From ecf84096a526f2632ee85c32a3d05de3fa60ce80 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 9 Apr 2020 13:33:05 +0200 Subject: ubifs: remove broken lazytime support When "ubifs: introduce UBIFS_ATIME_SUPPORT to ubifs" introduced atime support to ubifs, it also added lazytime support. As far as I can tell the lazytime support is terminally broken, as it causes mark_inode_dirty_sync to be called from __writeback_single_inode, which will then trigger the locking assert in ubifs_dirty_inode. Just remove the broken lazytime support for now, it can be added back later, especially as some infrastructure changes should make that easier soon. Fixes: 8c1c5f263833 ("ubifs: introduce UBIFS_ATIME_SUPPORT to ubifs") Signed-off-by: Christoph Hellwig Signed-off-by: Richard Weinberger --- fs/ubifs/file.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c index 743928efffc1..49fe062ce45e 100644 --- a/fs/ubifs/file.c +++ b/fs/ubifs/file.c @@ -1375,7 +1375,6 @@ int ubifs_update_time(struct inode *inode, struct timespec64 *time, struct ubifs_info *c = inode->i_sb->s_fs_info; struct ubifs_budget_req req = { .dirtied_ino = 1, .dirtied_ino_d = ALIGN(ui->data_len, 8) }; - int iflags = I_DIRTY_TIME; int err, release; if (!IS_ENABLED(CONFIG_UBIFS_ATIME_SUPPORT)) @@ -1393,11 +1392,8 @@ int ubifs_update_time(struct inode *inode, struct timespec64 *time, if (flags & S_MTIME) inode->i_mtime = *time; - if (!(inode->i_sb->s_flags & SB_LAZYTIME)) - iflags |= I_DIRTY_SYNC; - release = ui->dirty; - __mark_inode_dirty(inode, iflags); + __mark_inode_dirty(inode, I_DIRTY_SYNC); mutex_unlock(&ui->ui_mutex); if (release) ubifs_release_budget(c, &req); -- cgit v1.2.3 From e67f0216939c048f02fe58dc1113738380480061 Mon Sep 17 00:00:00 2001 From: Vivek Goyal Date: Wed, 22 Apr 2020 09:08:49 -0400 Subject: ovl: clear ATTR_FILE from attr->ia_valid ovl_setattr() can be passed an attr which has ATTR_FILE set and attr->ia_file is a file pointer to overlay file. This is done in open(O_TRUNC) path. We should either replace with attr->ia_file with underlying file object or clear ATTR_FILE so that underlying filesystem does not end up using overlayfs file object pointer. There are no good use cases yet so for now clear ATTR_FILE. fuse seems to be one user which can use this. But it can work even without this. So it is not mandatory to pass ATTR_FILE to fuse. Signed-off-by: Vivek Goyal Fixes: bccece1ead36 ("ovl: allow remote upper") Signed-off-by: Miklos Szeredi --- fs/overlayfs/inode.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'fs') diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index b0d42ece4d7c..34bfe0f912e1 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -58,6 +58,13 @@ int ovl_setattr(struct dentry *dentry, struct iattr *attr) if (attr->ia_valid & (ATTR_KILL_SUID|ATTR_KILL_SGID)) attr->ia_valid &= ~ATTR_MODE; + /* + * We might have to translate ovl file into underlying file + * object once some use cases are there. For now, simply don't + * let underlying filesystem rely on attr->ia_file + */ + attr->ia_valid &= ~ATTR_FILE; + inode_lock(upperdentry->d_inode); old_cred = ovl_override_creds(dentry->d_sb); err = notify_change(upperdentry, attr, NULL); -- cgit v1.2.3 From 15fd2ea9f4f3d85fef787ba7db1b87939d0a2754 Mon Sep 17 00:00:00 2001 From: Vivek Goyal Date: Wed, 22 Apr 2020 09:08:50 -0400 Subject: ovl: clear ATTR_OPEN from attr->ia_valid As of now during open(), we don't pass bunch of flags to underlying filesystem. O_TRUNC is one of these. Normally this is not a problem as VFS calls ->setattr() with zero size and underlying filesystem sets file size to 0. But when overlayfs is running on top of virtiofs, it has an optimization where it does not send setattr request to server if dectects that truncation is part of open(O_TRUNC). It assumes that server already zeroed file size as part of open(O_TRUNC). fuse_do_setattr() { if (attr->ia_valid & ATTR_OPEN) { /* * No need to send request to userspace, since actual * truncation has already been done by OPEN. But still * need to truncate page cache. */ } } IOW, fuse expects O_TRUNC to be passed to it as part of open flags. But currently overlayfs does not pass O_TRUNC to underlying filesystem hence fuse/virtiofs breaks. Setup overlayfs on top of virtiofs and following does not zero the file size of a file is either upper only or has already been copied up. fd = open(foo.txt, O_TRUNC | O_WRONLY); There are two ways to fix this. Either pass O_TRUNC to underlying filesystem or clear ATTR_OPEN from attr->ia_valid so that fuse ends up sending a SETATTR request to server. Miklos is concerned that O_TRUNC might have side affects so it is better to clear ATTR_OPEN for now. Hence this patch clears ATTR_OPEN from attr->ia_valid. I found this problem while running unionmount-testsuite. With this patch, unionmount-testsuite passes with overlayfs on top of virtiofs. Signed-off-by: Vivek Goyal Fixes: bccece1ead36 ("ovl: allow remote upper") Signed-off-by: Miklos Szeredi --- fs/overlayfs/inode.c | 17 ++++++++++++++--- 1 file changed, 14 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c index 34bfe0f912e1..981f11ec51bc 100644 --- a/fs/overlayfs/inode.c +++ b/fs/overlayfs/inode.c @@ -59,12 +59,23 @@ int ovl_setattr(struct dentry *dentry, struct iattr *attr) attr->ia_valid &= ~ATTR_MODE; /* - * We might have to translate ovl file into underlying file - * object once some use cases are there. For now, simply don't - * let underlying filesystem rely on attr->ia_file + * We might have to translate ovl file into real file object + * once use cases emerge. For now, simply don't let underlying + * filesystem rely on attr->ia_file */ attr->ia_valid &= ~ATTR_FILE; + /* + * If open(O_TRUNC) is done, VFS calls ->setattr with ATTR_OPEN + * set. Overlayfs does not pass O_TRUNC flag to underlying + * filesystem during open -> do not pass ATTR_OPEN. This + * disables optimization in fuse which assumes open(O_TRUNC) + * already set file size to 0. But we never passed O_TRUNC to + * fuse. So by clearing ATTR_OPEN, fuse will be forced to send + * setattr request to server. + */ + attr->ia_valid &= ~ATTR_OPEN; + inode_lock(upperdentry->d_inode); old_cred = ovl_override_creds(dentry->d_sb); err = notify_change(upperdentry, attr, NULL); -- cgit v1.2.3 From c5f9d9db83d9f84d2b4aae5a1b29d9b582ccff2f Mon Sep 17 00:00:00 2001 From: David Howells Date: Mon, 4 May 2020 16:12:55 +0100 Subject: cachefiles: Fix corruption of the return value in cachefiles_read_or_alloc_pages() The patch which changed cachefiles from calling ->bmap() to using the bmap() wrapper overwrote the running return value with the result of calling bmap(). This causes an assertion failure elsewhere in the code. Fix this by using ret2 rather than ret to hold the return value. The oops looks like: kernel BUG at fs/nfs/fscache.c:468! ... RIP: 0010:__nfs_readpages_from_fscache+0x18b/0x190 [nfs] ... Call Trace: nfs_readpages+0xbf/0x1c0 [nfs] ? __alloc_pages_nodemask+0x16c/0x320 read_pages+0x67/0x1a0 __do_page_cache_readahead+0x1cf/0x1f0 ondemand_readahead+0x172/0x2b0 page_cache_async_readahead+0xaa/0xe0 generic_file_buffered_read+0x852/0xd50 ? mem_cgroup_commit_charge+0x6e/0x140 ? nfs4_have_delegation+0x19/0x30 [nfsv4] generic_file_read_iter+0x100/0x140 ? nfs_revalidate_mapping+0x176/0x2b0 [nfs] nfs_file_read+0x6d/0xc0 [nfs] new_sync_read+0x11a/0x1c0 __vfs_read+0x29/0x40 vfs_read+0x8e/0x140 ksys_read+0x61/0xd0 __x64_sys_read+0x1a/0x20 do_syscall_64+0x60/0x1e0 entry_SYSCALL_64_after_hwframe+0x44/0xa9 RIP: 0033:0x7f5d148267e0 Fixes: 10d83e11a582 ("cachefiles: drop direct usage of ->bmap method.") Reported-by: David Wysochanski Signed-off-by: David Howells Tested-by: David Wysochanski cc: Carlos Maiolino --- fs/cachefiles/rdwr.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c index 1dc97f2d6201..d3d78176b23c 100644 --- a/fs/cachefiles/rdwr.c +++ b/fs/cachefiles/rdwr.c @@ -398,7 +398,7 @@ int cachefiles_read_or_alloc_page(struct fscache_retrieval *op, struct inode *inode; sector_t block; unsigned shift; - int ret; + int ret, ret2; object = container_of(op->op.object, struct cachefiles_object, fscache); @@ -430,8 +430,8 @@ int cachefiles_read_or_alloc_page(struct fscache_retrieval *op, block = page->index; block <<= shift; - ret = bmap(inode, &block); - ASSERT(ret < 0); + ret2 = bmap(inode, &block); + ASSERT(ret2 == 0); _debug("%llx -> %llx", (unsigned long long) (page->index << shift), @@ -739,8 +739,8 @@ int cachefiles_read_or_alloc_pages(struct fscache_retrieval *op, block = page->index; block <<= shift; - ret = bmap(inode, &block); - ASSERT(!ret); + ret2 = bmap(inode, &block); + ASSERT(ret2 == 0); _debug("%llx -> %llx", (unsigned long long) (page->index << shift), -- cgit v1.2.3 From 3dc58df0e2436829971c41ffbb07167dda0979f8 Mon Sep 17 00:00:00 2001 From: Geert Uytterhoeven Date: Tue, 5 May 2020 15:43:35 +0200 Subject: CIFS: Spelling s/EACCESS/EACCES/ As per POSIX, the correct spelling is EACCES: include/uapi/asm-generic/errno-base.h:#define EACCES 13 /* Permission denied */ Fixes: b8f7442bc46e48fb ("CIFS: refactor cifs_get_inode_info()") Signed-off-by: Geert Uytterhoeven Signed-off-by: Steve French --- fs/cifs/inode.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cifs/inode.c b/fs/cifs/inode.c index 390d2b15ef6e..5d2965a23730 100644 --- a/fs/cifs/inode.c +++ b/fs/cifs/inode.c @@ -730,7 +730,7 @@ static __u64 simple_hashstr(const char *str) * cifs_backup_query_path_info - SMB1 fallback code to get ino * * Fallback code to get file metadata when we don't have access to - * @full_path (EACCESS) and have backup creds. + * @full_path (EACCES) and have backup creds. * * @data will be set to search info result buffer * @resp_buf will be set to cifs resp buf and needs to be freed with -- cgit v1.2.3 From ac9155842829a811c12d3e1868a133fdb8300df0 Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Wed, 22 Apr 2020 14:14:30 -0500 Subject: gfs2: fix withdraw sequence deadlock After a gfs2 file system withdraw, any attempt to read metadata is automatically rejected by function gfs2_meta_read() except for reads of the journal inode. This turns out to be a problem because function signal_our_withdraw() repeatedly calls check_journal_clean() which reads the metadata (both its dinode and indirect blocks) to see if the entire journal is mapped. The dinode read works, but reading the indirect blocks returns -EIO which gets sent back up and causes a consistency error. This results in withdraw-from-withdraw, which becomes a deadlock. This patch changes the test in gfs2_meta_read() to allow all metadata reads for the journal. Instead of checking the journal block, it now checks for the journal inode glock which is the same for all blocks in the journal. This allows check_journal_clean() to properly check the journal without trying to withdraw recursively. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/meta_io.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/meta_io.c b/fs/gfs2/meta_io.c index 4b72abcf83b2..9856cc2e0795 100644 --- a/fs/gfs2/meta_io.c +++ b/fs/gfs2/meta_io.c @@ -252,7 +252,7 @@ int gfs2_meta_read(struct gfs2_glock *gl, u64 blkno, int flags, int num = 0; if (unlikely(gfs2_withdrawn(sdp)) && - (!sdp->sd_jdesc || (blkno != sdp->sd_jdesc->jd_no_addr))) { + (!sdp->sd_jdesc || gl != sdp->sd_jinode_gl)) { *bhp = NULL; return -EIO; } -- cgit v1.2.3 From a8b7528b69d4dc7e94d0338851ff8c929231fc4b Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Thu, 23 Apr 2020 14:30:57 -0500 Subject: gfs2: Fix error exit in do_xmote Before this patch, if an error was detected from glock function go_sync by function do_xmote, it would return. But the function had temporarily unlocked the gl_lockref spin_lock, and it never re-locked it. When the caller of do_xmote tried to unlock it again, it was already unlocked, which resulted in a corrupted spin_lock value. This patch makes sure the gl_lockref spin_lock is re-locked after it is unlocked. Thanks to Wu Bo for reporting this problem. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/glock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c index 29f9b6684b74..a1c5f245553f 100644 --- a/fs/gfs2/glock.c +++ b/fs/gfs2/glock.c @@ -613,7 +613,7 @@ __acquires(&gl->gl_lockref.lock) fs_err(sdp, "Error %d syncing glock \n", ret); gfs2_dump_glock(NULL, gl, true); } - return; + goto out; } } if (test_bit(GLF_INVALIDATE_IN_PROGRESS, &gl->gl_flags)) { -- cgit v1.2.3 From 53af80ce0eaeb0fc4ce4b565c30e3a16e8e05de0 Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Fri, 24 Apr 2020 12:15:21 -0500 Subject: gfs2: Fix BUG during unmount after file system withdraw Before this patch, when the logd daemon was forced to withdraw, it would try to request its journal be recovered by another cluster node. However, in single-user cases with lock_nolock, there are no other nodes to recover the journal. Function signal_our_withdraw() was recognizing the lock_nolock situation, but not until after it had evicted its journal inode. Since the journal descriptor that points to the inode was never removed from the master list, when the unmount occurred, it did another iput on the evicted inode, which resulted in a BUG_ON(inode->i_state & I_CLEAR). This patch moves the check for this situation earlier in function signal_our_withdraw(), which avoids the extra iput, so the unmount may happen normally. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/util.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/util.c b/fs/gfs2/util.c index 9b64d40ab379..aa087a5675af 100644 --- a/fs/gfs2/util.c +++ b/fs/gfs2/util.c @@ -119,6 +119,12 @@ static void signal_our_withdraw(struct gfs2_sbd *sdp) if (!sb_rdonly(sdp->sd_vfs)) ret = gfs2_make_fs_ro(sdp); + if (sdp->sd_lockstruct.ls_ops->lm_lock == NULL) { /* lock_nolock */ + if (!ret) + ret = -EIO; + clear_bit(SDF_WITHDRAW_RECOVERY, &sdp->sd_flags); + goto skip_recovery; + } /* * Drop the glock for our journal so another node can recover it. */ @@ -159,10 +165,6 @@ static void signal_our_withdraw(struct gfs2_sbd *sdp) wait_on_bit(&gl->gl_flags, GLF_FREEING, TASK_UNINTERRUPTIBLE); } - if (sdp->sd_lockstruct.ls_ops->lm_lock == NULL) { /* lock_nolock */ - clear_bit(SDF_WITHDRAW_RECOVERY, &sdp->sd_flags); - goto skip_recovery; - } /* * Dequeue the "live" glock, but keep a reference so it's never freed. */ -- cgit v1.2.3 From d22f69a08dcb0f469170cda1976d5938cb0e5dcf Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Fri, 24 Apr 2020 12:17:33 -0500 Subject: gfs2: Fix use-after-free in gfs2_logd after withdraw When the gfs2_logd daemon withdrew, the withdraw sequence called into make_fs_ro() to make the file system read-only. That caused the journal descriptors to be freed. However, those journal descriptors were used by gfs2_logd's call to gfs2_ail_flush_reqd(). This caused a use-after free and NULL pointer dereference. This patch changes function gfs2_logd() so that it stops all logd work until the thread is told to stop. Once a withdraw is done, it only does an interruptible sleep. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/log.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs') diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c index 3a75843ae580..cf0b80c78c82 100644 --- a/fs/gfs2/log.c +++ b/fs/gfs2/log.c @@ -1131,6 +1131,10 @@ int gfs2_logd(void *data) while (!kthread_should_stop()) { + if (gfs2_withdrawn(sdp)) { + msleep_interruptible(HZ); + continue; + } /* Check for errors writing to the journal */ if (sdp->sd_log_error) { gfs2_lm(sdp, @@ -1139,6 +1143,7 @@ int gfs2_logd(void *data) "prevent further damage.