Re: Can't debug bash with gdb 8.2.1-1

[Ken Brown <kbrown at cornell dot edu>]

On 7/24/2019 10:32 AM, Ken Brown wrote:
> On 7/24/2019 8:33 AM, Ken Brown wrote:
>> $ cat getpgrp_test.c
>> #include <unistd.h>
>> #include <stdio.h>
>> #include <errno.h>
>>
>> int
>> main ()
>> {
>>      pid_t pid = getpid ();
>>      pid_t pgid = getpgid (0);
>>      printf ("getpid () = %d, getpgid (0) = %d, errno = %d\n", pid, pgid, errno);
>> }
>>
>> $ gcc -o getpgrp_test -Wall -g -O0 getpgrp_test.c
>>
>> $ ./getpgrp_test.exe
>> getpid () = 1424, getpgid (0) = 1424, errno = 0
>>
>> $ gdb ./getpgrp_test.exe
>> GNU gdb (GDB) (Cygwin 8.2.1-1) 8.2.1
>> [...]
>> (gdb) r
>> Starting program: /home/kbrown/Documents/programming/misc_examples/getpgrp_test.exe
>> [...]
>> getpid () = 1427, getpgid (0) = -1, errno = 0
>> [...]
>> [Inferior 1 (process 23872) exited normally]
>>
>> So getpgid (0) returns -1 without setting errno.  I think this means that
>> pinfo::init() didn't fail, but for some reason myself->pgid == -1.
> 
> I just ran the above gdb session under strace and found several lines like this:
> 
>      25 32828861 [main] gdb 1799 tcsetpgrp: 0 = tcsetpgrp(0, -1)
> 
> I haven't looked at the gdb code, but doesn't this mean that gdb is trying to
> set pgrp to -1?  If so, why?  And shouldn't tcsetpgrp return -1 with EINVAL?

I just looked at the gdb commit that Jon pointed to earlier in the thread.  It 
adds calls like this:

   result = tcsetpgrp (0, getpgid (inf->pid));

It then checks for errors and takes appropriate action.  But tcsetpgrp returns 
success.

So this all seems to boil down to the fact that Cygwin's tcsetpgrp() doesn't 
check the validity of its second argument.

Ken

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple