Dalibor Dukic | 1 Mar 2006 01:03
Picon

[uml-user] alarm_handler

Hi, 

I'm doing some debugging of IPSec in linux kernel with uml
(CONFIG_PT_PROXY enabled) and gdb for educational purpose.

The uml+gdb is starting very well.

When I put breakpoint on some function and I trigger that break I can't
step the subroutine code because gdb is stepping the code from
alarm_handler function defined in os/signal.c. And this is repeating
till CTRL+c. I will be glad if someone help me that I can step the
function which I break.

UML kernel is 2.6.14.4 and host kernel is 2.6.15.
Can anyone help ? 
                                      thnx in advance.
--

-- 
Dalibor Dukic <duke <at> x.pbf.hr>
Jeff Dike | 1 Mar 2006 03:19

Re: [uml-user] alarm_handler

On Wed, Mar 01, 2006 at 01:03:41AM +0100, Dalibor Dukic wrote:
> When I put breakpoint on some function and I trigger that break I can't
> step the subroutine code because gdb is stepping the code from
> alarm_handler function defined in os/signal.c. And this is repeating
> till CTRL+c. I will be glad if someone help me that I can step the
> function which I break.

Disabling CONFIG_UML_REAL_TIME_CLOCK should help this.

Are you doing machine instruction stepping or source stepping?  I
think source stepping ahould work the way you want, since gdb should
set a temporary breakpoint on the next line and continue.

If it's machine-level stepping, gdb is actually doing the right thing,
since it is stopping on the next executed instruction.

				Jeff

-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
Dalibor Dukic | 2 Mar 2006 01:44
Picon

[uml-user] alarm_handler

Hi, 

I'm doing some debugging of IPSec in linux kernel with uml
(CONFIG_PT_PROXY enabled) and gdb for educational purpose.

The uml kernel under gdb is starting very well.

When I put breakpoint on some function and I trigger that break I can't
step the subroutine code because gdb is stepping the code from
alarm_handler function defined in os/signal.c. And this is repeating
till CTRL+c. I will be glad if someone help me that I can step the
function which I break.

UML kernel is 2.6.14.4 and host kernel is 2.6.15.
Can anyone help ? 
                                      thnx in advance.
--

-- 
Dalibor Dukic <duke <at> x.pbf.hr>

-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
John Dykstra | 3 Mar 2006 16:31
Favicon

[uml-user] Soft lockups while debugging

Is it normal to get a "BUG: soft lockup detected on CPU#0!" when using gdb in SKAS0 mode? 

Can they be safely ignored;  i.e. is the process continuing on without changes afterwards?

  --  John

Jeff Dike | 3 Mar 2006 18:20

Re: [uml-user] Soft lockups while debugging

On Fri, Mar 03, 2006 at 09:31:34AM -0600, John Dykstra wrote:
> Is it normal to get a "BUG: soft lockup detected on CPU#0!" when using
> gdb in SKAS0 mode?  

Yup, that happens if 10 seconds go by without a timer interrrupt, so
you better not spend > 10 seconds with it sitting at a gdb prompt if
you want to avoid these.

Just disable soft lockup detection if you're going to be gdb-ing UML.

> Can they be safely ignored;  i.e. is the process continuing on without
> changes afterwards?

Yup, it's just an ugly stack dump.

				Jeff

-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
Phill Wombat | 7 Mar 2006 12:02

[uml-user] 2.6.15.5-bs3-tls (somewhat late) x86-32 host and guest

Hi Paolo,

This is somewhat late, however here it is for what it is worth:

login: root
Password:
Last login: Tue Mar  7 17:35:08 on tty0
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
load_TLS(O_FORCE) in fork_handler succeeded
[root <at> avg ~]#

FC4 host with skas3 patch. Guest has my hacks made to the rootfs to
avoid NPTL issues. As you can see the TLS stuff seems to be working! I
haven't tried it with a completely standard FC4 rootfs yet.

My only issue is the messages.... I like them but.... :-)

Now to download the latest binaries. Do these include the TLS stuff?

Cheers
Phill.

-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
Phill Wombat | 7 Mar 2006 12:05

[uml-user] Oops read 2.6.15.5-bs1-tls not the latest bs3-tls

Sorry.

Phill.

-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
Phill Wombat | 7 Mar 2006 12:15

[uml-user] 2.6.15.5-bs3-tls

Just downloaded the binary for
http://www.user-mode-linux.org/~blaisorblade/binaries/2.6.15.5-bs3-tls/uml-release-2.6.15.5-bs3-tls.tar.bz2

This works well for me and fixes the issue I just posted.

Sorry to pepper the list with this. (Just saw the new bin today < 24
hours old).

Still haven't tried a plain vanilla FC4 rootfs. If I can run fsck then I
know everything works.