\n", sdp->sd_fsname, sdp->sd_log_error); gfs2_withdraw(sdp); + continue; } did_flush = false; -- cgit v1.2.3 From 566a2ab3c9005f62e784bd39022d58d34ef4365c Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Mon, 20 Apr 2020 19:42:04 +0200 Subject: gfs2: Another gfs2_walk_metadata fix Make sure we don't walk past the end of the metadata in gfs2_walk_metadata: the inode holds fewer pointers than indirect blocks. Slightly clean up gfs2_iomap_get. Fixes: a27a0c9b6a20 ("gfs2: gfs2_walk_metadata fix") Cc: stable@vger.kernel.org # v5.3+ Signed-off-by: Andreas Gruenbacher Signed-off-by: Bob Peterson --- fs/gfs2/bmap.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/bmap.c b/fs/gfs2/bmap.c index 936a8ec6b48e..6306eaae378b 100644 --- a/fs/gfs2/bmap.c +++ b/fs/gfs2/bmap.c @@ -528,10 +528,12 @@ lower_metapath: /* Advance in metadata tree. */ (mp->mp_list[hgt])++; - if (mp->mp_list[hgt] >= sdp->sd_inptrs) { - if (!hgt) + if (hgt) { + if (mp->mp_list[hgt] >= sdp->sd_inptrs) + goto lower_metapath; + } else { + if (mp->mp_list[hgt] >= sdp->sd_diptrs) break; - goto lower_metapath; } fill_up_metapath: @@ -876,10 +878,9 @@ static int gfs2_iomap_get(struct inode *inode, loff_t pos, loff_t length, ret = -ENOENT; goto unlock; } else { - /* report a hole */ iomap->offset = pos; iomap->length = length; - goto do_alloc; + goto hole_found; } } iomap->length = size; @@ -933,8 +934,6 @@ unlock: return ret; do_alloc: - iomap->addr = IOMAP_NULL_ADDR; - iomap->type = IOMAP_HOLE; if (flags & IOMAP_REPORT) { if (pos >= size) ret = -ENOENT; @@ -956,6 +955,9 @@ do_alloc: if (pos < size && height == ip->i_height) ret = gfs2_hole_size(inode, lblock, len, mp, iomap); } +hole_found: + iomap->addr = IOMAP_NULL_ADDR; + iomap->type = IOMAP_HOLE; goto out; } -- cgit v1.2.3 From aa83da7f47b26c9587bade6c4bc4736ffa308f0a Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Tue, 28 Apr 2020 01:15:41 +0200 Subject: gfs2: More gfs2_find_jhead fixes It turns out that when extending an existing bio, gfs2_find_jhead fails to check if the block number is consecutive, which leads to incorrect reads for fragmented journals. In addition, limit the maximum bio size to an arbitrary value of 2 megabytes: since commit 07173c3ec276 ("block: enable multipage bvecs"), if we just keep adding pages until bio_add_page fails, bios will grow much larger than useful, which pins more memory than necessary with barely any additional performance gains. Fixes: f4686c26ecc3 ("gfs2: read journal in large chunks") Cc: stable@vger.kernel.org # v5.2+ Signed-off-by: Andreas Gruenbacher Signed-off-by: Bob Peterson --- fs/gfs2/lops.c | 19 ++++++++++++------- 1 file changed, 12 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/lops.c b/fs/gfs2/lops.c index 5ea96757afc4..48b54ec1c793 100644 --- a/fs/gfs2/lops.c +++ b/fs/gfs2/lops.c @@ -263,7 +263,7 @@ static struct bio *gfs2_log_alloc_bio(struct gfs2_sbd *sdp, u64 blkno, struct super_block *sb = sdp->sd_vfs; struct bio *bio = bio_alloc(GFP_NOIO, BIO_MAX_PAGES); - bio->bi_iter.bi_sector = blkno << (sb->s_blocksize_bits - 9); + bio->bi_iter.bi_sector = blkno << sdp->sd_fsb2bb_shift; bio_set_dev(bio, sb->s_bdev); bio->bi_end_io = end_io; bio->bi_private = sdp; @@ -509,7 +509,7 @@ int gfs2_find_jhead(struct gfs2_jdesc *jd, struct gfs2_log_header_host *head, unsigned int bsize = sdp->sd_sb.sb_bsize, off; unsigned int bsize_shift = sdp->sd_sb.sb_bsize_shift; unsigned int shift = PAGE_SHIFT - bsize_shift; - unsigned int readahead_blocks = BIO_MAX_PAGES << shift; + unsigned int max_bio_size = 2 * 1024 * 1024; struct gfs2_journal_extent *je; int sz, ret = 0; struct bio *bio = NULL; @@ -537,12 +537,17 @@ int gfs2_find_jhead(struct gfs2_jdesc *jd, struct gfs2_log_header_host *head, off = 0; } - if (!bio || (bio_chained && !off)) { + if (!bio || (bio_chained && !off) || + bio->bi_iter.bi_size >= max_bio_size) { /* start new bio */ } else { - sz = bio_add_page(bio, page, bsize, off); - if (sz == bsize) - goto block_added; + sector_t sector = dblock << sdp->sd_fsb2bb_shift; + + if (bio_end_sector(bio) == sector) { + sz = bio_add_page(bio, page, bsize, off); + if (sz == bsize) + goto block_added; + } if (off) { unsigned int blocks = (PAGE_SIZE - off) >> bsize_shift; @@ -568,7 +573,7 @@ block_added: off += bsize; if (off == PAGE_SIZE) page = NULL; - if (blocks_submitted < blocks_read + readahead_blocks) { + if (blocks_submitted < 2 * max_bio_size >> bsize_shift) { /* Keep at least one bio in flight */ continue; } -- cgit v1.2.3 From 2297ab6144c2e85c418d0fd47b2f24e294b55dca Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Mon, 4 May 2020 10:18:43 -0500 Subject: gfs2: Fix problems regarding gfs2_qa_get and _put This patch fixes a couple of places in which gfs2_qa_get and gfs2_qa_put are not balanced: we now keep references around whenever a file is open for writing (see gfs2_open_common and gfs2_release), so we need to put all references we grab in function gfs2_create_inode. This was broken in the successful case and on one error path. This also means that we don't have a reference to put in gfs2_evict_inode. In addition, gfs2_qa_put was called for the wrong inode in gfs2_link. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/inode.c | 7 ++++--- fs/gfs2/super.c | 1 - 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/inode.c b/fs/gfs2/inode.c index 70b2d3a1e866..5acd3ce30759 100644 --- a/fs/gfs2/inode.c +++ b/fs/gfs2/inode.c @@ -622,7 +622,7 @@ static int gfs2_create_inode(struct inode *dir, struct dentry *dentry, error = finish_no_open(file, NULL); } gfs2_glock_dq_uninit(ghs); - return error; + goto fail; } else if (error != -ENOENT) { goto fail_gunlock; } @@ -764,9 +764,11 @@ static int gfs2_create_inode(struct inode *dir, struct dentry *dentry, error = finish_open(file, dentry, gfs2_open_common); } gfs2_glock_dq_uninit(ghs); + gfs2_qa_put(ip); gfs2_glock_dq_uninit(ghs + 1); clear_bit(GLF_INODE_CREATING, &io_gl->gl_flags); gfs2_glock_put(io_gl); + gfs2_qa_put(dip); return error; fail_gunlock3: @@ -776,7 +778,6 @@ fail_gunlock2: clear_bit(GLF_INODE_CREATING, &io_gl->gl_flags); gfs2_glock_put(io_gl); fail_free_inode: - gfs2_qa_put(ip); if (ip->i_gl) { glock_clear_object(ip->i_gl, ip); gfs2_glock_put(ip->i_gl); @@ -1005,7 +1006,7 @@ out_gunlock: out_child: gfs2_glock_dq(ghs); out_parent: - gfs2_qa_put(ip); + gfs2_qa_put(dip); gfs2_holder_uninit(ghs); gfs2_holder_uninit(ghs + 1); return error; diff --git a/fs/gfs2/super.c b/fs/gfs2/super.c index 37fc41632aa2..956fced0a8ec 100644 --- a/fs/gfs2/super.c +++ b/fs/gfs2/super.c @@ -1404,7 +1404,6 @@ out: if (ip->i_qadata) gfs2_assert_warn(sdp, ip->i_qadata->qa_ref == 0); gfs2_rs_delete(ip, NULL); - gfs2_qa_put(ip); gfs2_ordered_del_inode(ip); clear_inode(inode); gfs2_dir_hash_inval(ip); -- cgit v1.2.3 From f9615fe3113f1067093f7f68d46d281b4c9a5978 Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Tue, 5 May 2020 11:50:24 -0500 Subject: gfs2: Change BUG_ON to an assert_withdraw in gfs2_quota_change Before this patch, gfs2_quota_change() would BUG_ON if the qa_ref counter was not a positive number. This patch changes it to be a withdraw instead. That way we can debug things more easily. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/quota.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/quota.c b/fs/gfs2/quota.c index cc0c4b5800be..a62be4267c17 100644 --- a/fs/gfs2/quota.c +++ b/fs/gfs2/quota.c @@ -1270,7 +1270,9 @@ void gfs2_quota_change(struct gfs2_inode *ip, s64 change, if (ip->i_diskflags & GFS2_DIF_SYSTEM) return; - BUG_ON(ip->i_qadata->qa_ref <= 0); + if (gfs2_assert_withdraw(sdp, ip->i_qadata && + ip->i_qadata->qa_ref > 0)) + return; for (x = 0; x < ip->i_qadata->qa_qd_num; x++) { qd = ip->i_qadata->qa_qd[x]; -- cgit v1.2.3 From e6ce26e571a813e6992c5148b27f2a4b17952080 Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Tue, 5 May 2020 11:52:51 -0500 Subject: gfs2: remove check for quotas on in gfs2_quota_check This patch removes a check from gfs2_quota_check for whether quotas are enabled by the superblock. There is a test just prior for the GIF_QD_LOCKED bit in the inode, and that can only be set by functions that already check that quotas are enabled in the superblock. Therefore, the check is redundant. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/quota.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/quota.c b/fs/gfs2/quota.c index a62be4267c17..c997cadc8d9a 100644 --- a/fs/gfs2/quota.c +++ b/fs/gfs2/quota.c @@ -1210,9 +1210,6 @@ int gfs2_quota_check(struct gfs2_inode *ip, kuid_t uid, kgid_t gid, if (!test_bit(GIF_QD_LOCKED, &ip->i_flags)) return 0; - if (sdp->sd_args.ar_quota != GFS2_QUOTA_ON) - return 0; - for (x = 0; x < ip->i_qadata->qa_qd_num; x++) { qd = ip->i_qadata->qa_qd[x]; -- cgit v1.2.3 From 4ed0c30811cb4d30ef89850b787a53a84d5d2bcb Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Tue, 5 May 2020 11:53:21 -0500 Subject: gfs2: move privileged user check to gfs2_quota_lock_check Before this patch, function gfs2_quota_lock checked if it was called from a privileged user, and if so, it bypassed the quota check: superuser can operate outside the quotas. That's the wrong place for the check because the lock/unlock functions are separate from the lock_check function, and you can do lock and unlock without actually checking the quotas. This patch moves the check to gfs2_quota_lock_check. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/quota.c | 3 +-- fs/gfs2/quota.h | 3 ++- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/quota.c b/fs/gfs2/quota.c index c997cadc8d9a..ed2e488f98b3 100644 --- a/fs/gfs2/quota.c +++ b/fs/gfs2/quota.c @@ -1051,8 +1051,7 @@ int gfs2_quota_lock(struct gfs2_inode *ip, kuid_t uid, kgid_t gid) u32 x; int error = 0; - if (capable(CAP_SYS_RESOURCE) || - sdp->sd_args.ar_quota != GFS2_QUOTA_ON) + if (sdp->sd_args.ar_quota != GFS2_QUOTA_ON) return 0; error = gfs2_quota_hold(ip, uid, gid); diff --git a/fs/gfs2/quota.h b/fs/gfs2/quota.h index 7f9ca8ef40fc..21ada332d555 100644 --- a/fs/gfs2/quota.h +++ b/fs/gfs2/quota.h @@ -44,7 +44,8 @@ static inline int gfs2_quota_lock_check(struct gfs2_inode *ip, int ret; ap->allowed = UINT_MAX; /* Assume we are permitted a whole lot */ - if (sdp->sd_args.ar_quota == GFS2_QUOTA_OFF) + if (capable(CAP_SYS_RESOURCE) || + sdp->sd_args.ar_quota == GFS2_QUOTA_OFF) return 0; ret = gfs2_quota_lock(ip, NO_UID_QUOTA_CHANGE, NO_GID_QUOTA_CHANGE); if (ret) -- cgit v1.2.3 From c9cb9e381985bbbe8acd2695bbe6bd24bf06b81c Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Tue, 5 May 2020 11:55:03 -0500 Subject: gfs2: don't call quota_unhold if quotas are not locked Before this patch, function gfs2_quota_unlock checked if quotas are turned off, and if so, it branched to label out, which called gfs2_quota_unhold. With the new system of gfs2_qa_get and put, we no longer want to call gfs2_quota_unhold or we won't balance our gets and puts. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/quota.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/quota.c b/fs/gfs2/quota.c index ed2e488f98b3..8259fef3f986 100644 --- a/fs/gfs2/quota.c +++ b/fs/gfs2/quota.c @@ -1124,7 +1124,7 @@ void gfs2_quota_unlock(struct gfs2_inode *ip) int found; if (!test_and_clear_bit(GIF_QD_LOCKED, &ip->i_flags)) - goto out; + return; for (x = 0; x < ip->i_qadata->qa_qd_num; x++) { struct gfs2_quota_data *qd; @@ -1161,7 +1161,6 @@ void gfs2_quota_unlock(struct gfs2_inode *ip) qd_unlock(qda[x]); } -out: gfs2_quota_unhold(ip); } -- cgit v1.2.3 From f4e2f5e1a527ce58fc9f85145b03704779a3123e Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Tue, 5 May 2020 11:56:46 -0500 Subject: gfs2: Grab glock reference sooner in gfs2_add_revoke This patch rearranges gfs2_add_revoke so that the extra glock reference is added earlier on in the function to avoid races in which the glock is freed before the new reference is taken. Signed-off-by: Andreas Gruenbacher Signed-off-by: Bob Peterson --- fs/gfs2/log.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/log.c b/fs/gfs2/log.c index cf0b80c78c82..0644e58c6191 100644 --- a/fs/gfs2/log.c +++ b/fs/gfs2/log.c @@ -669,13 +669,13 @@ void gfs2_add_revoke(struct gfs2_sbd *sdp, struct gfs2_bufdata *bd) struct buffer_head *bh = bd->bd_bh; struct gfs2_glock *gl = bd->bd_gl; + sdp->sd_log_num_revoke++; + if (atomic_inc_return(&gl->gl_revokes) == 1) + gfs2_glock_hold(gl); bh->b_private = NULL; bd->bd_blkno = bh->b_blocknr; gfs2_remove_from_ail(bd); /* drops ref on bh */ bd->bd_bh = NULL; - sdp->sd_log_num_revoke++; - if (atomic_inc_return(&gl->gl_revokes) == 1) - gfs2_glock_hold(gl); set_bit(GLF_LFLUSH, &gl->gl_flags); list_add(&bd->bd_list, &sdp->sd_log_revokes); } -- cgit v1.2.3 From b11e1a84f370866a8f47e85040687b49c1eb8705 Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Thu, 7 May 2020 12:12:23 -0500 Subject: gfs2: If go_sync returns error, withdraw but skip invalidate Before this patch, if the go_sync operation returned an error during the do_xmote process (such as unable to sync metadata to the journal) the code did goto out. That kept the glock locked, so it could not be given away, which correctly avoids file system corruption. However, it never set the withdraw bit or requeueing the glock work. So it would hang forever, unable to ever demote the glock. This patch changes to goto to a new label, skip_inval, so that errors from go_sync are treated the same way as errors from go_inval: The delayed withdraw bit is set and the work is requeued. That way, the logd should eventually figure out there's a problem and withdraw properly there. Signed-off-by: Bob Peterson Signed-off-by: Andreas Gruenbacher --- fs/gfs2/glock.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c index a1c5f245553f..5239098fcce6 100644 --- a/fs/gfs2/glock.c +++ b/fs/gfs2/glock.c @@ -613,7 +613,7 @@ __acquires(&gl->gl_lockref.lock) fs_err(sdp, "Error %d syncing glock \n", ret); gfs2_dump_glock(NULL, gl, true); } - goto out; + goto skip_inval; } } if (test_bit(GLF_INVALIDATE_IN_PROGRESS, &gl->gl_flags)) { @@ -633,6 +633,7 @@ __acquires(&gl->gl_lockref.lock) clear_bit(GLF_INVALIDATE_IN_PROGRESS, &gl->gl_flags); } +skip_inval: gfs2_glock_hold(gl); /* * Check for an error encountered since we called go_sync and go_inval. -- cgit v1.2.3 From b14c94908b1b884276a6608dea3d0b1b510338b7 Mon Sep 17 00:00:00 2001 From: Bob Peterson Date: Fri, 8 May 2020 15:01:25 -0500 Subject: Revert "gfs2: Don't demote a glock until its revokes are written" This reverts commit df5db5f9ee112e76b5202fbc331f990a0fc316d6. This patch fixes a regression: patch df5db5f9ee112 allowed function run_queue() to bypass its call to do_xmote() if revokes were queued for the glock. That's wrong because its call to do_xmote() is what is responsible for calling the go_sync() glops functions to sync both the ail list and any revokes queued for it. By bypassing the call, gfs2 could get into a stand-off where the glock could not be demoted until its revokes are written back, but the revokes would not be written back because do_xmote() was never called. It "sort of" works, however, because there are other mechanisms like the log flush daemon (logd) that can sync the ail items and revokes, if it deems it necessary. The problem is: without file system pressure, it might never deem it necessary. Signed-off-by: Bob Peterson --- fs/gfs2/glock.