* [COMMITTED] PowerPC: Add support for RFC02653 - Dense Math Facility
@ 2022-10-28 0:35 Peter Bergner
0 siblings, 0 replies; 6+ messages in thread
From: Peter Bergner @ 2022-10-28 0:35 UTC (permalink / raw)
To: Binutils; +Cc: Alan Modra
The following patch adds support for Power RFC02653 which is an extension
of the MMA support in power10. When or even if this will ever show up in
hardware is not determined or guaranteed, therefore this is enabled using
the -mfuture gas option.
Peter
PowerPC: Add support for RFC02653 - Dense Math Facility
gas/
* config/tc-ppc.c (pre_defined_registers): Add dense math registers.
(md_assemble): Check dmr specified in correct operand.
* testsuite/gas/ppc/outerprod.s <dmsetaccz, dmxvbf16ger2,
dmxvbf16ger2nn, dmxvbf16ger2np, dmxvbf16ger2pn, dmxvbf16ger2pp,
dmxvf16ger2, dmxvf16ger2nn, dmxvf16ger2np, dmxvf16ger2pn, dmxvf16ger2pp,
dmxvf32ger, dmxvf32gernn, dmxvf32gernp, dmxvf32gerpn, dmxvf32gerpp,
dmxvf64ger, dmxvf64gernn, dmxvf64gernp, dmxvf64gerpn, dmxvf64gerpp,
dmxvi16ger2, dmxvi16ger2pp, dmxvi16ger2s, dmxvi16ger2spp, dmxvi4ger8,
dmxvi4ger8pp, dmxvi8ger4, dmxvi8ger4pp, dmxvi8ger4spp, dmxxmfacc,
dmxxmtacc, pmdmxvbf16ger2, pmdmxvbf16ger2nn, pmdmxvbf16ger2np,
pmdmxvbf16ger2pn, pmdmxvbf16ger2pp, pmdmxvf16ger2, pmdmxvf16ger2nn,
pmdmxvf16ger2np, pmdmxvf16ger2pn, pmdmxvf16ger2pp, pmdmxvf32ger,
pmdmxvf32gernn, pmdmxvf32gernp, pmdmxvf32gerpn, pmdmxvf32gerpp,
pmdmxvf64ger, pmdmxvf64gernn, pmdmxvf64gernp, pmdmxvf64gerpn,
pmdmxvf64gerpp, pmdmxvi16ger2, pmdmxvi16ger2pp, pmdmxvi16ger2s,
pmdmxvi16ger2spp, pmdmxvi4ger8, pmdmxvi4ger8pp, pmdmxvi8ger4,
pmdmxvi8ger4pp, pmdmxvi8ger4spp>: Add new tests.
* testsuite/gas/ppc/outerprod.d: Likewise.
* testsuite/gas/ppc/rfc02653.s: New test.
* testsuite/gas/ppc/rfc02653.d: Likewise.
* testsuite/gas/ppc/ppc.exp: Run it.
include/
* opcode/ppc.h (PPC_OPERAND_DMR): Define. Renumber following
PPC_OPERAND defines.
opcodes/
* ppc-dis.c (print_insn_powerpc): Prepend 'dm' when printing DMR regs.
* ppc-opc.c (insert_p2, (extract_p2, (insert_xa5, (extract_xa5,
insert_xb5, (extract_xb5): New functions.
(insert_xa6a, extract_xa6a, insert_xb6a, extract_xb6a): Disallow
operand overlap only on Power10.
(DMR, DMRAB, P1, P2, XA5p, XB5p, XDMR_MASK, XDMRDMR_MASK, XX2ACC_MASK,
XX2DMR_MASK, XX3DMR_MASK): New defines.
(powerpc_opcodes): Add dmmr, dmsetaccz, dmsetdmrz, dmxor, dmxvbf16ger2,
dmxvbf16ger2nn, dmxvbf16ger2np, dmxvbf16ger2pn, dmxvbf16ger2pp,
dmxvf16ger2, dmxvf16ger2nn, dmxvf16ger2np, dmxvf16ger2pn, dmxvf16ger2pp,
dmxvf32ger, dmxvf32gernn, dmxvf32gernp, dmxvf32gerpn, dmxvf32gerpp,
dmxvf64ger, dmxvf64gernn, dmxvf64gernp, dmxvf64gerpn, dmxvf64gerpp,
dmxvi16ger2, dmxvi16ger2pp, dmxvi16ger2s, dmxvi16ger2spp, dmxvi4ger8,
dmxvi4ger8pp, dmxvi8ger4, dmxvi8ger4pp, dmxvi8ger4spp, dmxxextfdmr256,
dmxxextfdmr512, dmxxinstdmr256, dmxxinstdmr512, dmxxmfacc, dmxxmtacc,
pmdmxvbf16ger2, pmdmxvbf16ger2nn, pmdmxvbf16ger2np, pmdmxvbf16ger2pn,
pmdmxvbf16ger2pp, pmdmxvf16ger2, pmdmxvf16ger2nn, pmdmxvf16ger2np,
pmdmxvf16ger2pn, pmdmxvf16ger2pp, pmdmxvf32ger, pmdmxvf32gernn,
pmdmxvf32gernp, pmdmxvf32gerpn, pmdmxvf32gerpp, pmdmxvf64ger,
pmdmxvf64gernn, pmdmxvf64gernp, pmdmxvf64gerpn, pmdmxvf64gerpp,
pmdmxvi16ger2, pmdmxvi16ger2pp, pmdmxvi16ger2s, pmdmxvi16ger2spp,
pmdmxvi4ger8, pmdmxvi4ger8pp, pmdmxvi8ger4, pmdmxvi8ger4pp,
pmdmxvi8ger4spp.
---
gas/config/tc-ppc.c | 13 +-
gas/testsuite/gas/ppc/outerprod.d | 215 +++++++++++++++++++++---------
gas/testsuite/gas/ppc/outerprod.s | 61 +++++++++
gas/testsuite/gas/ppc/ppc.exp | 1 +
gas/testsuite/gas/ppc/rfc02653.d | 27 ++++
gas/testsuite/gas/ppc/rfc02653.s | 18 +++
include/opcode/ppc.h | 33 ++---
opcodes/ppc-dis.c | 3 +
opcodes/ppc-opc.c | 214 +++++++++++++++++++++++++----
9 files changed, 479 insertions(+), 106 deletions(-)
create mode 100644 gas/testsuite/gas/ppc/rfc02653.d
create mode 100644 gas/testsuite/gas/ppc/rfc02653.s
diff --git a/gas/config/tc-ppc.c b/gas/config/tc-ppc.c
index 97ad782012c..1acbba1791a 100644
--- a/gas/config/tc-ppc.c
+++ b/gas/config/tc-ppc.c
@@ -353,6 +353,16 @@ static const struct pd_reg pre_defined_registers[] =
{ "dec", 22, PPC_OPERAND_SPR },
{ "dsisr", 18, PPC_OPERAND_SPR },
+ /* Dense Math Registers. */
+ { "dm0", 0, PPC_OPERAND_DMR },
+ { "dm1", 1, PPC_OPERAND_DMR },
+ { "dm2", 2, PPC_OPERAND_DMR },
+ { "dm3", 3, PPC_OPERAND_DMR },
+ { "dm4", 4, PPC_OPERAND_DMR },
+ { "dm5", 5, PPC_OPERAND_DMR },
+ { "dm6", 6, PPC_OPERAND_DMR },
+ { "dm7", 7, PPC_OPERAND_DMR },
+
/* Floating point registers */
{ "f.0", 0, PPC_OPERAND_FPR },
{ "f.1", 1, PPC_OPERAND_FPR },
@@ -3475,7 +3485,8 @@ md_assemble (char *str)
& ~operand->flags
& (PPC_OPERAND_GPR | PPC_OPERAND_FPR | PPC_OPERAND_VR
| PPC_OPERAND_VSR | PPC_OPERAND_CR_BIT | PPC_OPERAND_CR_REG
- | PPC_OPERAND_SPR | PPC_OPERAND_GQR | PPC_OPERAND_ACC)) != 0
+ | PPC_OPERAND_SPR | PPC_OPERAND_GQR | PPC_OPERAND_ACC
+ | PPC_OPERAND_DMR)) != 0
&& !((ex.X_md & PPC_OPERAND_GPR) != 0
&& ex.X_add_number != 0
&& (operand->flags & PPC_OPERAND_GPR_0) != 0))
diff --git a/gas/testsuite/gas/ppc/outerprod.d b/gas/testsuite/gas/ppc/outerprod.d
index 613fb189966..332102b1318 100644
--- a/gas/testsuite/gas/ppc/outerprod.d
+++ b/gas/testsuite/gas/ppc/outerprod.d
@@ -8,97 +8,184 @@
Disassembly of section \.text:
0+0 <_start>:
-.*: (7e 80 01 62|62 01 80 7e) xxmfacc a5
-.*: (7f 01 01 62|62 01 01 7f) xxmtacc a6
-.*: (7f 83 01 62|62 01 83 7f) xxsetaccz a7
-.*: (ec 1f f1 1e|1e f1 1f ec) xvi4ger8 a0,vs63,vs62
-.*: (ec 9d e1 16|16 e1 9d ec) xvi4ger8pp a1,vs61,vs60
-.*: (07 90 ff fe|fe ff 90 07) pmxvi4ger8 a2,vs59,vs58,15,14,255
+.*: (7e 80 01 62|62 01 80 7e) dmxxmfacc a5
+.*: (7e 80 01 62|62 01 80 7e) dmxxmfacc a5
+.*: (7f 01 01 62|62 01 01 7f) dmxxmtacc a6
+.*: (7f 01 01 62|62 01 01 7f) dmxxmtacc a6
+.*: (7f 83 01 62|62 01 83 7f) dmsetaccz a7
+.*: (7f 83 01 62|62 01 83 7f) dmsetaccz a7
+.*: (ec 1f f1 1e|1e f1 1f ec) dmxvi4ger8 a0,vs63,vs62
+.*: (ec 1f f1 1e|1e f1 1f ec) dmxvi4ger8 a0,vs63,vs62
+.*: (ec 9d e1 16|16 e1 9d ec) dmxvi4ger8pp a1,vs61,vs60
+.*: (ec 9d e1 16|16 e1 9d ec) dmxvi4ger8pp a1,vs61,vs60
+.*: (07 90 ff fe|fe ff 90 07) pmdmxvi4ger8 a2,vs59,vs58,15,14,255
.*: (ed 1b d1 1e|1e d1 1b ed)
-.*: (07 90 80 78|78 80 90 07) pmxvi4ger8pp a3,vs57,vs56,7,8,128
+.*: (07 90 ff fe|fe ff 90 07) pmdmxvi4ger8 a2,vs59,vs58,15,14,255
+.*: (ed 1b d1 1e|1e d1 1b ed)
+.*: (07 90 80 78|78 80 90 07) pmdmxvi4ger8pp a3,vs57,vs56,7,8,128
+.*: (ed 99 c1 16|16 c1 99 ed)
+.*: (07 90 80 78|78 80 90 07) pmdmxvi4ger8pp a3,vs57,vs56,7,8,128
.*: (ed 99 c1 16|16 c1 99 ed)
-.*: (ee 17 b0 1e|1e b0 17 ee) xvi8ger4 a4,vs55,vs54
