[PATCH] PowerPC: Support for Prefixed Add Immediate Shifted Instruction (RFC02686)
Peter Bergner
bergner@linux.ibm.com
Mon Apr 21 22:22:42 GMT 2025
On 3/5/25 11:24 AM, Surya Kumari Jangala wrote:
> opcodes/
> * ppc-opc.c (insert_si32, extract_si32, insert_nsi32,
> extract_nsi32): New functions.
> (SI32, NSI32, P_D_SI32_MASK, P_DRAPCREL_SI32_MASK): New macros.
> (IM32): Update for new macros.
s/IM32/IMM32/
> diff --git a/gas/testsuite/gas/ppc/future.s b/gas/testsuite/gas/ppc/future.s
> index fec99e348bf..90e5e894bc1 100644
> --- a/gas/testsuite/gas/ppc/future.s
> +++ b/gas/testsuite/gas/ppc/future.s
> @@ -49,3 +49,16 @@ _start:
> xxgfmul128gcm 1, 2, 3
> xxgfmul128 31, 32, 33, 1
> xxgfmul128xts 31, 32, 33
> + paddis 12, 9, 15, 0
> + paddis 12, 9, ~(1<<15), 0
> + paddis 9, 0, 25, 1
> + paddis 24, 0, 2147483647, 0
> + plis 24, 2147483647
> + paddis 30, 10, -2147483648, 0
> + psubis 30, 10, 2147483648, 0
> + nop
> + paddis 30, 10, 2147483647, 0
> + psubis 30, 10, -2147483647, 0
> + paddis 15, 0, 2147483647, 1
> + psubis 15, 0, -2147483647, 1
For the tests that end with ',0', ie, the R=0 versions, please add an extra
test without the ',0' too. So for the test:
paddis 12, 9, 15, 0
add an extra test:
paddis 12, 9, 15
The paddis instruction has an extended mnemonic for the R=0 case and I
like to include those in the tests too using the same operands as the
normal mnemonic so they come out identically in the future.d expected
output.
> +/* The same as P_D_MASK, but with the RA and PCREL fields specified. */
> +#define P_DRAPCREL_SI32_MASK (P_D_SI32_MASK | PCREL_MASK | RA_MASK)
As Jan mentioned s/P_D_MASK/P_D_SI32_MASK/
> {"psubi", PMLS|OP(14), P_D_MASK, POWER10, EXT, {RT, RA0, NSI34, PCREL}},
> +
> +{"plis", PMLS|OP(15), P_DRAPCREL_SI32_MASK, FUTURE, EXT, {RT, SI32}},
> +{"paddis", PMLS|OP(15), P_D_SI32_MASK, FUTURE, 0, {RT, RA0, SI32, PCREL}},
> +{"psubis", PMLS|OP(15), P_D_SI32_MASK, FUTURE, EXT, {RT, RA0, NSI32, PCREL}},
> +
> {"xxsplti32dx", P8RR|VSOP(32,0), P_VSI_MASK, POWER10, 0, {XTS, IX, IMM32}},
Please remove the 2 extra blank lines.
Ok for mainline with the above changes. Sorry for the late review!
Peter
More information about the Binutils
mailing list