This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: RFC: Correct "paddr_t" in gdb_proc_service.h


On Wed, Mar 15, 2006 at 11:34:07PM +0100, Mark Kettenis wrote:
> > Date: Wed, 15 Mar 2006 11:36:50 -0500
> > From: Daniel Jacobowitz <drow@false.org>
> > 
> > If there is no <proc_service.h>, GDB will typedef unsigned long to paddr_t,
> > which is then used in various libthread_db interfaces.  Even when this code
> > was designed to work on Solaris systems in addition to GNU/Linux, this
> > wasn't right; Solaris doesn't provide a paddr_t that has anything to do with
> > libthread_db, only an unrelated one that deals with physical addressing.
> 
> Yes indeed.  On BSD paddr_t is an usnigned integer type used to store
> physical memory addresses
> 
> > The type that glibc's libthread_db uses is psaddr_t, which is a pointer
> > type.
> > 
> > When I wrote this patch (last year) I went through the original changelog
> > entries for this file from 2000; they suggest that this change is correct
> > and there should be no paddr_t references in GDB.
> 
> I think that's right, and therefore I don't understand your change at all :(.

Well, we do have paddr_t references.  Want me to replace them all with
psaddr_t references while I'm here?  (The same casts will still be
necessary, since again, psaddr_t is a pointer type.)

-- 
Daniel Jacobowitz
CodeSourcery


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]