[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-tar] [patch] O_NOATIME support
From: |
Paul Eggert |
Subject: |
Re: [Bug-tar] [patch] O_NOATIME support |
Date: |
Wed, 21 Sep 2005 14:30:48 -0700 |
User-agent: |
Gnus/5.1007 (Gnus v5.10.7) Emacs/21.4 (gnu/linux) |
Ian Turner <address@hidden> writes:
> If the file is already in the NFS cache, then there will be no read on
> the server, and therefore no atime update.
OK, but in that case st_atime won't be updated at all, right? And
that is what we want. So we're OK. I was mainly worried about the
possibility of a delayed update to st_atime; I don't see how that
could happen.
It should be really cheap to employ the heuristic that I mentioned,
since tar is already doing the fstat at the end anyway. No extra
system calls should be needed.
> But then you have to wrap any reference to O_NOATIME around #ifdefs.
Not if you put this into system.h (a similar thing is already done for
O_BINARY):
#ifndef O_NOATIME
# define O_NOATIME 0
#endif
- [Bug-tar] [patch] O_NOATIME support, Ian Turner, 2005/09/19
- [Bug-tar] [patch] O_NOATIME support, Ian Turner, 2005/09/19
- Re: [Bug-tar] [patch] O_NOATIME support, Paul Eggert, 2005/09/20
- Re: [Bug-tar] [patch] O_NOATIME support, Ian Turner, 2005/09/22
- Re: [Bug-tar] [patch] O_NOATIME support, Paul Eggert, 2005/09/20
- Re: [Bug-tar] [patch] O_NOATIME support, Ian Turner, 2005/09/20
- Re: [Bug-tar] [patch] O_NOATIME support,
Paul Eggert <=
- Message not available
- [Bug-tar] Re: [patch] O_NOATIME support, Paul Eggert, 2005/09/21
- [Bug-tar] Re: [patch] O_NOATIME support, Ian Turner, 2005/09/22
- [Bug-tar] Re: [patch] O_NOATIME support, Paul Eggert, 2005/09/22
- [Bug-tar] Re: [patch] O_NOATIME support, Ian Turner, 2005/09/23
- [Bug-tar] Re: [patch] O_NOATIME support, Paul Eggert, 2005/09/23
- [Bug-tar] Re: [patch] O_NOATIME support, Ian Turner, 2005/09/27
- Message not available
- [Bug-tar] Re: [patch] O_NOATIME support, Ian Turner, 2005/09/27