Kernel idles in loop after loading IPv4
Hello UML developers,
I'm trying to use UML to set up a simulator for a high-availability cluster. I encounter the following unexpected behavior -
On any kernel that I compile with IPv4 support, the kernel will stop intializing drivers after loading IPv4 support. The last message I see from the kernel is:
NET: Registered protocol family 2
Compiling with debug allows me to single-step through the kernel. I've traced execution from inet_init(), to synchronize_net(), to synchronize_kernel(), to a call to wait_for_completion(), to switch_to_tt(), after which the kernel spends all its time in the idle loop. This happens whether I'm using SKAS or TT mode.
Host kernel is 2.6.1 with Stephen William's 2.6.1 SKAS patch (thanks Stephen). Guest kernel is 2.6.1 with the latest patch from Sourceforge (January 15th), without modules, and with nearly everything disabled at compile except core functionality and IPv4.
I've tried to find similar cases in the FAQ's and the list archives, but I'm not seeing anything. It seems like the kernel is waiting for an event related to IPv4 initialization that it never gets, but I'm not familiar enough with the kernel proper to know what that might be. Does anyone have any ideas?
Thanks in advance.
Signature · 16 lines
Phil Nadeau Software Services Manager [email protected] (916)932-3200, (800)921-5513 ------------------------------------------------------- The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. http://www.eclipsecon.org/osdn _______________________________________________ User-mode-linux-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
Greetings,
I ran into that as well, but didn't have time to do much about it. As a further datapoint, if I disable all net support and pass it init=/bin/bash, I get the bash prompt (ubd filesystems do get checked/mounted) but then it goes into the idle loop and stays there.
G'day,
sjames
-------------------------steven james, director of research, linux labs
... ........ ..... .... 230 peachtree st nw ste 2701
the original linux labs atlanta.ga.us 30303
-since 1995 http://www.linuxlabs.com
office 404.577.7747 fax 404.577.7743
-----------------------------------------------------------------------
Quoted 37 lines · Phil Nadeau
Hello UML developers,
I'm trying to use UML to set up a simulator for a high-availability cluster. I encounter the following unexpected behavior -
On any kernel that I compile with IPv4 support, the kernel will stop intializing drivers after loading IPv4 support. The last message I see from the kernel is:
NET: Registered protocol family 2Compiling with debug allows me to single-step through the kernel. I've traced execution from
inet_init(), tosynchronize_net(), tosynchronize_kernel(), to a call towait_for_completion(), toswitch_to_tt(), after which the kernel spends all its time in the idle loop. This happens whether I'm using SKAS or TT mode.Host kernel is 2.6.1 with Stephen William's 2.6.1 SKAS patch (thanks Stephen). Guest kernel is 2.6.1 with the latest patch from Sourceforge (January 15th), without modules, and with nearly everything disabled at compile except core functionality and IPv4.
I've tried to find similar cases in the FAQ's and the list archives, but I'm not seeing anything. It seems like the kernel is waiting for an event related to IPv4 initialization that it never gets, but I'm not familiar enough with the kernel proper to know what that might be. Does anyone have any ideas?
Thanks in advance.
--
Phil Nadeau
Software Services Manager
[email protected]
(916)932-3200, (800)921-5513
Quoted 10 lines
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
Quoted 4 lines · [email protected] said:
I ran into that as well, but didn't have time to do much about it. As a further datapoint, if I disable all net support and pass it init=/ bin/bash, I get the bash prompt (ubd filesystems do get checked/ mounted) but then it goes into the idle loop and stays there.
Hmmm.
If you could get a stack trace from the last call to schedule, that would be handy. You could automate it with
comm <n>
bt
c
end
with breakpoint <n> attached to schedule().
Jeff
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
Quoted 10 lines · Jeff Dike
If you could get a stack trace from the last call to schedule, that would be handy. You could automate it with
comm <n>
bt
c
endwith breakpoint <n> attached to
schedule().Jeff
Here's the last part of a debug session that I ran which produced the error. The full transcript is attached. Thanks in advance.
NET: Registered protocol family 2
Breakpoint 1, schedule () at kernel/sched.c:1489
1489 if (likely(!(current->state & (TASK_DEAD | TASK_ZOMBIE)))) {
#0 schedule () at kernel/sched.c:1489 #1 0xa002db4e in wait_for_completion (x=0xa1fefb64) at kernel/sched.c:1758 #2 0xa0041104 in synchronize_kernel () at kernel/rcupdate.c:262 #3 0xa000b5dc in inet_init () at net/ipv4/af_inet.c:1144 #4 0xa00024e8 in do_initcalls () at init/main.c:499 #5 0xa000c679 in init (unused=0x0) at init/main.c:581 #6 0xa00141d3 in run_kernel_thread (fn=0xa000c658 <init>, arg=0x0,
jmp_ptr=0xa08bfab4) at arch/um/kernel/process.c:241 #7 0xa001fe23 in new_thread_handler (sig=10) at arch/um/kernel/skas/process_kern.c:71
#8 <signal handler called>
#9 0xa0155fed in syscall () at seq_file.h:49 Previous frame inner to this frame (corrupt stack?) (gdb) c
Continuing.
