Menu
▾
▴
colinux-devel
|
From: IWAMURO M. <vm...@ni...> - 2004-03-14 03:30:57
|
Hello, I found that SMP system freeze completely at boot time of coLinux with TAP-Win32. (If disable TAP-Win32, then not freeze) http://sourceforge.net/tracker/index.php?func=detail&aid=915488&group_id=98788&atid=622063 |
|
From: Florent C. <fc...@wa...> - 2004-03-30 19:31:11
|
Hello, I hope my feedback will help. I'm using colinux 0.6.0 with the provided debian image. I use the default kernel provided with colinux. My config is Pentium 3.0C HT on windows xp pro SP1. I encounter the same random "crash" at boot time (and sometime on linux shutdown). colinux daemon does not answer anymore, and a few seconds later, windows freezes (In fact not exactly, I can move the mouse but can do no action, the keyboard does not answer anymore (num lock key has no action, the led remains on), and if I keep pressing keys, my motherboard starts beeping on each key press. When the problem occurs, the colinux daemon says: (I can't remember but it is network related). and the linux console says : syslog starting. Another precision: even when colinux doesn't crash, this step takes a lot of time to execute. The crash seems to be network related, as I had no crash before configuring the network. But I'm sure this is not a TAP-Win32 problem because I simply do not use it. I even desinstalled the tap driver. I use a winpcap bridge instead. I cannot send any log or crash dump because when the problem occurs, windows crash immediately. Last, when colinux boot succeeds, I can use colinux during hours with no problem. Could you please make a release that binds the the 1st processor until the SMP problem is resolved ? Because the multiple crashs make me loose some data, and I will not use colinux anymore until this problem is solved. Florent |
|
From: Alejandro R. S. <as...@MI...> - 2004-03-30 20:51:35
|
I had the same problem on my work machine (P4, HT). Until a more permanent fix is made, what you can do is use imagecfg (google: imagecfg) to set the colinux binaries to be uniprocessor. in your colinux dir: imagecfg -u *.exe -Alejandro On Tue, 2004-03-30 at 14:31, Florent CUETO wrote: > > > Hello, > > > > I hope my feedback will help. I’m using colinux 0.6.0 with the > provided debian image. > > I use the default kernel provided with colinux. > > My config is Pentium 3.0C HT on windows xp pro SP1. > > > > I encounter the same random “crash” at boot time (and sometime on > linux shutdown). > > colinux daemon does not answer anymore, and a few seconds later, > windows freezes (In fact not exactly, I can move the mouse but can do > no action, the keyboard does not answer anymore (num lock key has no > action, the led remains on), and if I keep pressing keys, my > motherboard starts beeping on each key press. > > > > When the problem occurs, the colinux daemon says: (I can’t remember > but it is network related). > > and the linux console says : syslog starting. > > > > Another precision: even when colinux doesn’t crash, this step takes a > lot of time to execute. > > > > The crash seems to be network related, as I had no crash before > configuring the network. > > But I’m sure this is not a TAP-Win32 problem because I simply do not > use it. I even desinstalled the tap driver. I use a winpcap bridge > instead. > > > > I cannot send any log or crash dump because when the problem occurs, > windows crash immediately… > > > > Last, when colinux boot succeeds, I can use colinux during hours with > no problem… > > Could you please make a release that binds the the 1st processor until > the SMP problem is resolved ? Because the multiple crashs make me > loose some data, and I will not use colinux anymore until this problem > is solved… > > > > Florent > > > > |
|
From: Dan A. <da...@co...> - 2004-04-02 14:11:06
|
On Tue, Mar 30, 2004 at 03:51:29PM -0500, Alejandro R. Sedeno wrote: > I had the same problem on my work machine (P4, HT). > > Until a more permanent fix is made, what you can do is use imagecfg > (google: imagecfg) to set the colinux binaries to be uniprocessor. > > in your colinux dir: > imagecfg -u *.exe Are you saying that applying this modification to the executables allows to run coLinux on SMP/HT setups? If by tying the daemon process to one CPU the sympthoms stop from occuring, it may give a clue about what is causing the problem. -- Dan Aloni da...@co... |
|
From: Alejandro R. S. <as...@MI...> - 2004-04-02 14:36:02
