|
From: Rich C. <rc...@wi...> - 2013-03-01 21:17:54
|
Here's the 10.8 output. Similar in that main() is still missing.
strchr() has main().
I suspect if sendmsg.c had a return() instead of exit(), that main would
be listed too.
=================================================
./valgrind-new/memcheck/tests/sendmsg.stderr.diff
=================================================
--- sendmsg.stderr.exp 2013-03-01 07:35:12.000000000 -0600
+++ sendmsg.stderr.out 2013-03-01 07:46:58.000000000 -0600
@@ -1,6 +1,6 @@
Syscall param sendmsg(msg) points to uninitialised byte(s)
at 0x........: sendmsg (in /...libc...)
- by 0x........: main (sendmsg.c:45)
+ ...
Address 0x........ is on thread 1's stack
sendmsg: 6
On Fri, 01 Mar 2013 14:21:34 +0100
Mark Wielaard <mj...@re...> wrote:
> On Fri, 2013-03-01 at 07:10 -0600, Rich Coe wrote:
> > --- sendmsg.stderr.exp 2013-02-28 23:56:38.000000000 -0600
> > +++ sendmsg.stderr.out 2013-03-01 00:07:44.000000000 -0600
> > @@ -1,6 +1,6 @@
> > Syscall param sendmsg(msg) points to uninitialised byte(s)
> > - at 0x........: sendmsg (in /...libc...)
> > - by 0x........: main (sendmsg.c:45)
> > + at 0x........: sendmsg$UNIX2003 (in /...libc...)
> > + ...
> > Address 0x........ is on thread 1's stack
> >
> > sendmsg: 6
>
> Aha, thanks. So the test works as intended and even produces the right
> error, but the backtrace needs some munging to make it look more like
> what is in the .exp file. You should be able to add a new rule like
> "s/sendmsg\$UNIX2003/sendmsg/" to memcheck/tests/filter_stderr to make
> it so (or maybe something a bit more generic if there are other foobar
> $UNIXxxxx symbols).
>
> Cheers,
>
> Mark
>
--
Rich Coe rc...@wi...
|