Signature · 4 lines
Phil Nadeau Software Services Manager [email protected] (916)932-3200, (800)921-5513
hi !
i have the same problem.
my uml is 2.6.3-rc2 with jeff`s latest uml patch. host is 2.6.0-test11 with skas patch
with 2.6.0-test11 + uml patch i didn`t have such problem. any hints?
regards
roland
----- Original Message -----
From: "Phil Nadeau" <[email protected]>
To: <[email protected]>
Sent: Tuesday, January 27, 2004 9:54 PM
Subject: [uml-devel] Kernel idles in loop after loading IPv4
Quoted 21 lines
Hello UML developers,
I'm trying to use UML to set up a simulator for a high-availability cluster. I encounter the following unexpected behavior -
On any kernel that I compile with IPv4 support, the kernel will stop intializing drivers after loading IPv4 support. The last message I see from the kernel is:
NET: Registered protocol family 2Compiling with debug allows me to single-step through the kernel. I've traced execution from
inet_init(), tosynchronize_net(), tosynchronize_kernel(), to a call towait_for_completion(), toswitch_to_tt(), after which the kernel spends all its time in the idle loop. This happens whether I'm using SKAS or TT mode.Host kernel is 2.6.1 with Stephen William's 2.6.1 SKAS patch (thanks Stephen). Guest kernel is 2.6.1 with the latest patch from Sourceforge (January 15th), without modules, and with nearly everything disabled at compile except core functionality and IPv4.
Quoted 26 lines
I've tried to find similar cases in the FAQ's and the list archives, but I'm not seeing anything. It seems like the kernel is waiting for an event related to IPv4 initialization that it never gets, but I'm not familiar enough with the kernel proper to know what that might be. Does anyone have any ideas?
Thanks in advance.
Signature · 16 lines
Phil Nadeau Software Services Manager [email protected] (916)932-3200, (800)921-5513 ------------------------------------------------------- The SF.Net email is sponsored by EclipseCon 2004 Premiere Conference on Open Tools Development and Integration See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. http://www.eclipsecon.org/osdn _______________________________________________ User-mode-linux-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
Alle 22:23, venerdì 30 gennaio 2004, Phil Nadeau ha scritto:
Quoted 4 lines · Jeff Dike
Here's the last part of a debug session that I ran which produced the error. The full transcript is attached. Thanks in advance.
Could you try to reproduce this bug running onto a 2.4 host-skas kernel with the same guest kernel? Most people have no problem with the code; also, the host-skas patch for 2.6 that you used has more than one bug (somebody even reported kernel Oopses with that applied). Also, I'd like more info on the host configuration: glibc/gcc version and distro used.
Quoted 8 lines
NET: Registered protocol family 2
Breakpoint 1, schedule () at
kernel/sched.c:1489
1489 if (likely(!(current->state & (TASK_DEAD | TASK_ZOMBIE))))
{ #0 schedule () atkernel/sched.c:1489#1 0xa002db4e in wait_for_completion (x=0xa1fefb64) at kernel/sched.c:1758 #2 0xa0041104 in synchronize_kernel () at kernel/rcupdate.c:262 #3 0xa000b5dc in inet_init () at net/ipv4/af_inet.c:1144
An old message, "[uml-devel] wait queues broken?" on 19 November 2003, spoke of something that could be related to this. Since everything is about scheduling, this error message I got could be of interest:
Kernel panic: Kernel mode fault at addr 0x1f, ip 0x400d1397 Kernel panic: kernel BUG at kernel/exit.c:793!
The second message is of interest because the BUG line of interest reads as:
schedule();
BUG();
and this is fairly interesting.
To get this, I've run this command:
while /bin/true ; do /bin/true ; done
However, the kernel I was running went on after the "NET: Registered protocol family 2" message.
Signature · 15 lines
Paolo Giarrusso, aka Blaisorblade Linux registered user n. 292729 ------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id56&alloc_id438&opÌk _______________________________________________ User-mode-linux-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
Quoted 12 lines · BlaisorBlade
Alle 22:23, venerdì 30 gennaio 2004, Phil Nadeau ha scritto:
Quoted 4 lines · Jeff Dike
Here's the last part of a debug session that I ran which produced the error. The full transcript is attached. Thanks in advance.
Could you try to reproduce this bug running onto a 2.4 host-skas kernel with the same guest kernel? Most people have no problem with the code; also, the host-skas patch for 2.6 that you used has more than one bug (somebody even reported kernel Oopses with that applied). Also, I'd like more info on the host configuration: glibc/gcc version and distro used.
gcc (GCC) 3.3.2 20031022 (Red Hat Linux 3.3.2-1)
glibc-2.3.2
libpthread-0.10
Kernel is Linux version 2.6.1 ([email protected]) (gcc version 3.3.2 20031022 (Red Hat Linux 3.3.2-1))
Based on other things I've seen on the list, I suspect it's an issue with the host using NPTL. This occurs with 2.4.22 Fedora Core kernels and vanilla 2.6.0 and 2.6.1 kernels compiled with and without SKAS. Since NPTL seems to be a Known Issue, I was leaving it alone for now. Any other ideas?
Signature · 16 lines
Phil Nadeau Software Services Manager [email protected] (916)932-3200, (800)921-5513 ------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id56&alloc_id438&opÌk _______________________________________________ User-mode-linux-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel
I started from Fedora Core 1, which I upgraded to a 2.6.1 kernel the skas3. I also suspected NPTL was causing the troubles, so I tried moving /lib/tls to /lib/tls.off and still got the hang. I also tried rebuilding glibc-2.3.2 with linux threads, then building and running UML against that, but I still got the hang.
I obtained the skas3 patch from sdw.st. Alegedly, it has problems, but it was the only skas3 patch I could apply to 2.6.1 cleanly.
Chris.
Quoted 9 lines · --- Phil Nadeau
Based on other things I've seen on the list, I suspect it's an issue
with the host using NPTL. This occurs with 2.4.22 Fedora Core kernels
and vanilla 2.6.0 and 2.6.1 kernels compiled with and without SKAS.
Since NPTL seems to be a Known Issue, I was leaving it alone for now.
Any other ideas?
__________________________________
Do you Yahoo!?
Yahoo! Finance: Get your refund fast by filing online.
http://taxes.yahoo.com/filing.html
-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click
_______________________________________________
User-mode-linux-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel