Is there a way to enable core dumping?

[ethician]

Can someone please point me in the right direction?

Setting `ulimit -c unlimited` and /proc/sys/kernel/core_* and /proc/sys/fs/suid_dumpable doesn't do the trick.

Thanks!

 

[ethician]

For those interested:

Solved by modifying `target.mak` to add DEBUG=y (and GDB=y if you are interested in working with the core file afterwards) under the respective device section.

After system boots up:
echo /tmp/core > /proc/sys/kernel/core_pattern
echo 1 > /proc/sys/kernel/core_uses_pid
echo 2 > /proc/sys/fs/suid_dumpable

Will figure out how to make it persistent over reboots.

 

[RMerlin]

For those interested:

Solved by modifying `target.mak` to add DEBUG=y (and GDB=y if you are interested in working with the core file afterwards) under the respective device section.

After system boots up:
echo /tmp/core > /proc/sys/kernel/core_pattern
echo 1 > /proc/sys/kernel/core_uses_pid
echo 2 > /proc/sys/fs/suid_dumpable

Will figure out how to make it persistent over reboots.

Do it using an init-start user script.

 

[ethician]

Ah, thanks, will do!

 

[ethician]

I've added this to the `/jfffs/scripts/init-start`. What gives me trouble is `ulimit -c unlimited` as it seems to be shell specific. Is there a way to set this system-wide?

 

[RMerlin]

I've added this to the `/jfffs/scripts/init-start`. What gives me trouble is `ulimit -c unlimited` as it seems to be shell specific. Is there a way to set this system-wide?

I thought it was user-specific, not shell specific.

You could also try inserting it in the user profile, but not sure it will apply to services started by rc at boot time.

 

[ethician]

Well tell me if I test it wrong, but ssh-ing into the box, setting `ulimit -c unlimited`, disconnecting, and reconnecting again has ulimit -c back to 0. Forks do inherit parent's setting, though.

 

[RMerlin]

Well tell me if I test it wrong, but ssh-ing into the box, setting `ulimit -c unlimited`, disconnecting, and reconnecting again has ulimit -c back to 0. Forks do inherit parent's setting, though.

I never played with ulimit so I wouldn't know for sure. Just going with what I could find as documentation through an online search.

It's possible it's not persistent because the router has a rather limited Linux environment, unlike a full distro that has, for example, the /etc/security/limits file.

 

[ethician]

Right. I've found a solution in cases where one can insert a piece of code into the source: use setrlimit(). It also needs to be run with elevated privileges.

 

[RMerlin]

Right. I've found a solution in cases where one can insert a piece of code into the source: use setrlimit(). It also needs to be run with elevated privileges.

If the program is launched by the firmware, you could use a wrapper script to launch the process.