Thread (18 messages) 18 messages, 5 authors, 2018-08-01

Re: [PATCH v3 5/9] powerpc/traps: Print signal name for unhandled signals

From: Joe Perches <joe@perches.com>
Date: 2018-08-01 07:03:59
Also in: lkml

On Wed, 2018-08-01 at 08:37 +0200, Christophe LEROY wrote:
Le 31/07/2018 à 16:50, Murilo Opsfelder Araujo a écrit :
quoted
This adds a human-readable name in the unhandled signal message.
Before this patch, a page fault looked like:
   pandafault[6303]: unhandled signal 11 at 100007d0 nip 1000061c lr 7fff93c55100 code 2 in pandafault[10000000+10000]
After this patch, a page fault looks like:
   pandafault[6352]: segfault (11) at 13a2a09f8 nip 13a2a086c lr 7fffb63e5100 code 2 in pandafault[13a2a0000+10000]
]]
quoted
diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
[]
quoted
@@ -96,6 +96,41 @@ EXPORT_SYMBOL(__debugger_fault_handler);
  #define TM_DEBUG(x...) do { } while(0)
  #endif
  
+static const char *signames[SIGRTMIN + 1] = {
+	"UNKNOWN",
+	"SIGHUP",			// 1
+	"SIGINT",			// 2
[]
I don't think is is worth having that full table when we only use a few 
of them. (As discussed in v1 https://patchwork.ozlabs.org/patch/948802/)

I would suggest to instead use a function like this:

static const char *signame(int signr)
{
	if (signr == SIGBUS)
		return "bus error";
	if (signr == SIGFPE)
		return "floating point exception";
	if (signr == SIGILL)
		return "illegal instruction";
	if (signr == SIGILL)
		return "segfault";
	if (signr == SIGTRAP)
		return "unhandled trap";
	return "unknown signal";
}
trivia:

Unless the if tests are ordered most to least likely,
perhaps it would be better to use a switch/case and
let the compiler decide.

	switch (signr) {
	case SIGBUS:	return "bus error";
	case SIGFPE:	return "floating point exception";
	case SIGILL:	return "illegal instruction";
	case SIGSEGV:	return "segfault";
	case SIGTRAP:	return "unhandled trap";
	}
	return "unknown signal";
}
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help