From 664609e49f1c84fc97987b2bf64544e586b8849c Mon Sep 17 00:00:00 2001 From: Yue Hu Date: Wed, 5 Oct 2022 09:35:28 +0800 Subject: [PATCH 1/4] erofs: fix illegal unmapped accesses in z_erofs_fill_inode_lazy() Note that we are still accessing 'h_idata_size' and 'h_fragmentoff' after calling erofs_put_metabuf(), that is not correct. Fix it. Fixes: ab92184ff8f1 ("erofs: add on-disk compressed tail-packing inline support") Fixes: b15b2e307c3a ("erofs: support on-disk compressed fragments data") Signed-off-by: Yue Hu Reviewed-by: Gao Xiang Reviewed-by: Chao Yu Link: https://lore.kernel.org/r/20221005013528.62977-1-zbestahu@163.com Signed-off-by: Gao Xiang --- fs/erofs/zmap.c | 22 ++++++++++------------ 1 file changed, 10 insertions(+), 12 deletions(-) diff --git a/fs/erofs/zmap.c b/fs/erofs/zmap.c index 44c27ef39c43..0bb66927e3d0 100644 --- a/fs/erofs/zmap.c +++ b/fs/erofs/zmap.c @@ -57,8 +57,7 @@ static int z_erofs_fill_inode_lazy(struct inode *inode) pos = ALIGN(iloc(EROFS_SB(sb), vi->nid) + vi->inode_isize + vi->xattr_isize, 8); - kaddr = erofs_read_metabuf(&buf, sb, erofs_blknr(pos), - EROFS_KMAP_ATOMIC); + kaddr = erofs_read_metabuf(&buf, sb, erofs_blknr(pos), EROFS_KMAP); if (IS_ERR(kaddr)) { err = PTR_ERR(kaddr); goto out_unlock; @@ -73,7 +72,7 @@ static int z_erofs_fill_inode_lazy(struct inode *inode) vi->z_advise = Z_EROFS_ADVISE_FRAGMENT_PCLUSTER; vi->z_fragmentoff = le64_to_cpu(*(__le64 *)h) ^ (1ULL << 63); vi->z_tailextent_headlcn = 0; - goto unmap_done; + goto done; } vi->z_advise = le16_to_cpu(h->h_advise); vi->z_algorithmtype[0] = h->h_algorithmtype & 15; @@ -85,7 +84,7 @@ static int z_erofs_fill_inode_lazy(struct inode *inode) erofs_err(sb, "unknown HEAD%u format %u for nid %llu, please upgrade kernel", headnr + 1, vi->z_algorithmtype[headnr], vi->nid); err = -EOPNOTSUPP; - goto unmap_done; + goto out_put_metabuf; } vi->z_logical_clusterbits = LOG_BLOCK_SIZE + (h->h_clusterbits & 7); @@ -95,7 +94,7 @@ static int z_erofs_fill_inode_lazy(struct inode *inode) erofs_err(sb, "per-inode big pcluster without sb feature for nid %llu", vi->nid); err = -EFSCORRUPTED; - goto unmap_done; + goto out_put_metabuf; } if (vi->datalayout == EROFS_INODE_FLAT_COMPRESSION && !(vi->z_advise & Z_EROFS_ADVISE_BIG_PCLUSTER_1) ^ @@ -103,12 +102,8 @@ static int z_erofs_fill_inode_lazy(struct inode *inode) erofs_err(sb, "big pcluster head1/2 of compact indexes should be consistent for nid %llu", vi->nid); err = -EFSCORRUPTED; - goto unmap_done; + goto out_put_metabuf; } -unmap_done: - erofs_put_metabuf(&buf); - if (err) - goto out_unlock; if (vi->z_advise & Z_EROFS_ADVISE_INLINE_PCLUSTER) { struct erofs_map_blocks map = { @@ -127,7 +122,7 @@ unmap_done: err = -EFSCORRUPTED; } if (err < 0) - goto out_unlock; + goto out_put_metabuf; } if (vi->z_advise & Z_EROFS_ADVISE_FRAGMENT_PCLUSTER && @@ -141,11 +136,14 @@ unmap_done: EROFS_GET_BLOCKS_FINDTAIL); erofs_put_metabuf(&map.buf); if (err < 0) - goto out_unlock; + goto out_put_metabuf; } +done: /* paired with smp_mb() at the beginning of the function */ smp_mb(); set_bit(EROFS_I_Z_INITED_BIT, &vi->flags); +out_put_metabuf: + erofs_put_metabuf(&buf); out_unlock: clear_and_wake_up_bit(EROFS_I_BL_Z_BIT, &vi->flags); return err; From 63bbb85658ea43dd35dbfde6d4150b47c407fc87 Mon Sep 17 00:00:00 2001 From: Gao Xiang Date: Wed, 12 Oct 2022 12:50:56 +0800 Subject: [PATCH 2/4] erofs: shouldn't churn the mapping page for duplicated copies If other duplicated copies exist in one decompression shot, should leave the old page as is rather than replace it with the new duplicated one. Otherwise, the following cold path to deal with duplicated copies will use the invalid bvec. It impacts compressed data deduplication. Also, shift the onlinepage EIO bit to avoid touching the signed bit. Fixes: 267f2492c8f7 ("erofs: introduce multi-reference pclusters (fully-referenced)") Reviewed-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20221012045056.13421-1-hsiangkao@linux.alibaba.com --- fs/erofs/zdata.c | 8 +++----- fs/erofs/zdata.h | 6 +++--- 2 files changed, 6 insertions(+), 8 deletions(-) diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c index 559380a535af..4553be650968 100644 --- a/fs/erofs/zdata.c +++ b/fs/erofs/zdata.c @@ -888,15 +888,13 @@ static void z_erofs_do_decompressed_bvec(struct z_erofs_decompress_backend *be, if (!