-.*: (ee 95 a0 16|16 a0 95 ee) xvi8ger4pp a5,vs53,vs52
-.*: (07 90 b0 dc|dc b0 90 07) pmxvi8ger4 a6,vs51,vs50,13,12,11
+.*: (ee 17 b0 1e|1e b0 17 ee) dmxvi8ger4 a4,vs55,vs54
+.*: (ee 17 b0 1e|1e b0 17 ee) dmxvi8ger4 a4,vs55,vs54
+.*: (ee 95 a0 16|16 a0 95 ee) dmxvi8ger4pp a5,vs53,vs52
+.*: (ee 95 a0 16|16 a0 95 ee) dmxvi8ger4pp a5,vs53,vs52
+.*: (07 90 b0 dc|dc b0 90 07) pmdmxvi8ger4 a6,vs51,vs50,13,12,11
.*: (ef 13 90 1e|1e 90 13 ef)
-.*: (07 90 80 a9|a9 80 90 07) pmxvi8ger4pp a7,vs49,vs48,10,9,8
+.*: (07 90 b0 dc|dc b0 90 07) pmdmxvi8ger4 a6,vs51,vs50,13,12,11
+.*: (ef 13 90 1e|1e 90 13 ef)
+.*: (07 90 80 a9|a9 80 90 07) pmdmxvi8ger4pp a7,vs49,vs48,10,9,8
+.*: (ef 91 80 16|16 80 91 ef)
+.*: (07 90 80 a9|a9 80 90 07) pmdmxvi8ger4pp a7,vs49,vs48,10,9,8
.*: (ef 91 80 16|16 80 91 ef)
-.*: (ec 0f 71 5e|5e 71 0f ec) xvi16ger2s a0,vs47,vs46
-.*: (ec 8d 61 56|56 61 8d ec) xvi16ger2spp a1,vs45,vs44
-.*: (07 90 c0 76|76 c0 90 07) pmxvi16ger2s a2,vs43,vs42,7,6,3
+.*: (ec 0f 71 5e|5e 71 0f ec) dmxvi16ger2s a0,vs47,vs46
+.*: (ec 0f 71 5e|5e 71 0f ec) dmxvi16ger2s a0,vs47,vs46
+.*: (ec 8d 61 56|56 61 8d ec) dmxvi16ger2spp a1,vs45,vs44
+.*: (ec 8d 61 56|56 61 8d ec) dmxvi16ger2spp a1,vs45,vs44
+.*: (07 90 c0 76|76 c0 90 07) pmdmxvi16ger2s a2,vs43,vs42,7,6,3
.*: (ed 0b 51 5e|5e 51 0b ed)
-.*: (07 90 80 54|54 80 90 07) pmxvi16ger2spp a3,vs41,vs40,5,4,2
+.*: (07 90 c0 76|76 c0 90 07) pmdmxvi16ger2s a2,vs43,vs42,7,6,3
+.*: (ed 0b 51 5e|5e 51 0b ed)
+.*: (07 90 80 54|54 80 90 07) pmdmxvi16ger2spp a3,vs41,vs40,5,4,2
+.*: (ed 89 41 56|56 41 89 ed)
+.*: (07 90 80 54|54 80 90 07) pmdmxvi16ger2spp a3,vs41,vs40,5,4,2
.*: (ed 89 41 56|56 41 89 ed)
-.*: (ee 07 30 9e|9e 30 07 ee) xvf16ger2 a4,vs39,vs38
-.*: (ee 85 20 96|96 20 85 ee) xvf16ger2pp a5,vs37,vs36
-.*: (ef 03 14 96|96 14 03 ef) xvf16ger2pn a6,vs35,vs34
-.*: (ef 81 02 96|96 02 81 ef) xvf16ger2np a7,vs33,vs32
-.*: (ec 04 2e 90|90 2e 04 ec) xvf16ger2nn a0,vs4,vs5
-.*: (07 90 40 32|32 40 90 07) pmxvf16ger2 a1,vs2,vs3,3,2,1
+.*: (ee 07 30 9e|9e 30 07 ee) dmxvf16ger2 a4,vs39,vs38
+.*: (ee 07 30 9e|9e 30 07 ee) dmxvf16ger2 a4,vs39,vs38
+.*: (ee 85 20 96|96 20 85 ee) dmxvf16ger2pp a5,vs37,vs36
+.*: (ee 85 20 96|96 20 85 ee) dmxvf16ger2pp a5,vs37,vs36
+.*: (ef 03 14 96|96 14 03 ef) dmxvf16ger2pn a6,vs35,vs34
+.*: (ef 03 14 96|96 14 03 ef) dmxvf16ger2pn a6,vs35,vs34
+.*: (ef 81 02 96|96 02 81 ef) dmxvf16ger2np a7,vs33,vs32
+.*: (ef 81 02 96|96 02 81 ef) dmxvf16ger2np a7,vs33,vs32
+.*: (ec 04 2e 90|90 2e 04 ec) dmxvf16ger2nn a0,vs4,vs5
+.*: (ec 04 2e 90|90 2e 04 ec) dmxvf16ger2nn a0,vs4,vs5
+.*: (07 90 40 32|32 40 90 07) pmdmxvf16ger2 a1,vs2,vs3,3,2,1
.*: (ec 82 18 98|98 18 82 ec)
-.*: (07 90 00 10|10 00 90 07) pmxvf16ger2pp a2,vs4,vs5,1,0,0
+.*: (07 90 40 32|32 40 90 07) pmdmxvf16ger2 a1,vs2,vs3,3,2,1
+.*: (ec 82 18 98|98 18 82 ec)
+.*: (07 90 00 10|10 00 90 07) pmdmxvf16ger2pp a2,vs4,vs5,1,0,0
+.*: (ed 04 28 90|90 28 04 ed)
+.*: (07 90 00 10|10 00 90 07) pmdmxvf16ger2pp a2,vs4,vs5,1,0,0
.*: (ed 04 28 90|90 28 04 ed)
-.*: (07 90 c0 fe|fe c0 90 07) pmxvf16ger2pn a3,vs6,vs7,15,14,3
+.*: (07 90 c0 fe|fe c0 90 07) pmdmxvf16ger2pn a3,vs6,vs7,15,14,3
.*: (ed 86 3c 90|90 3c 86 ed)
-.*: (07 90 80 dc|dc 80 90 07) pmxvf16ger2np a4,vs8,vs9,13,12,2
+.*: (07 90 c0 fe|fe c0 90 07) pmdmxvf16ger2pn a3,vs6,vs7,15,14,3
+.*: (ed 86 3c 90|90 3c 86 ed)
+.*: (07 90 80 dc|dc 80 90 07) pmdmxvf16ger2np a4,vs8,vs9,13,12,2
+.*: (ee 08 4a 90|90 4a 08 ee)
+.*: (07 90 80 dc|dc 80 90 07) pmdmxvf16ger2np a4,vs8,vs9,13,12,2
.*: (ee 08 4a 90|90 4a 08 ee)
-.*: (07 90 40 ba|ba 40 90 07) pmxvf16ger2nn a5,vs10,vs11,11,10,1
+.*: (07 90 40 ba|ba 40 90 07) pmdmxvf16ger2nn a5,vs10,vs11,11,10,1
.*: (ee 8a 5e 90|90 5e 8a ee)
-.*: (ef 0c 68 d8|d8 68 0c ef) xvf32ger a6,vs12,vs13
-.*: (ef 8e 78 d0|d0 78 8e ef) xvf32gerpp a7,vs14,vs15
-.*: (ec 10 8c d0|d0 8c 10 ec) xvf32gerpn a0,vs16,vs17
-.*: (ec 92 9a d0|d0 9a 92 ec) xvf32gernp a1,vs18,vs19
-.*: (ed 14 ae d0|d0 ae 14 ed) xvf32gernn a2,vs20,vs21
-.*: (07 90 00 98|98 00 90 07) pmxvf32ger a3,vs22,vs23,9,8
+.*: (07 90 40 ba|ba 40 90 07) pmdmxvf16ger2nn a5,vs10,vs11,11,10,1
+.*: (ee 8a 5e 90|90 5e 8a ee)
+.*: (ef 0c 68 d8|d8 68 0c ef) dmxvf32ger a6,vs12,vs13
+.*: (ef 0c 68 d8|d8 68 0c ef) dmxvf32ger a6,vs12,vs13
+.*: (ef 8e 78 d0|d0 78 8e ef) dmxvf32gerpp a7,vs14,vs15
+.*: (ef 8e 78 d0|d0 78 8e ef) dmxvf32gerpp a7,vs14,vs15
+.*: (ec 10 8c d0|d0 8c 10 ec) dmxvf32gerpn a0,vs16,vs17
+.*: (ec 10 8c d0|d0 8c 10 ec) dmxvf32gerpn a0,vs16,vs17
+.*: (ec 92 9a d0|d0 9a 92 ec) dmxvf32gernp a1,vs18,vs19
+.*: (ec 92 9a d0|d0 9a 92 ec) dmxvf32gernp a1,vs18,vs19
+.*: (ed 14 ae d0|d0 ae 14 ed) dmxvf32gernn a2,vs20,vs21
+.*: (ed 14 ae d0|d0 ae 14 ed) dmxvf32gernn a2,vs20,vs21
+.*: (07 90 00 98|98 00 90 07) pmdmxvf32ger a3,vs22,vs23,9,8
+.*: (ed 96 b8 d8|d8 b8 96 ed)
+.*: (07 90 00 98|98 00 90 07) pmdmxvf32ger a3,vs22,vs23,9,8
.*: (ed 96 b8 d8|d8 b8 96 ed)
-.*: (07 90 00 76|76 00 90 07) pmxvf32gerpp a4,vs24,vs25,7,6
+.*: (07 90 00 76|76 00 90 07) pmdmxvf32gerpp a4,vs24,vs25,7,6
.*: (ee 18 c8 d0|d0 c8 18 ee)
-.*: (07 90 00 54|54 00 90 07) pmxvf32gerpn a5,vs26,vs27,5,4
+.*: (07 90 00 76|76 00 90 07) pmdmxvf32gerpp a4,vs24,vs25,7,6
+.*: (ee 18 c8 d0|d0 c8 18 ee)
+.*: (07 90 00 54|54 00 90 07) pmdmxvf32gerpn a5,vs26,vs27,5,4
+.*: (ee 9a dc d0|d0 dc 9a ee)
+.*: (07 90 00 54|54 00 90 07) pmdmxvf32gerpn a5,vs26,vs27,5,4
.*: (ee 9a dc d0|d0 dc 9a ee)
-.*: (60 00 00 00|00 00 00 60) nop
-.*: (07 90 00 32|32 00 90 07) pmxvf32gernp a6,vs28,vs29,3,2
+.*: (07 90 00 32|32 00 90 07) pmdmxvf32gernp a6,vs28,vs29,3,2
+.*: (ef 1c ea d0|d0 ea 1c ef)
+.*: (07 90 00 32|32 00 90 07) pmdmxvf32gernp a6,vs28,vs29,3,2
.*: (ef 1c ea d0|d0 ea 1c ef)
-.*: (07 90 00 10|10 00 90 07) pmxvf32gernn a7,vs0,vs1,1,0
+.*: (07 90 00 10|10 00 90 07) pmdmxvf32gernn a7,vs0,vs1,1,0
.*: (ef 80 0e d0|d0 0e 80 ef)
-.*: (ec 04 29 d8|d8 29 04 ec) xvf64ger a0,vs4,vs5
-.*: (ec 88 49 d0|d0 49 88 ec) xvf64gerpp a1,vs8,vs9
-.*: (ed 02 15 d0|d0 15 02 ed) xvf64gerpn a2,vs2,vs2
-.*: (ed 84 1b d0|d0 1b 84 ed) xvf64gernp a3,vs4,vs3
-.*: (ee 04 27 d0|d0 27 04 ee) xvf64gernn a4,vs4,vs4
-.*: (07 90 00 f0|f0 00 90 07) pmxvf64ger a5,vs6,vs5,15,0
+.*: (07 90 00 10|10 00 90 07) pmdmxvf32gernn a7,vs0,vs1,1,0
+.*: (ef 80 0e d0|d0 0e 80 ef)
+.*: (ec 04 29 d8|d8 29 04 ec) dmxvf64ger a0,vs4,vs5
+.*: (ec 04 29 d8|d8 29 04 ec) dmxvf64ger a0,vs4,vs5
+.*: (ec 88 49 d0|d0 49 88 ec) dmxvf64gerpp a1,vs8,vs9
+.*: (ec 88 49 d0|d0 49 88 ec) dmxvf64gerpp a1,vs8,vs9
+.*: (ed 02 15 d0|d0 15 02 ed) dmxvf64gerpn a2,vs2,vs2
+.*: (ed 02 15 d0|d0 15 02 ed) dmxvf64gerpn a2,vs2,vs2
+.*: (ed 84 1b d0|d0 1b 84 ed) dmxvf64gernp a3,vs4,vs3
+.*: (ed 84 1b d0|d0 1b 84 ed) dmxvf64gernp a3,vs4,vs3
+.*: (ee 04 27 d0|d0 27 04 ee) dmxvf64gernn a4,vs4,vs4
+.*: (ee 04 27 d0|d0 27 04 ee) dmxvf64gernn a4,vs4,vs4
+.*: (07 90 00 f0|f0 00 90 07) pmdmxvf64ger a5,vs6,vs5,15,0
+.*: (ee 86 29 d8|d8 29 86 ee)
+.*: (07 90 00 f0|f0 00 90 07) pmdmxvf64ger a5,vs6,vs5,15,0
.*: (ee 86 29 d8|d8 29 86 ee)
-.*: (07 90 00 e4|e4 00 90 07) pmxvf64gerpp a6,vs6,vs6,14,1
+.*: (07 90 00 e4|e4 00 90 07) pmdmxvf64gerpp a6,vs6,vs6,14,1
.*: (ef 06 31 d0|d0 31 06 ef)
-.*: (07 90 00 d8|d8 00 90 07) pmxvf64gerpn a7,vs8,vs7,13,2
+.*: (07 90 00 e4|e4 00 90 07) pmdmxvf64gerpp a6,vs6,vs6,14,1
+.*: (ef 06 31 d0|d0 31 06 ef)
+.*: (07 90 00 d8|d8 00 90 07) pmdmxvf64gerpn a7,vs8,vs7,13,2
+.*: (ef 88 3d d0|d0 3d 88 ef)
+.*: (07 90 00 d8|d8 00 90 07) pmdmxvf64gerpn a7,vs8,vs7,13,2
.*: (ef 88 3d d0|d0 3d 88 ef)
-.*: (60 00 00 00|00 00 00 60) nop
-.*: (07 90 00 cc|cc 00 90 07) pmxvf64gernp a0,vs4,vs5,12,3
+.*: (07 90 00 cc|cc 00 90 07) pmdmxvf64gernp a0,vs4,vs5,12,3
.*: (ec 04 2b d0|d0 2b 04 ec)
-.*: (07 90 00 a0|a0 00 90 07) pmxvf64gernn a1,vs2,vs1,10,0
+.*: (07 90 00 cc|cc 00 90 07) pmdmxvf64gernp a0,vs4,vs5,12,3
+.*: (ec 04 2b d0|d0 2b 04 ec)
+.*: (07 90 00 a0|a0 00 90 07) pmdmxvf64gernn a1,vs2,vs1,10,0
+.*: (ec 82 0f d0|d0 0f 82 ec)
+.*: (07 90 00 a0|a0 00 90 07) pmdmxvf64gernn a1,vs2,vs1,10,0
.*: (ec 82 0f d0|d0 0f 82 ec)
-.*: (ed 03 21 90|90 21 03 ed) xvbf16ger2pp a2,vs3,vs4
-.*: (ed 84 29 98|98 29 84 ed) xvbf16ger2 a3,vs4,vs5
-.*: (ee 05 33 90|90 33 05 ee) xvbf16ger2np a4,vs5,vs6
-.*: (ee 86 3d 90|90 3d 86 ee) xvbf16ger2pn a5,vs6,vs7
-.*: (ef 07 47 90|90 47 07 ef) xvbf16ger2nn a6,vs7,vs8
-.*: (07 90 c0 ff|ff c0 90 07) pmxvbf16ger2pp a7,vs8,vs9,15,15,3
+.*: (ed 03 21 90|90 21 03 ed) dmxvbf16ger2pp a2,vs3,vs4
+.*: (ed 03 21 90|90 21 03 ed) dmxvbf16ger2pp a2,vs3,vs4
+.*: (ed 84 29 98|98 29 84 ed) dmxvbf16ger2 a3,vs4,vs5
+.*: (ed 84 29 98|98 29 84 ed) dmxvbf16ger2 a3,vs4,vs5
+.*: (ee 05 33 90|90 33 05 ee) dmxvbf16ger2np a4,vs5,vs6
+.*: (ee 05 33 90|90 33 05 ee) dmxvbf16ger2np a4,vs5,vs6
+.*: (ee 86 3d 90|90 3d 86 ee) dmxvbf16ger2pn a5,vs6,vs7
+.*: (ee 86 3d 90|90 3d 86 ee) dmxvbf16ger2pn a5,vs6,vs7
+.*: (ef 07 47 90|90 47 07 ef) dmxvbf16ger2nn a6,vs7,vs8
+.*: (ef 07 47 90|90 47 07 ef) dmxvbf16ger2nn a6,vs7,vs8
+.*: (07 90 c0 ff|ff c0 90 07) pmdmxvbf16ger2pp a7,vs8,vs9,15,15,3
+.*: (ef 88 49 90|90 49 88 ef)
+.*: (07 90 c0 ff|ff c0 90 07) pmdmxvbf16ger2pp a7,vs8,vs9,15,15,3
.*: (ef 88 49 90|90 49 88 ef)
-.*: (07 90 80 cc|cc 80 90 07) pmxvbf16ger2 a0,vs9,vs10,12,12,2
+.*: (07 90 80 cc|cc 80 90 07) pmdmxvbf16ger2 a0,vs9,vs10,12,12,2
.*: (ec 09 51 98|98 51 09 ec)
-.*: (07 90 40 aa|aa 40 90 07) pmxvbf16ger2np a1,vs10,vs11,10,10,1
+.*: (07 90 80 cc|cc 80 90 07) pmdmxvbf16ger2 a0,vs9,vs10,12,12,2
+.*: (ec 09 51 98|98 51 09 ec)
+.*: (07 90 40 aa|aa 40 90 07) pmdmxvbf16ger2np a1,vs10,vs11,10,10,1
+.*: (ec 8a 5b 90|90 5b 8a ec)
+.*: (07 90 40 aa|aa 40 90 07) pmdmxvbf16ger2np a1,vs10,vs11,10,10,1
.*: (ec 8a 5b 90|90 5b 8a ec)
-.*: (60 00 00 00|00 00 00 60) nop
-.*: (07 90 00 dd|dd 00 90 07) pmxvbf16ger2pn a2,vs12,vs13,13,13,0
+.*: (07 90 00 dd|dd 00 90 07) pmdmxvbf16ger2pn a2,vs12,vs13,13,13,0
.*: (ed 0c 6d 90|90 6d 0c ed)
-.*: (07 90 c0 ee|ee c0 90 07) pmxvbf16ger2nn a3,vs16,vs17,14,14,3
+.*: (07 90 00 dd|dd 00 90 07) pmdmxvbf16ger2pn a2,vs12,vs13,13,13,0
+.*: (ed 0c 6d 90|90 6d 0c ed)
+.*: (07 90 c0 ee|ee c0 90 07) pmdmxvbf16ger2nn a3,vs16,vs17,14,14,3
+.*: (ed 90 8f 90|90 8f 90 ed)
+.*: (07 90 c0 ee|ee c0 90 07) pmdmxvbf16ger2nn a3,vs16,vs17,14,14,3
.*: (ed 90 8f 90|90 8f 90 ed)
-.*: (ee 00 0b 1e|1e 0b 00 ee) xvi8ger4spp a4,vs32,vs33
-.*: (07 90 f0 ff|ff f0 90 07) pmxvi8ger4spp a5,vs34,vs35,15,15,15
+.*: (ee 00 0b 1e|1e 0b 00 ee) dmxvi8ger4spp a4,vs32,vs33
+.*: (ee 00 0b 1e|1e 0b 00 ee) dmxvi8ger4spp a4,vs32,vs33
+.*: (07 90 f0 ff|ff f0 90 07) pmdmxvi8ger4spp a5,vs34,vs35,15,15,15
.*: (ee 82 1b 1e|1e 1b 82 ee)
-.*: (ef 04 2a 5e|5e 2a 04 ef) xvi16ger2 a6,vs36,vs37
-.*: (ef 86 3b 5e|5e 3b 86 ef) xvi16ger2pp a7,vs38,vs39
-.*: (07 90 40 ff|ff 40 90 07) pmxvi16ger2 a0,vs38,vs39,15,15,1
+.*: (07 90 f0 ff|ff f0 90 07) pmdmxvi8ger4spp a5,vs34,vs35,15,15,15
+.*: (ee 82 1b 1e|1e 1b 82 ee)
+.*: (ef 04 2a 5e|5e 2a 04 ef) dmxvi16ger2 a6,vs36,vs37
+.*: (ef 04 2a 5e|5e 2a 04 ef) dmxvi16ger2 a6,vs36,vs37
+.*: (ef 86 3b 5e|5e 3b 86 ef) dmxvi16ger2pp a7,vs38,vs39
+.*: (ef 86 3b 5e|5e 3b 86 ef) dmxvi16ger2pp a7,vs38,vs39
+.*: (07 90 40 ff|ff 40 90 07) pmdmxvi16ger2 a0,vs38,vs39,15,15,1
+.*: (ec 06 3a 5e|5e 3a 06 ec)
+.*: (07 90 40 ff|ff 40 90 07) pmdmxvi16ger2 a0,vs38,vs39,15,15,1
.*: (ec 06 3a 5e|5e 3a 06 ec)
-.*: (07 90 80 cc|cc 80 90 07) pmxvi16ger2pp a1,vs40,vs41,12,12,2
+.*: (07 90 80 cc|cc 80 90 07) pmdmxvi16ger2pp a1,vs40,vs41,12,12,2
+.*: (ec 88 4b 5e|5e 4b 88 ec)
+.*: (07 90 80 cc|cc 80 90 07) pmdmxvi16ger2pp a1,vs40,vs41,12,12,2
.*: (ec 88 4b 5e|5e 4b 88 ec)
#pass
diff --git a/gas/testsuite/gas/ppc/outerprod.s b/gas/testsuite/gas/ppc/outerprod.s
index 1f02c158daa..dd947fe9720 100644
--- a/gas/testsuite/gas/ppc/outerprod.s
+++ b/gas/testsuite/gas/ppc/outerprod.s
@@ -1,63 +1,124 @@
.text
_start:
xxmfacc 5
+ dmxxmfacc 5
xxmtacc 6
+ dmxxmtacc 6
xxsetaccz 7
+ dmsetaccz 7
xvi4ger8 0,63,62
+ dmxvi4ger8 0,63,62
xvi4ger8pp 1,61,60
+ dmxvi4ger8pp 1,61,60
pmxvi4ger8 2,59,58,15,14,255
+ pmdmxvi4ger8 2,59,58,15,14,255
pmxvi4ger8pp 3,57,56,7,8,128
+ pmdmxvi4ger8pp 3,57,56,7,8,128
xvi8ger4 4,55,54
+ dmxvi8ger4 4,55,54
xvi8ger4pp 5,53,52
+ dmxvi8ger4pp 5,53,52
pmxvi8ger4 6,51,50,13,12,11
+ pmdmxvi8ger4 6,51,50,13,12,11
pmxvi8ger4pp 7,49,48,10,9,8
+ pmdmxvi8ger4pp 7,49,48,10,9,8
xvi16ger2s 0,47,46
+ dmxvi16ger2s 0,47,46
xvi16ger2spp 1,45,44
+ dmxvi16ger2spp 1,45,44
pmxvi16ger2s 2,43,42,7,6,3
+ pmdmxvi16ger2s 2,43,42,7,6,3
pmxvi16ger2spp 3,41,40,5,4,2
+ pmdmxvi16ger2spp 3,41,40,5,4,2
xvf16ger2 4,39,38
+ dmxvf16ger2 4,39,38
xvf16ger2pp 5,37,36
+ dmxvf16ger2pp 5,37,36
xvf16ger2pn 6,35,34
+ dmxvf16ger2pn 6,35,34
xvf16ger2np 7,33,32
+ dmxvf16ger2np 7,33,32
xvf16ger2nn 0,4,5
+ dmxvf16ger2nn 0,4,5
pmxvf16ger2 1,2,3,3,2,1
+ pmdmxvf16ger2 1,2,3,3,2,1
pmxvf16ger2pp 2,4,5,1,0,0
+ pmdmxvf16ger2pp 2,4,5,1,0,0
pmxvf16ger2pn 3,6,7,15,14,3
+ pmdmxvf16ger2pn 3,6,7,15,14,3
pmxvf16ger2np 4,8,9,13,12,2
+ pmdmxvf16ger2np 4,8,9,13,12,2
pmxvf16ger2nn 5,10,11,11,10,1
+ pmdmxvf16ger2nn 5,10,11,11,10,1
xvf32ger 6,12,13
+ dmxvf32ger 6,12,13
xvf32gerpp 7,14,15
+ dmxvf32gerpp 7,14,15
xvf32gerpn 0,16,17
+ dmxvf32gerpn 0,16,17
xvf32gernp 1,18,19
+ dmxvf32gernp 1,18,19
xvf32gernn 2,20,21
+ dmxvf32gernn 2,20,21
