This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
Re: Systemtap probe for Signals
- From: Li Guanglei <guanglei at cn dot ibm dot com>
- To: "systemtap at sourceware dot org" <systemtap at sourceware dot org>
- Cc: Manoj S Pattabhiraman <mpattabh at in dot ibm dot com>
- Date: Fri, 28 Jul 2006 16:06:04 +0800
- Subject: Re: Systemtap probe for Signals
- Organization: IBM CSTL
- References: <OF5B05ABF6.75FEE7A3-ON652571A1.003C3F54-652571A1.003C19A3@in.ibm.com> <44C71635.60902@cn.ibm.com>
Hi,
Beside the probe points in your tapset, I'd like to probe the
following signal related kernel functions:
1. do_sigpending
It is used to suspend signals. Called by sys_rt_sigpending and
sys_sigpending
2. handle_signal
It is called when the kernel want to invoke the signal handlers.
Called by do_signal.
3. do_sigaction
It is called by sys_sigaction, compat_sys_rt_sigaction,
compat_sys_rt_sigaction, sys_rt_sigaction, sys_signal. It is
corresponding to the signal() and sigaction() calles from use app.
4. sigprocmask
It is called by sys_sigprocmask, sys_rt_sigprocmask.
And I found you put a point on sig_ignored. But I don't think
sig_ignored is used to ignore a signal. It is only used to check
whether or not a signal has been ignored.
And another comment is about signal.sendsig and
signal.send_sig_queue. The difference between send_signal and
send*sigqueue is that send_signal will alloc a sigqueue before adding
a signal into that queue. But their intention are almost the same:
sending a signal to a process. I am not sure whether we could combine
them into one probe or leave it unchanged.
- Guanglei