This is the mail archive of the gdb@sources.redhat.com mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Debuggee cmd-line arguments bug


On Mon, Jul 25, 2005 at 05:59:31PM +0400, Konstantin Karganov wrote:
> Hello.
> 
> Playing with the program startup arguments I've found the following 
> "feature" of gdb. I have a program that prints its argc/argv parameters.
> 
> (gdb) set args a b c d
> (gdb) run
> Starting program: /home/kostik/Work/a.out a b c d
> Argc=5, Argv=/home/kostik/Work/a.out,a,b,c,d
>  
> Program exited normally.
> (gdb) set args a b c d
> (gdb) run >out
> Starting program: /home/kostik/Work/a.out >out
> 
> Program exited normally.
> (gdb)
> 
> In file out there are the folowing:
> 
> Argc=1, Argv=/home/kostik/Work/a.out
> 
> So when trying to redirect debuggee IO we've lost all cmd-line parameters.
> The same of cource happens with MI (and breaks the debugging startup for 
> MPI program!).
> 
> Is it a bug?

No.  If you provide any options on the "run" command line, you are
replacing the arguments set by set args.  I think the text in section
4.3, "Your program's arguments", is pretty clear; please let us know if
it isn't.

> And what is about my previous posts (still unanswered) on:

Please don't reply to list messages to post to the list, by the way. 
It messes up threading; you can see the problem in the list archives.

> - running out of source code (stepping past main)
>   http://sources.redhat.com/ml/gdb/2005-07/msg00152.html

Your assumption that with step-mode set the PC will always be in user
code is incorrect.  For instance, you could be stopped by a signal, or
by a user interrupt, or by a breakpoint in a function without source
code available.

> - incorrect breakpoint diagnostics in MI
>   http://sources.redhat.com/ml/gdb/2005-07/msg00175.html

If someone knew the answer, they would have responded.  Sorry, that
just happens sometimes.

> - vanishing stopping reason in MI
>   http://sources.redhat.com/ml/gdb/2005-07/msg00173.html
> 
> "And there's no doing anything about it" ?

You _got_ an answer.  Bob agreed that there ought to be a reason.

-- 
Daniel Jacobowitz
CodeSourcery, LLC


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]