kernel/linux.git/fs/ext2/dir.c, branch v6.1.168

ext2: unbugger ext2_empty_dir()

2023-01-07T10:11:40+00:00

commit 27e714c007e4ad01837bf0fac5c11913a38d7695 upstream. In 27cfa258951a "ext2: fix fs corruption when trying to remove a non-empty directory with IO error" a funny thing has happened: - page = ext2_get_page(inode, i, dir_has_error, &page_addr); + page = ext2_get_page(inode, i, 0, &page_addr); - if (IS_ERR(page)) { - dir_has_error = 1; - continue; - } + if (IS_ERR(page)) + goto not_empty; And at not_empty: we hit ext2_put_page(page, page_addr), which does put_page(page). Which, unless I'm very mistaken, should oops immediately when given ERR_PTR(-E...) as page. OK, shit happens, insufficiently tested patches included. But when commit in question describes the fault-injection test that exercised that particular failure exit... Ow. CC: stable@vger.kernel.org Fixes: 27cfa258951a ("ext2: fix fs corruption when trying to remove a non-empty directory with IO error") Signed-off-by: Al Viro Signed-off-by: Jan Kara Signed-off-by: Greg Kroah-Hartman

ext2: Use a folio in ext2_get_page()

2022-08-02T16:34:03+00:00

Remove a call to read_mapping_page(). Signed-off-by: Matthew Wilcox (Oracle)

ext2: Remove check for PageError

2022-06-29T12:51:06+00:00

If read_mapping_page() encounters an error, it returns an errno, not a page with PageError set, so this test is not needed. Signed-off-by: Matthew Wilcox (Oracle)

ext2: fix fs corruption when trying to remove a non-empty directory with IO error

2022-06-16T08:55:45+00:00

We got issue as follows: [home]# mount /dev/sdd test [home]# cd test [test]# ls dir1 lost+found [test]# rmdir dir1 ext2_empty_dir: inject fault [test]# ls lost+found [test]# cd .. [home]# umount test [home]# fsck.ext2 -fn /dev/sdd e2fsck 1.42.9 (28-Dec-2013) Pass 1: Checking inodes, blocks, and sizes Inode 4065, i_size is 0, should be 1024. Fix? no Pass 2: Checking directory structure Pass 3: Checking directory connectivity Unconnected directory inode 4065 (/???) Connect to /lost+found? no '..' in ... (4065) is / (2), should be (0). Fix? no Pass 4: Checking reference counts Inode 2 ref count is 3, should be 4. Fix? no Inode 4065 ref count is 2, should be 3. Fix? no Pass 5: Checking group summary information /dev/sdd: ********** WARNING: Filesystem still has errors ********** /dev/sdd: 14/128016 files (0.0% non-contiguous), 18477/512000 blocks Reason is same with commit 7aab5c84a0f6. We can't assume directory is empty when read directory entry failed. Link: https://lore.kernel.org/r/20220615090010.1544152-1-yebin10@huawei.com Signed-off-by: Ye Bin Signed-off-by: Jan Kara

fs/ext2: Avoid page_address on pages returned by ext2_get_page

2021-07-16T10:36:51+00:00

Commit 782b76d7abdf02b12c46ed6f1e9bf715569027f7 ("fs/ext2: Replace kmap() with kmap_local_page()") replaced the kmap/kunmap calls in ext2_get_page/ext2_put_page with kmap_local_page/kunmap_local for efficiency reasons. As a necessary side change, the commit also made ext2_get_page (and ext2_find_entry and ext2_dotdot) return the mapping address along with the page itself, as it is required for kunmap_local, and converted uses of page_address on such pages to use the newly returned address instead. However, uses of page_address on such pages were missed in ext2_check_page and ext2_delete_entry, which triggers oopses if kmap_local_page happens to return an address from high memory. Fix this now by converting the remaining uses of page_address to use the right address, as returned by kmap_local_page. Link: https://lore.kernel.org/r/20210714185448.8707ac239e9f12b3a7f5b9f9@urjc.es Reviewed-by: Ira Weiny Signed-off-by: Javier Pello Fixes: 782b76d7abdf ("fs/ext2: Replace kmap() with kmap_local_page()") Signed-off-by: Jan Kara

fs/ext2: Replace kmap() with kmap_local_page()

