Re: gdb + sigtraps

[Patrick Welche <prlw1%cam.ac.uk@localhost>]

On Mon, Mar 26, 2012 at 12:33:13PM -0400, Christos Zoulas wrote:
> On Mar 26,  5:11pm, prlw1%cam.ac.uk@localhost (Patrick Welche) wrote:
> -- Subject: Re: gdb + sigtraps
> 
> | I do - I did a update build with head - maybe I need to clear obj out...
> | 
> | quark: {1} gdb /usr/bin/dig
> | GNU gdb (GDB) 7.3.1
> | ...
> | Reading symbols from /usr/bin/dig...(no debugging symbols found)...done.
> | (gdb) break main
> | Breakpoint 1 at 0x408b59
> | (gdb) run
> | Starting program: /usr/bin/dig 
> | 
> | Breakpoint 1, 0x0000000000408b59 in main ()
> | (gdb) c
> | Continuing.
> | 
> | Program received signal SIGTRAP, Trace/breakpoint trap.
> | 0x00007f7ff58390ba in _sys___kevent50 () from /usr/lib/libc.so.12
> | (gdb) 
> 
> Ok, I do too now. Good. Let me see what's wrong now. Hopefully I have
> my printf filled files still around.

I just compared an MP amd64 of 31st March with an MP i386 of 2nd April, using
the same gdb /usr/bin/dig example.

On run, they both receive a SIGTRAP
After c, the amd64 gets another SIGRAP, but the i386 apparently runs
"RXl+", but dig consumes over 80% of one the cpus.  DiG 9.9.0a1 is never
printed. dig simply consumes cpu cycles. ctrl-C in an attempt to see
what dig is doing just prints "^C". kill -HUP the dig process then
shows (well, from the beginning on the i386):

(gdb) run
Starting program: /usr/bin/dig 

Program received signal SIGTRAP, Trace/breakpoint trap.
0xbb71f377 in _sys___kevent50 () from /usr/lib/libc.so.12
(gdb) c
Continuing.
^C
Program received signal SIGCONT, Continued.
0xbb71f377 in _sys___kevent50 () from /usr/lib/libc.so.12


On the amd64, "continuing" several times, one eventually sees the output
of dig... (though it is still impossible to step through main)


Cheers,

Patrick