Re: v2.0.30

Stephen Zedalis (
Tue, 15 Apr 1997 08:01:51 -0400 (EDT)

I beg your pardon. Check out the change summary for 2.0.30 on It clearly says that no_atime support is one of
the new changes for 2.0.30 (or at least it did). The current change
to /linux/include/linux/ext2_fs.h clearly adds the define
EXT2_MOUNT_NO_ATIME as a change over 2.0.29. This is the code that
makes the hunk fail on the current no_atime patch and is something
I just wanted to point out. I CAN read code. Your condescending
tone about "making sure the patch fragment does what you say it does"
was unwarranted. A simple, "oops we decided not to implement it yet, but
left that fragment in" would have been sufficient. As it is, several
if not most ISP's use the current no_atime patch in production servers,
clean or not. I wouldn't even have tried it, if it hadn't been for a
email discussion with Alan Cox as to how we could optimize our news
server. (The original discussion was over the shortcomings of the 2.0.x
SMP implementation). I'm glad you are working on a cleaner
implementation, but several people are using the current no_atime, and
the new fragment will cause patch to complain about patch hunk previously
applied. Thats all I was trying to say.

On Tue, 15 Apr 1997, Nigel Metheringham wrote:

>The no_atime patch was rejected by Linux because the implementation is
>unclean. If there is a fragment of the patch in 2.0.30 then thats due to
>something else being done - I would check to make sure that patch fragment
>does what you think it does.
>There is a cleaner implementation being worked on.