On 2020/12/8 9:26, Jaegeuk Kim wrote:
On 12/08, Chao Yu wrote:
On 2020/12/8 4:37, Eric Biggers wrote:
On Thu, Nov 26, 2020 at 06:32:09PM +0800, Chao Yu wrote:
+ if (!ret && fi->i_compress_flag & 1 << COMPRESS_CHKSUM) {
This really could use some parentheses. People shouldn't have to look up a
C operator precedence table to understand the code.
Will add parentheses to avoid misread.
+ u32 provided = le32_to_cpu(dic->cbuf->chksum);
+ u32 calculated = f2fs_crc32(sbi, dic->cbuf->cdata, dic->clen);
+
+ if (provided != calculated) {
+ if (!is_inode_flag_set(dic->inode,
FI_COMPRESS_CORRUPT)) {
+ set_inode_flag(dic->inode, FI_COMPRESS_CORRUPT);
+ printk_ratelimited(
+ "%sF2FS-fs (%s): checksum invalid, nid =
%lu, %x vs %x",
+ KERN_INFO, sbi->sb->s_id,
dic->inode->i_ino,
+ provided, calculated);
+ }
+ set_sbi_flag(sbi, SBI_NEED_FSCK);
+ WARN_ON_ONCE(1);
WARN, WARN_ON_ONCE, BUG, BUG_ON, etc. are only for kernel bugs, not for invalid
inputs from disk or userspace.
There is already a log message printed just above, so it seems this WARN_ON_ONCE
should just be removed.
Jaegeuk wants to give WARN_ON and marking a FSCK flag without returning
EFSCORRUPTED,
I think above printk_ratelimited should be enough.
Okay, so let me update the patch.
Thanks,
Jaegeuk, thoughts?
Thanks,
- Eric
.
.
