2.4.0test13 + Dual G4

7 messages, 3 authors, 2000-12-27 · open the first message on its own page

2.4.0test13 + Dual G4

From: Chris Backas <hidden>
Date: 2000-12-22 22:22:29

Hello all,

As per the advice from various list members (thanks all who responded!) regarding my random-coredumps under 2.2.18 in SMP mode, I've now upgraded to kernel 2.4.0
Test 13.  I was unable to build it myself (unresolved references in the link step that I couldn't figure out) I grabbed a pre-built SMP kernel from
ftp.linuxppc.org/users/kfukui.

The kernel worked fairly well, (and was REALLY REALLY FAST!) but something went wrong with XFree86 (4.0.1) and my screen went blank and wouldn't return.  Sooo... I
upgraded to Xfree86 4.0.2 from the same source.  Once I fixed a bad symlink that the RPM created (X linked to XPMac instead of XFree86) XFree worked again.

SO, getting to the point of the mail - with the newest kernel and newest XFree, my mysterious crashes still exist.  Mozilla Mail is the biggest culprit, nearly every time I open
it it dies.  No core, it just disappears.  This does not happen on a Uni-processor kernel.

My question is, when a program dies on an SMP kernel, but it fine under a Uni processor kernel, should I assume that the application has SMP issues and report it to the
author(s), or is it a kernel bug?

Thanks!
Chris Backas


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.0test13 + Dual G4

From: Kaoru Fukui <hidden>
Date: 2000-12-22 22:54:00

From: "Chris Backas" <redacted>
To: "linuxppc-dev@lists.linuxppc.org" <redacted>
Subject: 2.4.0test13 + Dual G4
Date: Fri, 22 Dec 2000 17:22:29 -0500 (EST)
Hello all,

As per the advice from various list members (thanks all who responded!) regarding
my random-coredumps under 2.2.18 in SMP mode, I've now upgraded to kernel 2.4.0
Test 13.  I was unable to build it myself (unresolved references in the link
step that I couldn't figure out) I grabbed a pre-built SMP kernel from
ftp.linuxppc.org/users/kfukui.
The link problem was fixed on 2.4.0-test13-pre3.
Get again bk tree linux_2_5

Kaoru


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.0test13 + Dual G4

From: Benjamin Herrenschmidt <hidden>
Date: 2000-12-26 18:29:40

As per the advice from various list members (thanks all who responded!)
regarding my random-coredumps under 2.2.18 in SMP mode, I've now upgraded
to kernel 2.4.0
Test 13.  I was unable to build it myself (unresolved references in the
link step that I couldn't figure out) I grabbed a pre-built SMP kernel from
ftp.linuxppc.org/users/kfukui.

The kernel worked fairly well, (and was REALLY REALLY FAST!) but
something went wrong with XFree86 (4.0.1) and my screen went blank and
wouldn't return.  Sooo... I
upgraded to Xfree86 4.0.2 from the same source.  Once I fixed a bad
symlink that the RPM created (X linked to XPMac instead of XFree86) XFree
worked again.

SO, getting to the point of the mail - with the newest kernel and newest
XFree, my mysterious crashes still exist.  Mozilla Mail is the biggest
culprit, nearly every time I open
it it dies.  No core, it just disappears.  This does not happen on a Uni-
processor kernel.
Two things come to my mind:

 - Paul fixed another SMP issues very recently. Make sure you get the
linuxppc_2_3 (or eventually _2_5) tree from hq.fsmlabs.com to get the fix

 - There are some issues with XFree 4. vs. Linux SMP (generic issues on
x86 too) that have been discussed several times on XFree list. XFree
people tend to say it's a kernel problem, while nobody (AFAIK) on the
kernel camp have made any comment.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.0test13 + Dual G4

From: Kaoru Fukui <hidden>
Date: 2000-12-26 19:13:16

From: Benjamin Herrenschmidt <redacted>
To: Chris Backas <redacted>, <redacted>,<
linuxppc-dev@lists.linuxppc.org>
 - Paul fixed another SMP issues very recently. Make sure you get the
linuxppc_2_3 (or eventually _2_5) tree from hq.fsmlabs.com to get the fix
Yesterday linux_2_5`s source was broken for SMP.
SingleCPU kernel was very well.

Now I'm puting kernel-2.4.0-test13pre3ac4 from linux_2_3.

Kaoru


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.0test13 + Dual G4

From: Benjamin Herrenschmidt <hidden>
Date: 2000-12-27 09:55:32

Yesterday linux_2_5`s source was broken for SMP.
SingleCPU kernel was very well.

Now I'm puting kernel-2.4.0-test13pre3ac4 from linux_2_3.
Be careful not to mismatch linux_2_3 and linuxppc_2_3. These are 2
different trees on fsmlabs.com. (linux_2_3 is linus tree while
linuxppc_2_3 is the PPC tree). It's possible that Paul's latest SMP fixes
did not yet make it into _2_5.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.0test13 + Dual G4

From: Kaoru Fukui <hidden>
Date: 2000-12-27 21:27:46

From: Benjamin Herrenschmidt <redacted>
To: Kaoru Fukui <redacted>, <redacted>
Subject: Re: 2.4.0test13 + Dual G4
Date: Wed, 27 Dec 2000 10:55:32 +0100
quoted
Yesterday linux_2_5`s source was broken for SMP.
SingleCPU kernel was very well.

Now I'm puting kernel-2.4.0-test13pre3ac4 from linux_2_3.
Be careful not to mismatch linux_2_3 and linuxppc_2_3. These are 2
different trees on fsmlabs.com. (linux_2_3 is linus tree while
linuxppc_2_3 is the PPC tree). It's possible that Paul's latest SMP fixes
did not yet make it into _2_5.

Ben.
Thanks Ben.

I'm struggling with bk tree 5000 linuxppc_2_3.

The source was able to run as singleCPU.
it's good working mono cpu.
The compilation is no problem,
 I set SMP flag ,then the compilation have done,
but the kernel say one cpu.

Kaoru


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: 2.4.0test13 + Dual G4

From: Kaoru Fukui <hidden>
Date: 2000-12-27 22:02:05

From: Kaoru Fukui <redacted>
 I set SMP flag ,then the compilation have done,
but the kernel say one cpu.
Sorry i didn't say my machine.

my machine is Dual g4

Kaoru


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help