2021-03-31T11:10:20+00:00

The k[un]map() calls in ext2_[get|put]_page() are localized to a single thread. kmap_local_page() is more efficient. Replace the kmap/kunmap calls with kmap_local_page()/kunmap_local(). kunmap_local() requires the mapping address so return that address from ext2_get_page() to be used in ext2_put_page(). This works well because many of the callers need the address anyway so it is not bad to return it along with the page. In addition, kmap_local_page()/kunmap_local() require strict nesting rules to be followed. Document the new nesting requirements of ext2_get_page() and ext2_put_page() as well as the relationship between ext2_get_page(), ext2_find_entry(), and ext2_dotdot(). Adjust one ext2_put_page() call site in ext2_rename() to ensure the new nesting requirements are met. Finally, adjust code style for checkpatch. To: Jan Kara Link: https://lore.kernel.org/r/20210329065402.3297092-3-ira.weiny@intel.com Signed-off-by: Ira Weiny Signed-off-by: Jan Kara

ext2: Match up ext2_put_page() with ext2_dotdot() and ext2_find_entry()

2021-03-31T11:02:08+00:00

ext2_dotdot() and ext2_find_entry() both require ext2_put_page() to be called after successful return. For some of the calls this corresponding put was hidden in ext2_set_link and ext2_delete_entry(). Match up ext2_put_page() with ext2_dotdot() and ext2_find_entry() in the functions which call them. This makes the code easier to follow regarding the get/put of the page. Clean up comments to match new behavior. To: Jan Kara Link: https://lore.kernel.org/r/20210329065402.3297092-2-ira.weiny@intel.com Signed-off-by: Ira Weiny Signed-off-by: Jan Kara

fs/ext2: Use ext2_put_page

2020-11-13T10:19:30+00:00

There are 3 places in namei.c where the equivalent of ext2_put_page() is open coded on a page which was returned from the ext2_get_page() call [through the use of ext2_find_entry() and ext2_dotdot()]. Move ext2_put_page() to ext2.h and use it in namei.c Also add a comment regarding the proper way to release the page returned from ext2_find_entry() and ext2_dotdot(). Link: https://lore.kernel.org/r/20201112174244.701325-1-ira.weiny@intel.com Signed-off-by: Ira Weiny Signed-off-by: Jan Kara

ext2: ext2_find_entry() return -ENOENT if no entry found

2020-07-09T06:14:00+00:00

Almost all callers of ext2_find_entry() transform NULL return value to -ENOENT, so just let ext2_find_entry() retuen -ENOENT instead of NULL if no valid entry found, and also switch to check the return value of ext2_inode_by_name() in ext2_lookup() and ext2_get_parent(). Link: https://lore.kernel.org/r/20200608034043.10451-2-yi.zhang@huawei.com Signed-off-by: zhangyi (F) Suggested-by: Jan Kara Signed-off-by: Jan Kara

ext2: propagate errors up to ext2_find_entry()'s callers

2020-07-09T06:14:00+00:00

The same to commit <36de928641ee4> (ext4: propagate errors up to ext4_find_entry()'s callers') in ext4, also return error instead of NULL pointer in case of some error happens in ext2_find_entry() (e.g. -ENOMEM or -EIO). This could avoid a negative dentry cache entry installed even it failed to read directory block due to IO error. Link: https://lore.kernel.org/r/20200608034043.10451-1-yi.zhang@huawei.com Signed-off-by: zhangyi (F) Signed-off-by: Jan Kara