clearing call stack when a new binary file is loaded

clearing call stack when a new binary file is loaded

[Jiju George T]

Hi,

  We are developing a graphical embedded debugger based on GDB mi interface.
UI part is based on Eclipse CDT and it communicates with GDB through mi
interface. We are facing a problem with call stacks after a new binary is
loaded to GDB. Our UI relays on the call stack information provided by GDB
to show the debug context. 

Steps to reproduce the issues: 

  1) Start GDB and connect to remote GDB server which is connected to an
embedded target. The embedded target is waiting at PC 0xBFC00000
  2) Now load an ELF file to GDB with entry point as 0xA0000000
  3) Now GDB shows $pc register value as 0xA0000000. But the call stack
still shows frames with 0xBFC00000
  
 
Is this GDB behaviour by design? Is there any mi interface to clear the
stack frame and update to the current $pc value? 

Thanks,
Jiju George

Re: clearing call stack when a new binary file is loaded

[Vladimir Prus]

Jiju George T wrote:

> Hi,
> 
> We are developing a graphical embedded debugger based on GDB mi interface.
> UI part is based on Eclipse CDT and it communicates with GDB through mi
> interface. We are facing a problem with call stacks after a new binary is
> loaded to GDB. Our UI relays on the call stack information provided by GDB
> to show the debug context.
> 
> Steps to reproduce the issues:
> 
>   1) Start GDB and connect to remote GDB server which is connected to an
> embedded target. The embedded target is waiting at PC 0xBFC00000
>   2) Now load an ELF file to GDB with entry point as 0xA0000000
>   3) Now GDB shows $pc register value as 0xA0000000. But the call stack
> still shows frames with 0xBFC00000
>   
>  
> Is this GDB behaviour by design? Is there any mi interface to clear the
> stack frame and update to the current $pc value?

Which version of GDB?

- Volodya

RE: clearing call stack when a new binary file is loaded

[Jiju George T]

I am using a cross compiled version of GDB for MIPS architecture
(mipsel-linux-gdb) compiled using cross tools. It is based on GNU GDB
version 6.6

Thanks,
Jiju

-----Original Message-----
From: gdb-owner@sourceware.org [mailto:gdb-owner@sourceware.org] On Behalf
Of Vladimir Prus
Sent: 23 April 2008 11:52
To: gdb@sources.redhat.com
Subject: Re: clearing call stack when a new binary file is loaded

Jiju George T wrote:

> Hi,
> 
> We are developing a graphical embedded debugger based on GDB mi interface.
> UI part is based on Eclipse CDT and it communicates with GDB through mi
> interface. We are facing a problem with call stacks after a new binary is
> loaded to GDB. Our UI relays on the call stack information provided by GDB
> to show the debug context.
> 
> Steps to reproduce the issues:
> 
>   1) Start GDB and connect to remote GDB server which is connected to an
> embedded target. The embedded target is waiting at PC 0xBFC00000
>   2) Now load an ELF file to GDB with entry point as 0xA0000000
>   3) Now GDB shows $pc register value as 0xA0000000. But the call stack
> still shows frames with 0xBFC00000
>   
>  
> Is this GDB behaviour by design? Is there any mi interface to clear the
> stack frame and update to the current $pc value?

Which version of GDB?

- Volodya

Re: clearing call stack when a new binary file is loaded

[Daniel Jacobowitz]

On Wed, Apr 23, 2008 at 02:51:35PM +0400, Vladimir Prus wrote:
> Which version of GDB?

I expect it still does this.  write_pc doesn't invalidate the frame
cache, so we should manually do so at the end of generic_load.

-- 
Daniel Jacobowitz
CodeSourcery

RE: clearing call stack when a new binary file is loaded

[Jiju George T]

Is there any equivalent mi command to invalidate the frame cache? 

Thanks,
Jiju

-----Original Message-----
From: gdb-owner@sourceware.org [mailto:gdb-owner@sourceware.org] On Behalf
Of Daniel Jacobowitz
Sent: 23 April 2008 12:46
To: Vladimir Prus
Cc: gdb@sources.redhat.com
Subject: Re: clearing call stack when a new binary file is loaded

On Wed, Apr 23, 2008 at 02:51:35PM +0400, Vladimir Prus wrote:
> Which version of GDB?

I expect it still does this.  write_pc doesn't invalidate the frame
cache, so we should manually do so at the end of generic_load.

-- 
Daniel Jacobowitz
CodeSourcery

Re: clearing call stack when a new binary file is loaded

['Daniel Jacobowitz']

On Wed, Apr 23, 2008 at 01:53:46PM +0100, Jiju George T wrote:
> Is there any equivalent mi command to invalidate the frame cache? 

No.  If you use GDB 6.8 the CLI command "flushregs" will hack around
the problem but we simply need to fix it.

-- 
Daniel Jacobowitz
CodeSourcery

RE: clearing call stack when a new binary file is loaded

[Jiju George T]

Will it be fixed in the next release? Do I have to file a bug on this? 

-----Original Message-----
From: 'Daniel Jacobowitz' [mailto:drow@false.org] 
Sent: 23 April 2008 13:57
To: Jiju George T
Cc: 'Vladimir Prus'; gdb@sources.redhat.com
Subject: Re: clearing call stack when a new binary file is loaded

On Wed, Apr 23, 2008 at 01:53:46PM +0100, Jiju George T wrote:
> Is there any equivalent mi command to invalidate the frame cache? 

No.  If you use GDB 6.8 the CLI command "flushregs" will hack around
the problem but we simply need to fix it.

-- 
Daniel Jacobowitz
CodeSourcery