Re: [Qemu-devel] [PATCH 1/14] target/mips: Fix MIPS64 MFC0 UserLocal on BE host

[Aurelien Jarno]

On 2017-07-19 14:44, James Hogan wrote:
> On Wed, Jul 19, 2017 at 12:27:50PM +0200, Aurelien Jarno wrote:
> > On 2017-07-18 12:55, James Hogan wrote:
> > > Using MFC0 to read CP0_UserLocal uses tcg_gen_ld32s_tl, however
> > > CP0_UserLocal is a target_ulong. On a big endian host with a MIPS64
> > > target this reads and sign extends the more significant half of the
> > > 64-bit register.
> > > 
> > > Fix this by using ld_tl to load the whole target_ulong and ext32s_tl to
> > > sign extend it, as done for various other target_ulong COP0 registers.
> > > 
> > > Fixes: d279279e2b5c ("target-mips: implement UserLocal Register")
> > > Signed-off-by: James Hogan <address@hidden>
> > > Cc: Yongbok Kim <address@hidden>
> > > Cc: Aurelien Jarno <address@hidden>
> > > Cc: Petar Jovanovic <address@hidden>
> > > ---
> > > Changes in v2:
> > > - New patch.
> > > ---
> > >  target/mips/translate.c | 5 +++--
> > >  1 file changed, 3 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/target/mips/translate.c b/target/mips/translate.c
> > > index 3022f349cb2a..556aba969a12 100644
> > > --- a/target/mips/translate.c
> > > +++ b/target/mips/translate.c
> > > @@ -5138,8 +5138,9 @@ static void gen_mfc0(DisasContext *ctx, TCGv arg, 
> > > int reg, int sel)
> > >              goto cp0_unimplemented;
> > >          case 2:
> > >              CP0_CHECK(ctx->ulri);
> > > -            tcg_gen_ld32s_tl(arg, cpu_env,
> > > -                             offsetof(CPUMIPSState, 
> > > active_tc.CP0_UserLocal));
> > > +            tcg_gen_ld_tl(arg, cpu_env,
> > > +                          offsetof(CPUMIPSState, 
> > > active_tc.CP0_UserLocal));
> > > +            tcg_gen_ext32s_tl(arg, arg);
> > >              rn = "UserLocal";
> > >              break;
> > >          default:
> > 
> > I think this is what gen_mfc0_load64() does, that said this whole area
> 
> Ah yes, that could do with some wider use (and possibly s/64/tl/ or
> something).
> 
> > probably need a rework (see for example how inefficiently
> > gen_mfc0_load32 is implemented). So:
> 
> Erm, doesn't gen_mfc0_load32() fail to sign extend as it should when
> used for mfc0...?

Yes, this is correct. tcg_gen_ext_i32_tl sign-extends the value when
converting it from TCGv_i32 to TCGv_tl. tcg_gen_extu_i32_tl is the op
which zero-extends the value. Basically TCG sign-extends the value "by
default", like for MIPS.

Aurelien

-- 
Aurelien Jarno                          GPG: 4096R/1DDD8C9B
address@hidden                 http://www.aurel32.net

signature.asc
Description: PGP signature