|
On Fri, 2004-04-02 at 09:11, Dan Aloni wrote: > On Tue, Mar 30, 2004 at 03:51:29PM -0500, Alejandro R. Sedeno wrote: > > I had the same problem on my work machine (P4, HT). > > > > Until a more permanent fix is made, what you can do is use imagecfg > > (google: imagecfg) to set the colinux binaries to be uniprocessor. > > > > in your colinux dir: > > imagecfg -u *.exe > > Are you saying that applying this modification to the executables > allows to run coLinux on SMP/HT setups? > > If by tying the daemon process to one CPU the sympthoms stop from > occuring, it may give a clue about what is causing the problem. I'm doing that now at work, it seems to work most of the time, including all night last night. I've had a couple issues when starting and stopping it multiple times, but I'd venture to say that's a different problem. Note: I've not tested to see which specific binaries should be uniproc. I think the colinux-daemon would be sufficient, but I just opted to change them all. -Alejandro |
|
From: Florent C. <flo...@un...> - 2004-04-02 14:45:33
|
I have applied the imagecfg patch on the executables, now colinux works perfectly on my P4 3.0C HT. Like Alejandro, I've patched all .exe (I did not try to apply the patch on colinux-deamon only). Florent -----Message d'origine----- De=A0: Alejandro R. Sedeno [mailto:as...@MI...]=20 Envoy=E9=A0: vendredi 2 avril 2004 16:36 =C0=A0: Dan Aloni Cc=A0: Florent CUETO; col...@li... Objet=A0: Re: [coLinux-devel] SMP system freeze completely On Fri, 2004-04-02 at 09:11, Dan Aloni wrote: > On Tue, Mar 30, 2004 at 03:51:29PM -0500, Alejandro R. Sedeno wrote: > > I had the same problem on my work machine (P4, HT). > >=20 > > Until a more permanent fix is made, what you can do is use imagecfg > > (google: imagecfg) to set the colinux binaries to be uniprocessor. > >=20 > > in your colinux dir: > > imagecfg -u *.exe >=20 > Are you saying that applying this modification to the executables > allows to run coLinux on SMP/HT setups? >=20 > If by tying the daemon process to one CPU the sympthoms stop from=20 > occuring, it may give a clue about what is causing the problem. I'm doing that now at work, it seems to work most of the time, including all night last night. I've had a couple issues when starting and stopping it multiple times, but I'd venture to say that's a different problem. Note: I've not tested to see which specific binaries should be uniproc. I think the colinux-daemon would be sufficient, but I just opted to change them all. -Alejandro |
|
From: Clemmitt M. S. <sig...@bl...> - 2004-04-12 15:55:44
|
Hi,
On Fri, 2 Apr 2004, Florent CUETO wrote:
> I have applied the imagecfg patch on the executables, now colinux works
> perfectly on my P4 3.0C HT.
> Like Alejandro, I've patched all .exe (I did not try to apply the patch
> on colinux-deamon only).
Quick summary: I don't see stability problems with coLinux once
it's running, but I do see cL hangs on startup or shutdown which
cause the Windows host OS to stop working then hang. I experimented
with configuring some of the executables for uniprocessor
operation via imagecfg -u and got some interesting results. Some
combinations boot/shutdown without hanging, others hang with
regularity.
Long version: I finally had time to experiment with coLinux-0.6.0 on my
ancient SMP server box at home. It's an Intel N440BX machine with dual
PIII 600 CPUs and 768MB of RAM. My host OS is WinXP Pro with SP1a and
all the latest critical and driver updates installed. I did this using
the Red Hat 9 image with wget, openssh-clients, and samba-clients
installed, and using 64MB of RAM and a 256MB swap device. The network
was configured for "tap" access with Internet Connection Sharing
under Windows.
I stress-tested coLinux by running two network and one number-crunching
tests. The network tests were down/uploading linux-2.6.5.tar.bz2 and
patch-2.6.5.bz2 on my outside DSL connection using wget and scp, and
put/getting the same files to/from the host OS using smbclient. The
number-crunching program was a small threaded program to calculate
the digits of pi to arbitrary precision. I ran the network and
number-crunching tests for approx. 2 1/2 hrs. each to see if I
could find any instability. Under all cases (*.exe being configured
for either UP or MP) coLinux running the Red Hat image didn't
crash or hang. I didn't try running two tests simultaneously so
maybe I'm not stressing it hard enough.
However, I still see boot/shutdown hangs when running the
original executables on the SMP box. So I set up some of the
executables to be uniprocessor via imagecfg -u. Here are
my results (I didn't test colinux-console-nt or
colinux-bridged-net-daemon):
-------------------------------------------------------------------------
Test case | c-d | c-n-d | c-c-f | Result
-------------------------------------------------------------------------
A | U | U | U | No hangs seen
-------------------------------------------------------------------------
B | | U | U | Hangs within a few boots
-------------------------------------------------------------------------
C | U | | U | No hangs seen
-------------------------------------------------------------------------
D | U | U | | Ho hangs seen
-------------------------------------------------------------------------
E | U | | | Hangs within a few boots
-------------------------------------------------------------------------
Key:
" " == Run as original multiprocessor executable
"U" == Run as uniprocessor executable after `imagecfg -u'
"c-d" == colinux-daemon.exe
"c-n-d" == colinux-net-daemon.exe
"c-c-f" == colinux-console-fltk.exe
"No hangs seen" == Booted/shutdown 6 times consecutively without
a hang seen
"Hangs within a few boots" == Hangs within 1-3 consecutive
boots/shutdowns
Once case B boots, it seems to run fine. I couldn't induce any crashes
or hangs while running my network or computational stress-testing,
but maybe I'm not testing rigorously enough?
Case E is interesting. Running either c-n-d or c-c-f as an MP
.exe seems to boot OK, but running them *both* as MP .exe's caused
boots/shutdowns to hang after a few tries.
Again, this is just my own config and testing at home, YMMV :^)
Clemmitt
|
|
From: Dan A. <da...@co...> - 2004-04-12 17:14:22
|
On Mon, Apr 12, 2004 at 11:55:36AM -0400, Clemmitt M. Sigler wrote: > Hi, > > On Fri, 2 Apr 2004, Florent CUETO wrote: > > I have applied the imagecfg patch on the executables, now colinux works > > perfectly on my P4 3.0C HT. > > Like Alejandro, I've patched all .exe (I did not try to apply the patch > > on colinux-deamon only). > > [..snips..] > > Case E is interesting. Running either c-n-d or c-c-f as an MP > .exe seems to boot OK, but running them *both* as MP .exe's caused > boots/shutdowns to hang after a few tries. Your test cases are interesting. The one that I find most bizarre is the C case. It's the only case that involves a daemon communicating with a kernel driver (this case, the TAP driver), whose executable is not marked UP *and* it doesn't incur hangs. The only diff between C and E is only the UP flag of the FLTK console, and it's weird that only that change causes hangs, as the FLTK console process doesn't communicate with any driver. Please test it without networking at all, so in case where the TAP driver is faulty on SMP, it doesn't gets in our way when we try to find the problem with coLinux. -- Dan Aloni da...@co... |
|
From: Florent C. <flo...@un...> - 2004-04-16 16:28:03
|
As I said in my previous post, the problem comes not from the TAP drivers, because I had the problem on my P4 3.0 HT and I used a bridged win pcap network (not TAP driver installed). Hope it helps. Florent -----Message d'origine----- De=A0: Dan Aloni [mailto:da...@co...]=20 Envoy=E9=A0: lundi 12 avril 2004 19:14 =C0=A0: Clemmitt M. Sigler Cc=A0: Florent CUETO; col...@li... Objet=A0: Re: RE : [coLinux-devel] SMP system freeze completely On Mon, Apr 12, 2004 at 11:55:36AM -0400, Clemmitt M. Sigler wrote: > Hi, >=20 > On Fri, 2 Apr 2004, Florent CUETO wrote: > > I have applied the imagecfg patch on the executables, now colinux works > > perfectly on my P4 3.0C HT. > > Like Alejandro, I've patched all .exe (I did not try to apply the patch > > on colinux-deamon only). >=20 > [..snips..] >=20 > Case E is interesting. Running either c-n-d or c-c-f as an MP > .exe seems to boot OK, but running them *both* as MP .exe's caused > boots/shutdowns to hang after a few tries. Your test cases are interesting. The one that I find most bizarre is the C case. It's the only case that involves a daemon communicating with a=20 kernel driver (this case, the TAP driver), whose executable is not marked UP *and* it doesn't incur hangs. The only diff between C and E is only the UP flag of the FLTK console, and it's weird that only that change causes=20 hangs, as the FLTK console process doesn't communicate with any driver. Please test it without networking at all, so in case where the TAP=20 driver is faulty on SMP, it doesn't gets in our way when we try to find=20 the problem with coLinux. --=20 Dan Aloni da...@co... |
|
From: Dan A. <da...@co...> - 2004-03-14 05:25:33
|
On Sun, Mar 14, 2004 at 12:30:46PM +0900, IWAMURO Motonori wrote: > Hello, > > I found that SMP system freeze completely at boot time of coLinux > with TAP-Win32. (If disable TAP-Win32, then not freeze) > > http://sourceforge.net/tracker/index.php?func=detail&aid=915488&group_id=98788&atid=622063 Can you please try again with the latest TAP-Win32 driver from the OpenVPN package? -- Dan Aloni da...@co... |
|
From: IWAMURO M. <vm...@ni...> - 2004-03-14 09:39:01
|
At Sun, 14 Mar 2004 07:25:23 +0200, Dan Aloni wrote: > On Sun, Mar 14, 2004 at 12:30:46PM +0900, IWAMURO Motonori wrote: > > I found that SMP system freeze completely at boot time of coLinux > > with TAP-Win32. (If disable TAP-Win32, then not freeze) <snip> > Can you please try again with the latest TAP-Win32 driver from > the OpenVPN package? By your suggestion, I tried again with TAP-Win32 from OpenVPN 1.6_rc2. But, the problem was not solved. The timing of system crash is not fixed. occasionally, crash after coLinux shutdown. |
|
From: Benoit F. <ben...@li...> - 2004-03-17 09:44:33
|
I confirm the problem. On a p4, when hyperthreading is activated, colinux=20 freezes the whole system usually on syslog launch at boot time. The crash=20 seems to occur at random... Everything is fine if hyperthreading is=20 deactivated. Le Sunday 14 March 2004 10:38, IWAMURO Motonori a =E9crit : > At Sun, 14 Mar 2004 07:25:23 +0200, > > Dan Aloni wrote: > > On Sun, Mar 14, 2004 at 12:30:46PM +0900, IWAMURO Motonori wrote: > > > I found that SMP system freeze completely at boot time of coLinux > > > with TAP-Win32. (If disable TAP-Win32, then not freeze) > > <snip> > > > Can you please try again with the latest TAP-Win32 driver from > > the OpenVPN package? > > By your suggestion, I tried again with TAP-Win32 from OpenVPN 1.6_rc2. > But, the problem was not solved. > > The timing of system crash is not fixed. > occasionally, crash after coLinux shutdown. > > > ------------------------------------------------------- > This SF.Net email is sponsored by: IBM Linux Tutorials > Free Linux tutorial presented by Daniel Robbins, President and CEO of > GenToo technologies. Learn everything from fundamentals to system > administration.http://ads.osdn.com/?ad_id=3D1470&alloc_id=3D3638&op=3Dcli= ck > _______________________________________________ > coLinux-devel mailing list > coL...@li... > https://lists.sourceforge.net/lists/listinfo/colinux-devel =2D-=20 Benoit Favre tel: +33 (0)4 90 84 35 77 mobile: +33 (0)6 63 20 02 14 mail: ben...@li... |
|
From: Clemmitt M. S. <sig...@bl...> - 2004-03-17 22:49:20
|
On Wed, 17 Mar 2004, Benoit Favre wrote: > I confirm the problem. On a p4, when hyperthreading is activated, colinux > freezes the whole system usually on syslog launch at boot time. The crash > seems to occur at random... Everything is fine if hyperthreading is > deactivated. Aha! GENIUS! *This* is what's been causing my random and irregular coLinux lock-ups. I have a P4 2.80GHz that's been running hyperthreading under WinXP Pro. I turn off hyperthreading, and all lock-ups are gone! I'll update the Help forum where I asked for help with this info, and when I get a few minutes I'll add this as a Q&A to the Wiki FAQ :^) Clemmitt Sigler |
|
From: Ronald P. <pij...@ds...> - 2004-03-18 21:40:31
|
>>I confirm the problem. On a p4, when hyperthreading is activated, colinux >>freezes the whole system usually on syslog launch at boot time. The crash >>seems to occur at random... Everything is fine if hyperthreading is >>deactivated. > > > Aha! GENIUS! *This* is what's been causing my random and irregular > coLinux lock-ups. I have a P4 2.80GHz that's been running hyperthreading > under WinXP Pro. I turn off hyperthreading, and all lock-ups are gone! > I'll update the Help forum where I asked for help with this info, and > when I get a few minutes I'll add this as a Q&A to the Wiki FAQ :^) Isn't it a better idea to bind to the first processor as long as coLinux is not SMP safe? In that case, these kind of problems are easily solved? Ronald. |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X