Chung-Lin Tang wrote:
> I analyzed this testcase regression a while earlier; the
> direct cause of
> this is due to mips_order_regs_for_local_alloc(), which now serves as
> MIPS' ADJUST_REG_ALLOC_ORDER macro.
>
> The mips_order_regs_for_local_alloc() function seems to be written for
> the old local-alloc.c, still left as the deprecated
> ORDER_REGS_FOR_LOCAL_ALLOC macro after the transition to IRA (actually
> not called at all during then), and relatively recently
> 'revived' after
> a patch by Bernd that created the ADJUST_REG_ALLOC_ORDER
> macro went in.
>
> So you have a local-alloc.c heuristic working in IRA, which seemed to
> cause these regressions.
>
> Removing mips_order_regs_for_local_alloc() will let this
> testcase pass;
> of course the real fix should be to review the MIPS
> reg-ordering logic,
> left for you MIPS people...
I think your analysis is correct. We should just delete
mips_order_regs_for_local_alloc()
in mips.c and delete ADJUST_REG_ALLOC_ORDER in mips.h.
Then, 3 accumulators can be used in dspr2-MULT.c and dspr2-MULTU.c now. Thanks!
Ex:
/* Test MIPS32 DSP REV 2 MULT instruction. Tune for a CPU that has
pipelined mult. */
/* { dg-do compile } */
/* { dg-options "-mgp32 -mdspr2 -O2 -ffixed-hi -ffixed-lo -mtune=74kc" } */
/* { dg-final { scan-assembler "\tmult\t" } } */
/* { dg-final { scan-assembler "ac1" } } */
/* { dg-final { scan-assembler "ac2" } } */
/* { dg-final { scan-assembler "ac3" } } */
typedef long long a64;
NOMIPS16 a64 test (a64 *a, int *b, int *c)
{
a[0] = (a64) b[0] * c[0];
a[1] = (a64) b[1] * c[1];
a[2] = (a64) b[2] * c[2];
}
Ex:
/* Test MIPS32 DSP REV 2 MULTU instruction. Tune for a CPU that has
pipelined multu. */
/* { dg-do compile } */
/* { dg-options "-mgp32 -mdspr2 -O2 -ffixed-hi -ffixed-lo -mtune=74kc" } */
/* { dg-final { scan-assembler "\tmultu\t" } } */
/* { dg-final { scan-assembler "ac1" } } */
/* { dg-final { scan-assembler "ac2" } } */
/* { dg-final { scan-assembler "ac3" } } */
typedef unsigned long long a64;
NOMIPS16 a64 test (a64 *a, unsigned int *b, unsigned int *c)
{
a[0] = (a64) b[0] * c[0];
a[1] = (a64) b[1] * c[1];
a[2] = (a64) b[2] * c[2];
}
Regards,
Chao-ying
