Re: [RFA] mips 32/64 register/stack fix

[Andrew Cagney <cagney at gnu dot org>]

> Andrew Cagney wrote:
>
> > > #0  register_size (gdbarch=0x84093e8, regnum=120)
> >
> >
> >
> > 120 looks large enough to be cooked (but to confirm this can you look at "maint print raw-registers".  This will also tell us the register's type, and hence confirm the size.
>
>
> Yes, raw is from 0 to 89, and cooked is from 90 to 179.
> regcache->descr->sizeof_register is as follows:
> 8 for all raw
> 8 for the first 70 cooked
> 4 for the last 20 cooked
>
> register_type looks like:
> raw:
> 38 @ int64
> 32 @ double64
> 20 @ int64
> cooked:
> 38 @ int64
> 32 @ double64
> 20 @ int32

good.

> > Given that you see 8, this suggests that the register's type is wrong (see mips_register_type).
>
>
> It looks to me as if mips_register_type relies on mips_regsize.
> Which returns 8.  Yep, and mips_register_type (120) returns int64_t.
>
> But mips_saved_regsize (tdep) returns 4 (as it should, I guess).
>
> > Also, what information is available in the object file header?
>
>
> Can you be more specific?
[...]
> Here's your tdep info:
> {elf_flags = 1610625025, mips_abi = MIPS_ABI_EABI32,

I was looking to see if anything like that was present -> it is clearly 
identified as a 32-bit ABI binary.

I think the bug is in mips_register_type, the tail end should probably 
be changed to read something like:

  else if (regnum >= NUM_REGS
	   && gdbarch_tdep (gdbarch)->mips64_transfers_32bit_regs_p)
    /* The target, while using a 64-bit raw register buffer, is only
       transfering 32-bits of each integer register.  Reflect this in
       the cooked/pseudo register value.  */
    return builtin_type_int32;
  else if (regnum > NUM_REGS
	   && mips_saved_regsize (gdbarch) == 4)
    /* A 32-bit ABI such as o32 possibly running on a 64-bit ISA.  */
    return builtin_type_int32;
  else if (mips_regsize (gdbarch) == 8)
    /* 64-bit ISA.  */
    return builtin_type_int64;
  else
    /* 32-bit ISA.  */
    return builtin_type_int32;

Andrew