Re: [PATCH 1/2] f2fs: simplify node footer check in f2fs_cache_write_end_io

[Zhiguo Niu]

Hi Xusheng

Zhan Xusheng <zhanxusheng1024@xxxxxxxxx> 于2026年9月24日周四 15:23写道：
>
> On Thu, Sep 24, 2026 at 09:52:12AM +0800, Zhiguo Niu wrote:
> > -             f2fs_bug_on(sbi, entry->index != nid_of_node(sbi, entry));
> > +             f2fs_bug_on(sbi, f2fs_sanity_check_node_footer(sbi,
> > +                     entry, entry->index, NODE_TYPE_REGULAR, true));
>
> This also promotes every other footer inconsistency to an assertion.
> Before, only a nid mismatch reached f2fs_bug_on(); the rest went through
> the helper's own path, which sets SBI_NEED_FSCK, warns and calls
> f2fs_handle_error().  After, any of them BUG() under CONFIG_F2FS_CHECK_FS.
For NODE_TYPE_REGULAR,  add the  case of “if(is_inode && is_xnode)” to
this flow,
aim to check return value  of f2fs_sanity_check_node_footer && remove
redundant  footer inconsistency check。
>
> The other two callers went the other way.  __write_node_cache():
>
>         if (f2fs_sanity_check_node_footer(sbi, entry, entry->index,
>                                         NODE_TYPE_REGULAR, false)) {
>                 fserror_report_metadata(sbi->sb, -EFSCORRUPTED, GFP_NOFS);
>                 f2fs_stop_checkpoint(sbi, false, STOP_CP_REASON_CORRUPTED_NID);
>                 goto redirty_out;
>         }
>
> and the read completion turns it into BLK_STS_IOERR.  Worth a line in the
> changelog either way.
>
> FAULT_INCONSISTENT_FOOTER is injected inside that helper, so with
> CONFIG_F2FS_CHECK_FS=y injecting it now panics instead of exercising the
> error path.  Deliberate?
oh ,sorry, I missed this, thansk for pointing it.