image_info pii = { 0 }; u64 mapping_size; int ret; @@ -7059,12 +7074,14 @@ static int rbd_dev_refresh(struct rbd_device *rbd_dev) * mapped image getting flattened. */ if (rbd_dev->parent) { - ret = rbd_dev_v2_parent_info(rbd_dev); + ret = rbd_dev_v2_parent_info(rbd_dev, &pii); if (ret) goto out; } rbd_dev_update_header(rbd_dev, &header); + if (rbd_dev->parent) + rbd_dev_update_parent(rbd_dev, &pii); if (rbd_dev->spec->snap_id == CEPH_NOSNAP) { rbd_dev->mapping.size = rbd_dev->header.image_size; @@ -7078,6 +7095,7 @@ out: if (!ret && mapping_size != rbd_dev->mapping.size) rbd_dev_update_size(rbd_dev); + rbd_parent_info_cleanup(&pii); rbd_image_header_cleanup(&header); return ret; } From 6e37de4a1407e635d293a7593c9f881ba1cc11f9 Mon Sep 17 00:00:00 2001 From: Ilya Dryomov Date: Thu, 5 Oct 2023 17:29:53 +0200 Subject: [PATCH 4/8] rbd: take header_rwsem in rbd_dev_refresh() only when updating commit 0b207d02bd9ab8dcc31b262ca9f60dbc1822500d upstream. rbd_dev_refresh() has been holding header_rwsem across header and parent info read-in unnecessarily for ages. With commit 870611e4877e ("rbd: get snapshot context after exclusive lock is ensured to be held"), the potential for deadlocks became much more real owning to a) header_rwsem now nesting inside lock_rwsem and b) rw_semaphores not allowing new readers after a writer is registered. For example, assuming that I/O request 1, I/O request 2 and header read-in request all target the same OSD: 1. I/O request 1 comes in and gets submitted 2. watch error occurs 3. rbd_watch_errcb() takes lock_rwsem for write, clears owner_cid and releases lock_rwsem 4. after reestablishing the watch, rbd_reregister_watch() calls rbd_dev_refresh() which takes header_rwsem for write and submits a header read-in request 5. I/O request 2 comes in: after taking lock_rwsem for read in __rbd_img_handle_request(), it blocks trying to take header_rwsem for read in rbd_img_object_requests() 6. another watch error occurs 7. rbd_watch_errcb() blocks trying to take lock_rwsem for write 8. I/O request 1 completion is received by the messenger but can't be processed because lock_rwsem won't be granted anymore 9. header read-in request completion can't be received, let alone processed, because the messenger is stranded Change rbd_dev_refresh() to take header_rwsem only for actually updating rbd_dev->header. Header and parent info read-in don't need any locking. Cc: stable@vger.kernel.org # 0b035401c570: rbd: move rbd_dev_refresh() definition Cc: stable@vger.kernel.org # 510a7330c82a: rbd: decouple header read-in from updating rbd_dev->header Cc: stable@vger.kernel.org # c10311776f0a: rbd: decouple parent info read-in from updating rbd_dev Cc: stable@vger.kernel.org Fixes: 870611e4877e ("rbd: get snapshot context after exclusive lock is ensured to be held") Signed-off-by: Ilya Dryomov Reviewed-by: Dongsheng Yang [idryomov@gmail.com: backport to 5.4: open-code rbd_is_snap(), preserve rbd_exists_validate() call] Signed-off-by: Sasha Levin --- drivers/block/rbd.c | 31 ++++++++++++++++--------------- 1 file changed, 16 insertions(+), 15 deletions(-) diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c index e86dca49fae7..7117fa490243 100644 --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -7001,7 +7001,19 @@ static void rbd_dev_update_header(struct rbd_device *rbd_dev, rbd_assert(rbd_image_format_valid(rbd_dev->image_format)); rbd_assert(rbd_dev->header.object_prefix); /* !first_time */ - rbd_dev->header.image_size = header->image_size; + if (rbd_dev->header.image_size != header->image_size) { + rbd_dev->header.image_size = header->image_size; + + if (rbd_dev->spec->snap_id == CEPH_NOSNAP) { + rbd_dev->mapping.size = header->image_size; + rbd_dev_update_size(rbd_dev); + } + } + + if (rbd_dev->spec->snap_id != CEPH_NOSNAP) { + /* validate mapped snapshot's EXISTS flag */ + rbd_exists_validate(rbd_dev); + } ceph_put_snap_context(rbd_dev->header.snapc); rbd_dev->header.snapc = header->snapc; @@ -7059,11 +7071,9 @@ static int rbd_dev_refresh(struct rbd_device *rbd_dev) { struct rbd_image_header header = { 0 }; struct parent_image_info pii = { 0 }; - u64 mapping_size; int ret; - down_write(&rbd_dev->header_rwsem); - mapping_size = rbd_dev->mapping.size; + dout("%s rbd_dev %p\n", __func__, rbd_dev); ret = rbd_dev_header_info(rbd_dev, &header, false); if (ret) @@ -7079,22 +7089,13 @@ static int rbd_dev_refresh(struct rbd_device *rbd_dev) goto out; } + down_write(&rbd_dev->header_rwsem); rbd_dev_update_header(rbd_dev, &header); if (rbd_dev->parent) rbd_dev_update_parent(rbd_dev, &pii); - - if (rbd_dev->spec->snap_id == CEPH_NOSNAP) { - rbd_dev->mapping.size = rbd_dev->header.image_size; - } else { - /* validate mapped snapshot's EXISTS flag */ - rbd_exists_validate(rbd_dev); - } + up_write(&rbd_dev->header_rwsem); out: - up_write(&rbd_dev->header_rwsem); - if (!ret && mapping_size != rbd_dev->mapping.size) - rbd_dev_update_size(rbd_dev); - rbd_parent_info_cleanup(&pii); rbd_image_header_cleanup(&header); return ret; From e60272ab021cf7da0264f7fd7cee410aac519358 Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Sat, 7 Oct 2023 13:57:34 +0200 Subject: [PATCH 5/8] Revert "PCI: qcom: Disable write access to read only registers for IP v2.3.3" This reverts commit 35c95eda7b6d9883d1cc9bb1f89d454baa140ebc which is commit a33d700e8eea76c62120cb3dbf5e01328f18319a upstream. It was applied to the incorrect function as the original function the commit changed is not in this kernel branch. Reported-by: Ben Hutchings Link: https://lore.kernel.org/r/f23affddab4d8b3cc07508f2d8735d88d823821d.camel@decadent.org.uk Signed-off-by: Greg Kroah-Hartman --- drivers/pci/controller/dwc/pcie-qcom.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c index 24dbb6968831..17f411772f0c 100644 --- a/drivers/pci/controller/dwc/pcie-qcom.c +++ b/drivers/pci/controller/dwc/pcie-qcom.c @@ -807,8 +807,6 @@ static int qcom_pcie_get_resources_2_4_0(struct qcom_pcie *pcie) return PTR_ERR(res->phy_ahb_reset); } - dw_pcie_dbi_ro_wr_dis(pci); - return 0; } From 810248a12999f12757b4087c674c2b749d8c7b76 Mon Sep 17 00:00:00 2001 From: Dinghao Liu Date: Sat, 23 Sep 2023 18:37:23 +0800 Subject: [PATCH 6/8] scsi: zfcp: Fix a double put in zfcp_port_enqueue() commit b481f644d9174670b385c3a699617052cd2a79d3 upstream. When device_register() fails, zfcp_port_release() will be called after put_device(). As a result, zfcp_ccw_adapter_put() will be called twice: one in zfcp_port_release() and one in the error path after device_register(). So the reference on the adapter object is doubly put, which may lead to a premature free. Fix this by adjusting the error tag after device_register(). Fixes: f3450c7b9172 ("[SCSI] zfcp: Replace local reference counting with common kref") Signed-off-by: Dinghao Liu Link: https://lore.kernel.org/r/20230923103723.10320-1-dinghao.liu@zju.edu.cn Acked-by: Benjamin Block Cc: stable@vger.kernel.org # v2.6.33+ Signed-off-by: Martin K. Petersen Signed-off-by: Greg Kroah-Hartman --- drivers/s390/scsi/zfcp_aux.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/drivers/s390/scsi/zfcp_aux.c b/drivers/s390/scsi/zfcp_aux.c index e390f8c6d5f3..0d57432066ca 100644 --- a/drivers/s390/scsi/zfcp_aux.c +++ b/drivers/s390/scsi/zfcp_aux.c @@ -488,12 +488,12 @@ struct zfcp_port *zfcp_port_enqueue(struct zfcp_adapter *adapter, u64 wwpn, if (port) { put_device(&port->dev); retval = -EEXIST; - goto err_out; + goto err_put; } port = kzalloc(sizeof(struct zfcp_port), GFP_KERNEL); if (!port) - goto err_out; + goto err_put; rwlock_init(&port->unit_list_lock); INIT_LIST_HEAD(&port->unit_list); @@ -516,7 +516,7 @@ struct zfcp_port *zfcp_port_enqueue(struct zfcp_adapter *adapter, u64 wwpn, if (dev_set_name(&port->dev, "0x%016llx", (unsigned long long)wwpn)) { kfree(port); - goto err_out; + goto err_put; } retval = -EINVAL; @@ -533,7 +533,8 @@ struct zfcp_port *zfcp_port_enqueue(struct zfcp_adapter *adapter, u64 wwpn, return port; -err_out: +err_put: zfcp_ccw_adapter_put(adapter); +err_out: return ERR_PTR(retval); } From 2e608cede0ae752480b7521b3a17007acf30d338 Mon Sep 17 00:00:00 2001 From: "Gustavo A. R. Silva" Date: Sat, 23 Sep 2023 19:15:59 -0600 Subject: [PATCH 7/8] qed/red_ll2: Fix undefined behavior bug in struct qed_ll2_info commit eea03d18af9c44235865a4bc9bec4d780ef6cf21 upstream. The flexible structure (a structure that contains a flexible-array member at the end) `qed_ll2_tx_packet` is nested within the second layer of `struct qed_ll2_info`: struct qed_ll2_tx_packet { ... /* Flexible Array of bds_set determined by max_bds_per_packet */ struct { struct core_tx_bd *txq_bd; dma_addr_t tx_frag; u16 frag_len; } bds_set[]; }; struct qed_ll2_tx_queue { ... struct qed_ll2_tx_packet cur_completing_packet; }; struct qed_ll2_info { ... struct qed_ll2_tx_queue tx_queue; struct qed_ll2_cbs cbs; }; The problem is that member `cbs` in `struct qed_ll2_info` is placed just after an object of type `struct qed_ll2_tx_queue`, which is in itself an implicit flexible structure, which by definition ends in a flexible array member, in this case `bds_set`. This causes an undefined behavior bug at run-time when dynamic memory is allocated for `bds_set`, which could lead to a serious issue if `cbs` in `struct qed_ll2_info` is overwritten by the contents of `bds_set`. Notice that the type of `cbs` is a structure full of function pointers (and a cookie :) ): include/linux/qed/qed_ll2_if.h: 107 typedef 108 void (*qed_ll2_complete_rx_packet_cb)(void *cxt, 109 struct qed_ll2_comp_rx_data *data); 110 111 typedef 112 void (*qed_ll2_release_rx_packet_cb)(void *cxt, 113 u8 connection_handle, 114 void *cookie, 115 dma_addr_t rx_buf_addr, 116 bool b_last_packet); 117 118 typedef 119 void (*qed_ll2_complete_tx_packet_cb)(void *cxt, 120 u8 connection_handle, 121 void *cookie, 122 dma_addr_t first_frag_addr, 123 bool b_last_fragment, 124 bool b_last_packet); 125 126 typedef 127 void (*qed_ll2_release_tx_packet_cb)(void *cxt, 128 u8 connection_handle, 129 void *cookie, 130 dma_addr_t first_frag_addr, 131 bool b_last_fragment, bool b_last_packet); 132 133 typedef 134 void (*qed_ll2_slowpath_cb)(void *cxt, u8 connection_handle, 135 u32 opaque_data_0, u32 opaque_data_1); 136 137 struct qed_ll2_cbs { 138 qed_ll2_complete_rx_packet_cb rx_comp_cb; 139 qed_ll2_release_rx_packet_cb rx_release_cb; 140 qed_ll2_complete_tx_packet_cb tx_comp_cb; 141 qed_ll2_release_tx_packet_cb tx_release_cb; 142 qed_ll2_slowpath_cb slowpath_cb; 143 void *cookie; 144 }; Fix this by moving the declaration of `cbs` to the middle of its containing structure `qed_ll2_info`, preventing it from being overwritten by the contents of `bds_set` at run-time. This bug was introduced in 2017, when `bds_set` was converted to a one-element array, and started to be used as a Variable Length Object (VLO) at run-time. Fixes: f5823fe6897c ("qed: Add ll2 option to limit the number of bds per packet") Cc: stable@vger.kernel.org Signed-off-by: Gustavo A. R. Silva Reviewed-by: Kees Cook Reviewed-by: Simon Horman Link: https://lore.kernel.org/r/ZQ+Nz8DfPg56pIzr@work Signed-off-by: Paolo Abeni Signed-off-by: Greg Kroah-Hartman --- drivers/net/ethernet/qlogic/qed/qed_ll2.