From: Stefan Jeglinski <hidden> Date: 2001-10-21 04:02:12
6500, Linux/PPC 2KQ4 running 2.2.19, I compiled 2.4.13-pre5, rsynced
from bk yesterday. On boot, when it gets to "mounting proc
filesystem," I get a break into the mon debugger, but I have no
keyboard response (so no backtrace). The info I can copy is:
vector: 300 at pc = c02432fc lr = c00544e0 msr = b032 sp = c5dc3ef0
plus some more that I'm not sure will help much. For this output
alone, should I include System.map also?
Is this a known problem?
Stefan Jeglinski
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: Stefan Jeglinski <hidden> Date: 2001-10-21 04:25:25
6500, Linux/PPC 2KQ4 running 2.2.19, I compiled 2.4.13-pre5, rsynced
from bk yesterday. On boot, when it gets to "mounting proc
filesystem," I get a break into the mon debugger, but I have no
keyboard response (so no backtrace). The info I can copy is:
vector: 300 at pc = c02432fc lr = c00544e0 msr = b032 sp = c5dc3ef0
plus some more that I'm not sure will help much. For this output
alone, should I include System.map also?
Is this a known problem?
I've just found the same result on a quite-different machine: Power
Mac 9500 with G3 upgrade.
On both the 6500 and the 9500, I usually make a number of small mods
to the compile (compile sound in instead of module, turn off video
and ethernet drivers I don't need, turn off USB modem and PPP, stuff
like this), but I can't be totally sure I haven't turned something
important off because I didn't write it all down...
I've never seen a freeze at the proc filesystem mount. What should I
look for to fix this?
Stefan Jeglinski
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: Paul Mackerras <hidden> Date: 2001-10-21 12:26:15
Stefan Jeglinski writes:
6500, Linux/PPC 2KQ4 running 2.2.19, I compiled 2.4.13-pre5, rsynced
from bk yesterday.
linuxppc_2_4 or linuxppc_2_4_devel?
On boot, when it gets to "mounting proc
filesystem," I get a break into the mon debugger, but I have no
keyboard response (so no backtrace). The info I can copy is:
You have an ADB keyboard, no? I am surprised the keyboard doesn't
respond.
vector: 300 at pc = c02432fc lr = c00544e0 msr = b032 sp = c5dc3ef0
plus some more that I'm not sure will help much. For this output
alone, should I include System.map also?
Yes, or alternatively find out where those pc and lr addresses are in
the System.map list and tell us the nearest symbols and their values.
From: Stefan Jeglinski <hidden> Date: 2001-10-21 15:48:13
> 6500, Linux/PPC 2KQ4 running 2.2.19, I compiled 2.4.13-pre5, rsynced
quoted
from bk yesterday.
linuxppc_2_4 or linuxppc_2_4_devel?
linuxppc_2_4, and to be more specific I rsynced from
source.mvista.com, which I understand is the bk mirror.
> On boot, when it gets to "mounting proc
quoted
filesystem," I get a break into the mon debugger, but I have no
keyboard response (so no backtrace). The info I can copy is:
You have an ADB keyboard, no? I am surprised the keyboard doesn't
respond.
Yes, on both the 6500 and the 9500. On the 6500, where I have long
been having stability problems with 2.4 that are completely unrelated
to this crash, the keyboard uniformly has never responded in mon. I
tried to hack mon to automatically print a backtrace but could never
get that to work either. On the 9500, I only assumed it broke into
mon. What happened (besides the total lockup) was that what looked
like parts of the mon output overwrote the boot information that was
listed above the point of the freeze. IOW, I saw a pc= and other
parts randomly written over the console output up to that point. For
some reason, no numbers came out, so I had even less info than for
the 6500.
> vector: 300 at pc = c02432fc lr = c00544e0 msr = b032 sp = c5dc3ef0
quoted
plus some more that I'm not sure will help much. For this output
alone, should I include System.map also?
Yes, or alternatively find out where those pc and lr addresses are in
the System.map list and tell us the nearest symbols and their values.
System.map-2.4.13-pre5 from the 6500:
c0242994 ? __exitcall_netlink_proto_exit
c0242998 ? __exitcall_packet_exit
c024299c A __start___ex_table
c024299c A _edata
c0245b4c ? __ksymtab_rtc_lock
c0245b4c A __start___ksymtab
c00541d8 t do_remount
c00542f0 t do_add_mount
c0054448 t copy_mount_options
c0054560 T do_mount
c005471c T sys_mount
c0054808 t chroot_fs_refs
Thanks for the help, Paul.