pmxvf32ger 3,22,23,9,8
+ pmdmxvf32ger 3,22,23,9,8
pmxvf32gerpp 4,24,25,7,6
+ pmdmxvf32gerpp 4,24,25,7,6
pmxvf32gerpn 5,26,27,5,4
+ pmdmxvf32gerpn 5,26,27,5,4
pmxvf32gernp 6,28,29,3,2
+ pmdmxvf32gernp 6,28,29,3,2
pmxvf32gernn 7,0,1,1,0
+ pmdmxvf32gernn 7,0,1,1,0
xvf64ger 0,4,5
+ dmxvf64ger 0,4,5
xvf64gerpp 1,8,9
+ dmxvf64gerpp 1,8,9
xvf64gerpn 2,2,2
+ dmxvf64gerpn 2,2,2
xvf64gernp 3,4,3
+ dmxvf64gernp 3,4,3
xvf64gernn 4,4,4
+ dmxvf64gernn 4,4,4
pmxvf64ger 5,6,5,15,0
+ pmdmxvf64ger 5,6,5,15,0
pmxvf64gerpp 6,6,6,14,1
+ pmdmxvf64gerpp 6,6,6,14,1
pmxvf64gerpn 7,8,7,13,2
+ pmdmxvf64gerpn 7,8,7,13,2
pmxvf64gernp 0,4,5,12,3
+ pmdmxvf64gernp 0,4,5,12,3
pmxvf64gernn 1,2,1,10,0
+ pmdmxvf64gernn 1,2,1,10,0
xvbf16ger2pp 2,3,4
+ dmxvbf16ger2pp 2,3,4
xvbf16ger2 3,4,5
+ dmxvbf16ger2 3,4,5
xvbf16ger2np 4,5,6
+ dmxvbf16ger2np 4,5,6
xvbf16ger2pn 5,6,7
+ dmxvbf16ger2pn 5,6,7
xvbf16ger2nn 6,7,8
+ dmxvbf16ger2nn 6,7,8
pmxvbf16ger2pp 7,8,9,15,15,3
+ pmdmxvbf16ger2pp 7,8,9,15,15,3
pmxvbf16ger2 0,9,10,12,12,2
+ pmdmxvbf16ger2 0,9,10,12,12,2
pmxvbf16ger2np 1,10,11,10,10,1
+ pmdmxvbf16ger2np 1,10,11,10,10,1
pmxvbf16ger2pn 2,12,13,13,13,0
+ pmdmxvbf16ger2pn 2,12,13,13,13,0
pmxvbf16ger2nn 3,16,17,14,14,3
+ pmdmxvbf16ger2nn 3,16,17,14,14,3
xvi8ger4spp 4,32,33
+ dmxvi8ger4spp 4,32,33
pmxvi8ger4spp 5,34,35,15,15,15
+ pmdmxvi8ger4spp 5,34,35,15,15,15
xvi16ger2 6,36,37
+ dmxvi16ger2 6,36,37
xvi16ger2pp 7,38,39
+ dmxvi16ger2pp 7,38,39
pmxvi16ger2 0,38,39,15,15,1
+ pmdmxvi16ger2 0,38,39,15,15,1
pmxvi16ger2pp 1,40,41,12,12,2
+ pmdmxvi16ger2pp 1,40,41,12,12,2
diff --git a/gas/testsuite/gas/ppc/ppc.exp b/gas/testsuite/gas/ppc/ppc.exp
index ae8a7b61cde..f27a79cfd73 100644
--- a/gas/testsuite/gas/ppc/ppc.exp
+++ b/gas/testsuite/gas/ppc/ppc.exp
@@ -145,6 +145,7 @@ run_dump_test "rightmost"
run_dump_test "scalarquad"
run_dump_test "rop"
run_dump_test "rop-checks"
+run_dump_test "rfc02653"
run_dump_test "dcbt"
run_dump_test "pr27676"
diff --git a/gas/testsuite/gas/ppc/rfc02653.d b/gas/testsuite/gas/ppc/rfc02653.d
new file mode 100644
index 00000000000..6ad49df9c04
--- /dev/null
+++ b/gas/testsuite/gas/ppc/rfc02653.d
@@ -0,0 +1,27 @@
+#as: -mfuture
+#objdump: -dr -Mfuture
+#name: RFC02653 tests
+
+.*
+
+
+Disassembly of section \.text:
+
+0+0 <_start>:
+.*: (62 01 02 7c|7c 02 01 62) dmsetdmrz dm0
+.*: (62 41 86 7c|7c 86 41 62) dmmr dm1,dm2
+.*: (62 61 07 7d|7d 07 61 62) dmxor dm2,dm3
+.*: (10 17 00 f2|f2 00 17 10) dmxxextfdmr512 vs0,vs2,dm4,0
+.*: (10 37 85 f2|f2 85 37 10) dmxxextfdmr512 vs4,vs6,dm5,1
+.*: (50 57 08 f3|f3 08 57 50) dmxxinstdmr512 dm6,vs8,vs10,0
+.*: (50 57 89 f3|f3 89 57 50) dmxxinstdmr512 dm7,vs8,vs10,1
+.*: (90 67 00 f0|f0 00 67 90) dmxxextfdmr256 vs12,dm0,0
+.*: (90 7f 80 f0|f0 80 7f 90) dmxxextfdmr256 vs14,dm1,1
+.*: (90 87 01 f1|f1 01 87 90) dmxxextfdmr256 vs16,dm2,2
+.*: (90 9f 81 f1|f1 81 9f 90) dmxxextfdmr256 vs18,dm3,3
+.*: (94 a7 00 f2|f2 00 a7 94) dmxxinstdmr256 dm4,vs20,0
+.*: (94 bf 80 f2|f2 80 bf 94) dmxxinstdmr256 dm5,vs22,1
+.*: (94 c7 01 f3|f3 01 c7 94) dmxxinstdmr256 dm6,vs24,2
+.*: (94 df 81 f3|f3 81 df 94) dmxxinstdmr256 dm7,vs26,3
+.*: (18 09 00 ec|ec 00 09 18) dmxvi4ger8 a0,vs0,vs1
+#pass
diff --git a/gas/testsuite/gas/ppc/rfc02653.s b/gas/testsuite/gas/ppc/rfc02653.s
new file mode 100644
index 00000000000..8b343d6cbfc
--- /dev/null
+++ b/gas/testsuite/gas/ppc/rfc02653.s
@@ -0,0 +1,18 @@
+ .text
+_start:
+ dmsetdmrz 0
+ dmmr 1,2
+ dmxor 2,3
+ dmxxextfdmr512 0,2,4,0
+ dmxxextfdmr512 4,6,5,1
+ dmxxinstdmr512 6,8,10,0
+ dmxxinstdmr512 7,8,10,1
+ dmxxextfdmr256 12,0,0
+ dmxxextfdmr256 14,1,1
+ dmxxextfdmr256 16,2,2
+ dmxxextfdmr256 18,3,3
+ dmxxinstdmr256 4,20,0
+ dmxxinstdmr256 5,22,1
+ dmxxinstdmr256 6,24,2
+ dmxxinstdmr256 7,26,3
+ dmxvi4ger8 0,0,1 # VSRs can now overlap the ACCs
diff --git a/include/opcode/ppc.h b/include/opcode/ppc.h
index 930d13d3026..004b51db670 100644
--- a/include/opcode/ppc.h
+++ b/include/opcode/ppc.h
@@ -384,6 +384,9 @@ extern const unsigned int num_powerpc_operands;
/* This operand names a VSX accumulator. */
#define PPC_OPERAND_ACC (0x20)
+/* This operand names a dense math register. */
+#define PPC_OPERAND_DMR (0x40)
+
/* This operand may use the symbolic names for the CR fields (even
without -mregnames), which are
lt 0 gt 1 eq 2 so 3 un 3
@@ -391,60 +394,60 @@ extern const unsigned int num_powerpc_operands;
cr4 4 cr5 5 cr6 6 cr7 7
These may be combined arithmetically, as in cr2*4+gt. These are
only supported on the PowerPC, not the POWER. */
-#define PPC_OPERAND_CR_BIT (0x40)
+#define PPC_OPERAND_CR_BIT (0x80)
/* This is a CR FIELD that does not use symbolic names (unless
-mregnames is in effect). If both PPC_OPERAND_CR_BIT and
PPC_OPERAND_CR_REG are set then treat the field as per
PPC_OPERAND_CR_BIT for assembly, but as if neither of these
bits are set for disassembly. */
-#define PPC_OPERAND_CR_REG (0x80)
+#define PPC_OPERAND_CR_REG (0x100)
/* This operand names a special purpose register. */
-#define PPC_OPERAND_SPR (0x100)
+#define PPC_OPERAND_SPR (0x200)
/* This operand names a paired-single graphics quantization register. */
-#define PPC_OPERAND_GQR (0x200)
+#define PPC_OPERAND_GQR (0x400)
/* This operand is a relative branch displacement. The disassembler
prints these symbolically if possible. */
-#define PPC_OPERAND_RELATIVE (0x400)
+#define PPC_OPERAND_RELATIVE (0x800)
/* This operand is an absolute branch address. The disassembler
prints these symbolically if possible. */
-#define PPC_OPERAND_ABSOLUTE (0x800)
+#define PPC_OPERAND_ABSOLUTE (0x1000)
/* This operand takes signed values. */
-#define PPC_OPERAND_SIGNED (0x1000)
+#define PPC_OPERAND_SIGNED (0x2000)
/* This operand takes signed values, but also accepts a full positive
range of values when running in 32 bit mode. That is, if bits is
16, it takes any value from -0x8000 to 0xffff. In 64 bit mode,
this flag is ignored. */
-#define PPC_OPERAND_SIGNOPT (0x2000)
+#define PPC_OPERAND_SIGNOPT (0x4000)
/* The next operand should be wrapped in parentheses rather than
separated from this one by a comma. This is used for the load and
store instructions which want their operands to look like
reg,displacement(reg)
*/
-#define PPC_OPERAND_PARENS (0x4000)
+#define PPC_OPERAND_PARENS (0x8000)
/* This operand is for the DS field in a DS form instruction. */
-#define PPC_OPERAND_DS (0x8000)
+#define PPC_OPERAND_DS (0x10000)
/* This operand is for the DQ field in a DQ form instruction. */
-#define PPC_OPERAND_DQ (0x10000)
+#define PPC_OPERAND_DQ (0x20000)
/* This operand should be regarded as a negative number for the
