Re: [RHEL7 qemu-kvm PATCH 3/3] Fix tcg_out_op argument mismatch warning

[Miroslav Rezanina]

----- Original Message -----
> From: "Philippe Mathieu-Daudé" <philmd@redhat.com>
> To: mrezanin@redhat.com, qemu-devel@nongnu.org
> Cc: "Richard Henderson" <richard.henderson@linaro.org>, "Eric Blake" 
> <eblake@redhat.com>
> Sent: Monday, January 11, 2021 1:15:04 PM
> Subject: Re: [RHEL7 qemu-kvm PATCH 3/3] Fix tcg_out_op argument mismatch 
> warning
> 
> On 1/11/21 12:30 PM, mrezanin@redhat.com wrote:
> > From: Miroslav Rezanina <mrezanin@redhat.com>
> > 
> > There's prototype mismatch between tcg/tcg.c and
> > tcg/aarch/tcg-target.c.inc:
> > 
> > tcg.c:
> > 
> >     static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg
> >     *args,
> >                            const int *const_args);
> > 
> > tcg-target.c.inc:
> > 
> >     static void tcg_out_op(TCGContext *s, TCGOpcode opc,
> >                            const TCGArg args[TCG_MAX_OP_ARGS],
> >                            const int const_args[TCG_MAX_OP_ARGS])
> > 
> > This missmatch cause warnings on GCC 11:
> > 
> >     tcg/aarch64/tcg-target.c.inc:1855:37: error: argument 3 of type 'const
> >     TCGArg[16]' {aka 'const long unsigned int[16]'} with mismatched bound
> >     [-Werror=array-parameter=]
> >     tcg/aarch64/tcg-target.c.inc:1856:34: error: argument 4 of type 'const
> >     int[16]' with mismatched bound [-Werror=array-parameter=]
> 
> TIL. Interesting, compilers are getting smarter :)
> 
> > Only architectures with this definition are aarch and sparc. Fixing both
> > archs to use
> > proper argument type.
> > ---
> >  tcg/aarch64/tcg-target.c.inc | 3 +--
> >  tcg/sparc/tcg-target.c.inc   | 3 +--
> >  2 files changed, 2 insertions(+), 4 deletions(-)
> > 
> > diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
> > index 26f71cb599..fe6bdbf721 100644
> > --- a/tcg/aarch64/tcg-target.c.inc
> > +++ b/tcg/aarch64/tcg-target.c.inc
> > @@ -1852,8 +1852,7 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg
> > data_reg, TCGReg addr_reg,
> >  static tcg_insn_unit *tb_ret_addr;
> >  
> >  static void tcg_out_op(TCGContext *s, TCGOpcode opc,
> > -                       const TCGArg args[TCG_MAX_OP_ARGS],
> > -                       const int const_args[TCG_MAX_OP_ARGS])
> > +                       const TCGArg *args, const int *const_args)
> 
> Doing this way we loose information (that the array pointed has
> TCG_MAX_OP_ARGS elements). What about letting this prototype and
> fix the other uses?

I'm not author of the code so I went with smaller change - forward definition
in tcg.c and most tct-target.c.inc use this form. I would need someone more
familiar with this part to clarify whether it's ok to go with opposite change.

Mirek

> 
> >  {
> >      /* 99% of the time, we can signal the use of extension registers
> >         by looking to see if the opcode handles 64-bit data.  */
> > diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
> > index 6775bd30fc..976f0f05af 100644
> > --- a/tcg/sparc/tcg-target.c.inc
> > +++ b/tcg/sparc/tcg-target.c.inc
> > @@ -1294,8 +1294,7 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg
> > data, TCGReg addr,
> >  }
> >  
> >  static void tcg_out_op(TCGContext *s, TCGOpcode opc,
> > -                       const TCGArg args[TCG_MAX_OP_ARGS],
> > -                       const int const_args[TCG_MAX_OP_ARGS])
> > +                       const TCGArg *args, const int *const_args)
> >  {
> >      TCGArg a0, a1, a2;
> >      int c, c2;
> > 
> 
> 

-- 
Miroslav Rezanina
Software Engineer - Virtualization Team Maintainer