Stefan Jeglinski
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: Benjamin Herrenschmidt <benh@kernel.crashing.org> Date: 2001-10-21 18:58:12
quoted
On boot, when it gets to "mounting proc
filesystem," I get a break into the mon debugger, but I have no
keyboard response (so no backtrace). The info I can copy is:
You have an ADB keyboard, no? I am surprised the keyboard doesn't
respond.
Could be async probing not beeing complete. I could push the adb_sync
kernel arg to the bk trees, at least _devel.
quoted
vector: 300 at pc = c02432fc lr = c00544e0 msr = b032 sp = c5dc3ef0
plus some more that I'm not sure will help much. For this output
alone, should I include System.map also?
Yes, or alternatively find out where those pc and lr addresses are in
the System.map list and tell us the nearest symbols and their values.
quoted
Is this a known problem?
Not to me...
Not to me neither but I'm still running pre3.
Ben.
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: KOBAYASHI R. Taizo <hidden> Date: 2001-10-25 14:31:36
On Sun, 21 Oct 2001 00:02:12 -0400
Stefan Jeglinski [off-list ref] wrote:
Subject: 2.4.13-pre5 crash when mounting proc
6500, Linux/PPC 2KQ4 running 2.2.19, I compiled 2.4.13-pre5, rsynced
from bk yesterday. On boot, when it gets to "mounting proc
filesystem," I get a break into the mon debugger, but I have no
keyboard response (so no backtrace). The info I can copy is:
vector: 300 at pc = c02432fc lr = c00544e0 msr = b032 sp = c5dc3ef0
plus some more that I'm not sure will help much. For this output
alone, should I include System.map also?
My pb2400 got well again with following:
--- arch/ppc/lib/string.S.orig Thu Oct 25 19:09:31 2001+++ arch/ppc/lib/string.S Thu Oct 25 23:17:39 2001
@@ -62,8 +62,7 @@.long8## n ## 5b,9 ## n ## 1b; \.long8## n ## 6b,9 ## n ## 1b; \.long8## n ## 7b,9 ## n ## 1b; \--.text+.textCACHELINE_BYTES=L1_CACHE_LINE_SIZELG_CACHELINE_BYTES=LG_L1_CACHE_LINE_SIZE
On Thu, Oct 25, 2001 at 11:31:36PM +0900, KOBAYASHI R. Taizo wrote:
On Sun, 21 Oct 2001 00:02:12 -0400
Stefan Jeglinski [off-list ref] wrote:
Subject: 2.4.13-pre5 crash when mounting proc
quoted
6500, Linux/PPC 2KQ4 running 2.2.19, I compiled 2.4.13-pre5, rsynced
from bk yesterday. On boot, when it gets to "mounting proc
filesystem," I get a break into the mon debugger, but I have no
keyboard response (so no backtrace). The info I can copy is:
vector: 300 at pc = c02432fc lr = c00544e0 msr = b032 sp = c5dc3ef0
plus some more that I'm not sure will help much. For this output
alone, should I include System.map also?
My pb2400 got well again with following:
[snip]
Um. Is that _all_ that changed? That really shouldn't have any effect
at all... What compiler/binutils combo is this?
--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: Edward Swarthout <hidden> Date: 2001-10-25 16:53:27
Tom wrote:
Um. Is that _all_ that changed? That really shouldn't have any effect
at all... What compiler/binutils combo is this?
Well it's important for the COPY_16_BYTES_EXCODE macro to switch back to
the .text section or any following code will go in the __ex_table section!
It looks like the .text line was deleted in the 1.8 change set for linuxppc_2_4
If you notice the linuxppc_2_4_devel has two .text lines next to each other.
(I had fun debugging this same type failure when I added the .stabs lines
and left only one .text line)
-Ed
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: Stefan Jeglinski <hidden> Date: 2001-10-25 17:30:56
Well now I feel stupid... This should be fixed now.
Well, thanks to kobayashi (I think?) for pointing it out. I will
apply the fix locally and see if it now allows me to mount proc at
boot.
Stefan Jeglinski
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/
From: Erik van Roode <hidden> Date: 2001-10-26 05:44:08
At 10:09 -0700 10/25/01, Tom Rini wrote:
On Thu, Oct 25, 2001 at 11:53:27AM -0500, Edward Swarthout wrote:
quoted
Tom wrote:
>
> Um. Is that _all_ that changed? That really shouldn't have any effect
> at all... What compiler/binutils combo is this?
Well it's important for the COPY_16_BYTES_EXCODE macro to switch back to
the .text section or any following code will go in the __ex_table section!
Well now I feel stupid... This should be fixed now.
I updated my copy of the bk:linuxppc_2_4 tree to
2.4.14-pre1 and it boots properly again on my G4.
Erik
--
** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/