New ARI warning Wed Jun 23 01:54:57 UTC 2010

New ARI warning Wed Jun 23 01:54:57 UTC 2010

[GDB Administrator]

493a494,495
> gdb/i386-tdep.c:543: deprecate: read_memory: Replace read_memory() with regcache_read() et.al.
gdb/i386-tdep.c:543:  read_memory (from, buf, len);
> gdb/i386-tdep.c:557: deprecate: write_memory: Replace write_memory() with regcache_read() et.al.
gdb/i386-tdep.c:557:  write_memory (to, buf, len);

Re: New ARI warning Wed Jun 23 01:54:57 UTC 2010

[Daniel Jacobowitz]

On Wed, Jun 23, 2010 at 01:54:57AM +0000, GDB Administrator wrote:
> 493a494,495
> > gdb/i386-tdep.c:543: deprecate: read_memory: Replace read_memory() with regcache_read() et.al.
> gdb/i386-tdep.c:543:  read_memory (from, buf, len);
> > gdb/i386-tdep.c:557: deprecate: write_memory: Replace write_memory() with regcache_read() et.al.
> gdb/i386-tdep.c:557:  write_memory (to, buf, len);

Am I missing something here?  Why would you replace read_memory with
regcache_read?

-- 
Daniel Jacobowitz
CodeSourcery

RE: New ARI warning Wed Jun 23 01:54:57 UTC 2010

[Pierre Muller]

> -----Message d'origine-----
> De : gdb-patches-owner@sourceware.org [mailto:gdb-patches-
> owner@sourceware.org] De la part de Daniel Jacobowitz
> Envoyé : Thursday, June 24, 2010 4:56 PM
> À : gdb-patches@sourceware.org
> Objet : Re: New ARI warning Wed Jun 23 01:54:57 UTC 2010
> 
> On Wed, Jun 23, 2010 at 01:54:57AM +0000, GDB Administrator wrote:
> > 493a494,495
> > > gdb/i386-tdep.c:543: deprecate: read_memory: Replace read_memory()
> with regcache_read() et.al.
> > gdb/i386-tdep.c:543:  read_memory (from, buf, len);
> > > gdb/i386-tdep.c:557: deprecate: write_memory: Replace
> write_memory() with regcache_read() et.al.
> > gdb/i386-tdep.c:557:  write_memory (to, buf, len);


  I don't really know the origin of this rule,
but if you look up the sources, you will see that 'read_memory' 
is defined in gdbcore.h

  So this function should be specific for core management, no?

  Maybe it would be better to suggest using target_read_memory 
function in this rule.

Pierre

Re: New ARI warning Wed Jun 23 01:54:57 UTC 2010

[Daniel Jacobowitz]

On Thu, Jun 24, 2010 at 05:14:12PM +0200, Pierre Muller wrote:
>   I don't really know the origin of this rule,
> but if you look up the sources, you will see that 'read_memory' 
> is defined in gdbcore.h
> 
>   So this function should be specific for core management, no?

I don't see why... if it is, it should be renamed.

-- 
Daniel Jacobowitz
CodeSourcery

Re: New ARI warning Wed Jun 23 01:54:57 UTC 2010

[Ulrich Weigand]

Dan Jacobowitz:
> On Wed, Jun 23, 2010 at 01:54:57AM +0000, GDB Administrator wrote:
> > 493a494,495
> > > gdb/i386-tdep.c:543: deprecate: read_memory: Replace read_memory() with regcache_read() et.al.
> > gdb/i386-tdep.c:543:  read_memory (from, buf, len);
> > > gdb/i386-tdep.c:557: deprecate: write_memory: Replace write_memory() with regcache_read() et.al.
> > gdb/i386-tdep.c:557:  write_memory (to, buf, len);
> 
> Am I missing something here?  Why would you replace read_memory with
> regcache_read?

The ARI claims "read_memory" and "write_memory" as "to be deprecated"
interfaces.  These were added by Andrew Cagney 2003-05-24:
http://sources.redhat.com/ml/gdb/2003-05/msg00341.html
at about the time he introduced get_frame_memory:
http://sourceware.org/ml/gdb-patches/2003-06/msg00104.html

The idea seems to have been that all memory access was supposed to
identify the process/thread/target where the memory resided by
means of passing a frame.

However, this was never really fully pursues, it seems.  I'm not
really sure it still makes sense to keep these ARI entries ...

Bye,
Ulrich

-- 
  Dr. Ulrich Weigand
  GNU Toolchain for Linux on System z and Cell BE
  Ulrich.Weigand@de.ibm.com

Re: New ARI warning Wed Jun 23 01:54:57 UTC 2010

[Michael Snyder]

Daniel Jacobowitz wrote:
> On Thu, Jun 24, 2010 at 05:14:12PM +0200, Pierre Muller wrote:
>>   I don't really know the origin of this rule,
>> but if you look up the sources, you will see that 'read_memory' 
>> is defined in gdbcore.h
>>
>>   So this function should be specific for core management, no?
> 
> I don't see why... if it is, it should be renamed.
> 

read_memory is widely used in other parts of gdb.