This is the mail archive of the gdb@sources.redhat.com 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: Supplying regsets containing pseudoregisters


Jim Blandy <jimb@redhat.com> writes:

> Andrew Cagney <cagney@gnu.org> writes:
> > It sounds like the regcache layout needs to be changed.  I'd break the
> > raw EV registers down into hi:lo pairs so that each part can be
> > supplied separatly - this appears to more closely reflect the
> > underlying implementation.
> 
> Okay.  This is what I'd started doing thinking of it as a workaround,
> so it works for me.

Actually, maybe not.

The sim provides access to the GPRs as full 64-bit values.  It calls
sim_fetch_register, gets 64 bits, and then calls supply_register on
the full 64 bits.

But core files and libthread_db operate on gregset_t values, which
contain only the lower 32 bits of the GPRs.  The regset 'supply'
functions want to call regcache_raw_supply on only the lower halves.

So does gdbsim_fetch_register need to be tweaked somehow?  How would
it know how to distribute the 64-bit value across high/low pairs?

Do I need to use separate regcache layouts for sim and native?  How
would I even do that?


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