Need Help

Need Help

[Amarnath]

Hi,

I have got your IDs from the mailing list of redhat insight sources.

I am in need of installation package of bison version 1.875b. Could you
please help me in letting me know the site address / link from which I
can get the installation package of the same.

I am in need of it urgently. So please help me in this aspect.

Thank you.

With Regards,
Amarnath M

Make breakpoint pending

[Guillaume Savaton]

Hi,

I have been using previous versions of insight (5.x) for a while, and I have
noticed a new message in 6.4 when I attempt to run a program :
"Make breakpoint pending on future shared library load?"

This message is quite annoying since I am asked to confirm for each breakpoint
in my program individually.

Moreover, most of the time, I use insight to debug simple bare-metal programs
where there are no "shared libraries" at all. So the question is not relevant in
my case.

Is there a possibility to tell insight not to ask ?

Thanks

Guillaume Savaton

Re: Make breakpoint pending

[Keith Seitz]

Guillaume Savaton wrote:
> Hi,
> 
> I have been using previous versions of insight (5.x) for a while, and I 
> have
> noticed a new message in 6.4 when I attempt to run a program :
> "Make breakpoint pending on future shared library load?"

Yeah, that happens when gdb is asked to set a breakpoint someplace that 
it knows nothing about (so it assumes that you have not made a mistake 
but that the location exists in a shared library).

> This message is quite annoying since I am asked to confirm for each 
> breakpoint
> in my program individually.

Okay, I have not seen this before, and I cannot reproduce it with any 
testcase I've managed to devise. If you look at the breakpoints which 
were marked pending, are they part of your program? [While testing this, 
I just ran into a problem where I was running in the build directory. 
Insight added the breakpoints from .gdbinit to its list of installed 
breakpoints for the executable. Those buggers just coming out of the 
woodwork!]

> Is there a possibility to tell insight not to ask ?

Yes, you can add it to the ignored warnings list or teach insight that 
it is an ignorable warning (so it will only ask once). See 
gdbtk_tcl_warning in interface.tcl to globally ignore the warning. 
Unfortunately, teaching insight to optionally ignore the warning, i.e., 
"Don't show me this warning again", is slightly non-trivial (or at least 
more painful). GDB never did clean up the warning/error landscape. All 
we know is that it is a warning.

Ignoring gdb, we'd have to add logic in place (similar to 
gdbtk_tcl_warning) to flag warnings are optionally ignorable. [Gdb just 
still doesn't get it.]

Keith

Re: Make breakpoint pending

[Guillaume Savaton]

Keith Seitz wrote :
>> This message is quite annoying since I am asked to confirm for each 
>> breakpoint in my program individually.
> 
> If you look at the breakpoints which 
> were marked pending, are they part of your program?

Well, for instance, I have a "start" label at the beginning of my program.
Usually, in the "target settings" dialog, I tell insight to place a breakpoint
at label "start".

It is the only breakpoint set in my program, and it actually belongs to the program.
But before running, the "make breakpoint pending..." message pops up.

N.B: I am using insight compiled for arm-elf targets.
All my programs are written in assembly language.
The executable is generated using GNU as and ld.
The same behavior has been noticed whether the debug target is set to
"Simulator" or "Remote TCP"

 >> Is there a possibility to tell insight not to ask ?
 >
 > Yes, you can add it to the ignored warnings list or teach insight that
 > it is an ignorable warning (so it will only ask once). See
 > gdbtk_tcl_warning in interface.tcl to globally ignore the warning.

I will try this.

Thanks for your help.

Guillaume Savaton

Re: Make breakpoint pending

[Keith Seitz]

Guillaume Savaton wrote:

> All my programs are written in assembly language.
> The executable is generated using GNU as and ld.

Ah, okay, now I'm beginning to see what is happening.

If you type "break _start" (or whatever your label is called), I'll bet 
gdb asks you the same question. You really need to use "break *_start", 
since _start is only a LABEL, not a FUNCTION. [Okay, I am assuming you 
didn't write your assembler code make the distinction.]

Keith

Re: Make breakpoint pending

[Guillaume Savaton]

Keith Seitz a écrit :
> Ah, okay, now I'm beginning to see what is happening.

Well, it's still confused for me :
when debugging with plain gdb and typing "break start", there is no such warning !

Guillaume Savaton

Re: Make breakpoint pending

[Keith Seitz]

Guillaume Savaton wrote:
> Keith Seitz a écrit :
>> Ah, okay, now I'm beginning to see what is happening.
> 
> Well, it's still confused for me :
> when debugging with plain gdb and typing "break start", there is no such 
> warning !

Doh! Oh well, it was a good guess. In past lives, I've seen things like 
this happen with insight/gdb when people have hand-written assembly.

Still, it might not hurt to try entering your break as "*start" in the 
target preferences. If it still doesn't work, there's always the "ignore 
it!" option.

Sorry,
Keith