Re: Catch SIGSEGV from Linux kernel
- From: Rainer Weikusat <rweikusat@xxxxxxxxxxx>
- Date: Thu, 21 Feb 2008 08:05:42 +0100
"F.Julien" <stopspam@xxxxxxxxxxxx> writes:
I'd like to catch all processes that end because of a segmentation
fault. My goal is to print a kernel log when a process exit in that
case.
ARM Linux has a configuration option named CONFIG_DEBUG_USER
doing does exaclty that, which I have found very helpful to debug
applications. Maybe looking at this code could help with implementing
something similar for other architectures.
I did look into the kernel code and the only solution I found is to
patch the function "send_signal" (file kernel/signal.c).
I did also search on google an existing kernel patch but I only found
a very old patch for 2.2 kernel. I'm surprised that nobody wants to
monitor this (especially sysadmin).
Generally, people have no clue why a particular application would
segfault, except "that it just does this" and they would prefer this
effect to be as clandestine as possible, lest someone could actually
expect them to fix it (instead of 'working on exciting new features').
NB: That's an imputation, but not much of it. The 'established,
sensible way' of application development, ie use as much third party
code as humanly possible, whose actual interactions nobody
understands, pretty much leaves no other options.
.
- References:
- Catch SIGSEGV from Linux kernel
- From: F.Julien
- Catch SIGSEGV from Linux kernel
- Prev by Date: Journalling in VM?
- Next by Date: Re: Catch SIGSEGV from Linux kernel
- Previous by thread: Re: Catch SIGSEGV from Linux kernel
- Next by thread: Journalling in VM?
- Index(es):
Relevant Pages
|
