Hi,
This patch teaches GCC that zero-extending a MVE predicate from 16-bits
to 32-bits and then only using 16-bits is a no-op.
It does so in two steps:
- it lets gcc know that it can access any MVE predicate mode using any
other MVE predicate mode without needing to copy it, using the
TARGET_MODES_TIEABLE_P hook,
- it teaches simplify_subreg to optimize a subreg with a vector
outermode, by replacing this outermode with a same-sized integer mode
and trying the avalailable optimizations, then if successful it
surrounds the result with a subreg casting it back to the original
vector outermode.
This removes the unnecessary zero-extending shown on PR 107674 (though
it's a sign-extend there), that was introduced in gcc 11.
Bootstrapped on aarch64-none-linux-gnu and regression tested on
arm-none-eabi and armeb-none-eabi for armv8.1-m.main+mve.fp.
OK for trunk?
gcc/ChangeLog:
PR target/107674
* conig/arm/arm.cc (arm_hard_regno_mode_ok): Use new MACRO.
(arm_modes_tieable_p): Make MVE predicate modes tieable.
* config/arm/arm.h (VALID_MVE_PRED_MODE): New define.
* simplify-rtx.cc (simplify_context::simplify_subreg): Teach
simplify_subreg to simplify subregs where the outermode is not scalar.
gcc/testsuite/ChangeLog:
* gcc.target/arm/mve/mve_vpt.c: Change to remove unecessary
zero-extend.diff --git a/gcc/config/arm/arm.h b/gcc/config/arm/arm.h
index
6f7ecf9128047647fc41677e634cd9612a13242b..4352c830cb6d2e632a225edea861b5ceb35dd035
100644
--- a/gcc/config/arm/arm.h
+++ b/gcc/config/arm/arm.h
@@ -1091,6 +1091,10 @@ extern const int arm_arch_cde_coproc_bits[];
|| (MODE) == V16QImode || (MODE) == V8HFmode || (MODE) == V4SFmode \
|| (MODE) == V2DFmode)
+#define VALID_MVE_PRED_MODE(MODE) \
+ ((MODE) == HImode \
+ || (MODE) == V16BImode || (MODE) == V8BImode || (MODE) == V4BImode)
+
#define VALID_MVE_SI_MODE(MODE) \
((MODE) == V2DImode ||(MODE) == V4SImode || (MODE) == V8HImode \
|| (MODE) == V16QImode)
diff --git a/gcc/config/arm/arm.cc b/gcc/config/arm/arm.cc
index
3f171188de513e258369397e4726afe27bd9fdbf..18460ef5280be8c1df85eff424a1bf66d6019c0a
100644
--- a/gcc/config/arm/arm.cc
+++ b/gcc/config/arm/arm.cc
@@ -25564,10 +25564,7 @@ arm_hard_regno_mode_ok (unsigned int regno,
machine_mode mode)
return false;
if (IS_VPR_REGNUM (regno))
- return mode == HImode
- || mode == V16BImode
- || mode == V8BImode
- || mode == V4BImode;
+ return VALID_MVE_PRED_MODE (mode);
if (TARGET_THUMB1)
/* For the Thumb we only allow values bigger than SImode in
@@ -25646,6 +25643,10 @@ arm_modes_tieable_p (machine_mode mode1, machine_mode
mode2)
if (GET_MODE_CLASS (mode1) == GET_MODE_CLASS (mode2))
return true;
+ if (TARGET_HAVE_MVE
+ && (VALID_MVE_PRED_MODE (mode1) && VALID_MVE_PRED_MODE (mode2)))
+ return true;
+
/* We specifically want to allow elements of "structure" modes to
be tieable to the structure. This more general condition allows
other rarer situations too. */
diff --git a/gcc/simplify-rtx.cc b/gcc/simplify-rtx.cc
index
7fb1e97fbea4e7b8b091f5724ebe0cb61eee7ec3..a951272186585c0a5cc3e0155285e7a635865f42
100644
--- a/gcc/simplify-rtx.cc
+++ b/gcc/simplify-rtx.cc
@@ -7652,6 +7652,22 @@ simplify_context::simplify_subreg (machine_mode
outermode, rtx op,
}
}
+ /* Try simplifying a SUBREG expression of a non-integer OUTERMODE by using a
+ NEW_OUTERMODE of the same size instead, other simplifications rely on
+ integer to integer subregs and we'd potentially miss out on optimizations
+ otherwise. */
+ if (known_gt (GET_MODE_SIZE (innermode),
+ GET_MODE_SIZE (outermode))
+ && SCALAR_INT_MODE_P (innermode)
+ && !SCALAR_INT_MODE_P (outermode)
+ && int_mode_for_size (GET_MODE_BITSIZE (outermode),
+ 0).exists (&int_outermode))
+ {
+ rtx tem = simplify_subreg (int_outermode, op, innermode, byte);
+ if (tem)
+ return simplify_gen_subreg (outermode, tem, GET_MODE (tem), byte);
+ }
+
/* If OP is a vector comparison and the subreg is not changing the
number of elements or the size of the elements, change the result
of the comparison to the new mode. */
diff --git a/gcc/testsuite/gcc.target/arm/mve/mve_vpt.c
b/gcc/testsuite/gcc.target/arm/mve/mve_vpt.c
index
26a565b79dd1348e361b3aa23a1d6e6d13bffce8..8e562a9f065eff157f63ebd5acf9af0a2155b5c5
100644
--- a/gcc/testsuite/gcc.target/arm/mve/mve_vpt.c
+++ b/gcc/testsuite/gcc.target/arm/mve/mve_vpt.c
@@ -16,9 +16,6 @@ void test0 (uint8_t *a, uint8_t *b, uint8_t *c)
** vldrb.8 q2, \[r0\]
** vldrb.8 q1, \[r1\]
** vcmp.i8 eq, q2, q1
-** vmrs r3, p0 @ movhi
-** uxth r3, r3
-** vmsr p0, r3 @ movhi
** vpst
** vaddt.i8 q3, q2, q1
** vpst
