* kswapd and SMP still broke on 2.2.0-pre9
@ 1999-01-22 19:44 Troy Benjegerdes
1999-01-23 2:59 ` Paul Mackerras
0 siblings, 1 reply; 10+ messages in thread
From: Troy Benjegerdes @ 1999-01-22 19:44 UTC (permalink / raw)
To: linuxppc-dev
kswapd is still dieing on boot with SMP enabled (and this time the second
processor is actually active)
Does this indicate something is wrong with SMP and the MM code, or that
some PPC-specific part needs fixing? I've tried to keep up on
linux-kernel, but it's a little overwhelming ;)
ncr53c875-0: Downloading SCSI SCRIPTS.
scsi0 : ncr53c8xx - revision 3.1f
scsi : 1 host.
Trying to vfree() nonexistent vm area (ffffffff)
NIP: 00000000 XER: 20000000 LR: 00000000 REGS: c03d5e30 TRAP: 0700
MSR: 00089032 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11
TASK = c03d4000[3] 'kswapd' mm->pgd c012a000 Last syscall: -1
last math 00000000 CPU: 0 last CPU: 1
GPR00: 00000000 C03D5F20 C03D4000 00000031 00000001 00000000 C0140000 C0140C18
GPR08: 000F3F26 C0160000 000003F8 C03D5E50 00000000 07F719C0 00000000 00000000
GPR16: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
GPR24: 00000000 C0140000 C0135314 C0140000 C0180000 00000064 C0140000 C00FF064
Call backtrace:
00000000 80097188
Instruction DUMP:NIP: C0007290 XER: 20000000 LR: C0007278 REGS: c03d5cf0
TRAP: 0300
MSR: 00009032 EE: 1 PR: 0 FP: 0 ME: 1 IR/DR: 11
TASK = c03d4000[3] 'kswapd' mm->pgd c012a000 Last syscall: -1
last math 00000000 CPU: 0 last CPU: 1
GPR00: C0007278 C03D5DE0 C03D4000 00000011 00000000 00000000 00000000 C0140C18
GPR08: 000F3F26 C0160000 000003F8 C03D5D20 00000000 07F719C0 00000000 00000000
GPR16: 00000000 00000000 00000000 00000000 00009032 003D5E20 00000000 C0003A98
GPR24: C00057A8 C0140000 C0135314 C0140000 C0100000 00000000 FFFFFFF4 FFFFFFFD
Call backtrace:
C0007278 C00055D0 C00057EC C0003A98 00000000 80097188
Instruction DUMP: 309fffff 7c842110 548606bc <7cbee82e> 548406ba
60840020 60c6
--------------------------------------------------------------------------
| Troy Benjegerdes | troy@microux.com | hozer@drgw.net |
| Unix is user friendly... You just have to be friendly to it first. |
| This message composed with 100% free software. http://www.gnu.org |
--------------------------------------------------------------------------
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: kswapd and SMP still broke on 2.2.0-pre9
1999-01-22 19:44 kswapd and SMP still broke on 2.2.0-pre9 Troy Benjegerdes
@ 1999-01-23 2:59 ` Paul Mackerras
1999-01-23 5:54 ` Troy Benjegerdes
0 siblings, 1 reply; 10+ messages in thread
From: Paul Mackerras @ 1999-01-23 2:59 UTC (permalink / raw)
To: hozer; +Cc: linuxppc-dev
Troy Benjegerdes <hozer@drgw.net> wrote:
> kswapd is still dieing on boot with SMP enabled (and this time the second
> processor is actually active)
>
> Does this indicate something is wrong with SMP and the MM code, or that
> some PPC-specific part needs fixing? I've tried to keep up on
> linux-kernel, but it's a little overwhelming ;)
The SMP stuff in Linux/PPC currently only supports the Apple MP
architecture (I think they call it "powersurge"). Your MTX board
probably has entirely different ways of coordinating the CPUs. Have
you looked at arch/ppc/kernel/smp.c? I think that's where you'd have
to start changing things.
Paul.
