From: Julian Seward <jseward@acm.org>
Date: Wed, 15 Jun 2011 15:11:50 +0000 (+0000)
Subject: Add debugging hack, if 0'd, to print a stack trace at each syscall.
X-Git-Tag: svn/VALGRIND_3_7_0~422
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=01abe3ed83e310366791880c29157cb733677e7a;p=thirdparty%2Fvalgrind.git

Add debugging hack, if 0'd, to print a stack trace at each syscall.


git-svn-id: svn://svn.valgrind.org/valgrind/trunk@11815
---

diff --git a/coregrind/m_syswrap/syswrap-main.c b/coregrind/m_syswrap/syswrap-main.c
index 0757a41775..f50a57b158 100644
--- a/coregrind/m_syswrap/syswrap-main.c
+++ b/coregrind/m_syswrap/syswrap-main.c
@@ -1429,6 +1429,13 @@ void VG_(client_syscall) ( ThreadId tid, UInt trc )
       is interrupted by a signal. */
    sysno = sci->orig_args.sysno;
 
+   /* It's sometimes useful, as a crude debugging hack, to get a
+      stack trace at each (or selected) syscalls. */
+   if (0 /* sysno == __NR_mmap */) {
+      VG_(umsg)("\n");
+      VG_(get_and_pp_StackTrace)(tid, 10);
+   }
+
 #  if defined(VGO_darwin)
    /* Record syscall class.  But why?  Because the syscall might be
       interrupted by a signal, and in the signal handler (which will