cvs commit: src/sys/kern sys_generic.c
phk at phk.freebsd.dk
Tue Jan 25 07:39:46 PST 2005
In message <20050125153729.GF5662 at lum.celabo.org>, "Jacques A. Vidrine" writes:
>On Tue, Jan 25, 2005 at 12:42:26PM +0300, Andrey Chernov wrote:
>> On Tue, Jan 25, 2005 at 09:15:32AM +0000, Poul-Henning Kamp wrote:
>> > sys/kern sys_generic.c
>> > Log:
>> > Previously a read of zero bytes got handled in devfs:vop_read() but I
>> > missed that when the vnode bypass was introduced.
>> > Deal with zero length transfers before we even get to fo_ops->fo_read().
>> Please back it out. Zero bytes read may affect atime, now not.
>I do not think that a zero byte read should not affect atime.
Explain to me then why open+mmap doesn't affect atime isn't a bigger problem ?
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
More information about the cvs-all