| View previous topic :: View next topic |
| Author |
Message |
idl
Retired Dev
Joined: 24 Dec 2002
Posts: 1728
Location: Nottingham, UK
|
 Posted: Sat Jul 05, 2003 11:12 pm Post subject: |
 |
|
I get the pmd_offset error aswell 
_________________
a.k.a port001
Found a bug? Please report it: Gentoo Bugzilla |
|
| Back to top |
|
 |
idl
Retired Dev
Joined: 24 Dec 2002
Posts: 1728
Location: Nottingham, UK
|
| Posted: Sat Jul 05, 2003 11:39 pm Post subject: |
|
|
ok, I got rid of the pmd_offset error and the module loads fine  . But X doesn't start. It seems the code I commented out was critical 
_________________
a.k.a port001
Found a bug? Please report it: Gentoo Bugzilla |
|
| Back to top |
|
|
krazo
Tux's lil' helper
Joined: 19 Oct 2002
Posts: 90
|
| Posted: Sun Jul 06, 2003 12:40 am Post subject: D state processes |
|
|
Is anyone encountering processes that just freeze up when there is heavy IO? It seems I can reproduce it when I emerge sync, untar a large file (ie the kernel) or even compile.. I won't be able to kill it either..
Thanks.
_________________
Hey--hey how are ya? |
|
| Back to top |
|
|
Safrax
Guru
Joined: 23 Apr 2002
Posts: 422
|
| Posted: Sun Jul 06, 2003 2:25 am Post subject: Re: D state processes |
|
|
| krazo wrote: | Is anyone encountering processes that just freeze up when there is heavy IO? It seems I can reproduce it when I emerge sync, untar a large file (ie the kernel) or even compile.. I won't be able to kill it either..
Thanks. |
Yeah. I call this the disappearing program bug (even though the program really doesnt disappear...)
The problem is not present in 2.5.72-mm3.
When I went poking around in /proc looking at the (pid)/status file, I found that those processes were listed as asleep even though they shouldn't have been.
Anyone got a fix? |
|
| Back to top |
|
|
Freak_NL
Apprentice
Joined: 27 Jun 2002
Posts: 261
Location: The Netherlands
|
| Posted: Sun Jul 06, 2003 9:35 am Post subject: |
|
|
Yup, here too..
I used to have a XFS root filesystem, but that caused kernel panics in the XFS page buffer when some heavy IO was going on. With ReiserFS I only noticed problems with compiling and unpacking OpenOffice, but hey, at least I now have a Ximian OpenOffice 1.0.3 
Out of curiousity, what filesystem do you all use on your root partition? I've had XFS panics on two computers now, so I'm starting to doubt my kernel configuration skills a bit.  |
|
| Back to top |
|
|
krazo
Tux's lil' helper
Joined: 19 Oct 2002
Posts: 90
|
| Posted: Sun Jul 06, 2003 11:32 am Post subject: |
|
|
| Freak_NL wrote: | Yup, here too..
I used to have a XFS root filesystem, but that caused kernel panics in the XFS page buffer when some heavy IO was going on. With ReiserFS I only noticed problems with compiling and unpacking OpenOffice, but hey, at least I now have a Ximian OpenOffice 1.0.3
Out of curiousity, what filesystem do you all use on your root partition? I've had XFS panics on two computers now, so I'm starting to doubt my kernel configuration skills a bit. |
I use ReiserFS. Does this problem show up when you emerge sync too? In fact, it happens to Galeon as well now that I think about it... :\ It began in 2.5.73-mm2 I believe.
_________________
Hey--hey how are ya? |
|
| Back to top |
|
|
handsomepete
Guru
Joined: 21 Apr 2002
Posts: 548
Location: Kansas City, MO
|
| Posted: Sun Jul 06, 2003 12:23 pm Post subject: |
|
|
| krazo wrote: | | Freak_NL wrote: | Yup, here too..
I used to have a XFS root filesystem, but that caused kernel panics in the XFS page buffer when some heavy IO was going on. With ReiserFS I only noticed problems with compiling and unpacking OpenOffice, but hey, at least I now have a Ximian OpenOffice 1.0.3
Out of curiousity, what filesystem do you all use on your root partition? I've had XFS panics on two computers now, so I'm starting to doubt my kernel configuration skills a bit. |
I use ReiserFS. Does this problem show up when you emerge sync too? In fact, it happens to Galeon as well now that I think about it... :\ It began in 2.5.73-mm2 I believe. |
Ditto. I've got the same problem, especially if I emerge sync and do anything else (usually web browsing) at the same time - very reproducible. Glad to hear I'm not alone. Has anyone passed this along yet?
port001: Ever find a fix for the nvidia problem? |
|
| Back to top |
|
|
Peaceable Frood
Guru
Joined: 06 Apr 2003
Posts: 338
Location: New Jersey, USA
|
| Posted: Sun Jul 06, 2003 12:52 pm Post subject: |
|
|
My Broadcom 4401 isn't working, I heard that the the driver in the kernel is broken and has been in the 2.5 series for quite sometime. How do I fix this problem, as its my only one.