purposes of overflow checking (i.e., the normal most negative
number is disallowed and one more than the normal most positive
number is allowed). This flag will only be set for a signed
operand. */
-#define PPC_OPERAND_NEGATIVE (0x20000)
+#define PPC_OPERAND_NEGATIVE (0x40000)
/* Valid range of operand is 0..n rather than 0..n-1. */
-#define PPC_OPERAND_PLUS1 (0x40000)
+#define PPC_OPERAND_PLUS1 (0x80000)
/* This operand is optional, and is zero if omitted. This is used for
example, in the optional BF field in the comparison instructions. The
@@ -452,7 +455,7 @@ extern const unsigned int num_powerpc_operands;
and the number of operands remaining for the opcode, and decide
whether this operand is present or not. The disassembler should
print this operand out only if it is not zero. */
-#define PPC_OPERAND_OPTIONAL (0x80000)
+#define PPC_OPERAND_OPTIONAL (0x100000)
/* This flag is only used with PPC_OPERAND_OPTIONAL. If this operand
is omitted, then for the next operand use this operand value plus
@@ -460,7 +463,7 @@ extern const unsigned int num_powerpc_operands;
hack is needed because the Power rotate instructions can take
either 4 or 5 operands. The disassembler should print this operand
out regardless of the PPC_OPERAND_OPTIONAL field. */
-#define PPC_OPERAND_NEXT (0x100000)
+#define PPC_OPERAND_NEXT (0x200000)
/* This flag is only used with PPC_OPERAND_OPTIONAL. The operand is
only optional when generating 32-bit code. */
diff --git a/opcodes/ppc-dis.c b/opcodes/ppc-dis.c
index 33a96701ca8..69466661800 100644
--- a/opcodes/ppc-dis.c
+++ b/opcodes/ppc-dis.c
@@ -1097,6 +1097,9 @@ print_insn_powerpc (bfd_vma memaddr,
else if ((operand->flags & PPC_OPERAND_VSR) != 0)
(*info->fprintf_styled_func) (info->stream, dis_style_register,
"vs%" PRId64, value);
+ else if ((operand->flags & PPC_OPERAND_DMR) != 0)
+ (*info->fprintf_styled_func) (info->stream, dis_style_register,
+ "dm%" PRId64, value);
else if ((operand->flags & PPC_OPERAND_ACC) != 0)
(*info->fprintf_styled_func) (info->stream, dis_style_register,
"a%" PRId64, value);
diff --git a/opcodes/ppc-opc.c b/opcodes/ppc-opc.c
index b470ebd0aa9..c323a2761e3 100644
--- a/opcodes/ppc-opc.c
+++ b/opcodes/ppc-opc.c
@@ -1477,6 +1477,26 @@ extract_pl (uint64_t insn,
return value;
}
+/* The 2-bit P field in a MMA XX2-form instruction. This is split. */
+
+static uint64_t
+insert_p2 (uint64_t insn,
+ int64_t value,
+ ppc_cpu_t dialect ATTRIBUTE_UNUSED,
+ const char **errmsg ATTRIBUTE_UNUSED)
+{
+ return insn | ((value & 0x2) << 15) | ((value & 0x1) << 11);
+}
+
+static int64_t
+extract_p2 (uint64_t insn,
+ ppc_cpu_t dialect ATTRIBUTE_UNUSED,
+ int *invalid ATTRIBUTE_UNUSED)
+{
+ uint64_t value = ((insn >> 15) & 0x2) | ((insn >> 11) & 0x1);
+ return value;
+}
+
/* The RA field in a D or X form instruction which is an updating
load, which means that the RA field may not be zero and may not
equal the RT field. */
@@ -2129,6 +2149,25 @@ extract_xtq6 (uint64_t insn,
return ((insn << 2) & 0x20) | ((insn >> 21) & 0x1f);
}
+/* The 5-bit XAp field in an XX3 form instruction. This is split. */
+
+static uint64_t
+insert_xa5 (uint64_t insn,
+ int64_t value,
+ ppc_cpu_t dialect ATTRIBUTE_UNUSED,
+ const char **errmsg ATTRIBUTE_UNUSED)
+{
+ return insn | ((value & 0x1e) << 16) | ((value & 0x20) >> 3);
+}
+
+static int64_t
+extract_xa5 (uint64_t insn,
+ ppc_cpu_t dialect ATTRIBUTE_UNUSED,
+ int *invalid ATTRIBUTE_UNUSED)
+{
+ return ((insn << 3) & 0x20) | ((insn >> 16) & 0x1e);
+}
+
/* The XA field in an XX3 form instruction. This is split. */
static uint64_t
@@ -2158,7 +2197,9 @@ insert_xa6a (uint64_t insn,
const char **errmsg)
{
int64_t acc = (insn >> 23) & 0x7;
- if ((value >> 2) == acc)
+ /* Power10 doesn't allow VSRs to overlap ACCs in MMA instructions. */
+ if ((dialect & PPC_OPCODE_FUTURE) == 0
+ && (value >> 2) == acc)
*errmsg = _("VSR overlaps ACC operand");
return insert_xa6 (insn, value, dialect, errmsg);
}
@@ -2170,11 +2211,31 @@ extract_xa6a (uint64_t insn,
{
int64_t acc = (insn >> 23) & 0x7;
int64_t value = extract_xa6 (insn, dialect, invalid);
- if ((value >> 2) == acc)
+ /* Power10 doesn't allow VSRs to overlap ACCs in MMA instructions. */
+ if ((dialect & PPC_OPCODE_FUTURE) == 0
+ && (value >> 2) == acc)
*invalid = 1;
return value;
}
+/* The 5-bit XB field in an XX3 form instruction. This is split. */
+
+static uint64_t
+insert_xb5 (uint64_t insn,
+ int64_t value,
+ ppc_cpu_t dialect ATTRIBUTE_UNUSED,
+ const char **errmsg ATTRIBUTE_UNUSED)
+{
+ return insn | ((value & 0x1e) << 11) | ((value & 0x20) >> 4);
+}
+
+static int64_t
+extract_xb5 (uint64_t insn,
+ ppc_cpu_t dialect ATTRIBUTE_UNUSED,
+ int *invalid ATTRIBUTE_UNUSED)
+{
+ return ((insn << 4) & 0x20) | ((insn >> 11) & 0x1e);
+}
/* The XB field in an XX3 form instruction. This is split. */
static uint64_t
@@ -2204,7 +2265,9 @@ insert_xb6a (uint64_t insn,
const char **errmsg)
{
int64_t acc = (insn >> 23) & 0x7;
- if ((value >> 2) == acc)
+ /* Power10 doesn't allow VSRs to overlap ACCs in MMA instructions. */
+ if ((dialect & PPC_OPCODE_FUTURE) == 0
+ && (value >> 2) == acc)
*errmsg = _("VSR overlaps ACC operand");
return insert_xb6 (insn, value, dialect, errmsg);
}
@@ -2216,7 +2279,9 @@ extract_xb6a (uint64_t insn,
{
int64_t acc = (insn >> 23) & 0x7;
int64_t value = extract_xb6 (insn, dialect, invalid);
- if ((value >> 2) == acc)
+ /* Power10 doesn't allow VSRs to overlap ACCs in MMA instructions. */
+ if ((dialect & PPC_OPCODE_FUTURE) == 0
+ && (value >> 2) == acc)
*invalid = 1;
return value;
}
@@ -2824,9 +2889,17 @@ const struct powerpc_operand powerpc_operands[] =
#define ACC BFF + 1
{ 0x7, 23, NULL, NULL, PPC_OPERAND_ACC },
+ /* The DMR field in a MMA instruction. */
+#define DMR ACC + 1
+ { 0x7, 23, NULL, NULL, PPC_OPERAND_DMR },
+
+ /* The second DMR field in a two DMR operand MMA instruction. */
+#define DMRAB DMR + 1
+ { 0x7, 13, NULL, NULL, PPC_OPERAND_DMR },
+
/* An optional BF field. This is used for comparison instructions,
in which an omitted BF field is taken as zero. */
-#define OBF ACC + 1
+#define OBF DMRAB + 1
{ 0x7, 23, NULL, NULL, PPC_OPERAND_CR_REG | PPC_OPERAND_OPTIONAL },
/* The BFA field in an X or XL form instruction. */
@@ -3656,6 +3729,7 @@ const struct powerpc_operand powerpc_operands[] =
#define R RMC + 1
#define MP R
+#define P1 R
{ 0x1, 16, NULL, NULL, 0 },
#define RIC R + 1
@@ -3769,8 +3843,13 @@ const struct powerpc_operand powerpc_operands[] =
#define XA6ap XA6a + 1
{ 0x3e, PPC_OPSHIFT_INV, insert_xa6a, extract_xa6a, PPC_OPERAND_VSR },
+ /* The 5-bit XAp field in an MMA XX3 form instruction. This is split.