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: kswapd and SMP still broke on 2.2.0-pre9
1999-01-23 2:59 ` Paul Mackerras
@ 1999-01-23 5:54 ` Troy Benjegerdes
1999-01-23 9:05 ` Ryan Nielsen
` (2 more replies)
0 siblings, 3 replies; 10+ messages in thread
From: Troy Benjegerdes @ 1999-01-23 5:54 UTC (permalink / raw)
To: Paul.Mackerras; +Cc: linuxppc-dev
On Sat, 23 Jan 1999, Paul Mackerras wrote:
> Troy Benjegerdes <hozer@drgw.net> wrote:
>
> > kswapd is still dieing on boot with SMP enabled (and this time the second
> > processor is actually active)
> >
> > Does this indicate something is wrong with SMP and the MM code, or that
> > some PPC-specific part needs fixing? I've tried to keep up on
> > linux-kernel, but it's a little overwhelming ;)
>
> The SMP stuff in Linux/PPC currently only supports the Apple MP
> architecture (I think they call it "powersurge"). Your MTX board
> probably has entirely different ways of coordinating the CPUs. Have
> you looked at arch/ppc/kernel/smp.c? I think that's where you'd have
> to start changing things.
This is true, and I need to finish stuff in smp.c, but the same crash
occured from 2.2.0-pre5 on, without the second CPU being active. Patching
the kswapd portion of back to 2.2.0-pre4 fixed this.. Look for a message
in the archives from Ryan Nielson, I believe.
Can anyone verify 2.2.0-pre9 works on a SMP Pmac? (or for that matter,
vger.. if vger works I can figure out what changed)
--------------------------------------------------------------------------
| Troy Benjegerdes | troy@microux.com | hozer@drgw.net |
| Unix is user friendly... You just have to be friendly to it first. |
| This message composed with 100% free software. http://www.gnu.org |
--------------------------------------------------------------------------
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: kswapd and SMP still broke on 2.2.0-pre9
1999-01-23 5:54 ` Troy Benjegerdes
@ 1999-01-23 9:05 ` Ryan Nielsen
1999-01-23 18:25 ` Cort Dougan
1999-01-23 23:25 ` Paul Mackerras
2 siblings, 0 replies; 10+ messages in thread
From: Ryan Nielsen @ 1999-01-23 9:05 UTC (permalink / raw)
To: Troy Benjegerdes; +Cc: linuxppc-dev, Paul.Mackerras
Troy Benjegerdes wrote:
>
> On Sat, 23 Jan 1999, Paul Mackerras wrote:
>
> > Troy Benjegerdes <hozer@drgw.net> wrote:
> >
> > > kswapd is still dieing on boot with SMP enabled (and this time the second
> > > processor is actually active)
> > >
> > > Does this indicate something is wrong with SMP and the MM code, or that
> > > some PPC-specific part needs fixing? I've tried to keep up on
> > > linux-kernel, but it's a little overwhelming ;)
> >
> > The SMP stuff in Linux/PPC currently only supports the Apple MP
> > architecture (I think they call it "powersurge"). Your MTX board
> > probably has entirely different ways of coordinating the CPUs. Have
> > you looked at arch/ppc/kernel/smp.c? I think that's where you'd have
> > to start changing things.
>
> This is true, and I need to finish stuff in smp.c, but the same crash
> occured from 2.2.0-pre5 on, without the second CPU being active. Patching
> the kswapd portion of back to 2.2.0-pre4 fixed this.. Look for a message
> in the archives from Ryan Nielson, I believe.
>
> Can anyone verify 2.2.0-pre9 works on a SMP Pmac? (or for that matter,
> vger.. if vger works I can figure out what changed)
vger-pre8 and 2.2.0-pre9 dont work.
on an Apple 9600/200MP.
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: kswapd and SMP still broke on 2.2.0-pre9
1999-01-23 5:54 ` Troy Benjegerdes
1999-01-23 9:05 ` Ryan Nielsen
@ 1999-01-23 18:25 ` Cort Dougan
1999-01-23 23:25 ` Paul Mackerras
2 siblings, 0 replies; 10+ messages in thread
From: Cort Dougan @ 1999-01-23 18:25 UTC (permalink / raw)
To: Troy Benjegerdes; +Cc: Paul.Mackerras, linuxppc-dev
vger SMP is broken right now.
