user-mode-linux-user

Re: [User-mode-linux-user] UMLinux: kernel config constraints ?

[Jeff D. <jd...@ka...>]

yd...@al... said:
> Something like "If you intend to use one of the root filesystems we
> provide, you will need to built a kernel with devfs support, as
> devfs-specific paths are referenced in those archives (in /etc/
> fstab)." 

OK, that seems fine.

				Jeff

Re: [User-mode-linux-user] UMLinux: kernel config constraints ?

[Yann D. <yd...@al...>]

On Sun, Oct 15, 2000 at 09:03:36PM -0500, Jeff Dike wrote:
> yd...@al... said:
> > This could at least be mentionned on the "compiling" page, don't you
> > think ? 
> 
> OK.  What could I have put on the web site that would have prevented you from 
> being confused?

Something like "If you intend to use one of the root filesystems we provide,
you will need to built a kernel with devfs support, as devfs-specific paths
are referenced in those archives (in /etc/fstab)."

Regards,
-- 
Yann Dirson    <yd...@al...> |    Why make M$-Bill richer & richer ?
debian-email:   <di...@de...> |   Support Debian GNU/Linux:
                                    | Cheaper, more Powerful, more Stable !
http://ydirson.free.fr/             | Check <http://www.debian.org/>

Re: [User-mode-linux-user] UMLinux: kernel config constraints ?

[Jeff D. <jd...@ka...>]

yd...@al... said:
> This could at least be mentionned on the "compiling" page, don't you
> think ? 

OK.  What could I have put on the web site that would have prevented you from 
being confused?

> However, I feel it is quite unexpected for a 2.2-based root_fr (as
> Debian 2.2 is, and as others still are I guess) to require a
> 2.4-specific feature. 

What's the difference between a 2.2 filesystem and a 2.4 filesystem?  Not 
much.  The devfs names in fstab are about it, and if you wanted to boot 2.4 
without devfs, an existing 2.2 filesystem would probably work fine.

				Jeff

Re: [User-mode-linux-user] Child 1012 exited with signal 11

[Sylvain P. <syp...@ya...>]

Jeff Dike wrote:
> 
> syp...@ya... said:
> > Mounted devfs on /dev
> > Child 1012 exited with signal 11
> 
> This is what happens when you run UML (and a few other things) on a test8 host.
> 
> Switch to test7 or test9 and things will work better.
> 
>                                 Jeff
Many thanks, 
I tested on the test7, and it ran without problems.

But both test8 and test9 don't work for me.

Have other people managed to get a test9 host running with uml?
And has someone an idea what in the test8 breaks the uml ?
(should be in the signal handling)

				Sylvain

__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com

Re: [User-mode-linux-user] UMLinux: kernel config constraints ?

[Yann D. <yd...@al...>]

On Fri, Oct 13, 2000 at 08:16:54AM -0400, Jeff Dike wrote:
> yd...@al... said:
> > I supposed that what really depends on devfs are the root_fs provided,
> > not the kernel itself.  But then Debian 2.2 is based on a 2.2 kernel
> > without devfs, so if the provided Debian root_fs is not a hacked one
> > I'm a bit lost...
> 
> The devfs filesystem works with devfs.  The main requirement are that the 
> devices mentioned in fstab exist.  So, if it mentions devfs devices, devfs had
> better be running.

I easily understand that :)

However, I feel it is quite unexpected for a 2.2-based root_fr (as Debian
2.2 is, and as others still are I guess) to require a 2.4-specific feature.
This could at least be mentionned on the "compiling" page, don't you think ?

Thanks for your support !
-- 
Yann Dirson    <yd...@al...> |    Why make M$-Bill richer & richer ?
debian-email:   <di...@de...> |   Support Debian GNU/Linux:
                                    | Cheaper, more Powerful, more Stable !
http://ydirson.free.fr/             | Check <http://www.debian.org/>

Re: [User-mode-linux-user] Child 1012 exited with signal 11

[Jeff D. <jd...@ka...>]

syp...@ya... said:
> Mounted devfs on /dev 
> Child 1012 exited with signal 11 

This is what happens when you run UML (and a few other things) on a test8 host.

Switch to test7 or test9 and things will work better.

				Jeff

Re: [User-mode-linux-user] MySQL Segmentation Fault in UML

[Jeff D. <jd...@ka...>]

ma...@co... said:
> I've attached the PHP and Perl scripts I'm using.  

Thanks.  I'll look at it when I get back.

Are segfault messages showing up in the system log?

				Jeff

Re: [User-mode-linux-user] UMLinux: kernel config constraints ?