_________________
"Our enemies are innovative and resourceful, and so are we, they never stop thinking about new ways to harm our country and our people, and neither do we." George W. Bush |
|
| Back to top |
|
|
Freak_NL
Apprentice
Joined: 27 Jun 2002
Posts: 261
Location: The Netherlands
|
| Posted: Sun Jul 06, 2003 2:40 pm Post subject: |
|
|
I haven't passed it along yet, since I was having some problems with a dying HD that incidentally included my root partition, so I wasn't sure about what to blame. Now that my system is transferred to a fresh HD I was thinking using a spare partition to recreate the XFS setup and go for a full debug kernel and pass the results upstream.
I'm just a bit daunted by the task of having to manually typing the resulting kernel panic into a textfile..
With ReiserFS the only problem I encountered was with a loooooong OpenOffice emerge. Have you ever tried to chop up an ebuild so you could resume with the already created objects? It's scary...
It's just that ReiserFS doesn't go for the kernel panic, the problem seems to occur higher up and just hangs the cp or cc or tar or ld or whatever is doing heavy IO at the time. They then proceed to act like undead and refuse to die or live, they don't show up as zombies though.. |
|
| Back to top |
|
|
Safrax
Guru
Joined: 23 Apr 2002
Posts: 422
|
| Posted: Sun Jul 06, 2003 3:15 pm Post subject: |
|
|
| Freak_NL wrote: | | Out of curiousity, what filesystem do you all use on your root partition? I've had XFS panics on two computers now, so I'm starting to doubt my kernel configuration skills a bit. |
ReiserFS. |
|
| Back to top |
|
|
robmoss
Retired Dev
Joined: 27 May 2003
Posts: 2634
Location: Jesus College, Oxford
|
| Posted: Sun Jul 06, 2003 3:26 pm Post subject: |
|
|
Hmm, this is strange - I quite often have very heavy I/O tasks in progress, using ReiserFS, and am yet to have any sort of crash. Weird...
I'm currently compiling 2.5.74-mm2, and will follow that with a reboot and an attempt to compile the nvidia-kernel module. Wish me luck! If I get it working, I'll report back. |
|
| Back to top |
|
|
handsomepete
Guru
Joined: 21 Apr 2002
Posts: 548
Location: Kansas City, MO
|
| Posted: Sun Jul 06, 2003 3:51 pm Post subject: |
|
|
| robmoss2k wrote: | | Hmm, this is strange - I quite often have very heavy I/O tasks in progress, using ReiserFS, and am yet to have any sort of crash. Weird... |
If you're feeling adventurous, here's my steps to reproduce. Note this worked every time in 2.5.73-mm1, but less frequently in 2.5.74-mm1 (actually stalls but can sometimes recover).
1.) Open Mozilla. Open multiple tabs within mozilla
2.) emerge sync
3.) Towards the middle/end of the sync process, start loading new urls in multiple mozilla tabs (adding new tabs if you feel like it)
Result: Emerge will stall on updating the cache and the mozilla window will stop refreshing/responding. It doesn't exactly cause a crash, but the system stays unstable from that point forward. I think untarring a large file also does the trick. |
|
| Back to top |
|
|
thubble
Tux's lil' helper
Joined: 16 May 2003
Posts: 119
Location: Winnipeg, Canada
|
| Posted: Sun Jul 06, 2003 5:16 pm Post subject: Re: nvidia errors |
|
|
I believe I've found the cause of the pmd_offset error with nvidia drivers. It's caused by the new highpmd.patch introduced in 2.5.74-mm2. To remove it, apply this patch:
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.74/2.5.74-mm2/broken-out/highpmd.patch
and recompile your kernel. This patch probably won't do anything for you anyway unless you use highmem (>4GB physical memory).
I'm temporarily using the nv driver while compiling Ximian Desktop 2, but I'll try to test this as soon as possible. Please let me know if it works. Thanks. |
|
| Back to top |
|
|
robmoss
Retired Dev
Joined: 27 May 2003
Posts: 2634
Location: Jesus College, Oxford
|
| Posted: Sun Jul 06, 2003 5:47 pm Post subject: |
|
|
| handsomepete wrote: | | If you're feeling adventurous |
I was. I tried it; it still didn't crash! Maybe it's a kernel problem relating to specific hardware? Or maybe I'm just incredibly lucky?! |
|
| Back to top |
|
|
bssteph
l33t
Joined: 26 Feb 2003
Posts: 652
Location: Wisconsin
|
| Posted: Sun Jul 06, 2003 6:04 pm Post subject: Re: nvidia errors |
|
|
| thubble wrote: | I believe I've found the cause of the pmd_offset error with nvidia drivers. It's caused by the new highpmd.patch introduced in 2.5.74-mm2. To remove it, apply this patch:
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.74/2.5.74-mm2/broken-out/highpmd.patch
and recompile your kernel. This patch probably won't do anything for you anyway unless you use highmem (>4GB physical memory).
