Re: attach.exp failure

[Michael Elizabeth Chastain <mec@shout.net>]

Well now that's interesting.

  attach 1065
  Attaching to program: /gdb/mirror/src/gdb/testsuite/gdb.base/attach2, process 1065
  Reading symbols from /lib/i686/libm.so.6...done.
  Loaded symbols for /lib/i686/libm.so.6
  Reading symbols from /lib/i686/libc.so.6...done.
  Loaded symbols for /lib/i686/libc.so.6
  Reading symbols from /lib/ld-linux.so.2...done.
  Loaded symbols for /lib/ld-linux.so.2
  0x420b0205 in nanosleep () from /lib/i686/libc.so.6
  (gdb) PASS: gdb.base/attach.exp: attach call
  i r r3
  ERROR: Couldn't send p should_exit = 1 to GDB.
  UNRESOLVED: gdb.base/attach.exp: p should_exit = 1
  ERROR: Couldn't send c to GDB.
  UNRESOLVED: gdb.base/attach.exp: c

Looks like gdb crashed or went into a loop.  Probably crashed because
if it went into a loop, the test would take some long $TIMEOUT time
to run.  Are there any funky core files lying around?

The target program is inside nanosleep, probably called from sleep, and
I know that gdb screws up the frame info when it backtraces through
sleep.

But I have never gotten this behavior.  The connection between
'nanosleep' and a crash is just too tenuous.  Your problem has the
word 'nanosleep' in it but I just don't see a smoking gun linking
it to the 'sleep' frame info problem.  :(
 
What vendor/version of linux are you running on?

Can you disassemble 'sleep' from libc.so and post the first few
instructions?

Want to try this patch?

  http://sources.redhat.com/ml/gdb-patches/2003-08/msg00053.html