Re: mips-tdep.c: Fix sw watchpoint-out-of-scope events

[Daniel Jacobowitz <drow@false.org>]

On Tue, Sep 11, 2007 at 05:23:54PM +0100, Maciej W. Rozycki wrote:
>  This is a fix for software watchpoint-out-of-scope events for MIPS 
> platforms.  It implements a in_function_epilogue_p() hook which is used to 
> single-step a function's epilogue once a local watchpoint has been 
> destroyed (because of the restoration of the frame pointer).  As a result 
> the event is only reported once the caller has been reached back.

This patch is fine.  I wish it wasn't necessary, but for now it is.

> for the remote targets.  It does not fix them for Linux, because the o32 
> calling convention makes the restoration of the GOT pointer in the caller 
> ("lw gp,..." after "jalr") a part of the function call.  Therefore 
> execution stops in the right function, but one instruction too early, 
> still within the source line of the call.  It is still a step forward 
> though and less of nuisance for the user.

Hmm, do you think we should accept that line as a pass then?  The same
will happen on other architectures too.

-- 
Daniel Jacobowitz
CodeSourcery