Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
On Wed, 28 Mar 2001, Paul Cassella wrote:
> Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
I've been running -ac27 for over 5 days, and it's been fine, so this seems
to have been fixed.
--
Paul Cassella
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
On Wed, 28 Mar 2001, Paul Cassella wrote:
Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
I've been running -ac27 for over 5 days, and it's been fine, so this seems
to have been fixed.
--
Paul Cassella
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
Hello Paul Cassella,
Once you wrote about "Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.":
PC> [1.] One line summary of the problem:
PC> Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
I have similar problem with 2.4.0, 2.4.1, 2.4.2. I tried running -ac24,25,26
and 2.4.3-pre6 and I don't have any problems so far.
PC> [2.] Full description of the problem/report:
PC> I have had hangs under 2.4.2-ac18, -ac19, and -ac24, after uptimes of
PC> 36 hours, 12 hours, and 10 hours, respectively. -ac12 has twice run
PC> for a week without crashing. I didn't see anything in the later -ac
PC> changelogs that looks responsible, but I haven't actually tried them.
My uptimes were bigger, but each of them was 16 days + X hours (X being 0-20)
PC> All the crashes were under X. The machine did not respond to pings,
PC> and no sysrq keys other than B worked; I didn't hear disk activity
PC> after S, and the disks weren't unmounted. Nothing made it to the
PC> logs. In the -ac19 crash, I had run at the console for about 12
PC> hours, and then started X; it crashed within 15 minutes.
I also have all these troubles under X.
PC> In the one crash that happened while I was at the console, X
PC> completely froze, and sound output stopped. In the others, the
PC> monitor was in power-save mode and didn't wake up.
I had it twice.
PC> The hangs don't appear to be related to IO load or anything else I can
PC> think of besides X. Each time, there was a distributed.net client
PC> running, and nothing else that was in any way intensive. I don't
PC> believe any sort of updatedb or makewhatis was running during the
PC> crashes, and it never hung overnight when these jobs run.
No distributed.net client here ;)
PC> I ran with -ac12 with nearly 1300 lines of diff narrowed down from
PC> [...skip...]
PC> - i810, (Debian unstable) X 4.0.2, with DRI
I think that the problem might be somewhere he. I am running i810,
(RedHat 7...not original anymore :)) X 4.0.1.
PC> I'll be happy to try out patches, configuration changes, and other
PC> suggestions, but I won't be able to tell for three or four days
PC> whether or not it helped.
With regular uptime of 16 days I will be very slow responsive for the testing
phase, though I am willing to try too ;)
--
Best regards,
Leonid Mamtchenkov
System Administrator
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
On Thu, 29 Mar 2001, Alan Cox wrote: > Was anything between 12 and 18 stable ? I didn't actually try them; I jumped right from 12 to 18, and when that and 19 died, I went back to 12. But a quick look suggests that the entire patch I'd applied to 12 and got a hang with was in 13, including the pm.c change. I also haven't tried anything after 24; is it likely to have been fixed? -- Paul Cassella - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
> I have had hangs under 2.4.2-ac18, -ac19, and -ac24, after uptimes of > 36 hours, 12 hours, and 10 hours, respectively. -ac12 has twice run > for a week without crashing. I didn't see anything in the later -ac > changelogs that looks responsible, but I haven't actually tried them. Was anything between 12 and 18 stable ? > A few lines earlier in this function, inode->i_op->truncate() is called > without lock_kernel(). Should it also have a lock_kernel(), or is it not > needed there? Absolutely correct. The lock is missing. Bizarrely Al Viro just noticed this about 15 minutes ago - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
Earlier today, I wrote > and no sysrq keys other than B worked; I didn't hear disk activity > after S, and the disks weren't unmounted. Nothing made it to the Of course, when I rebooted this time (after SysRQ S,U,B), all the filesystems were clean. Nothing in the logs this time either though. > When I get home and reboot (following this most recent hang :( ), I'll > put the diff, .config, and more stuff from /proc at > http://manetheren.eigenray.com/~fortytwo/crash-12-18.2 This is now there. -- Paul Cassella - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
[1.] One line summary of the problem:
Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
[2.] Full description of the problem/report:
I have had hangs under 2.4.2-ac18, -ac19, and -ac24, after uptimes of
36 hours, 12 hours, and 10 hours, respectively. -ac12 has twice run
for a week without crashing. I didn't see anything in the later -ac
changelogs that looks responsible, but I haven't actually tried them.
