[PATCH-ping] VAX: Forced decoding of function entry masks (eg. for disassembling ROM images)

Nick Clifton nickc@redhat.com
Tue Mar 29 19:12:00 GMT 2005


Hi Jan-Benedict,

> Basically, I of course like the idea of parsing the options only
> once.  OTOH, this is one more location that's leaking. Maybe the
> time is come to *really* implement an initializer and termination
> function for the disassembler backends? But I'm not sure if it's
> worth it. Disassembling used to be a slow process, and it's a rare
> situation. Even on a VAX, I'd accept that it takes some time[tm]
> to disassemble something.  ...and with today's PCs, the overhead
> of parsing the options each time /one/ opcode is being
> disassembled is irrelevant.
> 
> So the pros and cons of your implementation are:
> 
> 	+ faster
> 	- leaks memory
> 	- modifies disassembler_info->disassembler_options
> 
> If somebody would implement a init/fini function for the backend, that
> would turn out to only being faster with no drawbacks...
> 
> Being somewhat undecided, I'd say let's push in your version and start
> a new thread about init/fini functions for the disassembler backend.
> I'll push it through the compiler in the mean time.

Ok - I have pushed my version, with a few fixes to address all of the 
issues you mentioned, except the memory leak.  (See the attached patch). 
  I think that this leak is OK for now, but we should not forget it. 
One day (soon I hope), I'll write the code to add the necessary hooks, 
so that the leak can be removed.

Cheers
   Nick


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: vax.dis.patch
URL: <https://sourceware.org/pipermail/binutils/attachments/20050329/b545809a/attachment.ksh>


More information about the Binutils mailing list