Dear all,
2.4.20-ac1 has hung my Laptop several times. May this be due to the
ATA-problem I found emails for in the archive?
Lukas
PS: I switched back to 2.4.19 which has been running without
problems...
--
Lukas Ruf
http://www.lpr.ch
Wanna know anything about raw ip?
Join [email protected] on http://www.rawip.org
On Mon, 2002-12-09 at 21:34, Lukas Ruf wrote:
> Dear all,
>
> 2.4.20-ac1 has hung my Laptop several times. May this be due to the
> ATA-problem I found emails for in the archive?
More likely to be something else. You'd get dumps from IDE problems, and
the thinkpad IDE is PIIX which is very well tested. Could well be a
.20-ac1 bug or a .20 bug though is vanilla 2.4.20 fine ?
I've been running on 2.4.20-ac1 since it came out with no problems.
ThinkPad 380XD. I can try it on an A20 if it is an issue.
cc me.
Garst
I have also stopped using 2.4.20-ac1 on my ThinkPad.
It performs much worse than 2.4.20-pre8-ac1 when running
large applications such as OpenOffice. (It never crashed
on me, though.) Having read the recent thread on changes
to the scheduler and yield(), I suspect that the problem
is related to that, since the 2.4-ac series has the O(1)
scheduler.
If my suspicion is correct, then I concur with the opinion
that O(1) should not be backported to mainline 2.4.
--
Thomas Hood
__________________________________________________
Do You Yahoo!?
Everything you'll ever need on one web page
from News and Sport to Email and Music Charts
http://uk.my.yahoo.com
On Tue, 2002-12-10 at 12:53, J.D. Hood wrote:
> I have also stopped using 2.4.20-ac1 on my ThinkPad.
> It performs much worse than 2.4.20-pre8-ac1 when running
> large applications such as OpenOffice. (It never crashed
> on me, though.) Having read the recent thread on changes
> to the scheduler and yield(), I suspect that the problem
> is related to that, since the 2.4-ac series has the O(1)
> scheduler.
Thats a bug in openoffice (or more accurately in the glibc thread
library). Our yield now does what it is *supposed* to do. Unfortunately
glibc is coded to rely on what used to happen. Andrea has a nice fix for
this which I'll probably merge soon