+ This is like XA6, but must be even. */
+#define XA5p XA6ap + 1
+ { 0x3e, PPC_OPSHIFT_INV, insert_xa5, extract_xa5, PPC_OPERAND_VSR },
+
/* The XB field in an XX2 or XX3 form instruction. This is split. */
-#define XB6 XA6ap + 1
+#define XB6 XA5p + 1
{ 0x3f, PPC_OPSHIFT_INV, insert_xb6, extract_xb6, PPC_OPERAND_VSR },
/* The XB field in an XX3 form instruction. This is split and
@@ -3778,9 +3857,14 @@ const struct powerpc_operand powerpc_operands[] =
#define XB6a XB6 + 1
{ 0x3f, PPC_OPSHIFT_INV, insert_xb6a, extract_xb6a, PPC_OPERAND_VSR },
+ /* The 5-bit XBp field in an MMA XX3 form instruction. This is split.
+ This is like XB6, but must be even. */
+#define XB5p XB6a + 1
+ { 0x3e, PPC_OPSHIFT_INV, insert_xb5, extract_xb5, PPC_OPERAND_VSR },
+
/* The XA and XB fields in an XX3 form instruction when they must be the same.
This is used in extended mnemonics like xvmovdp. This is split. */
-#define XAB6 XB6a + 1
+#define XAB6 XB5p + 1
{ 0x3f, PPC_OPSHIFT_INV, insert_xab6, extract_xab6, PPC_OPERAND_VSR },
/* The XC field in an XX4 form instruction. This is split. */
@@ -3815,8 +3899,11 @@ const struct powerpc_operand powerpc_operands[] =
#define PL SC2
{ 0x3, 16, insert_pl, extract_pl, PPC_OPERAND_OPTIONAL },
+#define P2 PL + 1
+ { 0x3, PPC_OPSHIFT_INV, insert_p2, extract_p2, 0 },
+
/* The 8-bit IMM8 field in a XX1 form instruction. */
-#define IMM8 SC2 + 1
+#define IMM8 P2 + 1
{ 0xff, 11, NULL, NULL, PPC_OPERAND_SIGNOPT },
#define VX_OFF IMM8 + 1
@@ -4452,15 +4539,22 @@ const unsigned int num_powerpc_operands = ARRAY_SIZE (powerpc_operands);
/* An X_MASK with an accumulator register and the RA and RB fields fixed. */
#define XACC_MASK (X_MASK | RA_MASK | RB_MASK | (3 << 21))
+#define XDMR_MASK XACC_MASK
-/* The mask for an XX3 form instruction with an accumulator register. */
-#define XX3ACC_MASK (XX3 (0x3f, 0xff) | (3 << 21) | (1))
+/* An X_MASK with two dense math register. */
+#define XDMRDMR_MASK (X_MASK | RA_MASK | (3 << 21) | (3 << 11))
/* The mask for an XX3 form instruction with the DM or SHW bits
specified. */
#define XX3DM_MASK (XX3 (0x3f, 0x1f) | (1 << 10))
#define XX3SHW_MASK XX3DM_MASK
+/* The masks for X* form instructions with an ACC/DMR register. */
+#define XX2ACC_MASK (XX2 (0x3f, 0x1ff) | (3 << 21) | 1)
+#define XX3ACC_MASK (XX3_MASK | (3 << 21) | 1)
+#define XX3DMR_MASK (XX3ACC_MASK | (1 << 11))
+#define XX2DMR_MASK (XX2ACC_MASK | (0xf << 17))
+
/* The mask for an XX4 form instruction. */
#define XX4_MASK XX4 (0x3f, 0x3)
@@ -7245,9 +7339,15 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"msgclrp", XRTRA(31,174,0,0), XRTRA_MASK, POWER8, 0, {RB}},
{"dcbtlse", X(31,174), X_MASK, PPCCHLK, E500MC, {CT, RA0, RB}},
+{"dmxxmfacc", XVA(31,177,0), XACC_MASK, POWER10, 0, {ACC}},
{"xxmfacc", XVA(31,177,0), XACC_MASK, POWER10, 0, {ACC}},
+{"dmxxmtacc", XVA(31,177,1), XACC_MASK, POWER10, 0, {ACC}},
{"xxmtacc", XVA(31,177,1), XACC_MASK, POWER10, 0, {ACC}},
+{"dmsetdmrz", XVA(31,177,2), XDMR_MASK, FUTURE, 0, {DMR}},
+{"dmsetaccz", XVA(31,177,3), XACC_MASK, POWER10, 0, {ACC}},
{"xxsetaccz", XVA(31,177,3), XACC_MASK, POWER10, 0, {ACC}},
+{"dmmr", XVA(31,177,6), XDMRDMR_MASK,FUTURE, 0, {DMR, DMRAB}},
+{"dmxor", XVA(31,177,7), XDMRDMR_MASK,FUTURE, 0, {DMR, DMRAB}},
{"mtmsrd", X(31,178), XRLARB_MASK, PPC64, 0, {RS, A_L}},
@@ -8888,7 +8988,9 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"dqua", ZRC(59,3,0), Z2_MASK, POWER6, PPCVLE, {FRT,FRA,FRB,RMC}},
{"dqua.", ZRC(59,3,1), Z2_MASK, POWER6, PPCVLE, {FRT,FRA,FRB,RMC}},
+{"dmxvi8ger4pp",XX3(59,2), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvi8ger4pp", XX3(59,2), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvi8ger4", XX3(59,3), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvi8ger4", XX3(59,3), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"fdivs", A(59,18,0), AFRC_MASK, PPC, PPCEFS|PPCVLE, {FRT, FRA, FRB}},
@@ -8940,8 +9042,10 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"dquai", ZRC(59,67,0), Z2_MASK, POWER6, PPCVLE, {TE, FRT,FRB,RMC}},
{"dquai.", ZRC(59,67,1), Z2_MASK, POWER6, PPCVLE, {TE, FRT,FRB,RMC}},
-{"xvf16ger2pp", XX3(59,18), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvf16ger2", XX3(59,19), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf16ger2pp",XX3(59,18), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvf16ger2pp", XX3(59,18), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf16ger2", XX3(59,19), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvf16ger2", XX3(59,19), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"dscri", ZRC(59,98,0), Z_MASK, POWER6, PPCVLE, {FRT, FRA, SH16}},
{"dscri.", ZRC(59,98,1), Z_MASK, POWER6, PPCVLE, {FRT, FRA, SH16}},
@@ -8949,30 +9053,40 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"drintx", ZRC(59,99,0), Z2_MASK, POWER6, PPCVLE, {R, FRT, FRB, RMC}},
{"drintx.", ZRC(59,99,1), Z2_MASK, POWER6, PPCVLE, {R, FRT, FRB, RMC}},
+{"dmxvf32gerpp",XX3(59,26), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvf32gerpp", XX3(59,26), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf32ger", XX3(59,27), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvf32ger", XX3(59,27), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"dcmpo", X(59,130), X_MASK, POWER6, PPCVLE, {BF, FRA, FRB}},
+{"dmxvi4ger8pp",XX3(59,34), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvi4ger8pp", XX3(59,34), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvi4ger8", XX3(59,35), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvi4ger8", XX3(59,35), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"dtstex", X(59,162), X_MASK, POWER6, PPCVLE, {BF, FRA, FRB}},
-{"xvi16ger2spp",XX3(59,42), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvi16ger2s", XX3(59,43), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvi16ger2spp",XX3(59,42), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvi16ger2spp", XX3(59,42), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvi16ger2s", XX3(59,43), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvi16ger2s", XX3(59,43), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"dtstdc", Z(59,194), Z_MASK, POWER6, PPCVLE, {BF, FRA, DCM}},
-{"xvbf16ger2pp",XX3(59,50), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvbf16ger2", XX3(59,51), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvbf16ger2pp",XX3(59,50), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvbf16ger2pp", XX3(59,50), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvbf16ger2", XX3(59,51), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvbf16ger2", XX3(59,51), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"dtstdg", Z(59,226), Z_MASK, POWER6, PPCVLE, {BF, FRA, DGM}},
{"drintn", ZRC(59,227,0), Z2_MASK, POWER6, PPCVLE, {R, FRT, FRB, RMC}},
{"drintn.", ZRC(59,227,1), Z2_MASK, POWER6, PPCVLE, {R, FRT, FRB, RMC}},
+{"dmxvf64gerpp",XX3(59,58), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
{"xvf64gerpp", XX3(59,58), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
+{"dmxvf64ger", XX3(59,59), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
{"xvf64ger", XX3(59,59), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
{"dctdp", XRC(59,258,0), X_MASK, POWER6, PPCVLE, {FRT, FRB}},
@@ -8984,22 +9098,29 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"ddedpd", XRC(59,322,0), X_MASK, POWER6, PPCVLE, {SP, FRT, FRB}},
{"ddedpd.", XRC(59,322,1), X_MASK, POWER6, PPCVLE, {SP, FRT, FRB}},
+{"dmxvi16ger2", XX3(59,75), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvi16ger2", XX3(59,75), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvf16ger2np", XX3(59,82), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf16ger2np", XX3(59,82), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvf16ger2np", XX3(59,82), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"dxex", XRC(59,354,0), X_MASK, POWER6, PPCVLE, {FRT, FRB}},
{"dxex.", XRC(59,354,1), X_MASK, POWER6, PPCVLE, {FRT, FRB}},
-{"xvf32gernp", XX3(59,90), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf32gernp", XX3(59,90), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvf32gernp", XX3(59,90), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvi8ger4spp", XX3(59,99), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvi8ger4spp", XX3(59,99), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvi8ger4spp", XX3(59,99), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvi16ger2pp", XX3(59,107), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvi16ger2pp", XX3(59,107), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvi16ger2pp", XX3(59,107), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvbf16ger2np",XX3(59,114), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvbf16ger2np",XX3(59,114), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvbf16ger2np", XX3(59,114), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvf64gernp", XX3(59,122), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
