fs: ext4fs: Fix: Data abort in ext4fs_log_gdt()
cover fs: ext4fs: Fix: Data abort in ext4fs_log_gdt()
Currently, the number of EXT4 journal entries is defined as a constant, MAX_JOURNAL_ENTRIES = 100. When the HDD is greater than 2TB in capacity, the number of blocks per gdt (no_blk_pergdt) is a lot larger than 100. So the function ext4fs_log_gdt causes an array index out-of-bound in the journal log (journal_ptr[MAX_JOURNAL_ENTRIES]), and results in a data abort.
This was tested with the Synology DS116 (32-bit Armada 385 SoC) board, and a 4TB Seagate HDD.
Tony Dinh (1):
fs: ext4fs: Fix: Data abort in ext4fs_log_gdt()
fs/ext4/ext4_journal.c | 7 +++++++ 1 file changed, 7 insertions(+)
Signature · 4 lines
2.39.5 base-commit: e9f852f7fd2481579bb3521c13663d33a7f22eef branch: master
PATCH 1/1 fs: ext4fs: Fix: Data abort in ext4fs_log_gdt()
Return ENOMEM in ext4fs_log_gdt when number of blocks per gdt is more than number of allocated journal entries.
---
fs/ext4/ext4_journal.c | 7 +++++++ 1 file changed, 7 insertions(+)
fs/ext4/ext4_journal.c +7 -0
@@ -131,6 +131,13 @@ int ext4fs_log_gdt(char *gd_table)struct ext_filesystem *fs =get_fs ();short i;long int var = fs->gdtable_blkno; + +/* Make sure there is enough journal entries */ +if (fs->no_blk_pergdt >MAX_JOURNAL_ENTRIES ) { +log_err ("*** Not enough journal entries allocated\n" ); +return -ENOMEM ; + } +for (i =0 ; i < fs->no_blk_pergdt; i++) { journal_ptr[gindex]->buf =zalloc (fs->blksz);if (!journal_ptr[gindex]->buf)
Signature · 1 line
2.39.5
Quoted 4 lines · Tony Dinh
Return ENOMEM in
ext4fs_log_gdtwhen number of blocks per gdt is more than number of allocated journal entries.
Applied to u-boot/next, thanks!
[1/1] fs: ext4fs: Fix: Data abort in ext4fs_log_gdt()
commit: 29cb98d515f96c2330455af4b8cace0220abd00c
Signature · 1 line
Tom