On Thu, Sep 11, 2025 at 7:13 AM Christoph Hellwig [off-list ref] wrote:
On Mon, Sep 08, 2025 at 11:51:11AM -0700, Joanne Koong wrote:
quoted
Propagate any error encountered in iomap_read_folio() back up to its
caller (otherwise a default -EIO will be passed up by
filemap_read_folio() to callers). This is standard behavior for how
other filesystems handle their ->read_folio() errors as well.
Remove the out of date comment about setting the folio error flag.
Folio error flags were removed in commit 1f56eedf7ff7 ("iomap:
Remove calls to set and clear folio error flag").
As mentioned last time I actually think this is a bad idea, and the most
common helpers (mpage_read_folio and block_read_full_folio) will not
return and error here.
I'll drop this. I interpreted Matthew's comment to mean the error
return isn't useful for ->readahead but is for ->read_folio.
If iomap_read_folio() doesn't do error returns and always just returns
0, then maybe we should just make it `void iomap_read_folio(...)`
instead of `int iomap_read_folio(...)` then.
Thanks,
Joanne