[Jeff D. <jd...@ka...>]

yd...@al... said:
> I supposed that what really depends on devfs are the root_fs provided,
> not the kernel itself.  But then Debian 2.2 is based on a 2.2 kernel
> without devfs, so if the provided Debian root_fs is not a hacked one
> I'm a bit lost...

The devfs filesystem works with devfs.  The main requirement are that the 
devices mentioned in fstab exist.  So, if it mentions devfs devices, devfs had 
better be running.

				Jeff

Re: [User-mode-linux-user] UMLinux: kernel config constraints ?

[Yann D. <yd...@al...>]

On Thu, Oct 12, 2000 at 08:34:25AM -0400, Jeff Dike wrote:
> yd...@al... said:
> > The precompiled 2.4.0-test9 kernel from the sourceforge works fine,
> > but I have real problems with my self-compiled kernel.   

Looks like the .config I sent is not the one that caused the problems.  It
is a modified version of the one that caused problems, and I had ommitted to
"make clean" in between, which made me think this one broke as well.

Just deselect support for devfs and the symptoms I described are back.

I supposed that what really depends on devfs are the root_fs provided, not
the kernel itself.  But then Debian 2.2 is based on a 2.2 kernel without
devfs, so if the provided Debian root_fs is not a hacked one I'm a bit
lost...

Regards,
-- 
Yann Dirson    <yd...@al...> |    Why make M$-Bill richer & richer ?
debian-email:   <di...@de...> |   Support Debian GNU/Linux:
                                    | Cheaper, more Powerful, more Stable !
http://ydirson.free.fr/             | Check <http://www.debian.org/>

Re: [User-mode-linux-user] MySQL Segmentation Fault in UML

[Matt C. <ma...@co...>]

I've been using a PHP script under Apache/PHP/MySQL to query the
database, and a Perl script to send the requests to the web server. 
After anywhere from 5000 to 30000 requests, MySQL gets a segmentation
fault or one of the user defined signals.  I've attached the PHP and
Perl scripts I'm using.  If you need more information, or even a
complete disk image for UML, just let me know.

To use the webtest.pl script, I've been running the following from UML:

webtest.pl http://localhost/db.php

I've got my db.php file in /usr/local/apache/htdocs/

I hope this helps you replicate the problem.

Jeff Dike wrote:
> 
> > /usr/local/mysql/bin/safe_mysqld: line 158:    88 Segmentation fault
> > /usr/local/mysql/bin/safe_mysqld: line 158:  5666 User defined signal 1
> 
> We have some problems here.  That segfault needs looking in to.
> 
> The SIGUSR1 might be something that I already know about.  In some cases, a
> process tries to send itself a signal, but kills a different process instead.
> 
> Do you have a script or something that can reproduce this?

-- 

 - Matt Clay
 - Cowboyz.com
 - (503) 241-1990

[User-mode-linux-user] Child 1012 exited with signal 11

[Sylvain P. <syp...@ya...>]

Hi,
Since several days, I am trying to solve a problem which makes me unable
to start the kernel.
I get the following message:

< ... >
ssl receive thread is pid 1017
devfs: devfs_register(): device already registered: "0"
devfs: v0.102 (20000622) Richard Gooch (rg...@at...)
devfs: devfs_debug: 0x0
devfs: boot_options: 0x0
VFS: Mounted root (ext2 filesystem) readonly.
Mounted devfs on /dev
Child 1012 exited with signal 11

The same problem appeared with the pre-compiled test9 binary, and with
my self-compiled kernel.
I tried with my own root_fs, Tomsrtb, and others.
The host kernel is also test9, and didn't see similar problems in the
list. 
So I was wondering what was special in my linux box, which makes it fail
here, but nowhere else.

Debugging informations:
----------------------
The segfalut seems to happen in the init thread (third pid from the
initial one).

First, this thread exec the init process:

(gdb) b exec_kern.c:88
(gdb) bt
#0  um_execve (file=0x100ef00d "/sbin/init", argv=0x10122140,
env=0x10122180) at exec_kern.c:88
#1  0x100013d8 in init (unused=0x0) at init/main.c:768
#2  0x100a2fd0 in new_thread_proc (t=0x50052000) at process_kern.c:128

It is this pid which will get signal 11 in the tracing thread next :
(gdb) call getpid()
$2 = 1012   

the thread is set to exec the init process (all is right here)
(gdb) p current_task->thread.request.op
$5 = 1 ( == OP_EXEC )

So everything seems normal, and as the SIGSEGV signal is handled, the
thread should not exit

(gdb) l 
88                      usr1_pid(getpid());

Finally, the thread sends itself a SIGUSR1, as normal.