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/ethernet/qlogic/qed/qed_ll2.h b/drivers/net/ethernet/qlogic/qed/qed_ll2.h index 5f01fbd3c073..1e0e2dbecc76 100644 --- a/drivers/net/ethernet/qlogic/qed/qed_ll2.h +++ b/drivers/net/ethernet/qlogic/qed/qed_ll2.h @@ -123,9 +123,9 @@ struct qed_ll2_info { enum core_tx_dest tx_dest; u8 tx_stats_en; bool main_func_queue; + struct qed_ll2_cbs cbs; struct qed_ll2_rx_queue rx_queue; struct qed_ll2_tx_queue tx_queue; - struct qed_ll2_cbs cbs; }; /** From 3c4bfa7a56c8b363511c90f95159e6f1b68a434a Mon Sep 17 00:00:00 2001 From: "Gustavo A. R. Silva" Date: Thu, 24 Aug 2023 21:06:51 -0600 Subject: [PATCH 8/8] wifi: mwifiex: Fix tlv_buf_left calculation commit eec679e4ac5f47507774956fb3479c206e761af7 upstream. In a TLV encoding scheme, the Length part represents the length after the header containing the values for type and length. In this case, `tlv_len` should be: tlv_len == (sizeof(*tlv_rxba) - 1) - sizeof(tlv_rxba->header) + tlv_bitmap_len Notice that the `- 1` accounts for the one-element array `bitmap`, which 1-byte size is already included in `sizeof(*tlv_rxba)`. So, if the above is correct, there is a double-counting of some members in `struct mwifiex_ie_types_rxba_sync`, when `tlv_buf_left` and `tmp` are calculated: 968 tlv_buf_left -= (sizeof(*tlv_rxba) + tlv_len); 969 tmp = (u8 *)tlv_rxba + tlv_len + sizeof(*tlv_rxba); in specific, members: drivers/net/wireless/marvell/mwifiex/fw.h:777 777 u8 mac[ETH_ALEN]; 778 u8 tid; 779 u8 reserved; 780 __le16 seq_num; 781 __le16 bitmap_len; This is clearly wrong, and affects the subsequent decoding of data in `event_buf` through `tlv_rxba`: 970 tlv_rxba = (struct mwifiex_ie_types_rxba_sync *)tmp; Fix this by using `sizeof(tlv_rxba->header)` instead of `sizeof(*tlv_rxba)` in the calculation of `tlv_buf_left` and `tmp`. This results in the following binary differences before/after changes: | drivers/net/wireless/marvell/mwifiex/11n_rxreorder.o | @@ -4698,11 +4698,11 @@ | drivers/net/wireless/marvell/mwifiex/11n_rxreorder.c:968 | tlv_buf_left -= (sizeof(tlv_rxba->header) + tlv_len); | - 1da7: lea -0x11(%rbx),%edx | + 1da7: lea -0x4(%rbx),%edx | 1daa: movzwl %bp,%eax | drivers/net/wireless/marvell/mwifiex/11n_rxreorder.c:969 | tmp = (u8 *)tlv_rxba + sizeof(tlv_rxba->header) + tlv_len; | - 1dad: lea 0x11(%r15,%rbp,1),%r15 | + 1dad: lea 0x4(%r15,%rbp,1),%r15 The above reflects the desired change: avoid counting 13 too many bytes; which is the total size of the double-counted members in `struct mwifiex_ie_types_rxba_sync`: $ pahole -C mwifiex_ie_types_rxba_sync drivers/net/wireless/marvell/mwifiex/11n_rxreorder.o struct mwifiex_ie_types_rxba_sync { struct mwifiex_ie_types_header header; /* 0 4 */ |----------------------------------------------------------------------- | u8 mac[6]; /* 4 6 */ | | u8 tid; /* 10 1 */ | | u8 reserved; /* 11 1 */ | | __le16 seq_num; /* 12 2 */ | | __le16 bitmap_len; /* 14 2 */ | | u8 bitmap[1]; /* 16 1 */ | |----------------------------------------------------------------------| | 13 bytes| ----------- /* size: 17, cachelines: 1, members: 7 */ /* last cacheline: 17 bytes */ } __attribute__((__packed__)); Fixes: 99ffe72cdae4 ("mwifiex: process rxba_sync event") Cc: stable@vger.kernel.org Signed-off-by: Gustavo A. R. Silva Reviewed-by: Kees Cook Signed-off-by: Kalle Valo Link: https://lore.kernel.org/r/06668edd68e7a26bbfeebd1201ae077a2a7a8bce.1692931954.git.gustavoars@kernel.org Signed-off-by: Greg Kroah-Hartman --- drivers/net/wireless/marvell/mwifiex/11n_rxreorder.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/net/wireless/marvell/mwifiex/11n_rxreorder.c b/drivers/net/wireless/marvell/mwifiex/11n_rxreorder.c index 05a3c61ac603..793be2835134 100644 --- a/drivers/net/wireless/marvell/mwifiex/11n_rxreorder.c +++ b/drivers/net/wireless/marvell/mwifiex/11n_rxreorder.c @@ -977,8 +977,8 @@ void mwifiex_11n_rxba_sync_event(struct mwifiex_private *priv, } } - tlv_buf_left -= (sizeof(*tlv_rxba) + tlv_len); - tmp = (u8 *)tlv_rxba + tlv_len + sizeof(*tlv_rxba); + tlv_buf_left -= (sizeof(tlv_rxba->header) + tlv_len); + tmp = (u8 *)tlv_rxba + sizeof(tlv_rxba->header) + tlv_len; tlv_rxba = (struct mwifiex_ie_types_rxba_sync *)tmp; } }