Christoph Hellwig <h...@infradead.org> writes:
> On Sat, Oct 18, 2014 at 07:21:26PM +0400, Dmitry Monakhov wrote:
>> sys_fadvise result in direct f_mode modification, which may be not
>> suitable for some unusual filesytems where file mode invariant is more
>> complex. In order to support such filesystems we have to delegate fadvise
>> logic to filesystem layer.
>
> Is there a real use case for it? So for it seems mostly about ecryptfs,
> and even that use is lacking a proper explanation.
Fairly to say original issue was with vzfs (COW stack filesystem for
containers from openvz). As far as I understand direct analog is
unionfs. We asserted v_file->mode == lower_file->f_mode, but
fadvise (POSIX_FADV_{RANDOM,RANDOM,SEQUENTIAL})changes v_file->mode directly.
>
> Also fadvice and set_flags seem entirely unrelated, I don't understand
> why you're throwing fadvice in thise series.
It has semantic relation. Both methods manipulate filp->XXX internals directly
No problem I can split this to separate patch set.
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
> the body of a message to majord...@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
pgpMiu9tTOapD.pgp
Description: PGP signature