But then, in the tracing thread, WIFSIGNALED(status) returns true
This means the thread has exited with a SIGSEGV, which it _should not_ :

VFS: Mounted root (ext2 filesystem) readonly.
Mounted devfs on /dev

Child 1012 exited with signal 11

And that's here I don't understand why it fails.


Thank you for any help

	Regards 

Sylvain

_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com

Re: [User-mode-linux-user] UMLinux: kernel config constraints ?

[Yann D. <yd...@al...>]

On Thu, Oct 12, 2000 at 08:34:25AM -0400, Jeff Dike wrote:
> yd...@al... said:
> > The precompiled 2.4.0-test9 kernel from the sourceforge works fine,
> > but I have real problems with my self-compiled kernel.   
> 
> Sending in your .config would certainly help us reproduce the problems.

Here it is.

Regards,
-- 
Yann Dirson    <yd...@al...> |    Why make M$-Bill richer & richer ?
debian-email:   <di...@de...> |   Support Debian GNU/Linux:
                                    | Cheaper, more Powerful, more Stable !
http://ydirson.free.fr/             | Check <http://www.debian.org/>

Re: [User-mode-linux-user] UMLinux: kernel config constraints ?

[Jeff D. <jd...@ka...>]

yd...@al... said:
> The precompiled 2.4.0-test9 kernel from the sourceforge works fine,
> but I have real problems with my self-compiled kernel.   

Sending in your .config would certainly help us reproduce the problems.

				Jeff

Re: [User-mode-linux-user] MySQL Segmentation Fault in UML

[Jeff D. <jd...@ka...>]

> /usr/local/mysql/bin/safe_mysqld: line 158:    88 Segmentation fault
> /usr/local/mysql/bin/safe_mysqld: line 158:  5666 User defined signal 1

We have some problems here.  That segfault needs looking in to.

The SIGUSR1 might be something that I already know about.  In some cases, a 
process tries to send itself a signal, but kills a different process instead.

Do you have a script or something that can reproduce this?

				Jeff

[User-mode-linux-user] UMLinux: kernel config constraints ?

[Yann D. <yd...@al...>]

Hi there,

The precompiled 2.4.0-test9 kernel from the sourceforge works fine,
but I have real problems with my self-compiled kernel.  

I suspect there are requirements wrt config options which my kernel
did not met.  Is there a centralized list of such requirements ?  Is
the .config file used for the precompiled kernel available somewhere ?


Differences include:

* using provided Debian image
- "bad superblock" on boot, hence single-user mode.
- init segfaults when I try to ping a second virtual host using the
new ethernet support

* using Tom's boot/root
- mine cannot have MAC address configured:

# ifconfig eth0 hw ether 0:0:10:0:0:1
0:0:10:0:0:1: Error 0

- precompiled refuses me to log in:

 console tomsrtbt login: root

 The default "root" password is "xxxx",
 edit /etc/passwd to change it, or edit
 settings.s to change it permanently...

Password: ile rev.2.01
ile: unable to allocate pty/tty pair


My .config file available on demand.

[please CC me on followup]

Best regards,
-- 
Yann Dirson    <yd...@al...> |    Why make M$-Bill richer & richer ?
debian-email:   <di...@de...> |   Support Debian GNU/Linux:
                                    | Cheaper, more Powerful, more Stable !
http://ydirson.free.fr/             | Check <http://www.debian.org/>

[User-mode-linux-user] MySQL Segmentation Fault in UML

[Matt C. <ma...@co...>]

I've been running MySQL/PHP/Apache in UML, doing some simple stability
tests to see how well they run under UML.  I'm getting the following
messages popping up in my UML console after a few hours of running
repeated queries through PHP to MySQL (output cleaned up for
readability):

---[ Begin Output ]---

usermode login:

/usr/local/mysql/bin/safe_mysqld: line 158:    88 Segmentation fault
nohup $ledir/mysqld --basedir=$MY_BASEDIR_VERSION --datadir=$DATADIR
--skip-locking "$@" >>$err_log 2>&1

Number of processes running now: 3
mysqld process hanging, pid 5634 - killed
mysqld process hanging, pid 5633 - killed
/usr/local/mysql/bin/safe_mysqld: kill: (5633) - No such pid
mysqld restarted on  Wed Oct 11 20:12:07 PDT 2000
/usr/local/mysql/bin/safe_mysqld: line 158:  5666 User defined signal 1
nohup $ledir/mysqld --basedir=$MY_BASEDIR_VERSION --datadir=$DATADIR
--skip-locking "$@" >>$err_log 2>&1

Number of processes running now: 3
mysqld process hanging, pid 5670 - killed
/usr/local/mysql/bin/safe_mysqld: kill: (5670) - No such pid
mysqld restarted on  Wed Oct 11 20:27:24 PDT 2000

---[ End Output ]---

I know this isn't much to go on, but I was wondering if anyone could
point me in the right direction, so I can work on tracking this problem
down.  Of course, if someone already has an idea as to what it is, I'd
like to know.

