Re: Inaccurate description of statx(2) stx_blksize field?

[Alejandro Colomar <alx@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 1207 bytes --]

[CC += Ted]

Hi,

On 2026-04-22T12:37:43+0200, Zeno Endemann wrote:
> Hi,
> 
> Currently, the documentation of stx_blksize in statx says:
> 
> > The "preferred" block size for efficient filesystem I/O.
> > (Writing to a file in smaller chunks may cause an
> > inefficient read-modify-rewrite.)
> 
> I believe this is misleading; For ext4 this field is set
> to the ext4 block size, but in that context "block size"
> refers to the "smallest allocation unit" (which is chosen
> at file system creation time, see the mkfs.ext4 -b option)
> and not the most efficient I/O size that avoids the need
> for read-modify-rewrite.
> 
> At least to my understanding, to avoid such a read-modify-
> rewrite you rather want to do writes in multiples of the
> page size (i.e. sysconf(PAGESIZE)), since that is the unit
> the page cache operates on.
> 
> While in many cases ext4 will have a block size equal to
> the page size, I think it is strictly better to use the
> page size directly for that purpose. Or am I mistaken?

I don't know.  You should CC someone expert about file systems when
asking fs questions.


Have a lovely day!
Alex

-- 
<https://www.alejandro-colomar.es>

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]