((bvec->offset + be->pcl->pageofs_out) & ~PAGE_MASK)) { unsigned int pgnr; - struct page *oldpage; pgnr = (bvec->offset + be->pcl->pageofs_out) >> PAGE_SHIFT; DBG_BUGON(pgnr >= be->nr_pages); - oldpage = be->decompressed_pages[pgnr]; - be->decompressed_pages[pgnr] = bvec->page; - - if (!oldpage) + if (!be->decompressed_pages[pgnr]) { + be->decompressed_pages[pgnr] = bvec->page; return; + } } /* (cold path) one pcluster is requested multiple times */ diff --git a/fs/erofs/zdata.h b/fs/erofs/zdata.h index e7f04c4fbb81..d98c95212985 100644 --- a/fs/erofs/zdata.h +++ b/fs/erofs/zdata.h @@ -126,10 +126,10 @@ static inline unsigned int z_erofs_pclusterpages(struct z_erofs_pcluster *pcl) } /* - * bit 31: I/O error occurred on this page - * bit 0 - 30: remaining parts to complete this page + * bit 30: I/O error occurred on this page + * bit 0 - 29: remaining parts to complete this page */ -#define Z_EROFS_PAGE_EIO (1 << 31) +#define Z_EROFS_PAGE_EIO (1 << 30) static inline void z_erofs_onlinepage_init(struct page *page) { From e7933278b442f97809b1ea84264586302bd08a03 Mon Sep 17 00:00:00 2001 From: Gao Xiang Date: Fri, 14 Oct 2022 14:49:15 +0800 Subject: [PATCH 3/4] erofs: fix up inplace decompression success rate Partial decompression should be checked after updating length. It's a new regression when introducing multi-reference pclusters. Fixes: 2bfab9c0edac ("erofs: record the longest decompressed size in this round") Reviewed-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20221014064915.8103-1-hsiangkao@linux.alibaba.com --- fs/erofs/zdata.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c index 4553be650968..c7f24fc7efd5 100644 --- a/fs/erofs/zdata.c +++ b/fs/erofs/zdata.c @@ -813,15 +813,14 @@ retry: ++spiltted; if (fe->pcl->pageofs_out != (map->m_la & ~PAGE_MASK)) fe->pcl->multibases = true; - - if ((map->m_flags & EROFS_MAP_FULL_MAPPED) && - !(map->m_flags & EROFS_MAP_PARTIAL_REF) && - fe->pcl->length == map->m_llen) - fe->pcl->partial = false; if (fe->pcl->length < offset + end - map->m_la) { fe->pcl->length = offset + end - map->m_la; fe->pcl->pageofs_out = map->m_la & ~PAGE_MASK; } + if ((map->m_flags & EROFS_MAP_FULL_MAPPED) && + !(map->m_flags & EROFS_MAP_PARTIAL_REF) && + fe->pcl->length == map->m_llen) + fe->pcl->partial = false; next_part: /* shorten the remaining extent to update progress */ map->m_llen = offset + cur - map->m_la; From ce4b815686573bef82d5ee53bf6f509bf20904dc Mon Sep 17 00:00:00 2001 From: Dawei Li Date: Mon, 17 Oct 2022 09:55:53 +0800 Subject: [PATCH 4/4] erofs: protect s_inodes with s_inode_list_lock for fscache s_inodes is superblock-specific resource, which should be protected by sb's specific lock s_inode_list_lock. Link: https://lore.kernel.org/r/TYCP286MB23238380DE3B74874E8D78ABCA299@TYCP286MB2323.JPNP286.PROD.OUTLOOK.COM Fixes: 7d41963759fe ("erofs: Support sharing cookies in the same domain") Reviewed-by: Yue Hu Reviewed-by: Jia Zhu Reviewed-by: Jingbo Xu Signed-off-by: Dawei Li Signed-off-by: Gao Xiang --- fs/erofs/fscache.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/fs/erofs/fscache.c b/fs/erofs/fscache.c index 998cd26a1b3b..fe05bc51f9f2 100644 --- a/fs/erofs/fscache.c +++ b/fs/erofs/fscache.c @@ -590,14 +590,17 @@ struct erofs_fscache *erofs_domain_register_cookie(struct super_block *sb, struct super_block *psb = erofs_pseudo_mnt->mnt_sb; mutex_lock(&erofs_domain_cookies_lock); + spin_lock(&psb->s_inode_list_lock); list_for_each_entry(inode, &psb->s_inodes, i_sb_list) { ctx = inode->i_private; if (!ctx || ctx->domain != domain || strcmp(ctx->name, name)) continue; igrab(inode); + spin_unlock(&psb->s_inode_list_lock); mutex_unlock(&erofs_domain_cookies_lock); return ctx; } + spin_unlock(&psb->s_inode_list_lock); ctx = erofs_fscache_domain_init_cookie(sb, name, need_inode); mutex_unlock(&erofs_domain_cookies_lock); return ctx;