+{"dmxvf64gernp", XX3(59,122), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
+{"xvf64gernp", XX3(59,122), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
{"dsub", XRC(59,514,0), X_MASK, POWER6, PPCVLE, {FRT, FRA, FRB}},
{"dsub.", XRC(59,514,1), X_MASK, POWER6, PPCVLE, {FRT, FRA, FRB}},
@@ -9007,8 +9128,10 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"ddiv", XRC(59,546,0), X_MASK, POWER6, PPCVLE, {FRT, FRA, FRB}},
{"ddiv.", XRC(59,546,1), X_MASK, POWER6, PPCVLE, {FRT, FRA, FRB}},
-{"xvf16ger2pn", XX3(59,146), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf16ger2pn", XX3(59,146), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvf16ger2pn", XX3(59,146), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf32gerpn",XX3(59,154), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvf32gerpn", XX3(59,154), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"dcmpu", X(59,642), X_MASK, POWER6, PPCVLE, {BF, FRA, FRB}},
@@ -9016,8 +9139,10 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"dtstsf", X(59,674), X_MASK, POWER6, PPCVLE, {BF, FRA, FRB}},
{"dtstsfi", X(59,675), X_MASK|1<<22,POWER9, PPCVLE, {BF, UIM6, FRB}},
-{"xvbf16ger2pn",XX3(59,178), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvbf16ger2pn",XX3(59,178), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvbf16ger2pn", XX3(59,178), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf64gerpn",XX3(59,186), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
{"xvf64gerpn", XX3(59,186), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
{"drsp", XRC(59,770,0), X_MASK, POWER6, PPCVLE, {FRT, FRB}},
@@ -9029,7 +9154,8 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"denbcd", XRC(59,834,0), X_MASK, POWER6, PPCVLE, {S, FRT, FRB}},
{"denbcd.", XRC(59,834,1), X_MASK, POWER6, PPCVLE, {S, FRT, FRB}},
-{"xvf16ger2nn", XX3(59,210), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvf16ger2nn", XX3(59,210), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvf16ger2nn", XX3(59,210), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"fcfids", XRC(59,846,0), XRA_MASK, POWER7|PPCA2, PPCVLE, {FRT, FRB}},
{"fcfids.", XRC(59,846,1), XRA_MASK, POWER7|PPCA2, PPCVLE, {FRT, FRB}},
@@ -9037,13 +9163,16 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"diex", XRC(59,866,0), X_MASK, POWER6, PPCVLE, {FRT, FRA, FRB}},
{"diex.", XRC(59,866,1), X_MASK, POWER6, PPCVLE, {FRT, FRA, FRB}},
+{"dmxvf32gernn",XX3(59,218), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"xvf32gernn", XX3(59,218), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
-{"xvbf16ger2nn",XX3(59,242), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"dmxvbf16ger2nn",XX3(59,242), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
+{"xvbf16ger2nn", XX3(59,242), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6a, XB6a}},
{"fcfidus", XRC(59,974,0), XRA_MASK, POWER7|PPCA2, PPCVLE, {FRT, FRB}},
{"fcfidus.", XRC(59,974,1), XRA_MASK, POWER7|PPCA2, PPCVLE, {FRT, FRB}},
+{"dmxvf64gernn",XX3(59,250), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
{"xvf64gernn", XX3(59,250), XX3ACC_MASK, POWER10, PPCVLE, {ACC, XA6ap, XB6a}},
{"xsaddsp", XX3(60,0), XX3_MASK, PPCVSX2, PPCVLE, {XT6, XA6, XB6}},
@@ -9220,6 +9349,7 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"xvnegsp", XX2(60,441), XX2_MASK, PPCVSX, PPCVLE, {XT6, XB6}},
{"xvmaxdp", XX3(60,224), XX3_MASK, PPCVSX, PPCVLE, {XT6, XA6, XB6}},
{"xvnmaddadp", XX3(60,225), XX3_MASK, PPCVSX, PPCVLE, {XT6, XA6, XB6}},
+{"dmxxextfdmr512",XX3(60,226), XX3DMR_MASK, FUTURE, PPCVLE, {XA5p, XB5p, DMR, P1}},
{"xvcvdpuxds", XX2(60,456), XX2_MASK, PPCVSX, PPCVLE, {XT6, XB6}},
{"xvcvspdp", XX2(60,457), XX2_MASK, PPCVSX, PPCVLE, {XT6, XB6}},
{"xxgenpcvbm", X(60,916), XX1_MASK, POWER10, PPCVLE, {XT6, VB, UIMM}},
@@ -9227,6 +9357,7 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"xsiexpdp", X(60,918), XX1_MASK, PPCVSX3, PPCVLE, {XT6, RA, RB}},
{"xvmindp", XX3(60,232), XX3_MASK, PPCVSX, PPCVLE, {XT6, XA6, XB6}},
{"xvnmaddmdp", XX3(60,233), XX3_MASK, PPCVSX, PPCVLE, {XT6, XA6, XB6}},
+{"dmxxinstdmr512",XX3(60,234), XX3DMR_MASK, FUTURE, PPCVLE, {DMR, XA5p, XB5p,P1}},
{"xvcvdpsxds", XX2(60,472), XX2_MASK, PPCVSX, PPCVLE, {XT6, XB6}},
{"xvabsdp", XX2(60,473), XX2_MASK, PPCVSX, PPCVLE, {XT6, XB6}},
{"xxgenpcvwm", X(60,948), XX1_MASK, POWER10, PPCVLE, {XT6, VB, UIMM}},
@@ -9247,6 +9378,8 @@ const struct powerpc_opcode powerpc_opcodes[] = {
{"xvmovdp", XX3(60,240), XX3_MASK, PPCVSX, PPCVLE|EXT, {XT6, XAB6}},
{"xvcpsgndp", XX3(60,240), XX3_MASK, PPCVSX, PPCVLE, {XT6, XA6, XB6}},
{"xvnmsubadp", XX3(60,241), XX3_MASK, PPCVSX, PPCVLE, {XT6, XA6, XB6}},
+{"dmxxextfdmr256",XX2(60,484), XX2DMR_MASK, FUTURE, PPCVLE, {XB5p, DMR, P2}},
+{"dmxxinstdmr256",XX2(60,485), XX2DMR_MASK, FUTURE, PPCVLE, {DMR, XB5p, P2}},
{"xvcvuxddp", XX2(60,488), XX2_MASK, PPCVSX, PPCVLE, {XT6, XB6}},
{"xvnabsdp", XX2(60,489), XX2_MASK, PPCVSX, PPCVLE, {XT6, XB6}},
{"xvtstdcdp", XX2(60,490), XX2DCMXS_MASK, PPCVSX3, PPCVLE, {XT6, XB6, DCMXS}},
@@ -9624,34 +9757,63 @@ const struct powerpc_opcode prefix_opcodes[] = {
{"plq", P8LS|OP(56), P_D_MASK, POWER10, 0, {RTQ, D34, PRAQ, PCREL}},
{"pld", P8LS|OP(57), P_D_MASK, POWER10, 0, {RT, D34, PRA0, PCREL}},
{"plxvp", P8LS|OP(58), P_D_MASK, POWER10, 0, {XTP, D34, PRA0, PCREL}},
+{"pmdmxvi8ger4pp",PMMIRR|XX3(59,2), P_GER4_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK4}},
{"pmxvi8ger4pp", PMMIRR|XX3(59,2), P_GER4_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK4}},
+{"pmdmxvi8ger4", PMMIRR|XX3(59,3), P_GER4_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK4}},
{"pmxvi8ger4", PMMIRR|XX3(59,3), P_GER4_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK4}},
+{"pmdmxvf16ger2pp",PMMIRR|XX3(59,18), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvf16ger2pp", PMMIRR|XX3(59,18), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf16ger2", PMMIRR|XX3(59,19), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvf16ger2", PMMIRR|XX3(59,19), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf32gerpp",PMMIRR|XX3(59,26), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
{"pmxvf32gerpp", PMMIRR|XX3(59,26), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
+{"pmdmxvf32ger", PMMIRR|XX3(59,27), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
{"pmxvf32ger", PMMIRR|XX3(59,27), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
+{"pmdmxvi4ger8pp",PMMIRR|XX3(59,34), P_GER8_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK8}},
{"pmxvi4ger8pp", PMMIRR|XX3(59,34), P_GER8_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK8}},
+{"pmdmxvi4ger8", PMMIRR|XX3(59,35), P_GER8_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK8}},
{"pmxvi4ger8", PMMIRR|XX3(59,35), P_GER8_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK8}},
+{"pmdmxvi16ger2spp",PMMIRR|XX3(59,42), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvi16ger2spp",PMMIRR|XX3(59,42), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvi16ger2s",PMMIRR|XX3(59,43), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvi16ger2s", PMMIRR|XX3(59,43), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvbf16ger2pp",PMMIRR|XX3(59,50), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvbf16ger2pp",PMMIRR|XX3(59,50), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvbf16ger2",PMMIRR|XX3(59,51), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvbf16ger2", PMMIRR|XX3(59,51), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf64gerpp",PMMIRR|XX3(59,58), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
{"pmxvf64gerpp", PMMIRR|XX3(59,58), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
+{"pmdmxvf64ger", PMMIRR|XX3(59,59), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
{"pmxvf64ger", PMMIRR|XX3(59,59), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
+{"pmdmxvi16ger2", PMMIRR|XX3(59,75), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvi16ger2", PMMIRR|XX3(59,75), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf16ger2np",PMMIRR|XX3(59,82), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvf16ger2np", PMMIRR|XX3(59,82), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf32gernp",PMMIRR|XX3(59,90), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
{"pmxvf32gernp", PMMIRR|XX3(59,90), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
+{"pmdmxvi8ger4spp",PMMIRR|XX3(59,99), P_GER4_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK4}},
{"pmxvi8ger4spp", PMMIRR|XX3(59,99), P_GER4_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK4}},