All the crashes were under X. The machine did not respond to pings,
and no sysrq keys other than B worked; I didn't hear disk activity
after S, and the disks weren't unmounted. Nothing made it to the
logs. In the -ac19 crash, I had run at the console for about 12
hours, and then started X; it crashed within 15 minutes.
In the one crash that happened while I was at the console, X
completely froze, and sound output stopped. In the others, the
monitor was in power-save mode and didn't wake up.
The hangs don't appear to be related to IO load or anything else I can
think of besides X. Each time, there was a distributed.net client
running, and nothing else that was in any way intensive. I don't
believe any sort of updatedb or makewhatis was running during the
crashes, and it never hung overnight when these jobs run.
I ran with -ac12 with nearly 1300 lines of diff narrowed down from
"interdiff -h ac12 ac18" for about 36 hours in console mode; it hung
within 3 hours of starting X.
When I get home and reboot (following this most recent hang :( ), I'll
put the diff, .config, and more stuff from /proc at
http://manetheren.eigenray.com/~fortytwo/crash-12-18.2
This should be sometime around 8PM CST. (If someone wants the diff
now, email me. I have it here, but I don't want to spam the list with
it.)
This diff wasn't "complete"; some modules (ide-cd, at least) weren't
able to load due to missing symbols.
The diff included all the changes referencing bust_spinlocks(), and
everything to do with the console_sem and the console tasklet/tq. This
included all the changes to printk.c.
It also included the following. In -ac18, this is a BUG(), not a
printk(), but I wanted something I could see while X was running. The
message never showed up. I didn't look to see what the effect of
returning -1 here is, though.
diff -u linux.ac/kernel/pm.c linux.ac/kernel/pm.c
--- linux.ac/kernel/pm.c
+++ linux.ac/kernel/pm.c
@@ -150,6 +154,10 @@
{
int status = 0;
int prev_state, next_state;
+
+ if (in_interrupt())
+ {printk("pm_send called from interrupt (0x%p)!\n",
+__builtin_return_address(0)); return -1; }
+
switch (rqst) {
case PM_SUSPEND:
case PM_RESUME:
AFAICT there was nothing else in the diff.
[7.1.] Software (add the output of the ver_linux script here)
Linux manetheren 2.4.2-ac12 #8 Mon Mar 5 20:02:30 CST 2001 i686 unknown
Gnu C 2.95.2
Gnu make 3.79.1
binutils 2.11.90.0.1
util-linux 2.11a
modutils 2.4.2
e2fsprogs 1.19
Linux C Library2.2.2
Dynamic linker (ldd) 2.2.2
Procps 2.0.7
Net-tools 1.59
Console-tools 0.2.3
Sh-utils 2.0.11
Modules Loaded usb-uhci parport_pc lp parport binfmt_misc rtc usbcore
Since I didn't think to copy my .config off the machine, I won't be
able to get to it until tonight. In the meantime, I do remember that
- It's a UP kernel on a UP box
- Celeron kernel and processor
- The hang happens with USB completely disabled
(Though I don't think I ever turned off hotplugging.)
- VTs, console on VT, and console on serial configured
(console was not on serial)
- i810, (Debian unstable) X 4.0.2, with DRI
- PIIX tuning enabled
- Auto-DMA
- No kernel debugging other than SysRq
- No SCSI
- APM was off; don't remember the other pm stuff.
- ecn was on, syncookies off.
- no ip masquerading or firewalling or anything fancy.
- 128M RAM; no HIGHMEM stuff.
I'll be happy to try out patches, configuration changes, and other
suggestions, but I won't be able to tell for three or four days
whether or not it helped.
[7.2.] Processor information (from /proc/cpuinfo):
Single processor,
cpu family : 6
model : 6
model name : Celeron (Mendocino) (466Mhz/66Mhz FSB)
stepping: 5
cpu MHz : 465.265
cache size : 128 KB
[7.3.] Module information (from /proc/modules):
The modules loaded at the -ac24 crash appear to have been
visor 8400 1
usbserial 17488 1 [visor]
parport_pc 18480 1 (autoclean)
lp 6096 1 (autoclean)
parport24704 1 (autoclean) [parport_pc lp]
uhci 21920 0 (unused)
binfmt_misc 5600 0
rtc 5056 0 (autoclean)
usbcore50480 1 (autoclean) [visor usbserial uhci]
from Debian's /var/log/ksymoops
[7.4.] Loaded driver a
Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
[1.] One line summary of the problem:
Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
[2.] Full description of the problem/report:
I have had hangs under 2.4.2-ac18, -ac19, and -ac24, after uptimes of
36 hours, 12 hours, and 10 hours, respectively. -ac12 has twice run
for a week without crashing. I didn't see anything in the later -ac
changelogs that looks responsible, but I haven't actually tried them.