Cheers
Phill.

-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
Blaisorblade | 7 Mar 2006 12:43
Picon
Favicon

Re: [uml-user] 2.6.15.5-bs3-tls

On Tuesday 07 March 2006 12:15, Phill Wombat wrote:
> Just downloaded the binary for
> http://www.user-mode-linux.org/~blaisorblade/binaries/2.6.15.5-bs3-tls/uml-
>release-2.6.15.5-bs3-tls.tar.bz2
>
> This works well for me and fixes the issue I just posted.
>
> Sorry to pepper the list with this. (Just saw the new bin today < 24
> hours old).
>
> Still haven't tried a plain vanilla FC4 rootfs. If I can run fsck then I
> know everything works.

I don't remember the testing ground (IIRC it was Debian), but I've seen 
reports that even multithreaded apps like MySQL do work well with this.

The code should be fairly complete, I have another couple of cleanups to do 
and some Jeff's patches to merge; when I'm done, I'll throw the result onto 
you and if ready send it for 2.6.17.

> Cheers
> Phill.

--

-- 
Inform me of my mistakes, so I can keep imitating Homer Simpson's "Doh!".
Paolo Giarrusso, aka Blaisorblade (Skype ID "PaoloGiarrusso", ICQ 215621894)
http://www.user-mode-linux.org/~blaisorblade

	

	
		
___________________________________ 
Yahoo! Mail: gratis 1GB per i messaggi e allegati da 10MB 
http://mail.yahoo.it

-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
Phill Wombat | 7 Mar 2006 14:21

Re: [uml-user] 2.6.15.5-bs3-tls

One issue when running the reboot command or just shutdown -h now:

nice -n 5 setarch i386 ../kernel/linux32-2.6.15.5-bs3-tls mem=256m
ubda=cow,../shared/root_fs.fc4.2 ubdb=swap root=/dev/ubda con=pty con0=fd:0,fd:1 eth0=tuntap,vm3,,

(host is x86-32, even though there is a setarch. Use this on x86-64 as well...).

Fedora Core release 4 (Stentz)
Kernel 2.6.15.5-bs3-tls on an i686

Unauthorized access to this system is strictly prohibited.
pdc login: root
Password:
Last login: Wed Mar  8 00:02:51 on tty0
[root <at> pdc ~]# shutdown -h now

Broadcast message from root (tty0) (Wed Mar  8 00:12:42 2006):

The system is going down for system halt NOW!
INIT: Switching to runlevel: 0
INIT: Sending processes the TERM signal
Stopping atd: [  OK  ]
Stopping sshd: [  OK  ]
Stopping crond: [  OK  ]
Shutting down kernel logger: [  OK  ]
Shutting down system logger: [  OK  ]
line_ioctl: tty0: ioctl TIOCLINUX called
Shutting down interface eth0:  line_ioctl: tty0: ioctl TIOCLINUX called
line_ioctl: tty0: ioctl TIOCLINUX called
[  OK  ]
Shutting down loopback interface:  line_ioctl: tty0: ioctl TIOCLINUX called
line_ioctl: tty0: ioctl TIOCLINUX called
[  OK  ]
Starting killall:  [  OK  ]
line_ioctl: tty0: ioctl TIOCLINUX called
Sending all processes the TERM signal...
Sending all processes the KILL signal...
Saving random seed:
Syncing hardware clock to system time
Turning off swap:
Turning off quotas:
Halting system...
Power down.
-bash: line 1: 11408 Segmentation fault 
[root <at> localhost vm3]#                              

So the machine has stopped cleanly, but it segfaults. Used to be able to
"reboot".

Cheers
Phill.

On Tue, 2006-03-07 at 12:43 +0100, Blaisorblade wrote:
> On Tuesday 07 March 2006 12:15, Phill Wombat wrote:
> > Just downloaded the binary for
> > http://www.user-mode-linux.org/~blaisorblade/binaries/2.6.15.5-bs3-tls/uml-
> >release-2.6.15.5-bs3-tls.tar.bz2
> >
> > This works well for me and fixes the issue I just posted.
> >
> > Sorry to pepper the list with this. (Just saw the new bin today < 24
> > hours old).
> >
> > Still haven't tried a plain vanilla FC4 rootfs. If I can run fsck then I
> > know everything works.
> 
> I don't remember the testing ground (IIRC it was Debian), but I've seen 
> reports that even multithreaded apps like MySQL do work well with this.
> 
> The code should be fairly complete, I have another couple of cleanups to do 
> and some Jeff's patches to merge; when I'm done, I'll throw the result onto 
> you and if ready send it for 2.6.17.
> 
> > Cheers
> > Phill.
> 

-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642

Gmane