msh and signals lost due to blocking

Jate Sujjavanich jsujjavanich at syntech-fuelmaster.com
Tue Jan 29 22:51:12 UTC 2008


I compiled busybox for x86, and the bug does not appear to manifest
itself. I've been reproducing it by running a ping and hitting ctrl-C
from the shell prompt. After about four successes, the signal block
occurs. With version 1.7.2 on x86, I can do this at will.

I fired up the kernel debugger. I ran ping and terminated it with a
ctrl-C. It looks like there's an explicit system call which turns on the
signal blocking. I haven't yet been able to determine if busybox msh is
the culprit.

The odd thing is that I added debug statements to call out sigprocmask
calls in busybox, and that allowed me to ctrl-c many times. That is
troubling.

I am fairly certain that this was not a problem with the last version
(whatever's in uClinux-dist) of busybox that I used. That version was
incompatible with the new toolchain/uClibc I got from codesourcery
(4.2-8). That led me to busybox 1.7.2 (where I am today).

I haven't been able to try any other versions yet.

- Jate S.

-----Original Message-----
From: Denis Vlasenko [mailto:vda.linux at googlemail.com] 
Sent: Tuesday, January 29, 2008 4:08 AM
To: busybox at busybox.net
Cc: Jate Sujjavanich
Subject: Re: msh and signals lost due to blocking

On Thursday 24 January 2008 22:54, Jate Sujjavanich wrote:
> I am using msh from busybox 1.7.2 and kernel 2.4.31-uc0 on m68knommu.
> The shell handles SIGINT properly one to four times. After these 
> successes, the process group has signals blocked, and it the processes

> never dequeue SIGINT. Restarting the shell resets the blocks. The 
> cycle repeats.
> 
> Does anyone familiar with msh know whether it activates signal
blocking?

Does it happen on desktop (x86 machine)? Can you post the .config?
Can you describe how to reproduce it?

msh seems to be relatively untouched for many versions, but it still
worth checking. Can you try older busybox versions, built with same
.config?

> I am also not sure whether the something in the library does this.

Not likely. Most probably it's msh bug.
--
vda



 
 
************************************************************************
************
This footnote confirms that this email message has been scanned by
PineApp Mail-SeCure for the presence of malicious code, vandals &
computer viruses.
************************************************************************
************



-------------- next part --------------
A non-text attachment was scrubbed...
Name: config.busybox
Type: application/octet-stream
Size: 20715 bytes
Desc: config.busybox
Url : http://lists.busybox.net/pipermail/busybox/attachments/20080129/7ae5612a/attachment-0002.obj 


More information about the busybox mailing list