All the crashes were under X. The machine did not respond to pings,
and no sysrq keys other than B worked; I didn't hear disk activity
after S, and the disks weren't unmounted. Nothing made it to the
logs. In the -ac19 crash, I had run at the console for about 12
hours, and then started X; it crashed within 15 minutes.
In the one crash that happened while I was at the console, X
completely froze, and sound output stopped. In the others, the
monitor was in power-save mode and didn't wake up.
The hangs don't appear to be related to IO load or anything else I can
think of besides X. Each time, there was a distributed.net client
running, and nothing else that was in any way intensive. I don't
believe any sort of updatedb or makewhatis was running during the
crashes, and it never hung overnight when these jobs run.
I ran with -ac12 with nearly 1300 lines of diff narrowed down from
"interdiff -h ac12 ac18" for about 36 hours in console mode; it hung
within 3 hours of starting X.
When I get home and reboot (following this most recent hang :( ), I'll
put the diff, .config, and more stuff from /proc at
http://manetheren.eigenray.com/~fortytwo/crash-12-18.2
This should be sometime around 8PM CST. (If someone wants the diff
now, email me. I have it here, but I don't want to spam the list with
it.)
This diff wasn't "complete"; some modules (ide-cd, at least) weren't
able to load due to missing symbols.
The diff included all the changes referencing bust_spinlocks(), and
everything to do with the console_sem and the console tasklet/tq. This
included all the changes to printk.c.
It also included the following. In -ac18, this is a BUG(), not a
printk(), but I wanted something I could see while X was running. The
message never showed up. I didn't look to see what the effect of
returning -1 here is, though.
diff -u linux.ac/kernel/pm.c linux.ac/kernel/pm.c
--- linux.ac/kernel/pm.c
+++ linux.ac/kernel/pm.c
@@ -150,6 +154,10 @@
{
int status = 0;
int prev_state, next_state;
+
+ if (in_interrupt())
+ {printk("pm_send called from interrupt (0x%p)!\n",
+__builtin_return_address(0)); return -1; }
+
switch (rqst) {
case PM_SUSPEND:
case PM_RESUME:
AFAICT there was nothing else in the diff.
[7.1.] Software (add the output of the ver_linux script here)
Linux manetheren 2.4.2-ac12 #8 Mon Mar 5 20:02:30 CST 2001 i686 unknown
Gnu C 2.95.2
Gnu make 3.79.1
binutils 2.11.90.0.1
util-linux 2.11a
modutils 2.4.2
e2fsprogs 1.19
Linux C Library2.2.2
Dynamic linker (ldd) 2.2.2
Procps 2.0.7
Net-tools 1.59
Console-tools 0.2.3
Sh-utils 2.0.11
Modules Loaded usb-uhci parport_pc lp parport binfmt_misc rtc usbcore
Since I didn't think to copy my .config off the machine, I won't be
able to get to it until tonight. In the meantime, I do remember that
- It's a UP kernel on a UP box
- Celeron kernel and processor
- The hang happens with USB completely disabled
(Though I don't think I ever turned off hotplugging.)
- VTs, console on VT, and console on serial configured
(console was not on serial)
- i810, (Debian unstable) X 4.0.2, with DRI
- PIIX tuning enabled
- Auto-DMA
- No kernel debugging other than SysRq
- No SCSI
- APM was off; don't remember the other pm stuff.
- ecn was on, syncookies off.
- no ip masquerading or firewalling or anything fancy.
- 128M RAM; no HIGHMEM stuff.
I'll be happy to try out patches, configuration changes, and other
suggestions, but I won't be able to tell for three or four days
whether or not it helped.