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'fs') diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c index 5239098fcce6..bf70e3b14938 100644 --- a/fs/gfs2/glock.c +++ b/fs/gfs2/glock.c @@ -723,9 +723,6 @@ __acquires(&gl->gl_lockref.lock) goto out_unlock; if (nonblock) goto out_sched; - smp_mb(); - if (atomic_read(&gl->gl_revokes) != 0) - goto out_sched; set_bit(GLF_DEMOTE_IN_PROGRESS, &gl->gl_flags); GLOCK_BUG_ON(gl, gl->gl_demote_state == LM_ST_EXCLUSIVE); gl->gl_target = gl->gl_demote_state; -- cgit v1.2.3 From d9bfced1fbcb35b28d8fbed4e785d2807055ed2b Mon Sep 17 00:00:00 2001 From: Dave Wysochanski Date: Wed, 15 Apr 2020 16:14:41 -0400 Subject: NFS: Fix fscache super_cookie index_key from changing after umount Commit 402cb8dda949 ("fscache: Attach the index key and aux data to the cookie") added the index_key and index_key_len parameters to fscache_acquire_cookie(), and updated the callers in the NFS client. One of the callers was inside nfs_fscache_get_super_cookie() and was changed to use the full struct nfs_fscache_key as the index_key. However, a couple members of this structure contain pointers and thus will change each time the same NFS share is remounted. Since index_key is used for fscache_cookie->key_hash and this subsequently is used to compare cookies, the effectiveness of fscache with NFS is reduced to the point at which a umount occurs. Any subsequent remount of the same share will cause a unique NFS super_block index_key and key_hash to be generated for the same data, rendering any prior fscache data unable to be found. A simple reproducer demonstrates the problem. 1. Mount share with 'fsc', create a file, drop page cache systemctl start cachefilesd mount -o vers=3,fsc 127.0.0.1:/export /mnt dd if=/dev/zero of=/mnt/file1.bin bs=4096 count=1 echo 3 > /proc/sys/vm/drop_caches 2. Read file into page cache and fscache, then unmount dd if=/mnt/file1.bin of=/dev/null bs=4096 count=1 umount /mnt 3. Remount and re-read which should come from fscache mount -o vers=3,fsc 127.0.0.1:/export /mnt echo 3 > /proc/sys/vm/drop_caches dd if=/mnt/file1.bin of=/dev/null bs=4096 count=1 4. Check for READ ops in mountstats - there should be none grep READ: /proc/self/mountstats Looking at the history and the removed function, nfs_super_get_key(), we should only use nfs_fscache_key.key plus any uniquifier, for the fscache index_key. Fixes: 402cb8dda949 ("fscache: Attach the index key and aux data to the cookie") Signed-off-by: Dave Wysochanski Signed-off-by: David Howells --- fs/nfs/fscache.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/fscache.c b/fs/nfs/fscache.c index 1abf126c2df4..8eff1fd806b1 100644 --- a/fs/nfs/fscache.c +++ b/fs/nfs/fscache.c @@ -188,7 +188,8 @@ void nfs_fscache_get_super_cookie(struct super_block *sb, const char *uniq, int /* create a cache index for looking up filehandles */ nfss->fscache = fscache_acquire_cookie(nfss->nfs_client->fscache, &nfs_fscache_super_index_def, - key, sizeof(*key) + ulen, + &key->key, + sizeof(key->key) + ulen, NULL, 0, nfss, 0, true); dfprintk(FSCACHE, "NFS: get superblock cookie (0x%p/0x%p)\n", -- cgit v1.2.3 From 15751612734ca0c419ac43ce986c9badcb5e2829 Mon Sep 17 00:00:00 2001 From: Dave Wysochanski Date: Wed, 15 Apr 2020 16:14:42 -0400 Subject: NFS: Fix fscache super_cookie allocation Commit f2aedb713c28 ("NFS: Add fs_context support.") reworked NFS mount code paths for fs_context support which included super_block initialization. In the process there was an extra return left in the code and so we never call nfs_fscache_get_super_cookie even if 'fsc' is given on as mount option. In addition, there is an extra check inside nfs_fscache_get_super_cookie for the NFS_OPTION_FSCACHE which is unnecessary since the only caller nfs_get_cache_cookie checks this flag. Fixes: f2aedb713c28 ("NFS: Add fs_context support.") Signed-off-by: Dave Wysochanski Signed-off-by: David Howells --- fs/nfs/fscache.c | 2 -- fs/nfs/super.c | 1 - 2 files changed, 3 deletions(-) (limited to 'fs') diff --git a/fs/nfs/fscache.c b/fs/nfs/fscache.c index 8eff1fd806b1..f51718415606 100644 --- a/fs/nfs/fscache.c +++ b/fs/nfs/fscache.c @@ -118,8 +118,6 @@ void nfs_fscache_get_super_cookie(struct super_block *sb, const char *uniq, int nfss->fscache_key = NULL; nfss->fscache = NULL; - if (!(nfss->options & NFS_OPTION_FSCACHE)) - return; if (!uniq) { uniq = ""; ulen = 1; diff --git a/fs/nfs/super.c b/fs/nfs/super.c index 59ef3b13ccca..cc34aa3a8ba4 100644 --- a/fs/nfs/super.c +++ b/fs/nfs/super.c @@ -1189,7 +1189,6 @@ static void nfs_get_cache_cookie(struct super_block *sb, uniq = ctx->fscache_uniq; ulen = strlen(ctx->fscache_uniq); } - return; } nfs_fscache_get_super_cookie(sb, uniq, ulen); -- cgit v1.2.3 From 50eaa652b54df1e2b48dc398d9e6114c9ed080eb Mon Sep 17 00:00:00 2001 From: Dave Wysochanski Date: Thu, 16 Apr 2020 06:06:08 -0400 Subject: NFSv4: Fix fscache cookie aux_data to ensure change_attr is included Commit 402cb8dda949 ("fscache: Attach the index key and aux data to the cookie") added the aux_data and aux_data_len to parameters to fscache_acquire_cookie(), and updated the callers in the NFS client. In the process it modified the aux_data to include the change_attr, but missed adding change_attr to a couple places where aux_data was used. Specifically, when opening a file and the change_attr is not added, the following attempt to lookup an object will fail inside cachefiles_check_object_xattr() = -116 due to nfs_fscache_inode_check_aux() failing memcmp on auxdata and returning FSCACHE_CHECKAUX_OBSOLETE. Fix this by adding nfs_fscache_update_auxdata() to set the auxdata from all relevant fields in the inode, including the change_attr. Fixes: 402cb8dda949 ("fscache: Attach the index key and aux data to the cookie") Signed-off-by: Dave Wysochanski Signed-off-by: David Howells --- fs/nfs/fscache.c | 34 ++++++++++++++++------------------ 1 file changed, 16 insertions(+), 18 deletions(-) (limited to 'fs') diff --git a/fs/nfs/fscache.c b/fs/nfs/fscache.c index f51718415606..a60df88efc40 100644 --- a/fs/nfs/fscache.c +++ b/fs/nfs/fscache.c @@ -225,6 +225,19 @@ void nfs_fscache_release_super_cookie(struct super_block *sb) } } +static void nfs_fscache_update_auxdata(struct nfs_fscache_inode_auxdata *auxdata, + struct nfs_inode *nfsi) +{ + memset(auxdata, 0, sizeof(*auxdata)); + auxdata->mtime_sec = nfsi->vfs_inode.i_mtime.tv_sec; + auxdata->mtime_nsec = nfsi->vfs_inode.i_mtime.tv_nsec; + auxdata->ctime_sec = nfsi->vfs_inode.i_ctime.tv_sec; + auxdata->ctime_nsec = nfsi->vfs_inode.i_ctime.tv_nsec; + + if (NFS_SERVER(&nfsi->vfs_inode)->nfs_client->rpc_ops->version == 4) + auxdata->change_attr = inode_peek_iversion_raw(&nfsi->vfs_inode); +} + /* * Initialise the per-inode cache cookie pointer for an NFS inode. */ @@ -238,14 +251,7 @@ void nfs_fscache_init_inode(struct inode *inode) if (!(nfss->fscache && S_ISREG(inode->i_mode))) return; - memset(&auxdata, 0, sizeof(auxdata)); - auxdata.mtime_sec = nfsi->vfs_inode.i_mtime.tv_sec; - auxdata.mtime_nsec = nfsi->vfs_inode.i_mtime.tv_nsec; - auxdata.ctime_sec = nfsi->vfs_inode.i_ctime.tv_sec; - auxdata.ctime_nsec = nfsi->vfs_inode.i_ctime.tv_nsec; - - if (NFS_SERVER(&nfsi->vfs_inode)->nfs_client->rpc_ops->version == 4) - auxdata.change_attr = inode_peek_iversion_raw(&nfsi->vfs_inode); + nfs_fscache_update_auxdata(&auxdata, nfsi); nfsi->fscache = fscache_acquire_cookie(NFS_SB(inode->i_sb)->fscache, &nfs_fscache_inode_object_def, @@ -265,11 +271,7 @@ void nfs_fscache_clear_inode(struct inode *inode) dfprintk(FSCACHE, "NFS: clear cookie (0x%p/0x%p)\n", nfsi, cookie); - memset(&auxdata, 0, sizeof(auxdata)); - auxdata.mtime_sec = nfsi->vfs_inode.i_mtime.tv_sec; - auxdata.mtime_nsec = nfsi->vfs_inode.i_mtime.tv_nsec; - auxdata.ctime_sec = nfsi->vfs_inode.i_ctime.tv_sec; - auxdata.ctime_nsec = nfsi->vfs_inode.i_ctime.tv_nsec; + nfs_fscache_update_auxdata(&auxdata, nfsi); fscache_relinquish_cookie(cookie, &auxdata, false); nfsi->fscache = NULL; } @@ -309,11 +311,7 @@ void nfs_fscache_open_file(struct inode *inode, struct file *filp) if (!fscache_cookie_valid(cookie)) return; - memset(&auxdata, 0, sizeof(auxdata)); - auxdata.mtime_sec = nfsi->vfs_inode.i_mtime.tv_sec; - auxdata.mtime_nsec = nfsi->vfs_inode.i_mtime.tv_nsec; - auxdata.ctime_sec = nfsi->vfs_inode.i_ctime.tv_sec; - auxdata.ctime_nsec = nfsi->vfs_inode.i_ctime.tv_nsec; + nfs_fscache_update_auxdata(&auxdata, nfsi); if (inode_is_open_for_write(inode)) { dfprintk(FSCACHE, "NFS: nfsi 0x%p disabling cache\n", nfsi); -- cgit v1.2.3 From 7bb0c5338436dae953622470d52689265867f032 Mon Sep 17 00:00:00 2001 From: Lei Xue Date: Thu, 7 May 2020 08:50:22 -0400 Subject: cachefiles: Fix race between read_waiter and read_copier involving op->to_do There is a potential race in fscache operation enqueuing for reading and copying multiple pages from cachefiles to netfs. The problem can be seen easily on a heavy loaded system (for example many processes reading files continually on an NFS share covered by fscache triggered this problem within a few minutes). The race is due to cachefiles_read_waiter() adding the op to the monitor to_do list and then then drop the object->work_lock spinlock before completing fscache_enqueue_operation(). Once the lock is dropped, cachefiles_read_copier() grabs the op, completes processing it, and makes it through fscache_retrieval_complete() which sets the op->state to the final state of FSCACHE_OP_ST_COMPLETE(4). When cachefiles_read_waiter() finally gets through the remainder of fscache_enqueue_operation() it sees the invalid state, and hits the ASSERTCMP and the following oops is seen: [ 2259.612361] FS-Cache: [ 2259.614785] FS-Cache: Assertion failed [ 2259.618639] FS-Cache: 4 == 5 is false [ 2259.622456] ------------[ cut here ]------------ [ 2259.627190] kernel BUG at fs/fscache/operation.c:70! ... [ 2259.791675] RIP: 0010:[] [] fscache_enqueue_operation+0xff/0x170 [fscache] [ 2259.802059] RSP: 0000:ffffa0263d543be0 EFLAGS: 00010046 [ 2259.807521] RAX: 0000000000000019 RBX: ffffa01a4d390480 RCX: 0000000000000006 [ 2259.814847] RDX: 0000000000000000 RSI: 0000000000000046 RDI: ffffa0263d553890 [ 2259.822176] RBP: ffffa0263d543be8 R08: 0000000000000000 R09: ffffa0263c2d8708 [ 2259.829502] R10: 0000000000001e7f R11: 0000000000000000 R12: ffffa01a4d390480 [ 2259.844483] R13: ffff9fa9546c5920 R14: ffffa0263d543c80 R15: ffffa0293ff9bf10 [ 2259.859554] FS: 00007f4b6efbd700(0000) GS:ffffa0263d540000(0000) knlGS:0000000000000000 [ 2259.875571] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 2259.889117] CR2: 00007f49e1624ff0 CR3: 0000012b38b38000 CR4: 00000000007607e0 [ 2259.904015] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 2259.918764] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 2259.933449] PKRU: 55555554 [ 2259.943654] Call Trace: [ 2259.953592] [ 2259.955577] [] cachefiles_read_waiter+0x92/0xf0 [cachefiles] [ 2259.978039] [] __wake_up_common+0x82/0x120 [ 2259.991392] [] __wake_up_common_lock+0x83/0xc0 [ 2260.004930] [] ? task_rq_unlock+0x20/0x20 [ 2260.017863] [] __wake_up+0x13/0x20 [ 2260.030230] [] __wake_up_bit+0x50/0x70 [ 2260.042535] [] unlock_page+0x2b/0x30 [ 2260.054495] [] page_endio+0x29/0x90 [ 2260.066184] [] mpage_end_io+0x51/0x80 CPU1 cachefiles_read_waiter() 20 static int cachefiles_read_waiter(wait_queue_entry_t *wait, unsigned mode, 21 int sync, void *_key) 22 { ... 61 spin_lock(&object->work_lock); 62 list_add_tail(&monitor->op_link, &op->to_do); 63 spin_unlock(&object->work_lock); 64 65 fscache_enqueue_retrieval(op); 182 static inline void fscache_enqueue_retrieval(struct fscache_retrieval *op) 183 { 184 fscache_enqueue_operation(&op->op); 185 } 58 void fscache_enqueue_operation(struct fscache_operation *op) 59 { 60 struct fscache_cookie *cookie = op->object->cookie; 61 62 _enter("{OBJ%x OP%x,%u}", 63 op->object->debug_id, op->debug_id, atomic_read(&op->usage)); 64 65 ASSERT(list_empty(&op->pend_link)); 66 ASSERT(op->processor != NULL); 67 ASSERT(fscache_object_is_available(op->object)); 68 ASSERTCMP(atomic_read(&op->usage), >, 0); CPU2 cachefiles_read_copier() 168 while (!list_empty(&op->to_do)) { ... 202 fscache_end_io(op, monitor->netfs_page, error); 203 put_page(monitor->netfs_page); 204 fscache_retrieval_complete(op, 1); CPU1 58 void fscache_enqueue_operation(struct fscache_operation *op) 59 { ... 69 ASSERTIFCMP(op->state != FSCACHE_OP_ST_IN_PROGRESS, 70 op->state, ==, FSCACHE_OP_ST_CANCELLED); Signed-off-by: Lei Xue Signed-off-by: Dave Wysochanski Signed-off-by: David Howells --- fs/cachefiles/rdwr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cachefiles/rdwr.c b/fs/cachefiles/rdwr.c index d3d78176b23c..e7726f5f1241 100644 --- a/fs/cachefiles/rdwr.c +++ b/fs/cachefiles/rdwr.c @@ -60,9 +60,9 @@ static int cachefiles_read_waiter(wait_queue_entry_t *wait, unsigned mode, object = container_of(op->op.object, struct cachefiles_object, fscache); spin_lock(&object->work_lock); list_add_tail(&monitor->op_link, &op->to_do); + fscache_enqueue_retrieval(op); spin_unlock(&object->work_lock); - fscache_enqueue_retrieval(op); fscache_put_retrieval(op); return 0; } -- cgit v1.2.3 From c96874265cd04b4bd4a8e114ac9af039a6d83cfe Mon Sep 17 00:00:00 2001 From: Pavel Begunkov Date: Mon, 4 May 2020 23:00:54 +0300 Subject: io_uring: fix zero len do_splice() do_splice() doesn't expect len to be 0. Just always return 0 in this case as splice(2) does. Fixes: 7d67af2c0134 ("io_uring: add splice(2) support") Reported-by: Jann Horn Signed-off-by: Pavel Begunkov Signed-off-by: Jens Axboe --- fs/io_uring.