I'm temporarily using the nv driver while compiling Ximian Desktop 2, but I'll try to test this as soon as possible. Please let me know if it works. Thanks. |
Perfect.
Unapplied the patch and nvidia modprobes without a hitch. Wonder if the patch will find its way into the kernel permanently...
Anyway, great, and thanks for the fix! :) |
|
| Back to top |
|
|
Freak_NL
Apprentice
Joined: 27 Jun 2002
Posts: 261
Location: The Netherlands
|
| Posted: Sun Jul 06, 2003 6:12 pm Post subject: |
|
|
Ya know.. Come to think of it, both systems are quite different (my younger brother's Athlon comp. and my desktop Celly system) but the common factor is the Via Apollo chipset.
handsomepete: you wouldn't happen to have a Via Apollo chipset?
I'd try this on my Intel chipset laptop, but 2.5 makes my network functionality go away (8139too, weird..), which is quite inconvenient. |
|
| Back to top |
|
|
handsomepete
Guru
Joined: 21 Apr 2002
Posts: 548
Location: Kansas City, MO
|
| Posted: Sun Jul 06, 2003 6:36 pm Post subject: |
|
|
| robmoss2k wrote: | | handsomepete wrote: | | If you're feeling adventurous |
I was. I tried it; it still didn't crash! Maybe it's a kernel problem relating to specific hardware? Or maybe I'm just incredibly lucky?! |
I have a feeling the people with problems are in the minority. I'm on a KT400 chipset. There appears to be a pair of possibly related reports on bugzilla, but with substantially different hardware. *shrug*
Thanks for the nvidia fix, thubble. |
|
| Back to top |
|
|
idl
Retired Dev
Joined: 24 Dec 2002
Posts: 1728
Location: Nottingham, UK
|
| Posted: Sun Jul 06, 2003 7:12 pm Post subject: |
|
|
| handsomepete wrote: | | robmoss2k wrote: | | handsomepete wrote: | | If you're feeling adventurous |
I was. I tried it; it still didn't crash! Maybe it's a kernel problem relating to specific hardware? Or maybe I'm just incredibly lucky?! |
I have a feeling the people with problems are in the minority. I'm on a KT400 chipset. There appears to be a pair of possibly related reports on bugzilla, but with substantially different hardware. *shrug*
Thanks for the nvidia fix, thubble. |
I've never had a problem with ReiserFS I just tried those steps to reproduce the problem but it didnt happen. I had xmms playing and the kernel compiling aswell, using 2.5.74-mm1.
_________________
a.k.a port001
Found a bug? Please report it: Gentoo Bugzilla |
|
| Back to top |
|
|
idl
Retired Dev
Joined: 24 Dec 2002
Posts: 1728
Location: Nottingham, UK
|
| Posted: Sun Jul 06, 2003 7:18 pm Post subject: Re: nvidia errors |
|
|
| bssteph wrote: | | thubble wrote: | I believe I've found the cause of the pmd_offset error with nvidia drivers. It's caused by the new highpmd.patch introduced in 2.5.74-mm2. To remove it, apply this patch:
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.5/2.5.74/2.5.74-mm2/broken-out/highpmd.patch
and recompile your kernel. This patch probably won't do anything for you anyway unless you use highmem (>4GB physical memory).
I'm temporarily using the nv driver while compiling Ximian Desktop 2, but I'll try to test this as soon as possible. Please let me know if it works. Thanks. |
Perfect.
Unapplied the patch and nvidia modprobes without a hitch. Wonder if the patch will find its way into the kernel permanently...