[7.2.] Processor information (from /proc/cpuinfo):
Single processor,
cpu family : 6
model : 6
model name : Celeron (Mendocino) (466Mhz/66Mhz FSB)
stepping: 5
cpu MHz : 465.265
cache size : 128 KB
[7.3.] Module information (from /proc/modules):
The modules loaded at the -ac24 crash appear to have been
visor 8400 1
usbserial 17488 1 [visor]
parport_pc 18480 1 (autoclean)
lp 6096 1 (autoclean)
parport24704 1 (autoclean) [parport_pc lp]
uhci 21920 0 (unused)
binfmt_misc 5600 0
rtc 5056 0 (autoclean)
usbcore50480 1 (autoclean) [visor usbserial uhci]
from Debian's /var/log/ksymoops
[7.4.] Loaded driver a
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
Earlier today, I wrote and no sysrq keys other than B worked; I didn't hear disk activity after S, and the disks weren't unmounted. Nothing made it to the Of course, when I rebooted this time (after SysRQ S,U,B), all the filesystems were clean. Nothing in the logs this time either though. When I get home and reboot (following this most recent hang :( ), I'll put the diff, .config, and more stuff from /proc at http://manetheren.eigenray.com/~fortytwo/crash-12-18.2 This is now there. -- Paul Cassella - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
I have had hangs under 2.4.2-ac18, -ac19, and -ac24, after uptimes of 36 hours, 12 hours, and 10 hours, respectively. -ac12 has twice run for a week without crashing. I didn't see anything in the later -ac changelogs that looks responsible, but I haven't actually tried them. Was anything between 12 and 18 stable ? A few lines earlier in this function, inode-i_op-truncate() is called without lock_kernel(). Should it also have a lock_kernel(), or is it not needed there? Absolutely correct. The lock is missing. Bizarrely Al Viro just noticed this about 15 minutes ago - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
On Thu, 29 Mar 2001, Alan Cox wrote: Was anything between 12 and 18 stable ? I didn't actually try them; I jumped right from 12 to 18, and when that and 19 died, I went back to 12. But a quick look suggests that the entire patch I'd applied to 12 and got a hang with was in 13, including the pm.c change. I also haven't tried anything after 24; is it likely to have been fixed? -- Paul Cassella - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
Hello Paul Cassella,
Once you wrote about "Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.":
PC [1.] One line summary of the problem:
PC Hangs under 2.4.2-ac{18,19,24} that do not happen under -ac12.
I have similar problem with 2.4.0, 2.4.1, 2.4.2. I tried running -ac24,25,26
and 2.4.3-pre6 and I don't have any problems so far.
PC [2.] Full description of the problem/report:
PC I have had hangs under 2.4.2-ac18, -ac19, and -ac24, after uptimes of
PC 36 hours, 12 hours, and 10 hours, respectively. -ac12 has twice run
PC for a week without crashing. I didn't see anything in the later -ac
PC changelogs that looks responsible, but I haven't actually tried them.
My uptimes were bigger, but each of them was 16 days + X hours (X being 0-20)
PC All the crashes were under X. The machine did not respond to pings,
PC and no sysrq keys other than B worked; I didn't hear disk activity
PC after S, and the disks weren't unmounted. Nothing made it to the
PC logs. In the -ac19 crash, I had run at the console for about 12
PC hours, and then started X; it crashed within 15 minutes.
I also have all these troubles under X.
PC In the one crash that happened while I was at the console, X
PC completely froze, and sound output stopped. In the others, the
PC monitor was in power-save mode and didn't wake up.
I had it twice.
PC The hangs don't appear to be related to IO load or anything else I can
PC think of besides X. Each time, there was a distributed.net client
PC running, and nothing else that was in any way intensive. I don't
PC believe any sort of updatedb or makewhatis was running during the
PC crashes, and it never hung overnight when these jobs run.
No distributed.net client here ;)
PC I ran with -ac12 with nearly 1300 lines of diff narrowed down from
PC [...skip...]
PC - i810, (Debian unstable) X 4.0.2, with DRI
I think that the problem might be somewhere he. I am running i810,
(RedHat 7...not original anymore :)) X 4.0.1.
PC I'll be happy to try out patches, configuration changes, and other
PC suggestions, but I won't be able to tell for three or four days
PC whether or not it helped.
With regular uptime of 16 days I will be very slow responsive for the testing
phase, though I am willing to try too ;)
--
Best regards,
Leonid Mamtchenkov
System Administrator
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
