Re: [PATCH 14/14] ext4: use a byte granularity cursor in ext4_mpage_readpages
Eric Biggers <[email protected]>
| Newsgroups | org.kernel.vger.linux-fscrypt,net.sourceforge.lists.linux-f2fs-devel,org.kernel.vger.linux-ext4,org.kernel.vger.linux-fsdevel |
|---|---|
| Message-ID | <20260227222950.GC5357@quark> |
On Thu, Feb 26, 2026 at 06:49:34AM -0800, Christoph Hellwig wrote:
> + block_in_file = EXT4_PG_TO_LBLK(inode, folio->index);
> + pos = (loff_t)block_in_file << blkbits;
The EXT4_PG_TO_LBLK() expands to:
(((loff_t)(folio->index) << PAGE_SHIFT) >> (inode)->i_blkbits)
So it calculates the pos as an intermediate step, and we end up with the
redundant pos = (pos >> blkbits) << blkbits.
It probably would make more sense to calculate the pos first, similar to
what other places in this series do:
pos = folio_pos(folio);
block_in_file = pos >> blkbits;
- Eric