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index 979d9f977409..4c2fe06ae20b 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -2767,16 +2767,19 @@ static int io_splice(struct io_kiocb *req, bool force_nonblock) struct file *out = sp->file_out; unsigned int flags = sp->flags & ~SPLICE_F_FD_IN_FIXED; loff_t *poff_in, *poff_out; - long ret; + long ret = 0; if (force_nonblock) return -EAGAIN; poff_in = (sp->off_in == -1) ? NULL : &sp->off_in; poff_out = (sp->off_out == -1) ? NULL : &sp->off_out; - ret = do_splice(in, poff_in, out, poff_out, sp->len, flags); - if (force_nonblock && ret == -EAGAIN) - return -EAGAIN; + + if (sp->len) { + ret = do_splice(in, poff_in, out, poff_out, sp->len, flags); + if (force_nonblock && ret == -EAGAIN) + return -EAGAIN; + } io_put_file(req, in, (sp->flags & SPLICE_F_FD_IN_FIXED)); req->flags &= ~REQ_F_NEED_CLEANUP; -- cgit v1.2.3 From c410bf01933e5e09d142c66c3df9ad470a7eec13 Mon Sep 17 00:00:00 2001 From: David Howells Date: Mon, 11 May 2020 14:54:34 +0100 Subject: rxrpc: Fix the excessive initial retransmission timeout rxrpc currently uses a fixed 4s retransmission timeout until the RTT is sufficiently sampled. This can cause problems with some fileservers with calls to the cache manager in the afs filesystem being dropped from the fileserver because a packet goes missing and the retransmission timeout is greater than the call expiry timeout. Fix this by: (1) Copying the RTT/RTO calculation code from Linux's TCP implementation and altering it to fit rxrpc. (2) Altering the various users of the RTT to make use of the new SRTT value. (3) Replacing the use of rxrpc_resend_timeout to use the calculated RTO value instead (which is needed in jiffies), along with a backoff. Notes: (1) rxrpc provides RTT samples by matching the serial numbers on outgoing DATA packets that have the RXRPC_REQUEST_ACK set and PING ACK packets against the reference serial number in incoming REQUESTED ACK and PING-RESPONSE ACK packets. (2) Each packet that is transmitted on an rxrpc connection gets a new per-connection serial number, even for retransmissions, so an ACK can be cross-referenced to a specific trigger packet. This allows RTT information to be drawn from retransmitted DATA packets also. (3) rxrpc maintains the RTT/RTO state on the rxrpc_peer record rather than on an rxrpc_call because many RPC calls won't live long enough to generate more than one sample. (4) The calculated SRTT value is in units of 8ths of a microsecond rather than nanoseconds. The (S)RTT and RTO values are displayed in /proc/net/rxrpc/peers. Fixes: 17926a79320a ([AF_RXRPC]: Provide secure RxRPC sockets for use by userspace and kernel both"") Signed-off-by: David Howells --- fs/afs/fs_probe.c | 18 ++-- fs/afs/vl_probe.c | 18 ++-- include/net/af_rxrpc.h | 2 +- include/trace/events/rxrpc.h | 17 ++-- net/rxrpc/Makefile | 1 + net/rxrpc/ar-internal.h | 25 ++++-- net/rxrpc/call_accept.c | 2 +- net/rxrpc/call_event.c | 22 ++--- net/rxrpc/input.c | 6 +- net/rxrpc/misc.c | 5 -- net/rxrpc/output.c | 9 +- net/rxrpc/peer_event.c | 46 ---------- net/rxrpc/peer_object.c | 12 +-- net/rxrpc/proc.c | 8 +- net/rxrpc/rtt.c | 195 +++++++++++++++++++++++++++++++++++++++++++ net/rxrpc/sendmsg.c | 26 ++---- net/rxrpc/sysctl.c | 9 -- 17 files changed, 266 insertions(+), 155 deletions(-) create mode 100644 net/rxrpc/rtt.c (limited to 'fs') diff --git a/fs/afs/fs_probe.c b/fs/afs/fs_probe.c index a587767b6ae1..237352d3cb53 100644 --- a/fs/afs/fs_probe.c +++ b/fs/afs/fs_probe.c @@ -32,9 +32,8 @@ void afs_fileserver_probe_result(struct afs_call *call) struct afs_server *server = call->server; unsigned int server_index = call->server_index; unsigned int index = call->addr_ix; - unsigned int rtt = UINT_MAX; + unsigned int rtt_us; bool have_result = false; - u64 _rtt; int ret = call->error; _enter("%pU,%u", &server->uuid, index); @@ -93,15 +92,9 @@ responded: } } - /* Get the RTT and scale it to fit into a 32-bit value that represents - * over a minute of time so that we can access it with one instruction - * on a 32-bit system. - */ - _rtt = rxrpc_kernel_get_rtt(call->net->socket, call->rxcall); - _rtt /= 64; - rtt = (_rtt > UINT_MAX) ? UINT_MAX : _rtt; - if (rtt < server->probe.rtt) { - server->probe.rtt = rtt; + rtt_us = rxrpc_kernel_get_srtt(call->net->socket, call->rxcall); + if (rtt_us < server->probe.rtt) { + server->probe.rtt = rtt_us; alist->preferred = index; have_result = true; } @@ -113,8 +106,7 @@ out: spin_unlock(&server->probe_lock); _debug("probe [%u][%u] %pISpc rtt=%u ret=%d", - server_index, index, &alist->addrs[index].transport, - (unsigned int)rtt, ret); + server_index, index, &alist->addrs[index].transport, rtt_us, ret); have_result |= afs_fs_probe_done(server); if (have_result) diff --git a/fs/afs/vl_probe.c b/fs/afs/vl_probe.c index 858498cc1b05..e3aa013c2177 100644 --- a/fs/afs/vl_probe.c +++ b/fs/afs/vl_probe.c @@ -31,10 +31,9 @@ void afs_vlserver_probe_result(struct afs_call *call) struct afs_addr_list *alist = call->alist; struct afs_vlserver *server = call->vlserver; unsigned int server_index = call->server_index; + unsigned int rtt_us = 0; unsigned int index = call->addr_ix; - unsigned int rtt = UINT_MAX; bool have_result = false; - u64 _rtt; int ret = call->error; _enter("%s,%u,%u,%d,%d", server->name, server_index, index, ret, call->abort_code); @@ -93,15 +92,9 @@ responded: } } - /* Get the RTT and scale it to fit into a 32-bit value that represents - * over a minute of time so that we can access it with one instruction - * on a 32-bit system. - */ - _rtt = rxrpc_kernel_get_rtt(call->net->socket, call->rxcall); - _rtt /= 64; - rtt = (_rtt > UINT_MAX) ? UINT_MAX : _rtt; - if (rtt < server->probe.rtt) { - server->probe.rtt = rtt; + rtt_us = rxrpc_kernel_get_srtt(call->net->socket, call->rxcall); + if (rtt_us < server->probe.rtt) { + server->probe.rtt = rtt_us; alist->preferred = index; have_result = true; } @@ -113,8 +106,7 @@ out: spin_unlock(&server->probe_lock); _debug("probe [%u][%u] %pISpc rtt=%u ret=%d", - server_index, index, &alist->addrs[index].transport, - (unsigned int)rtt, ret); + server_index, index, &alist->addrs[index].transport, rtt_us, ret); have_result |= afs_vl_probe_done(server); if (have_result) { diff --git a/include/net/af_rxrpc.h b/include/net/af_rxrpc.h index 04e97bab6f28..ab988940bf04 100644 --- a/include/net/af_rxrpc.h +++ b/include/net/af_rxrpc.h @@ -59,7 +59,7 @@ bool rxrpc_kernel_abort_call(struct socket *, struct rxrpc_call *, void rxrpc_kernel_end_call(struct socket *, struct rxrpc_call *); void rxrpc_kernel_get_peer(struct socket *, struct rxrpc_call *, struct sockaddr_rxrpc *); -u64 rxrpc_kernel_get_rtt(struct socket *, struct rxrpc_call *); +u32 rxrpc_kernel_get_srtt(struct socket *, struct rxrpc_call *); int rxrpc_kernel_charge_accept(struct socket *, rxrpc_notify_rx_t, rxrpc_user_attach_call_t, unsigned long, gfp_t, unsigned int); diff --git a/include/trace/events/rxrpc.h b/include/trace/events/rxrpc.h index 191fe447f990..ab75f261f04a 100644 --- a/include/trace/events/rxrpc.h +++ b/include/trace/events/rxrpc.h @@ -1112,18 +1112,17 @@ TRACE_EVENT(rxrpc_rtt_tx, TRACE_EVENT(rxrpc_rtt_rx, TP_PROTO(struct rxrpc_call *call, enum rxrpc_rtt_rx_trace why, rxrpc_serial_t send_serial, rxrpc_serial_t resp_serial, - s64 rtt, u8 nr, s64 avg), + u32 rtt, u32 rto), - TP_ARGS(call, why, send_serial, resp_serial, rtt, nr, avg), + TP_ARGS(call, why, send_serial, resp_serial, rtt, rto), TP_STRUCT__entry( __field(unsigned int, call ) __field(enum rxrpc_rtt_rx_trace, why ) - __field(u8, nr ) __field(rxrpc_serial_t, send_serial ) __field(rxrpc_serial_t, resp_serial ) - __field(s64, rtt ) - __field(u64, avg ) + __field(u32, rtt ) + __field(u32, rto ) ), TP_fast_assign( @@ -1132,18 +1131,16 @@ TRACE_EVENT(rxrpc_rtt_rx, __entry->send_serial = send_serial; __entry->resp_serial = resp_serial; __entry->rtt = rtt; - __entry->nr = nr; - __entry->avg = avg; + __entry->rto = rto; ), - TP_printk("c=%08x %s sr=%08x rr=%08x rtt=%lld nr=%u avg=%lld", + TP_printk("c=%08x %s sr=%08x rr=%08x rtt=%u rto=%u", __entry->call, __print_symbolic(__entry->why, rxrpc_rtt_rx_traces), __entry->send_serial, __entry->resp_serial, __entry->rtt, - __entry->nr, - __entry->avg) + __entry->rto) ); TRACE_EVENT(rxrpc_timer, diff --git a/net/rxrpc/Makefile b/net/rxrpc/Makefile index 6ffb7e9887ce..ddd0f95713a9 100644 --- a/net/rxrpc/Makefile +++ b/net/rxrpc/Makefile @@ -25,6 +25,7 @@ rxrpc-y := \ peer_event.o \ peer_object.o \ recvmsg.o \ + rtt.o \ security.o \ sendmsg.o \ skbuff.o \ diff --git a/net/rxrpc/ar-internal.h b/net/rxrpc/ar-internal.h index 3eb1ab40ca5c..9fe264bec70c 100644 --- a/net/rxrpc/ar-internal.h +++ b/net/rxrpc/ar-internal.h @@ -7,6 +7,7 @@ #include #include +#include #include #include #include @@ -311,11 +312,14 @@ struct rxrpc_peer { #define RXRPC_RTT_CACHE_SIZE 32 spinlock_t rtt_input_lock; /* RTT lock for input routine */ ktime_t rtt_last_req; /* Time of last RTT request */ - u64 rtt; /* Current RTT estimate (in nS) */ - u64 rtt_sum; /* Sum of cache contents */ - u64 rtt_cache[RXRPC_RTT_CACHE_SIZE]; /* Determined RTT cache */ - u8 rtt_cursor; /* next entry at which to insert */ - u8 rtt_usage; /* amount of cache actually used */ + unsigned int rtt_count; /* Number of samples we've got */ + + u32 srtt_us; /* smoothed round trip time << 3 in usecs */ + u32 mdev_us; /* medium deviation */ + u32 mdev_max_us; /* maximal mdev for the last rtt period */ + u32 rttvar_us; /* smoothed mdev_max */ + u32 rto_j; /* Retransmission timeout in jiffies */ + u8 backoff; /* Backoff timeout */ u8 cong_cwnd; /* Congestion window size */ }; @@ -1041,7 +1045,6 @@ extern unsigned long rxrpc_idle_ack_delay; extern unsigned int rxrpc_rx_window_size; extern unsigned int rxrpc_rx_mtu; extern unsigned int rxrpc_rx_jumbo_max; -extern unsigned long rxrpc_resend_timeout; extern const s8 rxrpc_ack_priority[]; @@ -1069,8 +1072,6 @@ void rxrpc_send_keepalive(struct rxrpc_peer *); * peer_event.c */ void rxrpc_error_report(struct sock *); -void rxrpc_peer_add_rtt(struct rxrpc_call *, enum rxrpc_rtt_rx_trace, - rxrpc_serial_t, rxrpc_serial_t, ktime_t, ktime_t); void rxrpc_peer_keepalive_worker(struct work_struct *); /* @@ -1102,6 +1103,14 @@ extern const struct seq_operations rxrpc_peer_seq_ops; void rxrpc_notify_socket(struct rxrpc_call *); int rxrpc_recvmsg(struct socket *, struct msghdr *, size_t, int); +/* + * rtt.c + */ +void rxrpc_peer_add_rtt(struct rxrpc_call *, enum rxrpc_rtt_rx_trace, + rxrpc_serial_t, rxrpc_serial_t, ktime_t, ktime_t); +unsigned long rxrpc_get_rto_backoff(struct rxrpc_peer *, bool); +void rxrpc_peer_init_rtt(struct rxrpc_peer *); + /* * rxkad.c */ diff --git a/net/rxrpc/call_accept.c b/net/rxrpc/call_accept.c index 70e44abf106c..b7611cc159e5 100644 --- a/net/rxrpc/call_accept.c +++ b/net/rxrpc/call_accept.c @@ -248,7 +248,7 @@ static void rxrpc_send_ping(struct rxrpc_call *call, struct sk_buff *skb) struct rxrpc_skb_priv *sp = rxrpc_skb(skb); ktime_t now = skb->tstamp; - if (call->peer->rtt_usage < 3 || + if (call->peer->rtt_count < 3 || ktime_before(ktime_add_ms(call->peer->rtt_last_req, 1000), now)) rxrpc_propose_ACK(call, RXRPC_ACK_PING, sp->hdr.serial, true, true, diff --git a/net/rxrpc/call_event.c b/net/rxrpc/call_event.c index cedbbb3a7c2e..2a65ac41055f 100644 --- a/net/rxrpc/call_event.c +++ b/net/rxrpc/call_event.c @@ -111,8 +111,8 @@ static void __rxrpc_propose_ACK(struct rxrpc_call *call, u8 ack_reason, } else { unsigned long now = jiffies, ack_at; - if (call->peer->rtt_usage > 0) - ack_at = nsecs_to_jiffies(call->peer->rtt); + if (call->peer->srtt_us != 0) + ack_at = usecs_to_jiffies(call->peer->srtt_us >> 3); else ack_at = expiry; @@ -157,24 +157,18 @@ static void rxrpc_congestion_timeout(struct rxrpc_call *call) static void rxrpc_resend(struct rxrpc_call *call, unsigned long now_j) { struct sk_buff *skb; - unsigned long resend_at; + unsigned long resend_at, rto_j; rxrpc_seq_t cursor, seq, top; - ktime_t now, max_age, oldest, ack_ts, timeout, min_timeo; + ktime_t now, max_age, oldest, ack_ts; int ix; u8 annotation, anno_type, retrans = 0, unacked = 0; _enter("{%d,%d}", call->tx_hard_ack, call->tx_top); - if (call->peer->rtt_usage > 1) - timeout = ns_to_ktime(call->peer->rtt * 3 / 2); - else - timeout = ms_to_ktime(rxrpc_resend_timeout); - min_timeo = ns_to_ktime((1000000000 / HZ) * 4); - if (ktime_before(timeout, min_timeo)) - timeout = min_timeo; + rto_j = call->peer->rto_j; now = ktime_get_real(); - max_age = ktime_sub(now, timeout); + max_age = ktime_sub(now, jiffies_to_usecs(rto_j)); spin_lock_bh(&call->lock); @@ -219,7 +213,7 @@ static void rxrpc_resend(struct rxrpc_call *call, unsigned long now_j) } resend_at = nsecs_to_jiffies(ktime_to_ns(ktime_sub(now, oldest))); - resend_at += jiffies + rxrpc_resend_timeout; + resend_at += jiffies + rto_j; WRITE_ONCE(call->resend_at, resend_at); if (unacked) @@ -234,7 +228,7 @@ static void rxrpc_resend(struct rxrpc_call *call, unsigned long now_j) rxrpc_timer_set_for_resend); spin_unlock_bh(&call->lock); ack_ts = ktime_sub(now, call->acks_latest_ts); - if (ktime_to_ns(ack_ts) < call->peer->rtt) + if (ktime_to_us(ack_ts) < (call->peer->srtt_us >> 3)) goto out; rxrpc_propose_ACK(call, RXRPC_ACK_PING, 0, true, false, rxrpc_propose_ack_ping_for_lost_ack); diff --git a/net/rxrpc/input.c b/net/rxrpc/input.c index 69e09d69c896..e438bfd3fdf5 100644 --- a/net/rxrpc/input.c +++ b/net/rxrpc/input.c @@ -91,11 +91,11 @@ static void rxrpc_congestion_management(struct rxrpc_call *call, /* We analyse the number of packets that get ACK'd per RTT * period and increase the window if we managed to fill it. */ - if (call->peer->rtt_usage == 0) + if (call->peer->rtt_count == 0) goto out; if (ktime_before(skb->tstamp, - ktime_add_ns(call->cong_tstamp, - call->peer->rtt))) + ktime_add_us(call->cong_tstamp, + call->peer->srtt_us >> 3))) goto out_no_clear_ca; change = rxrpc_cong_rtt_window_end; call->cong_tstamp = skb->tstamp; diff --git a/net/rxrpc/misc.c b/net/rxrpc/misc.c index 214405f75346..d4144fd86f84 100644 --- a/net/rxrpc/misc.c +++ b/net/rxrpc/misc.c @@ -63,11 +63,6 @@ unsigned int rxrpc_rx_mtu = 5692; */ unsigned int rxrpc_rx_jumbo_max = 4; -/* - * Time till packet resend (in milliseconds). - */ -unsigned long rxrpc_resend_timeout = 4 * HZ; - const s8 rxrpc_ack_priority[] = { [0] = 0, [RXRPC_ACK_DELAY] = 1, diff --git a/net/rxrpc/output.c b/net/rxrpc/output.c index 90e263c6aa69..f8b632a5c619 100644 --- a/net/rxrpc/output.c +++ b/net/rxrpc/output.c @@ -369,7 +369,7 @@ int rxrpc_send_data_packet(struct rxrpc_call *call, struct sk_buff *skb, (test_and_clear_bit(RXRPC_CALL_EV_ACK_LOST, &call->events) || retrans || call->cong_mode == RXRPC_CALL_SLOW_START || - (call->peer->rtt_usage < 3 && sp->hdr.seq & 1) || + (call->peer->rtt_count < 3 && sp->hdr.seq & 1) || ktime_before(ktime_add_ms(call->peer->rtt_last_req, 1000), ktime_get_real()))) whdr.flags |= RXRPC_REQUEST_ACK; @@ -423,13 +423,10 @@ done: if (whdr.flags & RXRPC_REQUEST_ACK) { call->peer->rtt_last_req = skb->tstamp; trace_rxrpc_rtt_tx(call, rxrpc_rtt_tx_data, serial); - if (call->peer->rtt_usage > 1) { + if (call->peer->rtt_count > 1) { unsigned long nowj = jiffies, ack_lost_at; - ack_lost_at = nsecs_to_jiffies(2 * call->peer->rtt); - if (ack_lost_at < 1) - ack_lost_at = 1; - + ack_lost_at = rxrpc_get_rto_backoff(call->peer, retrans); ack_lost_at += nowj; WRITE_ONCE(call->ack_lost_at, ack_lost_at); rxrpc_reduce_call_timer(call, ack_lost_at, nowj, diff --git a/net/rxrpc/peer_event.c b/net/rxrpc/peer_event.c index 923b263c401b..b1449d971883 100644 --- a/net/rxrpc/peer_event.c +++ b/net/rxrpc/peer_event.c @@ -295,52 +295,6 @@ static void rxrpc_distribute_error(struct rxrpc_peer *peer, int error, } } -/* - * Add RTT information to cache. This is called in softirq mode and has - * exclusive access to the peer RTT data. - */ -void rxrpc_peer_add_rtt(struct rxrpc_call *call, enum rxrpc_rtt_rx_trace why, - rxrpc_serial_t send_serial, rxrpc_serial_t resp_serial, - ktime_t send_time, ktime_t resp_time) -{ - struct rxrpc_peer *peer = call->peer; - s64 rtt; - u64 sum = peer->rtt_sum, avg; - u8 cursor = peer->rtt_cursor, usage = peer->rtt_usage; - - rtt = ktime_to_ns(ktime_sub(resp_time, send_time)); - if (rtt < 0) - return; - - spin_lock(&peer->rtt_input_lock); - - /* Replace the oldest datum in the RTT buffer */ - sum -= peer->rtt_cache[cursor]; - sum += rtt; - peer->rtt_cache[cursor] = rtt; - peer->rtt_cursor = (cursor + 1) & (RXRPC_RTT_CACHE_SIZE - 1); - peer->rtt_sum = sum; - if (usage < RXRPC_RTT_CACHE_SIZE) { - usage++; - peer->rtt_usage = usage; - } - - spin_unlock(&peer->rtt_input_lock); - - /* Now recalculate the average */ - if (usage == RXRPC_RTT_CACHE_SIZE) { - avg = sum / RXRPC_RTT_CACHE_SIZE; - } else { - avg = sum; - do_div(avg, usage); - } - - /* Don't need to update this under lock */ - peer->rtt = avg; - trace_rxrpc_rtt_rx(call, why, send_serial, resp_serial, rtt, - usage, avg); -} - /* * Perform keep-alive pings. */ diff --git a/net/rxrpc/peer_object.c b/net/rxrpc/peer_object.c index 452163eadb98..ca29976bb193 100644 --- a/net/rxrpc/peer_object.c +++ b/net/rxrpc/peer_object.c @@ -225,6 +225,8 @@ struct rxrpc_peer *rxrpc_alloc_peer(struct rxrpc_local *local, gfp_t gfp) spin_lock_init(&peer->rtt_input_lock); peer->debug_id = atomic_inc_return(&rxrpc_debug_id); + rxrpc_peer_init_rtt(peer); + if (RXRPC_TX_SMSS > 2190) peer->cong_cwnd = 2; else if (RXRPC_TX_SMSS > 1095) @@ -497,14 +499,14 @@ void rxrpc_kernel_get_peer(struct socket *sock, struct rxrpc_call *call, EXPORT_SYMBOL(rxrpc_kernel_get_peer); /** - * rxrpc_kernel_get_rtt - Get a call's peer RTT + * rxrpc_kernel_get_srtt - Get a call's peer smoothed RTT * @sock: The socket on which the call is in progress. * @call: The call to query * - * Get the call's peer RTT. + * Get the call's peer smoothed RTT. */ -u64 rxrpc_kernel_get_rtt(struct socket *sock, struct rxrpc_call *call) +u32 rxrpc_kernel_get_srtt(struct socket *sock, struct rxrpc_call *call) { - return call->peer->rtt; + return call->peer->srtt_us >> 3; } -EXPORT_SYMBOL(rxrpc_kernel_get_rtt); +EXPORT_SYMBOL(rxrpc_kernel_get_srtt); diff --git a/net/rxrpc/proc.c b/net/rxrpc/proc.c index b9d053e42821..8b179e3c802a 100644 --- a/net/rxrpc/proc.c +++ b/net/rxrpc/proc.c @@ -222,7 +222,7 @@ static int rxrpc_peer_seq_show(struct seq_file *seq, void *v) seq_puts(seq, "Proto Local " " Remote " - " Use CW MTU LastUse RTT Rc\n" + " Use CW MTU LastUse RTT RTO\n" ); return 0; } @@ -236,15 +236,15 @@ static int rxrpc_peer_seq_show(struct seq_file *seq, void *v) now = ktime_get_seconds(); seq_printf(seq, "UDP %-47.47s %-47.47s %3u" - " %3u %5u %6llus %12llu %2u\n", + " %3u %5u %6llus %8u %8u\n", lbuff, rbuff, atomic_read(&peer->usage), peer->cong_cwnd, peer->mtu, now - peer->last_tx_at, - peer->rtt, - peer->rtt_cursor); + peer->srtt_us >> 3, + jiffies_to_usecs(peer->rto_j)); return 0; } diff --git a/net/rxrpc/rtt.c b/net/rxrpc/rtt.c new file mode 100644 index 000000000000..928d8b34a3ee --- /dev/null +++ b/net/rxrpc/rtt.c @@ -0,0 +1,195 @@ +// SPDX-License-Identifier: GPL-2.0 +/* RTT/RTO calculation. + * + * Adapted from TCP for AF_RXRPC by David Howells (dhowells@redhat.com) + * + * https://tools.ietf.org/html/rfc6298 + * https://tools.ietf.org/html/rfc1122#section-4.2.3.1 + * http://ccr.sigcomm.org/archive/1995/jan95/ccr-9501-partridge87.pdf + */ + +#include +#include "ar-internal.h" + +#define RXRPC_RTO_MAX ((unsigned)(120 * HZ)) +#define RXRPC_TIMEOUT_INIT ((unsigned)(1*HZ)) /* RFC6298 2.1 initial RTO value */ +#define rxrpc_jiffies32 ((u32)jiffies) /* As rxrpc_jiffies32 */ +#define rxrpc_min_rtt_wlen 300 /* As sysctl_tcp_min_rtt_wlen */ + +static u32 rxrpc_rto_min_us(struct rxrpc_peer *peer) +{ + return 200; +} + +static u32 __rxrpc_set_rto(const struct rxrpc_peer *peer) +{ + return _usecs_to_jiffies((peer->srtt_us >> 3) + peer->rttvar_us); +} + +static u32 rxrpc_bound_rto(u32 rto) +{ + return min(rto, RXRPC_RTO_MAX); +} + +/* + * Called to compute a smoothed rtt estimate. The data fed to this + * routine either comes from timestamps, or from segments that were + * known _not_ to have been retransmitted [see Karn/Partridge + * Proceedings SIGCOMM 87]. The algorithm is from the SIGCOMM 88 + * piece by Van Jacobson. + * NOTE: the next three routines used to be one big routine. + * To save cycles in the RFC 1323 implementation it was better to break + * it up into three procedures. -- erics + */ +static void rxrpc_rtt_estimator(struct rxrpc_peer *peer, long sample_rtt_us) +{ + long m = sample_rtt_us; /* RTT */ + u32 srtt = peer->srtt_us; + + /* The following amusing code comes from Jacobson's + * article in SIGCOMM '88. Note that rtt and mdev + * are scaled versions of rtt and mean deviation. + * This is designed to be as fast as possible + * m stands for "measurement". + * + * On a 1990 paper the rto value is changed to: + * RTO = rtt + 4 * mdev + * + * Funny. This algorithm seems to be very broken. + * These formulae increase RTO, when it should be decreased, increase + * too slowly, when it should be increased quickly, decrease too quickly + * etc. I guess in BSD RTO takes ONE value, so that it is absolutely + * does not matter how to _calculate_ it. Seems, it was trap + * that VJ failed to avoid. 8) + */ + if (srtt != 0) { + m -= (srtt >> 3); /* m is now error in rtt est */ + srtt += m; /* rtt = 7/8 rtt + 1/8 new */ + if (m < 0) { + m = -m; /* m is now abs(error) */ + m -= (peer->mdev_us >> 2); /* similar update on mdev */ + /* This is similar to one of Eifel findings. + * Eifel blocks mdev updates when rtt decreases. + * This solution is a bit different: we use finer gain + * for mdev in this case (alpha*beta). + * Like Eifel it also prevents growth of rto, + * but also it limits too fast rto decreases, + * happening in pure Eifel. + */ + if (m > 0) + m >>= 3; + } else { + m -= (peer->mdev_us >> 2); /* similar update on mdev */ + } + + peer->mdev_us += m; /* mdev = 3/4 mdev + 1/4 new */ + if (peer->mdev_us > peer->mdev_max_us) { + peer->mdev_max_us = peer->mdev_us; + if (peer->mdev_max_us > peer->rttvar_us) + peer->rttvar_us = peer->mdev_max_us; + } + } else { + /* no previous measure. */ + srtt = m << 3; /* take the measured time to be rtt */ + peer->mdev_us = m << 1; /* make sure rto = 3*rtt */ + peer->rttvar_us = max(peer->mdev_us, rxrpc_rto_min_us(peer)); + peer->mdev_max_us = peer->rttvar_us; + } + + peer->srtt_us = max(1U, srtt); +} + +/* + * Calculate rto without backoff. This is the second half of Van Jacobson's + * routine referred to above. + */ +static void rxrpc_set_rto(struct rxrpc_peer *peer) +{ + u32 rto; + + /* 1. If rtt variance happened to be less 50msec, it is hallucination. + * It cannot be less due to utterly erratic ACK generation made + * at least by solaris and freebsd. "Erratic ACKs" has _nothing_ + * to do with delayed acks, because at cwnd>2 true delack timeout + * is invisible. Actually, Linux-2.4 also generates erratic + * ACKs in some circumstances. + */ + rto = __rxrpc_set_rto(peer); + + /* 2. Fixups made earlier cannot be right. + * If we do not estimate RTO correctly without them, + * all the algo is pure shit and should be replaced + * with correct one. It is exactly, which we pretend to do. + */ + + /* NOTE: clamping at RXRPC_RTO_MIN is not required, current algo + * guarantees that rto is higher. + */ + peer->rto_j = rxrpc_bound_rto(rto); +} + +static void rxrpc_ack_update_rtt(struct rxrpc_peer *peer, long rtt_us) +{ + if (rtt_us < 0) + return; + + //rxrpc_update_rtt_min(peer, rtt_us); + rxrpc_rtt_estimator(peer, rtt_us); + rxrpc_set_rto(peer); + + /* RFC6298: only reset backoff on valid RTT measurement. */ + peer->backoff = 0; +} + +/* + * Add RTT information to cache. This is called in softirq mode and has + * exclusive access to the peer RTT data. + */ +void rxrpc_peer_add_rtt(struct rxrpc_call *call, enum rxrpc_rtt_rx_trace why, + rxrpc_serial_t send_serial, rxrpc_serial_t resp_serial, + ktime_t send_time, ktime_t resp_time) +{ + struct rxrpc_peer *peer = call->peer; + s64 rtt_us; + + rtt_us = ktime_to_us(ktime_sub(resp_time, send_time)); + if (rtt_us < 0) + return; + + spin_lock(&peer->rtt_input_lock); + rxrpc_ack_update_rtt(peer, rtt_us); + if (peer->rtt_count < 3) + peer->rtt_count++; + spin_unlock(&peer->rtt_input_lock); + + trace_rxrpc_rtt_rx(call, why, send_serial, resp_serial, + peer->srtt_us >> 3, peer->rto_j); +} + +/* + * Get the retransmission timeout to set in jiffies, backing it off each time + * we retransmit. + */ +unsigned long rxrpc_get_rto_backoff(struct rxrpc_peer *peer, bool retrans) +{ + u64 timo_j; + u8 backoff = READ_ONCE(peer->backoff); + + timo_j = peer->rto_j; + timo_j <<= backoff; + if (retrans && timo_j * 2 <= RXRPC_RTO_MAX) + WRITE_ONCE(peer->backoff, backoff + 1); + + if (timo_j < 1) + timo_j = 1; + + return timo_j; +} + +void rxrpc_peer_init_rtt(struct rxrpc_peer *peer) +{ + peer->rto_j = RXRPC_TIMEOUT_INIT; + peer->mdev_us = jiffies_to_usecs(RXRPC_TIMEOUT_INIT); + peer->backoff = 0; + //minmax_reset(&peer->rtt_min, rxrpc_jiffies32, ~0U); +} diff --git a/net/rxrpc/sendmsg.c b/net/rxrpc/sendmsg.c index 0fcf157aa09f..5e9c43d4a314 100644 --- a/net/rxrpc/sendmsg.c +++ b/net/rxrpc/sendmsg.c @@ -66,15 +66,14 @@ static int rxrpc_wait_for_tx_window_waitall(struct rxrpc_sock *rx, struct rxrpc_call *call) { rxrpc_seq_t tx_start, tx_win; - signed long rtt2, timeout; - u64 rtt; + signed long rtt, timeout; - rtt = READ_ONCE(call->peer->rtt); - rtt2 = nsecs_to_jiffies64(rtt) * 2; - if (rtt2 < 2) - rtt2 = 2; + rtt = READ_ONCE(call->peer->srtt_us) >> 3; + rtt = usecs_to_jiffies(rtt) * 2; + if (rtt < 2) + rtt = 2; - timeout = rtt2; + timeout = rtt; tx_start = READ_ONCE(call->tx_hard_ack); for (;;) { @@ -92,7 +91,7 @@ static int rxrpc_wait_for_tx_window_waitall(struct rxrpc_sock *rx, return -EINTR; if (tx_win != tx_start) { - timeout = rtt2; + timeout = rtt; tx_start = tx_win; } @@ -271,16 +270,9 @@ static int rxrpc_queue_packet(struct rxrpc_sock *rx, struct rxrpc_call *call, _debug("need instant resend %d", ret); rxrpc_instant_resend(call, ix); } else { - unsigned long now = jiffies, resend_at; + unsigned long now = jiffies; + unsigned long resend_at = now + call->peer->rto_j; - if (call->peer->rtt_usage > 1) - resend_at = nsecs_to_jiffies(call->peer->rtt * 3 / 2); - else - resend_at = rxrpc_resend_timeout; - if (resend_at < 1) - resend_at = 1; - - resend_at += now; WRITE_ONCE(call->resend_at, resend_at); rxrpc_reduce_call_timer(call, resend_at, now, rxrpc_timer_set_for_send); diff --git a/net/rxrpc/sysctl.c b/net/rxrpc/sysctl.c index 2bbb38161851..18dade4e6f9a 100644 --- a/net/rxrpc/sysctl.c +++ b/net/rxrpc/sysctl.c @@ -71,15 +71,6 @@ static struct ctl_table rxrpc_sysctl_table[] = { .extra1 = (void *)&one_jiffy, .extra2 = (void *)&max_jiffies, }, - { - .procname = "resend_timeout", - .data = &rxrpc_resend_timeout, - .maxlen = sizeof(unsigned long), - .mode = 0644, - .proc_handler = proc_doulongvec_ms_jiffies_minmax, - .extra1 = (void *)&one_jiffy, - .extra2 = (void *)&max_jiffies, - }, /* Non-time values */ { -- cgit v1.2.3 From 29fe839976266bc7c55b927360a1daae57477723 Mon Sep 17 00:00:00 2001 From: "J. Bruce Fields" Date: Mon, 11 May 2020 10:02:48 -0400 Subject: nfs: fix NULL deference in nfs4_get_valid_delegation We add the new state to the nfsi->open_states list, making it potentially visible to other threads, before we've finished initializing it. That wasn't a problem when all the readers were also taking the i_lock (as we do here), but since we switched to RCU, there's now a possibility that a reader could see the partially initialized state. Symptoms observed were a crash when another thread called nfs4_get_valid_delegation() on a NULL inode, resulting in an oops like: BUG: unable to handle page fault for address: ffffffffffffffb0 ... RIP: 0010:nfs4_get_valid_delegation+0x6/0x30 [nfsv4] ... Call Trace: nfs4_open_prepare+0x80/0x1c0 [nfsv4] __rpc_execute+0x75/0x390 [sunrpc] ? finish_task_switch+0x75/0x260 rpc_async_schedule+0x29/0x40 [sunrpc] process_one_work+0x1ad/0x370 worker_thread+0x30/0x390 ? create_worker+0x1a0/0x1a0 kthread+0x10c/0x130 ? kthread_park+0x80/0x80 ret_from_fork+0x22/0x30 Fixes: 9ae075fdd190 "NFSv4: Convert open state lookup to use RCU" Reviewed-by: Seiichi Ikarashi Tested-by: Daisuke Matsuda Tested-by: Masayoshi Mizuma Signed-off-by: J. Bruce Fields Cc: stable@vger.kernel.org # v4.20+ Signed-off-by: Trond Myklebust --- fs/nfs/nfs4state.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index ac93715c05a4..a8dc25ce48bb 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -734,9 +734,9 @@ nfs4_get_open_state(struct inode *inode, struct nfs4_state_owner *owner) state = new; state->owner = owner; atomic_inc(&owner->so_count); - list_add_rcu(&state->inode_states, &nfsi->open_states); ihold(inode); state->inode = inode; + list_add_rcu(&state->inode_states, &nfsi->open_states); spin_unlock(&inode->i_lock); /* Note: The reclaim code dictates that we add stateless * and read-only stateids to the end of the list */ -- cgit v1.2.3 From f304a809a9f1eec23e023f3aae6c768a08495244 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 11 May 2020 10:42:04 -0400 Subject: NFS: Don't use RPC_TASK_CRED_NOREF with delegreturn We are not guaranteed that the credential will remain pinned. Fixes: 612965072020 ("NFSv4: Avoid referencing the cred unnecessarily during NFSv4 I/O") Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index a0c1e653a935..9056f3dd380e 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -6347,7 +6347,7 @@ static int _nfs4_proc_delegreturn(struct inode *inode, const struct cred *cred, .rpc_client = server->client, .rpc_message = &msg, .callback_ops = &nfs4_delegreturn_ops, - .flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF | RPC_TASK_TIMEOUT, + .flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT, }; int status = 0; -- cgit v1.2.3 From 9aafc1b0187322fa4fd4eb905d0903172237206c Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Tue, 5 May 2020 21:33:31 +0300 Subject: ovl: potential crash in ovl_fid_to_fh() The "buflen" value comes from the user and there is a potential that it could be zero. In do_handle_to_path() we know that "handle->handle_bytes" is non-zero and we do: handle_dwords = handle->handle_bytes >> 2; So values 1-3 become zero. Then in ovl_fh_to_dentry() we do: int len = fh_len << 2; So now len is in the "0,4-128" range and a multiple of 4. But if "buflen" is zero it will try to copy negative bytes when we do the memcpy in ovl_fid_to_fh(). memcpy(&fh->fb, fid, buflen - OVL_FH_WIRE_OFFSET); And that will lead to a crash. Thanks to Amir Goldstein for his help with this patch. Fixes: cbe7fba8edfc ("ovl: make sure that real fid is 32bit aligned in memory") Signed-off-by: Dan Carpenter Reviewed-by: Amir Goldstein Cc: # v5.5 Signed-off-by: Miklos Szeredi --- fs/overlayfs/export.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'fs') diff --git a/fs/overlayfs/export.c b/fs/overlayfs/export.c index 475c61f53f0f..ed5c1078919c 100644 --- a/fs/overlayfs/export.c +++ b/fs/overlayfs/export.c @@ -783,6 +783,9 @@ static struct ovl_fh *ovl_fid_to_fh(struct fid *fid, int buflen, int fh_type) if (fh_type != OVL_FILEID_V0) return ERR_PTR(-EINVAL); + if (buflen <= OVL_FH_WIRE_OFFSET) + return ERR_PTR(-EINVAL); + fh = kzalloc(buflen, GFP_KERNEL); if (!fh) return ERR_PTR(-ENOMEM); -- cgit v1.2.3 From 4fa7ef69e2918eb89c95e57978d839c0a31f332c Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 13 May 2020 09:55:36 -0400 Subject: NFS/pnfs: Don't use RPC_TASK_CRED_NOREF with pnfs When we're doing pnfs then the credential being used for the RPC call is not necessarily the same as the one used in the open context, so don't use RPC_TASK_CRED_NOREF. Fixes: 612965072020 ("NFSv4: Avoid referencing the cred unnecessarily during NFSv4 I/O") Signed-off-by: Trond Myklebust --- fs/nfs/pagelist.c | 5 +++-- fs/nfs/pnfs_nfs.c | 3 ++- fs/nfs/write.c | 4 ++-- 3 files changed, 7 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c index f61f96603df7..6ca421cbe19c 100644 --- a/fs/nfs/pagelist.c +++ b/fs/nfs/pagelist.c @@ -752,7 +752,7 @@ int nfs_initiate_pgio(struct rpc_clnt *clnt, struct nfs_pgio_header *hdr, .callback_ops = call_ops, .callback_data = hdr, .workqueue = nfsiod_workqueue, - .flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF | flags, + .flags = RPC_TASK_ASYNC | flags, }; hdr->rw_ops->rw_initiate(hdr, &msg, rpc_ops, &task_setup_data, how); @@ -950,7 +950,8 @@ static int nfs_generic_pg_pgios(struct nfs_pageio_descriptor *desc) hdr->cred, NFS_PROTO(hdr->inode), desc->pg_rpc_callops, - desc->pg_ioflags, 0); + desc->pg_ioflags, + RPC_TASK_CRED_NOREF); return ret; } diff --git a/fs/nfs/pnfs_nfs.c b/fs/nfs/pnfs_nfs.c index e7ddbce48321..679767ac258d 100644 --- a/fs/nfs/pnfs_nfs.c +++ b/fs/nfs/pnfs_nfs.c @@ -536,7 +536,8 @@ pnfs_generic_commit_pagelist(struct inode *inode, struct list_head *mds_pages, nfs_init_commit(data, NULL, NULL, cinfo); nfs_initiate_commit(NFS_CLIENT(inode), data, NFS_PROTO(data->inode), - data->mds_ops, how, 0); + data->mds_ops, how, + RPC_TASK_CRED_NOREF); } else { nfs_init_commit(data, NULL, data->lseg, cinfo); initiate_commit(data, how); diff --git a/fs/nfs/write.c b/fs/nfs/write.c index df4b87c30ac9..1e767f779c49 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -1695,7 +1695,7 @@ int nfs_initiate_commit(struct rpc_clnt *clnt, struct nfs_commit_data *data, .callback_ops = call_ops, .callback_data = data, .workqueue = nfsiod_workqueue, - .flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF | flags, + .flags = RPC_TASK_ASYNC | flags, .priority = priority, }; /* Set up the initial task struct. */ @@ -1813,7 +1813,7 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how, nfs_init_commit(data, head, NULL, cinfo); atomic_inc(&cinfo->mds->rpcs_out); return nfs_initiate_commit(NFS_CLIENT(inode), data, NFS_PROTO(inode), - data->mds_ops, how, 0); + data->mds_ops, how, RPC_TASK_CRED_NOREF); } /* -- cgit v1.2.3 From 9d9e88a24c1f20ebfc2f28b1762ce78c0b9e1cb3 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Wed, 13 May 2020 12:53:19 -0600 Subject: io_uring: polled fixed file must go through free iteration When we changed the file registration handling, it became important to iterate the bulk request freeing list for fixed files as well, or we miss dropping the fixed file reference. If not, we're leaking references, and we'll get a kworker stuck waiting for file references to disappear. This also means we can remove the special casing of fixed vs non-fixed files, we need to iterate for both and we can just rely on __io_req_aux_free() doing io_put_file() instead of doing it manually. Fixes: 055895537302 ("io_uring: refactor file register/unregister/update handling") Signed-off-by: Jens Axboe --- fs/io_uring.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index 4c2fe06ae20b..70ae7e840c85 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -1394,10 +1394,6 @@ static void io_free_req_many(struct io_ring_ctx *ctx, struct req_batch *rb) for (i = 0; i < rb->to_free; i++) { struct io_kiocb *req = rb->reqs[i]; - if (req->flags & REQ_F_FIXED_FILE) { - req->file = NULL; - percpu_ref_put(req->fixed_file_refs); - } if (req->flags & REQ_F_INFLIGHT) inflight++; __io_req_aux_free(req); @@ -1670,7 +1666,7 @@ static inline bool io_req_multi_free(struct req_batch *rb, struct io_kiocb *req) if ((req->flags & REQ_F_LINK_HEAD) || io_is_fallback_req(req)) return false; - if (!(req->flags & REQ_F_FIXED_FILE) || req->io) + if (req->file || req->io) rb->need_iter++; rb->reqs[rb->to_free++] = req; -- cgit v1.2.3 From 9bd21d4b1a767c3abebec203342f3820dcb84662 Mon Sep 17 00:00:00 2001 From: Steve French Date: Wed, 13 May 2020 10:27:16 -0500 Subject: cifs: Fix null pointer check in cifs_read Coverity scan noted a redundant null check Coverity-id: 728517 Reported-by: Coverity Signed-off-by: Steve French Reviewed-by: Shyam Prasad N --- fs/cifs/file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cifs/file.c b/fs/cifs/file.c index 0b1528edebcf..75ddce8ef456 100644 --- a/fs/cifs/file.c +++ b/fs/cifs/file.c @@ -4060,7 +4060,7 @@ cifs_read(struct file *file, char *read_data, size_t read_size, loff_t *offset) * than it negotiated since it will refuse the read * then. */ - if ((tcon->ses) && !(tcon->ses->capabilities & + if (!(tcon->ses->capabilities & tcon->ses->server->vals->cap_large_files)) { current_read_size = min_t(uint, current_read_size, CIFSMaxBufSize); -- cgit v1.2.3 From 65759097d804d2a9ad2b687db436319704ba7019 Mon Sep 17 00:00:00 2001 From: Roman Penyaev Date: Wed, 13 May 2020 17:50:38 -0700 Subject: epoll: call final ep_events_available() check under the lock There is a possible race when ep_scan_ready_list() leaves ->rdllist and ->obflist empty for a short period of time although some events are pending. It is quite likely that ep_events_available() observes empty lists and goes to sleep. Since commit 339ddb53d373 ("fs/epoll: remove unnecessary wakeups of nested epoll") we are conservative in wakeups (there is only one place for wakeup and this is ep_poll_callback()), thus ep_events_available() must always observe correct state of two lists. The easiest and correct way is to do the final check under the lock. This does not impact the performance, since lock is taken anyway for adding a wait entry to the wait queue. The discussion of the problem can be found here: https://lore.kernel.org/linux-fsdevel/a2f22c3c-c25a-4bda-8339-a7bdaf17849e@akamai.com/ In this patch barrierless __set_current_state() is used. This is safe since waitqueue_active() is called under the same lock on wakeup side. Short-circuit for fatal signals (i.e. fatal_signal_pending() check) is moved to the line just before actual events harvesting routine. This is fully compliant to what is said in the comment of the patch where the actual fatal_signal_pending() check was added: c257a340ede0 ("fs, epoll: short circuit fetching events if thread has been killed"). Fixes: 339ddb53d373 ("fs/epoll: remove unnecessary wakeups of nested epoll") Reported-by: Jason Baron Reported-by: Randy Dunlap Signed-off-by: Roman Penyaev Signed-off-by: Andrew Morton Reviewed-by: Jason Baron Cc: Khazhismel Kumykov Cc: Alexander Viro Cc: Link: http://lkml.kernel.org/r/20200505145609.1865152-1-rpenyaev@suse.de Signed-off-by: Linus Torvalds --- fs/eventpoll.c | 48 ++++++++++++++++++++++++++++-------------------- 1 file changed, 28 insertions(+), 20 deletions(-) (limited to 'fs') diff --git a/fs/eventpoll.c b/fs/eventpoll.c index aba03ee749f8..12eebcdea9c8 100644 --- a/fs/eventpoll.c +++ b/fs/eventpoll.c @@ -1879,34 +1879,33 @@ fetch_events: * event delivery. */ init_wait(&wait); - write_lock_irq(&ep->lock); - __add_wait_queue_exclusive(&ep->wq, &wait); - write_unlock_irq(&ep->lock); + write_lock_irq(&ep->lock); /* - * We don't want to sleep if the ep_poll_callback() sends us - * a wakeup in between. That's why we set the task state - * to TASK_INTERRUPTIBLE before doing the checks. + * Barrierless variant, waitqueue_active() is called under + * the same lock on wakeup ep_poll_callback() side, so it + * is safe to avoid an explicit barrier. */ - set_current_state(TASK_INTERRUPTIBLE); + __set_current_state(TASK_INTERRUPTIBLE); + /* - * Always short-circuit for fatal signals to allow - * threads to make a timely exit without the chance of - * finding more events available and fetching - * repeatedly. + * Do the final check under the lock. ep_scan_ready_list() + * plays with two lists (->rdllist and ->ovflist) and there + * is always a race when both lists are empty for short + * period of time although events are pending, so lock is + * important. */ - if (fatal_signal_pending(current)) { - res = -EINTR; - break; + eavail = ep_events_available(ep); + if (!eavail) { + if (signal_pending(current)) + res = -EINTR; + else + __add_wait_queue_exclusive(&ep->wq, &wait); } + write_unlock_irq(&ep->lock); - eavail = ep_events_available(ep); - if (eavail) - break; - if (signal_pending(current)) { - res = -EINTR; + if (eavail || res) break; - } if (!schedule_hrtimeout_range(to, slack, HRTIMER_MODE_ABS)) { timed_out = 1; @@ -1927,6 +1926,15 @@ fetch_events: } send_events: + if (fatal_signal_pending(current)) { + /* + * Always short-circuit for fatal signals to allow + * threads to make a timely exit without the chance of + * finding more events available and fetching + * repeatedly. + */ + res = -EINTR; + } /* * Try to transfer events to user space. In case we get 0 events and * there's still timeout left over, we go trying again in search of -- cgit v1.2.3 From 8eed292bc8cbf737e46fb1c119d4c8f6dcb00650 Mon Sep 17 00:00:00 2001 From: Olga Kornievskaia Date: Thu, 14 May 2020 16:09:40 -0400 Subject: NFSv3: fix rpc receive buffer size for MOUNT call Prior to commit e3d3ab64dd66 ("SUNRPC: Use au_rslack when computing reply buffer size"), there was enough slack in the reply buffer to commodate filehandles of size 60bytes. However, the real problem was that the reply buffer size for the MOUNT operation was not correctly calculated. Received buffer size used the filehandle size for NFSv2 (32bytes) which is much smaller than the allowed filehandle size for the v3 mounts. Fix the reply buffer size (decode arguments size) for the MNT command. Fixes: 2c94b8eca1a2 ("SUNRPC: Use au_rslack when computing reply buffer size") Signed-off-by: Olga Kornievskaia Signed-off-by: Trond Myklebust --- fs/nfs/mount_clnt.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/nfs/mount_clnt.c b/fs/nfs/mount_clnt.c index 35c8cb2d7637..dda5c3e65d8d 100644 --- a/fs/nfs/mount_clnt.c +++ b/fs/nfs/mount_clnt.c @@ -30,6 +30,7 @@ #define encode_dirpath_sz (1 + XDR_QUADLEN(MNTPATHLEN)) #define MNT_status_sz (1) #define MNT_fhandle_sz XDR_QUADLEN(NFS2_FHSIZE) +#define MNT_fhandlev3_sz XDR_QUADLEN(NFS3_FHSIZE) #define MNT_authflav3_sz (1 + NFS_MAX_SECFLAVORS) /* @@ -37,7 +38,7 @@ */ #define MNT_enc_dirpath_sz encode_dirpath_sz #define MNT_dec_mountres_sz (MNT_status_sz + MNT_fhandle_sz) -#define MNT_dec_mountres3_sz (MNT_status_sz + MNT_fhandle_sz + \ +#define MNT_dec_mountres3_sz (MNT_status_sz + MNT_fhandlev3_sz + \ MNT_authflav3_sz) /* -- cgit v1.2.3 From a48137996063d22ffba77e077425f49873856ca5 Mon Sep 17 00:00:00 2001 From: Adam McCoy Date: Wed, 13 May 2020 11:53:30 +0000 Subject: cifs: fix leaked reference on requeued write Failed async writes that are requeued may not clean up a refcount on the file, which can result in a leaked open. This scenario arises very reliably when using persistent handles and a reconnect occurs while writing. cifs_writev_requeue only releases the reference if the write fails (rc != 0). The server->ops->async_writev operation will take its own reference, so the initial reference can always be released. Signed-off-by: Adam McCoy Signed-off-by: Steve French CC: Stable Reviewed-by: Pavel Shilovsky --- fs/cifs/cifssmb.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/cifs/cifssmb.c b/fs/cifs/cifssmb.c index 182b864b3075..5014a82391ff 100644 --- a/fs/cifs/cifssmb.c +++ b/fs/cifs/cifssmb.c @@ -2152,8 +2152,8 @@ cifs_writev_requeue(struct cifs_writedata *wdata) } } + kref_put(&wdata2->refcount, cifs_writedata_release); if (rc) { - kref_put(&wdata2->refcount, cifs_writedata_release); if (is_retryable_error(rc)) continue; i += nr_pages; -- cgit v1.2.3 From 583863ed918136412ddf14de2e12534f17cfdc6f Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Sun, 17 May 2020 09:20:00 -0600 Subject: io_uring: initialize ctx->sqo_wait earlier Ensure that ctx->sqo_wait is initialized as soon as the ctx is allocated, instead of deferring it to the offload setup. This fixes a syzbot reported lockdep complaint, which is really due to trying to wake_up on an uninitialized wait queue: RSP: 002b:00007fffb1fb9aa8 EFLAGS: 00000246 ORIG_RAX: 00000000000001a9 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 0000000000441319 RDX: 0000000000000001 RSI: 0000000020000140 RDI: 000000000000047b RBP: 0000000000010475 R08: 0000000000000001 R09: 00000000004002c8 R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000402260 R13: 00000000004022f0 R14: 0000000000000000 R15: 0000000000000000 INFO: trying to register non-static key. the code is fine but needs lockdep annotation. turning off the locking correctness validator. CPU: 1 PID: 7090 Comm: syz-executor222 Not tainted 5.7.0-rc1-next-20200415-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 Call Trace: __dump_stack lib/dump_stack.c:77 [inline] dump_stack+0x188/0x20d lib/dump_stack.c:118 assign_lock_key kernel/locking/lockdep.c:913 [inline] register_lock_class+0x1664/0x1760 kernel/locking/lockdep.c:1225 __lock_acquire+0x104/0x4c50 kernel/locking/lockdep.c:4234 lock_acquire+0x1f2/0x8f0 kernel/locking/lockdep.c:4934 __raw_spin_lock_irqsave include/linux/spinlock_api_smp.h:110 [inline] _raw_spin_lock_irqsave+0x8c/0xbf kernel/locking/spinlock.c:159 __wake_up_common_lock+0xb4/0x130 kernel/sched/wait.c:122 io_cqring_ev_posted+0xa5/0x1e0 fs/io_uring.c:1160 io_poll_remove_all fs/io_uring.c:4357 [inline] io_ring_ctx_wait_and_kill+0x2bc/0x5a0 fs/io_uring.c:7305 io_uring_create fs/io_uring.c:7843 [inline] io_uring_setup+0x115e/0x22b0 fs/io_uring.c:7870 do_syscall_64+0xf6/0x7d0 arch/x86/entry/common.c:295 entry_SYSCALL_64_after_hwframe+0x49/0xb3 RIP: 0033:0x441319 Code: e8 5c ae 02 00 48 83 c4 18 c3 0f 1f 80 00 00 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 0f 83 bb 0a fc ff c3 66 2e 0f 1f 84 00 00 00 00 RSP: 002b:00007fffb1fb9aa8 EFLAGS: 00000246 ORIG_RAX: 00000000000001a9 Reported-by: syzbot+8c91f5d054e998721c57@syzkaller.appspotmail.com Signed-off-by: Jens Axboe --- fs/io_uring.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index 70ae7e840c85..79c90eb28c0d 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -924,6 +924,7 @@ static struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p) goto err; ctx->flags = p->flags; + init_waitqueue_head(&ctx->sqo_wait); init_waitqueue_head(&ctx->cq_wait); INIT_LIST_HEAD(&ctx->cq_overflow_list); init_completion(&ctx->completions[0]); @@ -6837,7 +6838,6 @@ static int io_sq_offload_start(struct io_ring_ctx *ctx, { int ret; - init_waitqueue_head(&ctx->sqo_wait); mmgrab(current->mm); ctx->sqo_mm = current->mm; -- cgit v1.2.3 From 650b548129b60b0d23508351800108196f4aa89f Mon Sep 17 00:00:00 2001 From: Pavel Begunkov Date: Sun, 17 May 2020 14:02:11 +0300 Subject: io_uring: don't prepare DRAIN reqs twice If req->io is not NULL, it's already prepared. Don't do it again, it's dangerous. Signed-off-by: Pavel Begunkov Signed-off-by: Jens Axboe --- fs/io_uring.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index 79c90eb28c0d..51be07390634 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -5014,12 +5014,13 @@ static int io_req_defer(struct io_kiocb *req, const struct io_uring_sqe *sqe) if (!req_need_defer(req) && list_empty_careful(&ctx->defer_list)) return 0; - if (!req->io && io_alloc_async_ctx(req)) - return -EAGAIN; - - ret = io_req_defer_prep(req, sqe); - if (ret < 0) - return ret; + if (!req->io) { + if (io_alloc_async_ctx(req)) + return -EAGAIN; + ret = io_req_defer_prep(req, sqe); + if (ret < 0) + return ret; + } spin_lock_irq(&ctx->completion_lock); if (!req_need_defer(req) && list_empty(&ctx->defer_list)) { -- cgit v1.2.3 From bd2ab18a1d6267446eae1b47dd839050452bdf7f Mon Sep 17 00:00:00 2001 From: Pavel Begunkov Date: Sun, 17 May 2020 14:02:12 +0300 Subject: io_uring: fix FORCE_ASYNC req preparation As for other not inlined requests, alloc req->io for FORCE_ASYNC reqs, so they can be prepared properly. Signed-off-by: Pavel Begunkov Signed-off-by: Jens Axboe --- fs/io_uring.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index 51be07390634..f18cd98be783 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -5608,9 +5608,15 @@ fail_req: io_double_put_req(req); } } else if (req->flags & REQ_F_FORCE_ASYNC) { - ret = io_req_defer_prep(req, sqe); - if (unlikely(ret < 0)) - goto fail_req; + if (!req->io) { + ret = -EAGAIN; + if (io_alloc_async_ctx(req)) + goto fail_req; + ret = io_req_defer_prep(req, sqe); + if (unlikely(ret < 0)) + goto fail_req; + } + /* * Never try inline submit of IOSQE_ASYNC is set, go straight * to async execution. -- cgit v1.2.3 From f87d1c9559164294040e58f5e3b74a162bf7c6e8 Mon Sep 17 00:00:00 2001 From: "Eric W. Biederman" Date: Sat, 16 May 2020 16:29:20 -0500 Subject: exec: Move would_dump into flush_old_exec I goofed when I added mm->user_ns support to would_dump. I missed the fact that in the case of binfmt_loader, binfmt_em86, binfmt_misc, and binfmt_script bprm->file is reassigned. Which made the move of would_dump from setup_new_exec to __do_execve_file before exec_binprm incorrect as it can result in would_dump running on the script instead of the interpreter of the script. The net result is that the code stopped making unreadable interpreters undumpable. Which allows them to be ptraced and written to disk without special permissions. Oops. The move was necessary because the call in set_new_exec was after bprm->mm was no longer valid. To correct this mistake move the misplaced would_dump from __do_execve_file into flos_old_exec, before exec_mmap is called. I tested and confirmed that without this fix I can attach with gdb to a script with an unreadable interpreter, and with this fix I can not. Cc: stable@vger.kernel.org Fixes: f84df2a6f268 ("exec: Ensure mm->user_ns contains the execed files") Signed-off-by: "Eric W. Biederman" --- fs/exec.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/exec.c b/fs/exec.c index 06b4c550af5d..2c465119affc 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -1317,6 +1317,8 @@ int flush_old_exec(struct linux_binprm * bprm) */ set_mm_exe_file(bprm->mm, bprm->file); + would_dump(bprm, bprm->file); + /* * Release all of the old mmap stuff */ @@ -1876,8 +1878,6 @@ static int __do_execve_file(int fd, struct filename *filename, if (retval < 0) goto out; - would_dump(bprm, bprm->file); - retval = exec_binprm(bprm); if (retval < 0) goto out; -- cgit v1.2.3 From 948a7749454b1712f1b2f2429f9493eb3e4a89b0 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Sun, 17 May 2020 14:21:38 -0600 Subject: io_uring: remove dead check in io_splice() We checked for 'force_nonblock' higher up, so it's definitely false at this point. Kill the check, it's a remnant of when we tried to do inline splice without always punting to async context. Fixes: 2fb3e82284fc ("io_uring: punt splice async because of inode mutex") Signed-off-by: Jens Axboe --- fs/io_uring.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index f18cd98be783..ecfd7f054ef6 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -2772,11 +2772,8 @@ static int io_splice(struct io_kiocb *req, bool force_nonblock) poff_in = (sp->off_in == -1) ? NULL : &sp->off_in; poff_out = (sp->off_out == -1) ? NULL : &sp->off_out; - if (sp->len) { + if (sp->len) ret = do_splice(in, poff_in, out, poff_out, sp->len, flags); - if (force_nonblock && ret == -EAGAIN) - return -EAGAIN; - } io_put_file(req, in, (sp->flags & SPLICE_F_FD_IN_FIXED)); req->flags &= ~REQ_F_NEED_CLEANUP; -- cgit v1.2.3 From 3c3c32f85b6cc05e5db78693457deff03ac0f434 Mon Sep 17 00:00:00 2001 From: Eric Biggers Date: Fri, 1 May 2020 22:59:45 -0700 Subject: ubifs: fix wrong use of crypto_shash_descsize() crypto_shash_descsize() returns the size of the shash_desc context needed to compute the hash, not the size of the hash itself. crypto_shash_digestsize() would be correct, or alternatively using c->hash_len and c->hmac_desc_len which already store the correct values. But actually it's simpler to just use stack arrays, so do that instead. Fixes: 49525e5eecca ("ubifs: Add helper functions for authentication support") Fixes: da8ef65f9573 ("ubifs: Authenticate replayed journal") Cc: # v4.20+ Cc: Sascha Hauer Signed-off-by: Eric Biggers Acked-by: Sascha Hauer Signed-off-by: Richard Weinberger --- fs/ubifs/auth.c | 17 ++++------------- fs/ubifs/replay.c | 13 ++----------- 2 files changed, 6 insertions(+), 24 deletions(-) (limited to 'fs') diff --git a/fs/ubifs/auth.c b/fs/ubifs/auth.c index 8cdbd53d780c..f985a3fbbb36 100644 --- a/fs/ubifs/auth.c +++ b/fs/ubifs/auth.c @@ -79,13 +79,9 @@ int ubifs_prepare_auth_node(struct ubifs_info *c, void *node, struct shash_desc *inhash) { struct ubifs_auth_node *auth = node; - u8 *hash; + u8 hash[UBIFS_HASH_ARR_SZ]; int err; - hash = kmalloc(crypto_shash_descsize(c->hash_tfm), GFP_NOFS); - if (!hash) - return -ENOMEM; - { SHASH_DESC_ON_STACK(hash_desc, c->hash_tfm); @@ -94,21 +90,16 @@ int ubifs_prepare_auth_node(struct ubifs_info *c, void *node, err = crypto_shash_final(hash_desc, hash); if (err) - goto out; + return err; } err = ubifs_hash_calc_hmac(c, hash, auth->hmac); if (err) - goto out; + return err; auth->ch.node_type = UBIFS_AUTH_NODE; ubifs_prepare_node(c, auth, ubifs_auth_node_sz(c), 0); - - err = 0; -out: - kfree(hash); - - return err; + return 0; } static struct shash_desc *ubifs_get_desc(const struct ubifs_info *c, diff --git a/fs/ubifs/replay.c b/fs/ubifs/replay.c index b28ac4dfb407..01fcf7975047 100644 --- a/fs/ubifs/replay.c +++ b/fs/ubifs/replay.c @@ -601,18 +601,12 @@ static int authenticate_sleb(struct ubifs_info *c, struct ubifs_scan_leb *sleb, struct ubifs_scan_node *snod; int n_nodes = 0; int err; - u8 *hash, *hmac; + u8 hash[UBIFS_HASH_ARR_SZ]; + u8 hmac[UBIFS_HMAC_ARR_SZ]; if (!ubifs_authenticated(c)) return sleb->nodes_cnt; - hash = kmalloc(crypto_shash_descsize(c->hash_tfm), GFP_NOFS); - hmac = kmalloc(c->hmac_desc_len, GFP_NOFS); - if (!hash || !hmac) { - err = -ENOMEM; - goto out; - } - list_for_each_entry(snod, &sleb->nodes, list) { n_nodes++; @@ -662,9 +656,6 @@ static int authenticate_sleb(struct ubifs_info *c, struct ubifs_scan_leb *sleb, err = 0; } out: - kfree(hash); - kfree(hmac); - return err ? err : n_nodes - n_not_auth; } -- cgit v1.2.3 From 035779483072ff7854943dc0cbae82c4e0070d15 Mon Sep 17 00:00:00 2001 From: Eric Sandeen Date: Fri, 1 May 2020 20:34:25 -0500 Subject: exfat: use iter_file_splice_write Doing copy_file_range() on exfat with a file opened for direct IO leads to an -EFAULT: # xfs_io -f -d -c "truncate 32768" \ -c "copy_range -d 16384 -l 16384 -f 0" /mnt/test/junk copy_range: Bad address and the reason seems to be that we go through: default_file_splice_write splice_from_pipe __splice_from_pipe write_pipe_buf __kernel_write new_sync_write generic_file_write_iter generic_file_direct_write exfat_direct_IO do_blockdev_direct_IO iov_iter_get_pages and land in iterate_all_kinds(), which does "return -EFAULT" for our kvec iter. Setting exfat's splice_write to iter_file_splice_write fixes this and lets fsx (which originally detected the problem) run to success from the xfstests harness. Signed-off-by: Eric Sandeen Signed-off-by: Namjae Jeon --- fs/exfat/file.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/exfat/file.c b/fs/exfat/file.c index 4f76764165cf..c9db8eb0cfc3 100644 --- a/fs/exfat/file.c +++ b/fs/exfat/file.c @@ -348,12 +348,13 @@ out: } const struct file_operations exfat_file_operations = { - .llseek = generic_file_llseek, - .read_iter = generic_file_read_iter, - .write_iter = generic_file_write_iter, - .mmap = generic_file_mmap, - .fsync = generic_file_fsync, - .splice_read = generic_file_splice_read, + .llseek = generic_file_llseek, + .read_iter = generic_file_read_iter, + .write_iter = generic_file_write_iter, + .mmap = generic_file_mmap, + .fsync = generic_file_fsync, + .splice_read = generic_file_splice_read, + .splice_write = iter_file_splice_write, }; const struct inode_operations exfat_file_inode_operations = { -- cgit v1.2.3 From 94182167ec730dadcaea5fbc6bb8f1136966ef66 Mon Sep 17 00:00:00 2001 From: Wei Yongjun Date: Wed, 6 May 2020 14:25:54 +0000 Subject: exfat: fix possible memory leak in exfat_find() 'es' is malloced from exfat_get_dentry_set() in exfat_find() and should be freed before leaving from the error handling cases, otherwise it will cause memory leak. Fixes: 5f2aa075070c ("exfat: add inode operations") Signed-off-by: Wei Yongjun Signed-off-by: Namjae Jeon --- fs/exfat/namei.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/exfat/namei.c b/fs/exfat/namei.c index b72d782568b8..a2659a8a68a1 100644 --- a/fs/exfat/namei.c +++ b/fs/exfat/namei.c @@ -692,6 +692,7 @@ static int exfat_find(struct inode *dir, struct qstr *qname, exfat_fs_error(sb, "non-zero size file starts with zero cluster (size : %llu, p_dir : %u, entry : 0x%08x)", i_size_read(dir), ei->dir.dir, ei->entry); + kfree(es); return -EIO; } -- cgit v1.2.3 From e3aabf9554fd04eb14cd44ae7583fc9d40edd250 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Mon, 18 May 2020 11:04:17 -0600 Subject: io_uring: cancel work if task_work_add() fails We currently move it to the io_wqe_manager for execution, but we cannot safely do so as we may lack some of the state to execute it out of context. As we cancel work anyway when the ring/task exits, just mark this request as canceled and io_async_task_func() will do the right thing. Fixes: aa96bf8a9ee3 ("io_uring: use io-wq manager as backup task if task is exiting") Signed-off-by: Jens Axboe --- fs/io_uring.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index ecfd7f054ef6..29aa53000def 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -4135,12 +4135,14 @@ static int __io_async_wake(struct io_kiocb *req, struct io_poll_iocb *poll, req->result = mask; init_task_work(&req->task_work, func); /* - * If this fails, then the task is exiting. Punt to one of the io-wq - * threads to ensure the work gets run, we can't always rely on exit - * cancelation taking care of this. + * If this fails, then the task is exiting. When a task exits, the + * work gets canceled, so just cancel this request as well instead + * of executing it. We can't safely execute it anyway, as we may not + * have the needed state needed for it anyway. */ ret = task_work_add(tsk, &req->task_work, true); if (unlikely(ret)) { + WRITE_ONCE(poll->canceled, true); tsk = io_wq_get_task(req->ctx->io_wq); task_work_add(tsk, &req->task_work, true); } -- cgit v1.2.3 From 9d1be4f4dc5ff1c66c86acfd2c35765d9e3776b3 Mon Sep 17 00:00:00 2001 From: David Howells Date: Sun, 17 May 2020 21:21:05 +0100 Subject: afs: Don't unlock fetched data pages until the op completes successfully Don't call req->page_done() on each page as we finish filling it with the data coming from the network. Whilst this might speed up the application a bit, it's a problem if there's a network failure and the operation has to be reissued. If this happens, an oops occurs because afs_readpages_page_done() clears the pointer to each page it unlocks and when a retry happens, the pointers to the pages it wants to fill are now NULL (and the pages have been unlocked anyway). Instead, wait till the operation completes successfully and only then release all the pages after clearing any terminal gap (the server can give us less data than we requested as we're allowed to ask for more than is available). KASAN produces a bug like the following, and even without KASAN, it can oops and panic. BUG: KASAN: wild-memory-access in _copy_to_iter+0x323/0x5f4 Write of size 1404 at addr 0005088000000000 by task md5sum/5235 CPU: 0 PID: 5235 Comm: md5sum Not tainted 5.7.0-rc3-fscache+ #250 Hardware name: ASUS All Series/H97-PLUS, BIOS 2306 10/09/2014 Call Trace: memcpy+0x39/0x58 _copy_to_iter+0x323/0x5f4 __skb_datagram_iter+0x89/0x2a6 skb_copy_datagram_iter+0x129/0x135 rxrpc_recvmsg_data.isra.0+0x615/0xd42 rxrpc_kernel_recv_data+0x1e9/0x3ae afs_extract_data+0x139/0x33a yfs_deliver_fs_fetch_data64+0x47a/0x91b afs_deliver_to_call+0x304/0x709 afs_wait_for_call_to_complete+0x1cc/0x4ad yfs_fs_fetch_data+0x279/0x288 afs_fetch_data+0x1e1/0x38d afs_readpages+0x593/0x72e read_pages+0xf5/0x21e __do_page_cache_readahead+0x128/0x23f ondemand_readahead+0x36e/0x37f generic_file_buffered_read+0x234/0x680 new_sync_read+0x109/0x17e vfs_read+0xe6/0x138 ksys_read+0xd8/0x14d do_syscall_64+0x6e/0x8a entry_SYSCALL_64_after_hwframe+0x49/0xb3 Fixes: 196ee9cd2d04 ("afs: Make afs_fs_fetch_data() take a list of pages") Fixes: 30062bd13e36 ("afs: Implement YFS support in the fs client") Signed-off-by: David Howells Reviewed-by: Matthew Wilcox (Oracle) Signed-off-by: Linus Torvalds --- fs/afs/fsclient.c | 8 ++++---- fs/afs/yfsclient.c | 8 ++++---- 2 files changed, 8 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/afs/fsclient.c b/fs/afs/fsclient.c index 68fc46634346..d2b3798c1932 100644 --- a/fs/afs/fsclient.c +++ b/fs/afs/fsclient.c @@ -385,8 +385,6 @@ static int afs_deliver_fs_fetch_data(struct afs_call *call) ASSERTCMP(req->offset, <=, PAGE_SIZE); if (req->offset == PAGE_SIZE) { req->offset = 0; - if (req->page_done) - req->page_done(req); req->index++; if (req->remain > 0) goto begin_page; @@ -440,11 +438,13 @@ static int afs_deliver_fs_fetch_data(struct afs_call *call) if (req->offset < PAGE_SIZE) zero_user_segment(req->pages[req->index], req->offset, PAGE_SIZE); - if (req->page_done) - req->page_done(req); req->offset = 0; } + if (req->page_done) + for (req->index = 0; req->index < req->nr_pages; req->index++) + req->page_done(req); + _leave(" = 0 [done]"); return 0; } diff --git a/fs/afs/yfsclient.c b/fs/afs/yfsclient.c index b5b45c57e1b1..fe413e7a5cf4 100644 --- a/fs/afs/yfsclient.c +++ b/fs/afs/yfsclient.c @@ -497,8 +497,6 @@ static int yfs_deliver_fs_fetch_data64(struct afs_call *call) ASSERTCMP(req->offset, <=, PAGE_SIZE); if (req->offset == PAGE_SIZE) { req->offset = 0; - if (req->page_done) - req->page_done(req); req->index++; if (req->remain > 0) goto begin_page; @@ -556,11 +554,13 @@ static int yfs_deliver_fs_fetch_data64(struct afs_call *call) if (req->offset < PAGE_SIZE) zero_user_segment(req->pages[req->index], req->offset, PAGE_SIZE); - if (req->page_done) - req->page_done(req); req->offset = 0; } + if (req->page_done) + for (req->index = 0; req->index < req->nr_pages; req->index++) + req->page_done(req); + _leave(" = 0 [done]"); return 0; } -- cgit v1.2.3 From 9f44eda19529b1c3eef50676dc54b8cd0aa86aa3 Mon Sep 17 00:00:00 2001 From: Ritesh Harjani Date: Tue, 5 May 2020 17:43:14 +0200 Subject: ext4: fix EXT4_MAX_LOGICAL_BLOCK macro ext4 supports max number of logical blocks in a file to be 0xffffffff. (This is since ext4_extent's ee_block is __le32). This means that EXT4_MAX_LOGICAL_BLOCK should be 0xfffffffe (starting from 0 logical offset). This patch fixes this. The issue was seen when ext4 moved to iomap_fiemap API and when overlayfs was mounted on top of ext4. Since overlayfs was missing filemap_check_ranges(), so it could pass a arbitrary huge length which lead to overflow of map.m_len logic. This patch fixes that. Fixes: d3b6f23f7167 ("ext4: move ext4_fiemap to use iomap framework") Reported-by: syzbot+77fa5bdb65cc39711820@syzkaller.appspotmail.com Signed-off-by: Ritesh Harjani Reviewed-by: Jan Kara Signed-off-by: Christoph Hellwig Link: https://lore.kernel.org/r/20200505154324.3226743-2-hch@lst.de Signed-off-by: Theodore Ts'o --- fs/ext4/ext4.