kernel/linux.git/fs/erofs/zdata.c, branch linux-6.0.y

erofs: Fix pcluster memleak when its block address is zero

2022-12-31T12:25:46+00:00

[ Upstream commit c42c0ffe81176940bd5dead474216b7198d77675 ] syzkaller reported a memleak: https://syzkaller.appspot.com/bug?id=62f37ff612f0021641eda5b17f056f1668aa9aed unreferenced object 0xffff88811009c7f8 (size 136): ... backtrace: [] z_erofs_do_read_page+0x99b/0x1740 [] z_erofs_readahead+0x24e/0x580 [] read_pages+0x86/0x3d0 ... syzkaller constructed a case: in z_erofs_register_pcluster(), ztailpacking = false and map->m_pa = zero. This makes pcl->obj.index be zero although pcl is not a inline pcluster. Then following path adds refcount for grp, but the refcount won't be put because pcl is inline. z_erofs_readahead() z_erofs_do_read_page() # for another page z_erofs_collector_begin() erofs_find_workgroup() erofs_workgroup_get() Since it's illegal for the block address of a non-inlined pcluster to be zero, add check here to avoid registering the pcluster which would be leaked. Fixes: cecf864d3d76 ("erofs: support inline data decompression") Reported-by: syzbot+6f8cd9a0155b366d227f@syzkaller.appspotmail.com Signed-off-by: Chen Zhongjin Reviewed-by: Yue Hu Reviewed-by: Gao Xiang Reviewed-by: Chao Yu Link: https://lore.kernel.org/r/Y42Kz6sVkf+XqJRB@debian Signed-off-by: Gao Xiang Signed-off-by: Sasha Levin

erofs: fix up inplace decompression success rate

2022-11-03T15:00:27+00:00

[ Upstream commit e7933278b442f97809b1ea84264586302bd08a03 ] 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 Signed-off-by: Sasha Levin

erofs: shouldn't churn the mapping page for duplicated copies

2022-10-29T08:08:33+00:00

[ Upstream commit 63bbb85658ea43dd35dbfde6d4150b47c407fc87 ] 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 Signed-off-by: Sasha Levin

erofs: introduce multi-reference pclusters (fully-referenced)

2022-07-22T13:44:27+00:00

Let's introduce multi-reference pclusters at runtime. In details, if one pcluster is requested by multiple extents at almost the same time (even belong to different files), the longest extent will be decompressed as representative and the other extents are actually copied from the longest one in one round. After this patch, fully-referenced extents can be correctly handled and the full decoding check needs to be bypassed for partial-referenced extents. Acked-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20220715154203.48093-17-hsiangkao@linux.alibaba.com

erofs: record the longest decompressed size in this round

2022-07-21T14:55:44+00:00

Currently, `pcl->length' records the longest decompressed length as long as the pcluster itself isn't reclaimed. However, such number is unneeded for the general cases since it doesn't indicate the exact decompressed size in this round. Instead, let's record the decompressed size for this round instead, thus `pcl->nr_pages' can be completely dropped and pageofs_out is also designed to be kept in sync with `pcl->length'. Acked-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20220715154203.48093-16-hsiangkao@linux.alibaba.com

erofs: introduce z_erofs_do_decompressed_bvec()

2022-07-21T14:55:37+00:00

Both out_bvecs and in_bvecs share the common logic for decompressed buffers. So let's make a helper for this. Acked-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20220715154203.48093-15-hsiangkao@linux.alibaba.com

erofs: try to leave (de)compressed_pages on stack if possible

2022-07-21T14:55:30+00:00

For the most cases, small pclusters can be decompressed with page arrays on stack. Try to leave both (de)compressed_pages on stack if possible as before. Acked-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20220715154203.48093-14-hsiangkao@linux.alibaba.com

erofs: introduce struct z_erofs_decompress_backend

2022-07-21T14:55:22+00:00

Let's introduce struct z_erofs_decompress_backend in order to pass on the decompression backend context between helper functions more easier and avoid too many arguments. Acked-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20220715154203.48093-13-hsiangkao@linux.alibaba.com

erofs: get rid of `z_pagemap_global'

2022-07-21T14:55:15+00:00

In order to introduce multi-reference pclusters for compressed data deduplication, let's get rid of the global page array for now since it needs to be re-designed then at least. Acked-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20220715154203.48093-12-hsiangkao@linux.alibaba.com

erofs: clean up `enum z_erofs_collectmode'

2022-07-21T14:55:07+00:00

`enum z_erofs_collectmode' is really ambiguous, but I'm not quite sure if there are better naming, basically it's used to judge whether inplace I/O can be used due to the current status of pclusters in the chain. Rename it as `enum z_erofs_pclustermode' instead. Acked-by: Chao Yu Signed-off-by: Gao Xiang Link: https://lore.kernel.org/r/20220715154203.48093-11-hsiangkao@linux.alibaba.com