Anyway, great, and thanks for the fix!  |
Hmm, kernel wont compile for me with the patch removed | Quote: | arch/i386/kernel/built-in.o(.text+0x4c07): In function `mark_screen_rdonly':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0x2d7): In function `set_pte_pfn':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0x380): In function `set_pmd_pfn':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0xa91): In function `do_page_fault':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0xa9f): In function `do_page_fault':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0xba6): more undefined references to `pmd_offset' follow
|
_________________
a.k.a port001
Found a bug? Please report it: Gentoo Bugzilla |
|
| Back to top |
|
|
handsomepete
Guru
Joined: 21 Apr 2002
Posts: 548
Location: Kansas City, MO
|
| Posted: Sun Jul 06, 2003 7:32 pm Post subject: |
|
|
| port001 wrote: | | I've never had a problem with ReiserFS I just tried those steps to reproduce the problem but it didnt happen. I had xmms playing and the kernel compiling aswell, using 2.5.74-mm1. |
I doubt it's a Reiser problem. I think it might be a scheduler thing, but like I said, it seems much improved with 2.5.74-mm1, so it may not be much of an issue anymore (has only cropped up a couple times since I booted .74 as opposed to 6+ times per day with .73). I'm going to give .74-mm2 a shot and see if I can recreate it later.
| Quote: | | Hmm, kernel wont compile for me with the patch removed |
Aw, nuts. |
|
| Back to top |
|
|
Kummer
n00b
Joined: 20 Nov 2002
Posts: 41
Location: Vancouver, Canada
|
| Posted: Sun Jul 06, 2003 8:03 pm Post subject: IO trouble. |
|
|
I have that IO hangup a lot in 5.73, mostly with mozilla and emerge in parallel.
It also gives me a garbled X-display when changing between virtual terminals.
Its a P4, with SIS645 chipset, nvidia-driver (with nvidia-gart and ACPI-IO on) , and reiserfs.
Couldn't try 74 yet, b/c of that module-problem. |
|
| Back to top |
|
|
bssteph
l33t
Joined: 26 Feb 2003
Posts: 652
Location: Wisconsin
|
| Posted: Sun Jul 06, 2003 8:38 pm Post subject: Re: nvidia errors |
|
|
| port001 wrote: |
Hmm, kernel wont compile for me with the patch removed :? | Quote: | arch/i386/kernel/built-in.o(.text+0x4c07): In function `mark_screen_rdonly':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0x2d7): In function `set_pte_pfn':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0x380): In function `set_pmd_pfn':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0xa91): In function `do_page_fault':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0xa9f): In function `do_page_fault':
: undefined reference to `pmd_offset'
arch/i386/mm/built-in.o(.text+0xba6): more undefined references to `pmd_offset' follow
|
|
:(
I dunno what would cause that, and unfortunately can't check right now. Maybe some other patch is interfering? I patched a mm2 kernel that I believe was priorly patched with granularity, but I don't see how that'd affect anything.
Maybe the patching went haywire? I did 'patch -p1 < highpmd.patch', which detected that the patch was previously applied, and forced me to manually answer "y" to all the questions it asked about undoing or something (I have very little experience with patching by hand :\ ). I could check more later when I get home, but that won't be until 12 hr from now. |
|
| Back to top |
|
|
krazo
Tux's lil' helper
Joined: 19 Oct 2002
Posts: 90
|
| Posted: Sun Jul 06, 2003 8:46 pm Post subject: |
|
|
| Freak_NL wrote: | Ya know.. Come to think of it, both systems are quite different (my younger brother's Athlon comp. and my desktop Celly system) but the common factor is the Via Apollo chipset.
handsomepete: you wouldn't happen to have a Via Apollo chipset?
I'd try this on my Intel chipset laptop, but 2.5 makes my network functionality go away (8139too, weird..), which is quite inconvenient. |
I'm on an Inspiron 8200, P4 with Intel chipset and it happens to me.. so it can't only be Via...
EDIT: I have ACPI enabled as well...
_________________
Hey--hey how are ya?
Last edited by krazo on Sun Jul 06, 2003 8:56 pm; edited 1 time in total |
|
| Back to top |
|
|
mark
Tux's lil' helper
Joined: 04 Jun 2002
Posts: 119
|
| Posted: Sun Jul 06, 2003 8:50 pm Post subject: processes freezing |
|
|
I was having my kde desktop lockup, usually when compiling and browsing the web under 2.5.73. Another forum member suggested I recompile the kernel without ACPI this seemed to resolve the problem, it may work for others.
Having said that I'm now running 2.5.74-mm1 with ACPI and so far no lockup. However, its early days .
I have an Abit vp6 smp board which I believe Appolo pro based.
Mark
_________________
Regards
Mark |
|
| Back to top |
|
|
Safrax
Guru
Joined: 23 Apr 2002
Posts: 422
|
| Posted: Sun Jul 06, 2003 9:30 pm Post subject: |
|
|
I removed the highpmd patch from 2.5.74-mm2 and I have yet to experience any process screw ups even when under extremely high-io conditions while compiling.
Edit: After an hour or so of testing the heavy-io bug appeared. So I got curious.. I tried removing the o1 and o2 patches, recompiled, rebooted, system felt faster over all, but that still didn't solve the problem. Arrrgghh this is so annoying. Processes seem to freeze whenever the system runs of out physical ram and begin to go to swap. |
|
| Back to top |
|
|
|
Kernel & Hardware