-- 

 - Matt Clay
 - Cowboyz.com
 - (503) 241-1990

Re: [User-mode-linux-user] Unknown HZ value! (20) Assume 100.

[William S. <wst...@po...>]

Good evening, Matt and David,

On Tue, 10 Oct 2000, David Coulson wrote:

> On Tue, 10 Oct 2000, Matt Clay wrote:
> 
> > When I run programs like ps or top in UML, I get the following message:
> > 
> > Unknown HZ value! (20) Assume 100.
> 
> It's to do with the kernel schedular, presumably it was lowered so it
> reduces the CPU usage of the virtual machine.
> 
> If you want to fix it, compile procps against a UML patched kernel.

	Once again we're caught in the age-old Linux software
dilemma.  From Riley Williams:

There is something frustrating about the quality and speed of Linux
development, ie., the quality is too high and the speed is too high,
in other words, I can implement this XXXX feature, but I bet someone
else has already done so and is just about to release their patch.

	From procps-2.0.7/proc/sysinfo.c:

#if 0 /* This ends up causing more harm than good.  :-( */                                                                                                           
    fprintf(stderr, "Unknown HZ value! (%ld) Assume %ld.\n", h, Hertz);                                                                                              
#endif                                                                                                                                                               

	As of the procps included in RedHat 7, this message is no longer
displayed.  If that rpm doesn't install on your system, download the .srpm
from a redhat 7 mirror, run "rpm --rebuild procps....src.rpm", and "rpm
-Uvh /usr/src/redhat/RPMS/i386/procps...i386.rpm --ignorearch".
	Cheers,
	- Bill

---------------------------------------------------------------------------
	"Analysts say they have little patience with LinuxOne, a virtually
unknown start-up in Mountain View, Calif., that recently launched a
Linux distribution that's seen as a transparent attempt to cash in on
the Linux hype. 
	'Stay away from the deal,' warns Hirschkorn. 
	Says DeGraw: 'It's an embarrassment that could hurt the whole
industry.'"
	-- Dominique Deckmyn
(Courtesy of http://www.computerworld.com/home/print.nsf/all/000306F3B2)
--------------------------------------------------------------------------
William Stearns (wst...@po...).  Mason, Buildkernel, named2hosts, 
and ipfwadm2ipchains are at:                http://www.pobox.com/~wstearns
LinuxMonth; articles for Linux Enthusiasts! http://www.linuxmonth.com
--------------------------------------------------------------------------

Re: [User-mode-linux-user] compiling error on 2.4.0-test9

[Jeff D. <jd...@ka...>]

> I don't know the reason but it seems to me my glibc
> environment is something wrong so I re-install it.

There's nothing wrong with your libc.  There's something wrong with my stupid
tricks...

> I'd like to debug it so I make with -g option.

In that case, I would try finding an environment in which you can 
sucessfully build.  I'm running on RH6.0 FWIW.

I can't look at this at the moment because I'm on my way to Atlanta, but 
I'll see about fixing it when I get back.

				Jeff

Re: [User-mode-linux-user] compiling error on 2.4.0-test9

[Hiro Y. <hyo...@mi...>]

Jeff,

Thanks for your help.

> > /usr/bin/ld: Warning: size of symbol `sprintf' changed from 30 to 52 in vmlinux.o
> > collect2: ld returned 1 exit status
> > make: *** [linux] Error 1
> >
> > I searched this mail list and found similar problem
> > but could not find an answer.
> >
> > Does any one have hints?
...
> For you, for some reason, sprintf is getting included in that link,
> and that's conflicting with the kernel sprintf.

I don't know the reason but it seems to me my glibc
environment is something wrong so I re-install it.

> If you just want to run UML and not debug it, you can go into the
> arch/um/kernel Makefile, and put unmap_fin.o (or unmap.o) in $OBJS and
> make unmap.c get compiled like any other file.

I'd like to debug it so I make with -g option.

