Re: [tytso-ext4:dev 21/41] fs/ext4/inode.c:4002:24: sparse: sparse: symbol 'ext4_iomap_buffered_read_ops' was not declared. Should it be static?

[Zhang Yi <yizhang089@gmail.com>]

On 10/10/2026 3:03 AM, Theodore Tso wrote:
> On Fri, Oct 09, 2026 at 10:55:18AM -0500, Zhang Yi wrote:
>> Hi Ted, would it be okay to replace the current version with my v7 on
>> ext4 dev branch after I send it out?
> 
> Sure, no problem.  I took a quick look and it looks like there were
> one or two new Sashiko issues?  I'll give time for folks to take a
> look, and I'll rerun my tests on the v7 patch series, so I might not
> pull it in right away.
> 
> 						- Ted

Hi Ted,

I've gone through the Sashiko and syzbot reports one by one, and apart
from some pre-existing issues and false positives, no new regressions
were found. However, there are currently two nits in the code that need
adjustment:

  - The WARN_ON_ONCE() check in patch 8 is too heavy and is better
    removed, since it seems to expose some corner pre-existing issues in
    the buffer_head path and causes syzbot noise[1]. This syzbot report
    doesn't belong to the current series and may need to be addressed
    separately.

  - The tracepoint in patch 19 may print an uninitialized map->m_pblk
    during delalloc writes, so it needs to be initialized. I'd prefer to
    initialize it separately in ext4_da_map_blocks().

Since you've already started testing this series, and these two nits
aren't critical I think, I can send fix patches to address them
separately, or let me know if you'd prefer that I send a v8 instead.

[1] 
https://lore.kernel.org/linux-ext4/34423908-aceb-4c57-9502-7da10b12e4a4@huaweicloud.com/

Cheers,
Yi.