kernel.git/fs/erofs, branch linux-5.19.y

erofs: use kill_anon_super() to kill super in fscache mode

2022-10-24T07:57:02Z

[ Upstream commit 1015c1016c231b26d4e2c9b3da65b6c043eb97a3 ] Use kill_anon_super() instead of generic_shutdown_super() since the mount() in erofs fscache mode uses get_tree_nodev() and associated anon bdev needs to be freed. Fixes: 9c0cc9c729657 ("erofs: add 'fsid' mount option") Suggested-by: Jingbo Xu Signed-off-by: Jia Zhu Reviewed-by: Jingbo Xu Link: https://lore.kernel.org/r/20220918043456.147-2-zhujia.zj@bytedance.com Signed-off-by: Gao Xiang Signed-off-by: Sasha Levin

erofs: fix order >= MAX_ORDER warning due to crafted negative i_size

2022-10-24T07:57:02Z

[ Upstream commit 1dd73601a1cba37a0ed5f89a8662c90191df5873 ] As syzbot reported [1], the root cause is that i_size field is a signed type, and negative i_size is also less than EROFS_BLKSIZ. As a consequence, it's handled as fast symlink unexpectedly. Let's fall back to the generic path to deal with such unusual i_size. [1] https://lore.kernel.org/r/000000000000ac8efa05e7feaa1f@google.com Reported-by: syzbot+f966c13b1b4fc0403b19@syzkaller.appspotmail.com Fixes: 431339ba9042 ("staging: erofs: add inode operations") Reviewed-by: Yue Hu Link: https://lore.kernel.org/r/20220909023948.28925-1-hsiangkao@linux.alibaba.com Signed-off-by: Gao Xiang Signed-off-by: Sasha Levin

erofs: fix pcluster use-after-free on UP platforms

2022-09-15T08:47:16Z

[ Upstream commit 2f44013e39984c127c6efedf70e6b5f4e9dcf315 ] During stress testing with CONFIG_SMP disabled, KASAN reports as below: ================================================================== BUG: KASAN: use-after-free in __mutex_lock+0xe5/0xc30 Read of size 8 at addr ffff8881094223f8 by task stress/7789 CPU: 0 PID: 7789 Comm: stress Not tainted 6.0.0-rc1-00002-g0d53d2e882f9 #3 Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011 Call Trace: .. __mutex_lock+0xe5/0xc30 .. z_erofs_do_read_page+0x8ce/0x1560 .. z_erofs_readahead+0x31c/0x580 .. Freed by task 7787 kasan_save_stack+0x1e/0x40 kasan_set_track+0x20/0x30 kasan_set_free_info+0x20/0x40 __kasan_slab_free+0x10c/0x190 kmem_cache_free+0xed/0x380 rcu_core+0x3d5/0xc90 __do_softirq+0x12d/0x389 Last potentially related work creation: kasan_save_stack+0x1e/0x40 __kasan_record_aux_stack+0x97/0xb0 call_rcu+0x3d/0x3f0 erofs_shrink_workstation+0x11f/0x210 erofs_shrink_scan+0xdc/0x170 shrink_slab.constprop.0+0x296/0x530 drop_slab+0x1c/0x70 drop_caches_sysctl_handler+0x70/0x80 proc_sys_call_handler+0x20a/0x2f0 vfs_write+0x555/0x6c0 ksys_write+0xbe/0x160 do_syscall_64+0x3b/0x90 The root cause is that erofs_workgroup_unfreeze() doesn't reset to orig_val thus it causes a race that the pcluster reuses unexpectedly before freeing. Since UP platforms are quite rare now, such path becomes unnecessary. Let's drop such specific-designed path directly instead. Fixes: 73f5c66df3e2 ("staging: erofs: fix `erofs_workgroup_{try_to_freeze, unfreeze}'") Reviewed-by: Yue Hu Reviewed-by: Chao Yu Link: https://lore.kernel.org/r/20220902045710.109530-1-hsiangkao@linux.alibaba.com Signed-off-by: Gao Xiang Signed-off-by: Sasha Levin

erofs: fix error return code in erofs_fscache_{meta_,}read_folio

2022-09-15T08:47:16Z

[ Upstream commit 5bd9628b784cc5e38e1c7ebb680bbd6ee741230e ] If erofs_fscache_alloc_request fail and then goto out, it will return 0. it should return a negative error code instead of 0. Fixes: d435d53228dd ("erofs: change to use asynchronous io for fscache readpage/readahead") Signed-off-by: Sun Ke Reviewed-by: Jingbo Xu Reviewed-by: Gao Xiang Reviewed-by: Chao Yu Link: https://lore.kernel.org/r/20220815034829.3940803-1-sunke32@huawei.com Signed-off-by: Gao Xiang Signed-off-by: Sasha Levin

erofs: update ctx->pos for every emitted dirent

2022-08-17T13:14:21Z

[ Upstream commit ecce9212d0fd7a2d4a4998f0c4623a66887e14c8 ] erofs_readdir update ctx->pos after filling a batch of dentries and it may cause dir/files duplication for NFS readdirplus which depends on ctx->pos to fill dir correctly. So update ctx->pos for every emitted dirent in erofs_fill_dentries to fix it. Also fix the update of ctx->pos when the initial file position has exceeded nameoff. Fixes: 3e917cc305c6 ("erofs: make filesystem exportable") Signed-off-by: Hongnan Li Signed-off-by: Jeffle Xu Reviewed-by: Gao Xiang Reviewed-by: Chao Yu Link: https://lore.kernel.org/r/20220722082732.30935-1-jefflexu@linux.alibaba.com Signed-off-by: Gao Xiang Signed-off-by: Sasha Levin

erofs: avoid consecutive detection for Highmem memory

2022-08-17T13:14:16Z

[ Upstream commit 448b5a1548d87c246c3d0c3df8480d3c6eb6c11a ] Currently, vmap()s are avoided if physical addresses are consecutive for decompressed buffers. I observed that is very common for 4KiB pclusters since the numbers of decompressed pages are almost 2 or 3. However, such detection doesn't work for Highmem pages on 32-bit machines, let's fix it now. Reported-by: Liu Jinbao Fixes: 7fc45dbc938a ("staging: erofs: introduce generic decompression backend") Link: https://lore.kernel.org/r/20220708101001.21242-1-hsiangkao@linux.alibaba.com Signed-off-by: Gao Xiang Signed-off-by: Sasha Levin

erofs: wake up all waiters after z_erofs_lzma_head ready

2022-08-17T13:14:16Z

[ Upstream commit 2df7c4bd7c1d2bc5ece5e9ed19dbd386810c2a65 ] When the user mounts the erofs second times, the decompression thread may hung. The problem happens due to a sequence of steps like the following: 1) Task A called z_erofs_load_lzma_config which obtain all of the node from the z_erofs_lzma_head. 2) At this time, task B called the z_erofs_lzma_decompress and wanted to get a node. But the z_erofs_lzma_head was empty, the Task B had to sleep. 3) Task A release nodes and push nodes into the z_erofs_lzma_head. But task B was still sleeping. One example report when the hung happens: task:kworker/u3:1 state:D stack:14384 pid: 86 ppid: 2 flags:0x00004000 Workqueue: erofs_unzipd z_erofs_decompressqueue_work Call Trace: __schedule+0x281/0x760 schedule+0x49/0xb0 z_erofs_lzma_decompress+0x4bc/0x580 ? cpu_core_flags+0x10/0x10 z_erofs_decompress_pcluster+0x49b/0xba0 ? __update_load_avg_se+0x2b0/0x330 ? __update_load_avg_se+0x2b0/0x330 ? update_load_avg+0x5f/0x690 ? update_load_avg+0x5f/0x690 ? set_next_entity+0xbd/0x110 ? _raw_spin_unlock+0xd/0x20 z_erofs_decompress_queue.isra.0+0x2e/0x50 z_erofs_decompressqueue_work+0x30/0x60 process_one_work+0x1d3/0x3a0 worker_thread+0x45/0x3a0 ? process_one_work+0x3a0/0x3a0 kthread+0xe2/0x110 ? kthread_complete_and_exit+0x20/0x20 ret_from_fork+0x22/0x30 Signed-off-by: Yuwen Chen Fixes: 622ceaddb764 ("erofs: lzma compression support") Reviewed-by: Gao Xiang Link: https://lore.kernel.org/r/20220626224041.4288-1-chenyuwen1@meizu.com Signed-off-by: Gao Xiang Signed-off-by: Sasha Levin

Merge tag 'erofs-for-5.19-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs

2022-06-01T18:54:29Z

Pull more erofs updates from Gao Xiang: "This is a follow-up to the main updates, including some fixes of fscache mode related to compressed inodes and a cachefiles tracepoint. There is also a patch to fix an unexpected decompression strategy change due to a cleanup in the past. All the fixes are quite small. Apart from these, documentation is also updated for a better description of recent new features. In addition, this has some trivial cleanups without actual code logic changes, so I could have a more recent codebase to work on folios and avoiding the PG_error page flag for the next cycle. Summary: - Leave compressed inodes unsupported in fscache mode for now - Avoid crash when using tracepoint cachefiles_prep_read - Fix `backmost' behavior due to a recent cleanup - Update documentation for better description of recent new features - Several decompression cleanups w/o logical change" * tag 'erofs-for-5.19-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs: erofs: fix 'backmost' member of z_erofs_decompress_frontend erofs: simplify z_erofs_pcluster_readmore() erofs: get rid of label `restart_now' erofs: get rid of `struct z_erofs_collection' erofs: update documentation erofs: fix crash when enable tracepoint cachefiles_prep_read erofs: leave compressed inodes unsupported in fscache mode for now

erofs: fix 'backmost' member of z_erofs_decompress_frontend

2022-05-31T15:15:30Z

Initialize 'backmost' to true in DECOMPRESS_FRONTEND_INIT. Fixes: 5c6dcc57e2e5 ("erofs: get rid of `struct z_erofs_collector'") Signed-off-by: Weizhao Ouyang Reviewed-by: Gao Xiang Reviewed-by: Yue Hu Reviewed-by: Chao Yu Link: https://lore.kernel.org/r/20220530075114.918874-1-o451686892@gmail.com Signed-off-by: Gao Xiang

erofs: simplify z_erofs_pcluster_readmore()

2022-05-31T15:15:21Z

Get rid of unnecessary label `skip'. No logic changes. Link: https://lore.kernel.org/r/20220529055425.226363-4-xiang@kernel.org Acked-by: Chao Yu Signed-off-by: Gao Xiang