Regards,
  Hiro

Re: [User-mode-linux-user] Unknown HZ value! (20) Assume 100.

[Jeff D. <jd...@ka...>]

> When I run programs like ps or top in UML, I get the following message:
> 
> Unknown HZ value! (20) Assume 100.
> 
> I'm wondering how to fix this, and what, if any, problems this may
> cause.

Don't worry, be happy.

You can locate the definition of HZ and change it to 100 if it really
bothers you.

I haven't noticed it causing any problems.

				Jeff

Re: [User-mode-linux-user] compiling error on 2.4.0-test9

[Jeff D. <jd...@ka...>]

> /usr/bin/ld: Warning: size of symbol `sprintf' changed from 30 to 52 in vmlinux.o
> collect2: ld returned 1 exit status
> make: *** [linux] Error 1
>
> I searched this mail list and found similar problem
> but could not find an answer.
>
> Does any one have hints?

You're being bit by the tricks I played in order to get the kernel debugger's
breakpoints to work.

The basic problem is that every thread in UML is in its own process.
When the debugger sets a breakpoint, I believe that it is set in that
one address space and not in the others.  To fix this, I copy the
kernel text into a file, unmap the text, and mmap that file shared in
its place.  This ensures that no copy-on-write happens with the code,
and when a breakpoint is set, it is set in all the processes.

The problem is that between the unmapping of the text and the mapping
in of the file, there is no code to execute because it was unmapped.
To get around this, I put that function in its own section of the
executable, and that section is not unmapped.

The file containing that code is linked in such a way that all symbols
are resolved, putting everything that that code needs in the
not-unmapped section.

For you, for some reason, sprintf is getting included in that link,
and that's conflicting with the kernel sprintf.

If you just want to run UML and not debug it, you can go into the
arch/um/kernel Makefile, and put unmap_fin.o (or unmap.o) in $OBJS and
make unmap.c get compiled like any other file.

				Jeff

[User-mode-linux-user] compiling error on 2.4.0-test9

[Hiro Y. <hyo...@mi...>]

Hi,

I'm a newbie of the user mode linux and trying to
compile the source but got the following error.

I'm using the linux-2.4.0-test9 + patch-2.4.0-test on
the Miracle Linux SE V1.0 which is a Turbo Linux Server
6.1J based linux.

gcc -Wl,-T,/usr/src/uml/linux/arch/um/link.ld -pg -DPROFILING -Wl,--wrap,__monstartup -o linux -static \
	/usr/src/uml/linux/arch/um/main.o vmlinux.o -L/usr/lib
vmlinux.o: In function `sprintf':
/usr/src/uml/linux/lib/vsprintf.c:315: multiple definition of `sprintf'
arch/um/kernel/unmap_fin.o:/usr/src/rpm/BUILD/glibc-2.1.3/stdio-common/sprintf.c:33: first defined here
/usr/bin/ld: Warning: size of symbol `sprintf' changed from 30 to 52 in vmlinux.o
collect2: ld returned 1 exit status
make: *** [linux] Error 1

I searched this mail list and found similar problem
but could not find an answer.

Does any one have hints?

Thanks in advance,
  Hiro

Re: [User-mode-linux-user] Re: User-mode linux stack overflow: could be generic problem

[David C. <tec...@li...>]

On Sun, 8 Oct 2000, Jeff Dike wrote:
> Anyway, the patch below removes 256 bytes from the set_signals frame.  It
> ought to alleviate things a bit.  I'll be looking for other things I can do,
> as well. Let me know how it works for you.

I tried to apply that patch to a checkout of the uml code from cvs, and it
failed.

Either it has been applied (doesn't look like it) to the code which is in
cvs already, or I'm doing something stupid.

Ideas?

-- 
David Coulson
tec...@so...

Re: [User-mode-linux-user] Unknown HZ value! (20) Assume 100.

[David C. <tec...@li...>]

On Tue, 10 Oct 2000, Matt Clay wrote:

> When I run programs like ps or top in UML, I get the following message:
> 
> Unknown HZ value! (20) Assume 100.

It's to do with the kernel schedular, presumably it was lowered so it
reduces the CPU usage of the virtual machine.

If you want to fix it, compile procps against a UML patched kernel.

-- 
David Coulson
tec...@so...

[User-mode-linux-user] Unknown HZ value! (20) Assume 100.

[Matt C. <ma...@co...>]

When I run programs like ps or top in UML, I get the following message:

Unknown HZ value! (20) Assume 100.

I'm wondering how to fix this, and what, if any, problems this may
cause.

 - Matt