Files
ftmemsim-valgrind/coregrind/m_stacktrace.c
Nicholas Nethercote 98386a7102 Fix up some stack trace inconsistencies:
- When printing suppressions, never print more entries than there are in the
  stack.  This avoids bogus suppressions in some cases!  (I haven't seen
  them on Linux, but I have seen them on Darwin.)

- When getting a stack trace, stop if we get an IP of zero or one;  that
  means we've hit the end of the stack.  And don't include that entry in the
  stack trace, because it's a guaranteed "???" if it's ever printed which is
  useless.

- In VG_(apply_StackTrace), we can now rely entirely on the n_ip parameter
  rather than looking for 0 or -1, because that check is done when the stack
  trace is first obtained.  In other words, stack traces all use an n_ip
  parameter to record their size, whereas previously they used an odd
  mixture of n_ip and null-termination.

- Rename 'n_ips' variables as 'max_n_ips' where appropriate;  those left as
  'n_ips' truly describe how many IPs there are in the stack trace.


git-svn-id: svn://svn.valgrind.org/valgrind/trunk@9793
2009-05-07 23:08:10 +00:00

22 KiB