}Can anyone verify 2.2.0-pre9 works on a SMP Pmac? (or for that matter,
}vger.. if vger works I can figure out what changed)
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: kswapd and SMP still broke on 2.2.0-pre9
1999-01-23 5:54 ` Troy Benjegerdes
1999-01-23 9:05 ` Ryan Nielsen
1999-01-23 18:25 ` Cort Dougan
@ 1999-01-23 23:25 ` Paul Mackerras
1999-01-24 7:02 ` Ryan Nielsen
1999-01-26 2:09 ` Cort Dougan
2 siblings, 2 replies; 10+ messages in thread
From: Paul Mackerras @ 1999-01-23 23:25 UTC (permalink / raw)
To: hozer; +Cc: linuxppc-dev
Troy Benjegerdes <hozer@drgw.net> wrote:
> Can anyone verify 2.2.0-pre9 works on a SMP Pmac? (or for that matter,
> vger.. if vger works I can figure out what changed)
Both 2.2.0-pre9 and vger work fine on my 2-processor powermac. I do
have some local changes in my tree to do with some extra memory
barriers etc., but I haven't changed kswapd. You could try rsyncing
my current tree from samba.anu.edu.au::linux-pmac-devel to see my
changes.
Paul.
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: kswapd and SMP still broke on 2.2.0-pre9
1999-01-23 23:25 ` Paul Mackerras
@ 1999-01-24 7:02 ` Ryan Nielsen
1999-01-24 21:30 ` Troy Benjegerdes
1999-01-26 2:09 ` Cort Dougan
1 sibling, 1 reply; 10+ messages in thread
From: Ryan Nielsen @ 1999-01-24 7:02 UTC (permalink / raw)
To: Paul.Mackerras; +Cc: linuxppc-dev
Paul Mackerras wrote:
>
> Troy Benjegerdes <hozer@drgw.net> wrote:
>
> > Can anyone verify 2.2.0-pre9 works on a SMP Pmac? (or for that matter,
> > vger.. if vger works I can figure out what changed)
>
> Both 2.2.0-pre9 and vger work fine on my 2-processor powermac. I do
> have some local changes in my tree to do with some extra memory
> barriers etc., but I haven't changed kswapd. You could try rsyncing
> my current tree from samba.anu.edu.au::linux-pmac-devel to see my
> changes.
do you use scsi on your 2-processor powermac?
with the rsync'd tree it stops somewhere after printing the last line of this type:
Detected scsi disk sd%d at scsi%d, channel %d, id %d, lun %d
all the disks are on the MESH.
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: kswapd and SMP still broke on 2.2.0-pre9
1999-01-24 7:02 ` Ryan Nielsen
@ 1999-01-24 21:30 ` Troy Benjegerdes
0 siblings, 0 replies; 10+ messages in thread
From: Troy Benjegerdes @ 1999-01-24 21:30 UTC (permalink / raw)
To: Ryan Nielsen; +Cc: Paul.Mackerras, linuxppc-dev
On Sat, 23 Jan 1999, Ryan Nielsen wrote:
>
> Paul Mackerras wrote:
> >
> > Troy Benjegerdes <hozer@drgw.net> wrote:
> >
> > > Can anyone verify 2.2.0-pre9 works on a SMP Pmac? (or for that matter,
> > > vger.. if vger works I can figure out what changed)
> >
> > Both 2.2.0-pre9 and vger work fine on my 2-processor powermac. I do
> > have some local changes in my tree to do with some extra memory
> > barriers etc., but I haven't changed kswapd. You could try rsyncing
> > my current tree from samba.anu.edu.au::linux-pmac-devel to see my
> > changes.
>
> do you use scsi on your 2-processor powermac?
> with the rsync'd tree it stops somewhere after printing the last line of this type:
> Detected scsi disk sd%d at scsi%d, channel %d, id %d, lun %d
> all the disks are on the MESH.
I haven't tried the rsync'd tree yet, my machine was crashing in the same
SCSI initialization. I have an on-board ncr53c8xx fast-wide chip, so this
leads me to believe there is some interaction between the memory
management code and the SCSI init code.
Is this something we should bring up on the regular linux-kernel list?
Maybe someone there would have an idea what is broken.
I don't know if this is related, but I also get crashes when doing nfs
access (checking out files from my local CVS archive, in particular).
I traced it one time and it appears a pointer is getting trashed in
rpciod.
--------------------------------------------------------------------------
| Troy Benjegerdes | troy@microux.com | hozer@drgw.net |
| Unix is user friendly... You just have to be friendly to it first. |
| This message composed with 100% free software. http://www.gnu.org |
--------------------------------------------------------------------------
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: kswapd and SMP still broke on 2.2.0-pre9
1999-01-23 23:25 ` Paul Mackerras
1999-01-24 7:02 ` Ryan Nielsen
@ 1999-01-26 2:09 ` Cort Dougan
1999-01-26 6:02 ` 2.2.0 SMP kswapd okay, but rpciod crashes Troy Benjegerdes
1 sibling, 1 reply; 10+ messages in thread
From: Cort Dougan @ 1999-01-26 2:09 UTC (permalink / raw)
To: Paul.Mackerras; +Cc: hozer, linuxppc-dev
I get failures. I get a hang right after MMU_init() with a MP kernel even
on a UP machine. Same crash I was getting on chrp and smp some time ago.
}Both 2.2.0-pre9 and vger work fine on my 2-processor powermac. I do
}have some local changes in my tree to do with some extra memory
}barriers etc., but I haven't changed kswapd. You could try rsyncing
}my current tree from samba.anu.edu.au::linux-pmac-devel to see my
}changes.
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
* 2.2.0 SMP kswapd okay, but rpciod crashes
1999-01-26 2:09 ` Cort Dougan
@ 1999-01-26 6:02 ` Troy Benjegerdes
0 siblings, 0 replies; 10+ messages in thread
From: Troy Benjegerdes @ 1999-01-26 6:02 UTC (permalink / raw)
To: Cort Dougan; +Cc: Paul.Mackerras, linuxppc-dev
2.2.0 fixes the kswapd crashes, but nfs has some problems in rpciod.
(as you can see I haven't finished the IPI support ;)
The crash occures at
<rpciod_tcp_dispatcher+1568> lwz r0,16256(r21)
This appears to be a problem with r20, r21, and r22 getting trashed
somewhere. This is supposed to be loading rpc_debug into r0.
If comment out #define RPC_DEBUG in linux/include/linux/sunrpc/debug.h,
and fix a couple of nfs-related stuff, nfs works without crashing again.
NIP: C0090724 XER: 20000000 LR: C00906D0 REGS: c6e8fe20 TRAP : 0300
MSR: 00001032 EE: 0 PR: 0 FP: 0 ME: 1 IR/DR: 11
TASK = c6e8e000[311] 'rpciod' mm->pgd c012a000 Last syscall: -1
last math 00000000 CPU: 0 last CPU: 1
GPR00: 00000000 C6E8FF10 C6E8E000 0000001F C0143ADC C0143ADC 00000002 C01450E0
GPR08: 00000000 FFFFFFF5 00000000 3FEBAF20 82242428 018529A0 00000001 00000000
GPR16: 00000000 00000000 0184D3A0 0184D3B0 00009032 0184D3E0 0184D3C0 00000000
GPR24: 0184D3C0 C0143F64 C0140000 C0140000 C0140000 FFFFFFF5 00000001 C0140000
Call backtrace:
C00906D0 C00093E8
Instruction DUMP: 40820008 901c0e0c 3ae00000 <80153f80> 70070001 4182001c
3c60c011 80bc0e0c 80dc0e18
Kernel panic: kernel access of bad area pc c0090724 lr c00906d0 address
1851360 ts
k rpciod/311
Message from syslogd@kalmia at Mon Jan 25 21:53:30 1999 ...
kalmia kernel: KOpenPIC IPI support not done yet!!
ernel panic: kerRebooting in 180 seconds..nel access of bad area pc
c0090724 lr c0
0906d0 address 1851360 tsk rpciod/311
--------------------------------------------------------------------------
| Troy Benjegerdes | troy@microux.com | hozer@drgw.net |
| Unix is user friendly... You just have to be friendly to it first. |
| This message composed with 100% free software. http://www.gnu.org |
--------------------------------------------------------------------------
[[ This message was sent via the linuxppc-dev mailing list. Replies are ]]
[[ not forced back to the list, so be sure to Cc linuxppc-dev if your ]]
[[ reply is of general interest. To unsubscribe from linuxppc-dev, send ]]
[[ the message 'unsubscribe' to linuxppc-dev-request@lists.linuxppc.org ]]
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~1999-01-26 6:02 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
1999-01-22 19:44 kswapd and SMP still broke on 2.2.0-pre9 Troy Benjegerdes
1999-01-23 2:59 ` Paul Mackerras
1999-01-23 5:54 ` Troy Benjegerdes
1999-01-23 9:05 ` Ryan Nielsen
1999-01-23 18:25 ` Cort Dougan
1999-01-23 23:25 ` Paul Mackerras
1999-01-24 7:02 ` Ryan Nielsen
1999-01-24 21:30 ` Troy Benjegerdes
1999-01-26 2:09 ` Cort Dougan
1999-01-26 6:02 ` 2.2.0 SMP kswapd okay, but rpciod crashes Troy Benjegerdes
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).