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index 91eb4381cae5..ad2dbf6e4924 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -722,7 +722,7 @@ enum { #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF /* Max logical block we can support */ -#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFF +#define EXT4_MAX_LOGICAL_BLOCK 0xFFFFFFFE /* * Structure of an inode on the disk -- cgit v1.2.3 From 959f7584512941a614113bfddb41b6812214169d Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 5 May 2020 17:43:15 +0200 Subject: ext4: fix fiemap size checks for bitmap files Add an extra validation of the len parameter, as for ext4 some files might have smaller file size limits than others. This also means the redundant size check in ext4_ioctl_get_es_cache can go away, as all size checking is done in the shared fiemap handler. Signed-off-by: Christoph Hellwig Reviewed-by: Ritesh Harjani Reviewed-by: Jan Kara Link: https://lore.kernel.org/r/20200505154324.3226743-3-hch@lst.de Signed-off-by: Theodore Ts'o --- fs/ext4/extents.c | 31 +++++++++++++++++++++++++++++++ fs/ext4/ioctl.c | 33 ++------------------------------- 2 files changed, 33 insertions(+), 31 deletions(-) (limited to 'fs') diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c index f2b577b315a0..2b4b94542e34 100644 --- a/fs/ext4/extents.c +++ b/fs/ext4/extents.c @@ -4832,6 +4832,28 @@ static const struct iomap_ops ext4_iomap_xattr_ops = { .iomap_begin = ext4_iomap_xattr_begin, }; +static int ext4_fiemap_check_ranges(struct inode *inode, u64 start, u64 *len) +{ + u64 maxbytes; + + if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) + maxbytes = inode->i_sb->s_maxbytes; + else + maxbytes = EXT4_SB(inode->i_sb)->s_bitmap_maxbytes; + + if (*len == 0) + return -EINVAL; + if (start > maxbytes) + return -EFBIG; + + /* + * Shrink request scope to what the fs can actually handle. + */ + if (*len > maxbytes || (maxbytes - *len) < start) + *len = maxbytes - start; + return 0; +} + static int _ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, __u64 start, __u64 len, bool from_es_cache) { @@ -4852,6 +4874,15 @@ static int _ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, if (fiemap_check_flags(fieinfo, ext4_fiemap_flags)) return -EBADR; + /* + * For bitmap files the maximum size limit could be smaller than + * s_maxbytes, so check len here manually instead of just relying on the + * generic check. + */ + error = ext4_fiemap_check_ranges(inode, start, &len); + if (error) + return error; + if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) { fieinfo->fi_flags &= ~FIEMAP_FLAG_XATTR; error = iomap_fiemap(inode, fieinfo, start, len, diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c index bfc1281fc4cb..0746532ba463 100644 --- a/fs/ext4/ioctl.c +++ b/fs/ext4/ioctl.c @@ -733,29 +733,6 @@ static void ext4_fill_fsxattr(struct inode *inode, struct fsxattr *fa) fa->fsx_projid = from_kprojid(&init_user_ns, ei->i_projid); } -/* copied from fs/ioctl.c */ -static int fiemap_check_ranges(struct super_block *sb, - u64 start, u64 len, u64 *new_len) -{ - u64 maxbytes = (u64) sb->s_maxbytes; - - *new_len = len; - - if (len == 0) - return -EINVAL; - - if (start > maxbytes) - return -EFBIG; - - /* - * Shrink request scope to what the fs can actually handle. - */ - if (len > maxbytes || (maxbytes - len) < start) - *new_len = maxbytes - start; - - return 0; -} - /* So that the fiemap access checks can't overflow on 32 bit machines. */ #define FIEMAP_MAX_EXTENTS (UINT_MAX / sizeof(struct fiemap_extent)) @@ -765,8 +742,6 @@ static int ext4_ioctl_get_es_cache(struct file *filp, unsigned long arg) struct fiemap __user *ufiemap = (struct fiemap __user *) arg; struct fiemap_extent_info fieinfo = { 0, }; struct inode *inode = file_inode(filp); - struct super_block *sb = inode->i_sb; - u64 len; int error; if (copy_from_user(&fiemap, ufiemap, sizeof(fiemap))) @@ -775,11 +750,6 @@ static int ext4_ioctl_get_es_cache(struct file *filp, unsigned long arg) if (fiemap.fm_extent_count > FIEMAP_MAX_EXTENTS) return -EINVAL; - error = fiemap_check_ranges(sb, fiemap.fm_start, fiemap.fm_length, - &len); - if (error) - return error; - fieinfo.fi_flags = fiemap.fm_flags; fieinfo.fi_extents_max = fiemap.fm_extent_count; fieinfo.fi_extents_start = ufiemap->fm_extents; @@ -792,7 +762,8 @@ static int ext4_ioctl_get_es_cache(struct file *filp, unsigned long arg) if (fieinfo.fi_flags & FIEMAP_FLAG_SYNC) filemap_write_and_wait(inode->i_mapping); - error = ext4_get_es_cache(inode, &fieinfo, fiemap.fm_start, len); + error = ext4_get_es_cache(inode, &fieinfo, fiemap.fm_start, + fiemap.fm_length); fiemap.fm_flags = fieinfo.fi_flags; fiemap.fm_mapped_extents = fieinfo.fi_extents_mapped; if (copy_to_user(ufiemap, &fiemap, sizeof(fiemap))) -- cgit v1.2.3 From 4f4eeba87cc731b200bff9372d14a80f5996b277 Mon Sep 17 00:00:00 2001 From: Bijan Mottahedeh Date: Tue, 19 May 2020 14:52:49 -0700 Subject: io_uring: don't use kiocb.private to store buf_index kiocb.private is used in iomap_dio_rw() so store buf_index separately. Signed-off-by: Bijan Mottahedeh Move 'buf_index' to a hole in io_kiocb. Signed-off-by: Jens Axboe --- fs/io_uring.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index 29aa53000def..d43f7e98e07a 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -619,6 +619,8 @@ struct io_kiocb { bool needs_fixed_file; u8 opcode; + u16 buf_index; + struct io_ring_ctx *ctx; struct list_head list; unsigned int flags; @@ -2101,9 +2103,7 @@ static int io_prep_rw(struct io_kiocb *req, const struct io_uring_sqe *sqe, req->rw.addr = READ_ONCE(sqe->addr); req->rw.len = READ_ONCE(sqe->len); - /* we own ->private, reuse it for the buffer index / buffer ID */ - req->rw.kiocb.private = (void *) (unsigned long) - READ_ONCE(sqe->buf_index); + req->buf_index = READ_ONCE(sqe->buf_index); return 0; } @@ -2146,7 +2146,7 @@ static ssize_t io_import_fixed(struct io_kiocb *req, int rw, struct io_ring_ctx *ctx = req->ctx; size_t len = req->rw.len; struct io_mapped_ubuf *imu; - unsigned index, buf_index; + u16 index, buf_index; size_t offset; u64 buf_addr; @@ -2154,7 +2154,7 @@ static ssize_t io_import_fixed(struct io_kiocb *req, int rw, if (unlikely(!ctx->user_bufs)) return -EFAULT; - buf_index = (unsigned long) req->rw.kiocb.private; + buf_index = req->buf_index; if (unlikely(buf_index >= ctx->nr_user_bufs)) return -EFAULT; @@ -2270,10 +2270,10 @@ static void __user *io_rw_buffer_select(struct io_kiocb *req, size_t *len, bool needs_lock) { struct io_buffer *kbuf; - int bgid; + u16 bgid; kbuf = (struct io_buffer *) (unsigned long) req->rw.addr; - bgid = (int) (unsigned long) req->rw.kiocb.private; + bgid = req->buf_index; kbuf = io_buffer_select(req, len, bgid, kbuf, needs_lock); if (IS_ERR(kbuf)) return kbuf; @@ -2364,7 +2364,7 @@ static ssize_t io_import_iovec(int rw, struct io_kiocb *req, } /* buffer index only valid with fixed read/write, or buffer select */ - if (req->rw.kiocb.private && !(req->flags & REQ_F_BUFFER_SELECT)) + if (req->buf_index && !(req->flags & REQ_F_BUFFER_SELECT)) return -EINVAL; if (opcode == IORING_OP_READ || opcode == IORING_OP_WRITE) { -- cgit v1.2.3 From 4e89b7210403fa4a8acafe7c602b6212b7af6c3b Mon Sep 17 00:00:00 2001 From: Al Viro Date: Tue, 19 May 2020 17:48:52 -0400 Subject: fix multiplication overflow in copy_fdtable() cpy and set really should be size_t; we won't get an overflow on that, since sysctl_nr_open can't be set above ~(size_t)0 / sizeof(void *), so nr that would've managed to overflow size_t on that multiplication won't get anywhere near copy_fdtable() - we'll fail with EMFILE before that. Cc: stable@kernel.org # v2.6.25+ Fixes: 9cfe015aa424 (get rid of NR_OPEN and introduce a sysctl_nr_open) Reported-by: Thiago Macieira Signed-off-by: Al Viro --- fs/file.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/file.c b/fs/file.c index c8a4e4c86e55..abb8b7081d7a 100644 --- a/fs/file.c +++ b/fs/file.c @@ -70,7 +70,7 @@ static void copy_fd_bitmaps(struct fdtable *nfdt, struct fdtable *ofdt, */ static void copy_fdtable(struct fdtable *nfdt, struct fdtable *ofdt) { - unsigned int cpy, set; + size_t cpy, set; BUG_ON(nfdt->max_fds < ofdt->max_fds); -- cgit v1.2.3 From b532576ed39efe3b351ae8320b2ab67a4c4c3719 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Tue, 19 May 2020 21:20:27 -0600 Subject: io_uring: don't add non-IO requests to iopoll pending list We normally disable any commands that aren't specifically poll commands for a ring that is setup for polling, but we do allow buffer provide and remove commands to support buffer selection for polled IO. Once a request is issued, we add it to the poll list to poll for completion. But we should not do that for non-IO commands, as those request complete inline immediately and aren't pollable. If we do, we can leave requests on the iopoll list after they are freed. Fixes: ddf0322db79c ("io_uring: add IORING_OP_PROVIDE_BUFFERS") Signed-off-by: Jens Axboe --- fs/io_uring.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index d43f7e98e07a..f9f79ac5ac7b 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -5306,7 +5306,8 @@ static int io_issue_sqe(struct io_kiocb *req, const struct io_uring_sqe *sqe, if (ret) return ret; - if (ctx->flags & IORING_SETUP_IOPOLL) { + /* If the op doesn't have a file, we're not polling for it */ + if ((ctx->flags & IORING_SETUP_IOPOLL) && req->file) { const bool in_async = io_wq_current_is_worker(); if (req->result == -EAGAIN) -- cgit v1.2.3 From d4ae271dfaae2a5f41c015f2f20d62a1deeec734 Mon Sep 17 00:00:00 2001 From: Xiaoguang Wang Date: Wed, 20 May 2020 21:24:35 +0800 Subject: io_uring: reset -EBUSY error when io sq thread is waken up In io_sq_thread(), currently if we get an -EBUSY error and go to sleep, we will won't clear it again, which will result in io_sq_thread() will never have a chance to submit sqes again. Below test program test.c can reveal this bug: int main(int argc, char *argv[]) { struct io_uring ring; int i, fd, ret; struct io_uring_sqe *sqe; struct io_uring_cqe *cqe; struct iovec *iovecs; void *buf; struct io_uring_params p; if (argc < 2) { printf("%s: file\n", argv[0]); return 1; } memset(&p, 0, sizeof(p)); p.flags = IORING_SETUP_SQPOLL; ret = io_uring_queue_init_params(4, &ring, &p); if (ret < 0) { fprintf(stderr, "queue_init: %s\n", strerror(-ret)); return 1; } fd = open(argv[1], O_RDONLY | O_DIRECT); if (fd < 0) { perror("open"); return 1; } iovecs = calloc(10, sizeof(struct iovec)); for (i = 0; i < 10; i++) { if (posix_memalign(&buf, 4096, 4096)) return 1; iovecs[i].iov_base = buf; iovecs[i].iov_len = 4096; } ret = io_uring_register_files(&ring, &fd, 1); if (ret < 0) { fprintf(stderr, "%s: register %d\n", __FUNCTION__, ret); return ret; } for (i = 0; i < 10; i++) { sqe = io_uring_get_sqe(&ring); if (!sqe) break; io_uring_prep_readv(sqe, 0, &iovecs[i], 1, 0); sqe->flags |= IOSQE_FIXED_FILE; ret = io_uring_submit(&ring); sleep(1); printf("submit %d\n", i); } for (i = 0; i < 10; i++) { io_uring_wait_cqe(&ring, &cqe); printf("receive: %d\n", i); if (cqe->res != 4096) { fprintf(stderr, "ret=%d, wanted 4096\n", cqe->res); ret = 1; } io_uring_cqe_seen(&ring, cqe); } close(fd); io_uring_queue_exit(&ring); return 0; } sudo ./test testfile above command will hang on the tenth request, to fix this bug, when io sq_thread is waken up, we reset the variable 'ret' to be zero. Suggested-by: Jens Axboe Signed-off-by: Xiaoguang Wang Signed-off-by: Jens Axboe --- fs/io_uring.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/io_uring.c b/fs/io_uring.c index f9f79ac5ac7b..bb25e3997d41 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -6032,6 +6032,7 @@ static int io_sq_thread(void *data) finish_wait(&ctx->sqo_wait, &wait); ctx->rings->sq_flags &= ~IORING_SQ_NEED_WAKEUP; + ret = 0; continue; } finish_wait(&ctx->sqo_wait, &wait); -- cgit v1.2.3 From 566d136289dc57816ac290de87a9a0f7d9bd3cbb Mon Sep 17 00:00:00 2001 From: Tetsuo Handa Date: Wed, 20 May 2020 08:51:59 +0900 Subject: pipe: Fix pipe_full() test in opipe_prep(). syzbot is reporting that splice()ing from non-empty read side to already-full write side causes unkillable task, for opipe_prep() is by error not inverting pipe_full() test. CPU: 0 PID: 9460 Comm: syz-executor.5 Not tainted 5.6.0-rc3-next-20200228-syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011 RIP: 0010:rol32 include/linux/bitops.h:105 [inline] RIP: 0010:iterate_chain_key kernel/locking/lockdep.c:369 [inline] RIP: 0010:__lock_acquire+0x6a3/0x5270 kernel/locking/lockdep.c:4178 Call Trace: lock_acquire+0x197/0x420 kernel/locking/lockdep.c:4720 __mutex_lock_common kernel/locking/mutex.c:956 [inline] __mutex_lock+0x156/0x13c0 kernel/locking/mutex.c:1103 pipe_lock_nested fs/pipe.c:66 [inline] pipe_double_lock+0x1a0/0x1e0 fs/pipe.c:104 splice_pipe_to_pipe fs/splice.c:1562 [inline] do_splice+0x35f/0x1520 fs/splice.c:1141 __do_sys_splice fs/splice.c:1447 [inline] __se_sys_splice fs/splice.c:1427 [inline] __x64_sys_splice+0x2b5/0x320 fs/splice.c:1427 do_syscall_64+0xf6/0x790 arch/x86/entry/common.c:295 entry_SYSCALL_64_after_hwframe+0x49/0xbe Reported-by: syzbot+b48daca8639150bc5e73@syzkaller.appspotmail.com Link: https://syzkaller.appspot.com/bug?id=9386d051e11e09973d5a4cf79af5e8cedf79386d Fixes: 8cefc107ca54c8b0 ("pipe: Use head and tail pointers for the ring, not cursor and length") Cc: stable@vger.kernel.org # 5.5+ Signed-off-by: Tetsuo Handa Signed-off-by: Linus Torvalds --- fs/splice.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/splice.c b/fs/splice.c index fd0a1e7e5959..4e53efbd621d 100644 --- a/fs/splice.c +++ b/fs/splice.c @@ -1494,7 +1494,7 @@ static int opipe_prep(struct pipe_inode_info *pipe, unsigned int flags) * Check pipe occupancy without the inode lock first. This function * is speculative anyways, so missing one is ok. */ - if (pipe_full(pipe->head, pipe->tail, pipe->max_usage)) + if (!pipe_full(pipe->head, pipe->tail, pipe->max_usage)) return 0; ret = 0; -- cgit v1.2.3 From 907fa893258ba6076f5fff32900a461decb9e8c5 Mon Sep 17 00:00:00 2001 From: Namjae Jeon Date: Fri, 22 May 2020 08:10:10 +0900 Subject: exfat: add the dummy mount options to be backward compatible with staging/exfat As Ubuntu and Fedora release new version used kernel version equal to or higher than v5.4, They started to support kernel exfat filesystem. Linus reported a mount error with new version of exfat on Fedora: exfat: Unknown parameter 'namecase' This is because there is a difference in mount option between old staging/exfat and new exfat. And utf8, debug, and codepage options as well as namecase have been removed from new exfat. This patch add the dummy mount options as deprecated option to be backward compatible with old one. Reported-by: Linus Torvalds Signed-off-by: Namjae Jeon Cc: Matthew Wilcox Cc: Al Viro Cc: Eric Sandeen Signed-off-by: Linus Torvalds --- fs/exfat/super.c | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) (limited to 'fs') diff --git a/fs/exfat/super.c b/fs/exfat/super.c index 0565d5539d57..a846ff555656 100644 --- a/fs/exfat/super.c +++ b/fs/exfat/super.c @@ -203,6 +203,12 @@ enum { Opt_errors, Opt_discard, Opt_time_offset, + + /* Deprecated options */ + Opt_utf8, + Opt_debug, + Opt_namecase, + Opt_codepage, }; static const struct constant_table exfat_param_enums[] = { @@ -223,6 +229,14 @@ static const struct fs_parameter_spec exfat_parameters[] = { fsparam_enum("errors", Opt_errors, exfat_param_enums), fsparam_flag("discard", Opt_discard), fsparam_s32("time_offset", Opt_time_offset), + __fsparam(NULL, "utf8", Opt_utf8, fs_param_deprecated, + NULL), + __fsparam(NULL, "debug", Opt_debug, fs_param_deprecated, + NULL), + __fsparam(fs_param_is_u32, "namecase", Opt_namecase, + fs_param_deprecated, NULL), + __fsparam(fs_param_is_u32, "codepage", Opt_codepage, + fs_param_deprecated, NULL), {} }; @@ -278,6 +292,11 @@ static int exfat_parse_param(struct fs_context *fc, struct fs_parameter *param) return -EINVAL; opts->time_offset = result.int_32; break; + case Opt_utf8: + case Opt_debug: + case Opt_namecase: + case Opt_codepage: + break; default: return -EINVAL; } -- cgit v1.2.3 From 8a1d24e1cc6d96a17f2dcb1400d370cadbfb7cb6 Mon Sep 17 00:00:00 2001 From: David Howells Date: Fri, 22 May 2020 23:58:28 +0100 Subject: rxrpc: Fix a warning Fix a warning due to an uninitialised variable. le included from ../fs/afs/fs_probe.c:11: ../fs/afs/fs_probe.c: In function 'afs_fileserver_probe_result': ../fs/afs/internal.h:1453:2: warning: 'rtt_us' may be used uninitialized in this function [-Wmaybe-uninitialized] 1453 | printk("[%-6.6s] "FMT"\n", current->comm ,##__VA_ARGS__) | ^~~~~~ ../fs/afs/fs_probe.c:35:15: note: 'rtt_us' was declared here Signed-off-by: David Howells --- fs/afs/fs_probe.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/afs/fs_probe.c b/fs/afs/fs_probe.c index 237352d3cb53..37d1bba57b00 100644 --- a/fs/afs/fs_probe.c +++ b/fs/afs/fs_probe.c @@ -32,7 +32,7 @@ void afs_fileserver_probe_result(struct afs_call *call) struct afs_server *server = call->server; unsigned int server_index = call->server_index; unsigned int index = call->addr_ix; - unsigned int rtt_us; + unsigned int rtt_us = 0; bool have_result = false; int ret = call->error; -- cgit v1.2.3