+{"pmdmxvi16ger2pp",PMMIRR|XX3(59,107), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvi16ger2pp", PMMIRR|XX3(59,107), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvbf16ger2np",PMMIRR|XX3(59,114),P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvbf16ger2np",PMMIRR|XX3(59,114), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf64gernp",PMMIRR|XX3(59,122), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
{"pmxvf64gernp", PMMIRR|XX3(59,122), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
+{"pmdmxvf16ger2pn",PMMIRR|XX3(59,146), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvf16ger2pn", PMMIRR|XX3(59,146), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf32gerpn",PMMIRR|XX3(59,154), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
{"pmxvf32gerpn", PMMIRR|XX3(59,154), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
+{"pmdmxvbf16ger2pn",PMMIRR|XX3(59,178),P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvbf16ger2pn",PMMIRR|XX3(59,178), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf64gerpn",PMMIRR|XX3(59,186), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
{"pmxvf64gerpn", PMMIRR|XX3(59,186), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
+{"pmdmxvf16ger2nn",PMMIRR|XX3(59,210), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvf16ger2nn", PMMIRR|XX3(59,210), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf32gernn",PMMIRR|XX3(59,218), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
{"pmxvf32gernn", PMMIRR|XX3(59,218), P_GER_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK}},
+{"pmdmxvbf16ger2nn",PMMIRR|XX3(59,242),P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
{"pmxvbf16ger2nn",PMMIRR|XX3(59,242), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
+{"pmdmxvf64gernn",PMMIRR|XX3(59,250), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
{"pmxvf64gernn", PMMIRR|XX3(59,250), P_GER64_MASK, POWER10, 0, {ACC, XA6ap, XB6a, XMSK, YMSK2}},
{"pstq", P8LS|OP(60), P_D_MASK, POWER10, 0, {RSQ, D34, PRA0, PCREL}},
{"pstd", P8LS|OP(61), P_D_MASK, POWER10, 0, {RS, D34, PRA0, PCREL}},
--
2.27.0
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [COMMITTED] PowerPC: Add support for RFC02653 - Dense Math Facility
2022-11-02 19:25 ` Peter Bergner
@ 2022-11-03 6:33 ` Tom de Vries
0 siblings, 0 replies; 6+ messages in thread
From: Tom de Vries @ 2022-11-03 6:33 UTC (permalink / raw)
To: Peter Bergner, Binutils; +Cc: Carl Love
On 11/2/22 20:25, Peter Bergner wrote:
> On 11/2/22 1:41 PM, Tom de Vries wrote:
>> On 11/2/22 19:32, Peter Bergner wrote:
>>> On 11/2/22 1:27 PM, Tom de Vries wrote:
>>>> The disassembly shows:
>>>> ...
>>>> 0x000000000000063c <+1596>: pmdmxvbf16ger2 a4,vs0,vs1,0,0,0^M
>>>> ...
>>>> while the test expects pmxvbf16ger2.
>>>
>>> Yes, Carl notified me of the fallout. I gave him a patch that updates
>>> the expected output to look for the new name. Carl has agreed to
>>> shepherd the patch to fix the fallout.
>>
>> Could someone explain why that is the right solution?
>>
>> I don't see any dm<n> registers in the instruction "pmxvbf16ger2 a4,vs0,vs1,0,0,0".
>>
>> Why then should the pmdmxvbf16ger2 name be used?
>
> a4 (ie, accumulator register 4) is a DMR register...more specifically,
> half of a DMR register. The two RFCs which I pushed support for changed
> the "official" ISA mnemonic to the *dm* name because that is more
> architecturally correct. We kept the old names as extended mnemonics,
> so the only visible change to users is that the disassembly behavior
> has changed, hence the need for the change to the expected output.
Understood, thanks for the explanation.
- Tom
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [COMMITTED] PowerPC: Add support for RFC02653 - Dense Math Facility
2022-11-02 18:41 ` Tom de Vries
@ 2022-11-02 19:25 ` Peter Bergner
2022-11-03 6:33 ` Tom de Vries
0 siblings, 1 reply; 6+ messages in thread
From: Peter Bergner @ 2022-11-02 19:25 UTC (permalink / raw)
To: Tom de Vries, Binutils; +Cc: Carl Love
On 11/2/22 1:41 PM, Tom de Vries wrote:
> On 11/2/22 19:32, Peter Bergner wrote:
>> On 11/2/22 1:27 PM, Tom de Vries wrote:
>>> The disassembly shows:
>>> ...
>>> 0x000000000000063c <+1596>: pmdmxvbf16ger2 a4,vs0,vs1,0,0,0^M
>>> ...
>>> while the test expects pmxvbf16ger2.
>>
>> Yes, Carl notified me of the fallout. I gave him a patch that updates
>> the expected output to look for the new name. Carl has agreed to
>> shepherd the patch to fix the fallout.
>
> Could someone explain why that is the right solution?
>
> I don't see any dm<n> registers in the instruction "pmxvbf16ger2 a4,vs0,vs1,0,0,0".
>
> Why then should the pmdmxvbf16ger2 name be used?
a4 (ie, accumulator register 4) is a DMR register...more specifically,
half of a DMR register. The two RFCs which I pushed support for changed
the "official" ISA mnemonic to the *dm* name because that is more
architecturally correct. We kept the old names as extended mnemonics,
so the only visible change to users is that the disassembly behavior
has changed, hence the need for the change to the expected output.
Peter
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [COMMITTED] PowerPC: Add support for RFC02653 - Dense Math Facility
2022-11-02 18:32 ` Peter Bergner
@ 2022-11-02 18:41 ` Tom de Vries
2022-11-02 19:25 ` Peter Bergner
0 siblings, 1 reply; 6+ messages in thread
From: Tom de Vries @ 2022-11-02 18:41 UTC (permalink / raw)
To: Peter Bergner, Binutils; +Cc: Carl Love
On 11/2/22 19:32, Peter Bergner wrote:
> On 11/2/22 1:27 PM, Tom de Vries wrote:
>>> +{"pmdmxvbf16ger2",PMMIRR|XX3(59,51), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
>>> {"pmxvbf16ger2", PMMIRR|XX3(59,51), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
>>
>> Hi,
>>
>> I noticed a regression in a gdb test-case:
>> ...
>> FAIL: gdb.arch/powerpc-power10.exp: found: pmxvbf16ger2 a4,vs0,vs1,0,0,0
>> ...
>> and I suspect it's because of this commit.
>>
>>
>> The disassembly shows:
>> ...
>> 0x000000000000063c <+1596>: pmdmxvbf16ger2 a4,vs0,vs1,0,0,0^M
>> ...
>> while the test expects pmxvbf16ger2.
>
> Yes, Carl notified me of the fallout. I gave him a patch that updates
> the expected output to look for the new name. Carl has agreed to
> shepherd the patch to fix the fallout.
Could someone explain why that is the right solution?
I don't see any dm<n> registers in the instruction "pmxvbf16ger2
a4,vs0,vs1,0,0,0".
Why then should the pmdmxvbf16ger2 name be used?
Thanks,
- Tom
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [COMMITTED] PowerPC: Add support for RFC02653 - Dense Math Facility
2022-11-02 18:27 Tom de Vries
@ 2022-11-02 18:32 ` Peter Bergner
2022-11-02 18:41 ` Tom de Vries
0 siblings, 1 reply; 6+ messages in thread
From: Peter Bergner @ 2022-11-02 18:32 UTC (permalink / raw)
To: Tom de Vries, Binutils; +Cc: Carl Love
On 11/2/22 1:27 PM, Tom de Vries wrote:
>> +{"pmdmxvbf16ger2",PMMIRR|XX3(59,51), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
>> {"pmxvbf16ger2", PMMIRR|XX3(59,51), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
>
> Hi,
>
> I noticed a regression in a gdb test-case:
> ...
> FAIL: gdb.arch/powerpc-power10.exp: found: pmxvbf16ger2 a4,vs0,vs1,0,0,0
> ...
> and I suspect it's because of this commit.
>
>
> The disassembly shows:
> ...
> 0x000000000000063c <+1596>: pmdmxvbf16ger2 a4,vs0,vs1,0,0,0^M
> ...
> while the test expects pmxvbf16ger2.
Yes, Carl notified me of the fallout. I gave him a patch that updates
the expected output to look for the new name. Carl has agreed to
shepherd the patch to fix the fallout.
Peter
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [COMMITTED] PowerPC: Add support for RFC02653 - Dense Math Facility
@ 2022-11-02 18:27 Tom de Vries
2022-11-02 18:32 ` Peter Bergner
0 siblings, 1 reply; 6+ messages in thread
From: Tom de Vries @ 2022-11-02 18:27 UTC (permalink / raw)
To: Binutils, Peter Bergner; +Cc: Carl Love
> +{"pmdmxvbf16ger2",PMMIRR|XX3(59,51), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
> {"pmxvbf16ger2", PMMIRR|XX3(59,51), P_GER2_MASK, POWER10, 0, {ACC, XA6a, XB6a, XMSK, YMSK, PMSK2}},
Hi,
I noticed a regression in a gdb test-case:
...
FAIL: gdb.arch/powerpc-power10.exp: found: pmxvbf16ger2 a4,vs0,vs1,0,0,0
...
and I suspect it's because of this commit.
The disassembly shows:
...
0x000000000000063c <+1596>: pmdmxvbf16ger2 a4,vs0,vs1,0,0,0^M
...
while the test expects pmxvbf16ger2.
Thanks,
- Tom
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2022-11-03 6:33 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-10-28 0:35 [COMMITTED] PowerPC: Add support for RFC02653 - Dense Math Facility Peter Bergner
2022-11-02 18:27 Tom de Vries
2022-11-02 18:32 ` Peter Bergner
2022-11-02 18:41 ` Tom de Vries
2022-11-02 19:25 ` Peter Bergner
2022-11-03 6:33 ` Tom de Vries
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).