Skip to contentSkip to conversation

Do not overwrite pre-existing SIGDEBUG handler in xeno_skin_bind

PATCH In review Jan Kiszka 2 participants 3 messages Mar 13, 2013 – Mar 14, 2013 Generic include/asm header files

lore
Share & tools
lore

Jan Kiszka

author
#0 lore raw

PATCH 1/1 Do not overwrite pre-existing SIGDEBUG handler in xeno_skin_bind

In case the application already set a SIGDEBUG handler, let that one handle any potential SIGDEBUG_NOMLOCK events and keep our handler (xeno_handle_mlock_alert) out of the loop. This helps in scenarios where skin libraries are loaded belatedly via dlopen.

---

Anything that speaks against this? We ran into that issue with a large application, and that can be one way to solve but should even make sense in general.

 include/asm-generic/bits/bind.h |    5 +++++
 1 files changed, 5 insertions(+), 0 deletions(-)
include/asm-generic/bits/bind.h +5 -0
index 269be95..f39c853 100644
@@ -23,6 +23,11 @@ xeno_bind_skin(unsigned skin_magic, const char *skin, const char *module)
 		exit(EXIT_FAILURE);
 	}
 
+	/* do not overwrite an already installed SIGDEBUG handler */
+	sigaction(SIGXCPU, NULL, &sa);
+	if (sa.sa_handler != SIG_DFL)
+		return muxid;
+
 	sa.sa_sigaction = xeno_handle_mlock_alert;
 	sigemptyset(&sa.sa_mask);
 	sa.sa_flags = SA_SIGINFO;
Signature · 1 line
1.7.3.4
#1 lore raw
Quoted 4 lines · Jan Kiszka

In case the application already set a SIGDEBUG handler, let that one handle any potential SIGDEBUG_NOMLOCK events and keep our handler (xeno_handle_mlock_alert) out of the loop. This helps in scenarios where skin libraries are loaded belatedly via dlopen.

Another option would be to save the previous handler, and call it when we detect the signal is not due to an mlock alert.

Signature · 1 line
                                                                Gilles.
author
#2 lore raw
Quoted 10 lines · Gilles Chanteperdrix
Quoted 4 lines · Jan Kiszka

In case the application already set a SIGDEBUG handler, let that one handle any potential SIGDEBUG_NOMLOCK events and keep our handler (xeno_handle_mlock_alert) out of the loop. This helps in scenarios where skin libraries are loaded belatedly via dlopen.

Another option would be to save the previous handler, and call it when we detect the signal is not due to an mlock alert.

I've no preference. Pushed that variant now into my queue.

Jan

Signature · 2 lines
Siemens AG, Corporate Technology, CT RTC ITP SDP-DE
Corporate Competence Center Embedded Linux