Intel AVX512 FP16 instructions use maps 3, 5 and 6. Maps 5 and 6 use 3 bits
in the EVEX.mmm field (0b101, 0b110). Map 5 is for instructions that were FP32
in map 1 (0Fxx). Map 6 is for instructions that were FP32 in map 2 (0F38xx).
There are some exceptions to this rule. Some things in map 1 (0Fxx) with imm8
operands predated our current conventions; those instructions moved to map 3.
FP32 things in map 3 (0F3Axx) found new opcodes in map3 for FP16 because map3
is very sparsely populated. Most of the FP16 instructions share opcodes and
prefix (EVEX.pp) bits with the related FP32 operations.
Intel AVX512 FP16 instructions has new displacements scaling rules, please refer
to the public software developer manual for detail information.
gas/
2021-08-05 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
H.J. Lu <hongjiu.lu@intel.com>
Wei Xiao <wei3.xiao@intel.com>
Lili Cui <lili.cui@intel.com>
* config/tc-i386.c (struct Broadcast_Operation): Adjust comment.
(cpu_arch): Add .avx512_fp16.
(cpu_noarch): Add noavx512_fp16.
(pte): Add evexmap5 and evexmap6.
(build_evex_prefix): Handle EVEXMAP5 and EVEXMAP6.
(check_VecOperations): Handle {1to32}.
(check_VecOperands): Handle CheckRegNumb.
(check_word_reg): Handle Toqword.
(i386_error): Add invalid_dest_and_src_register_set.
(match_template): Handle invalid_dest_and_src_register_set.
* doc/c-i386.texi: Document avx512_fp16, noavx512_fp16.
opcodes/
2021-08-05 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
H.J. Lu <hongjiu.lu@intel.com>
Wei Xiao <wei3.xiao@intel.com>
Lili Cui <lili.cui@intel.com>
* i386-dis.c (EXwScalarS): New.
(EXxh): Ditto.
(EXxhc): Ditto.
(EXxmmqh): Ditto.
(EXxmmqdh): Ditto.
(EXEvexXwb): Ditto.
(DistinctDest_Fixup): Ditto.
(enum): Add xh_mode, evex_half_bcst_xmmqh_mode, evex_half_bcst_xmmqdh_mode
and w_swap_mode.
(enum): Add PREFIX_EVEX_0F3A08_W_0, PREFIX_EVEX_0F3A0A_W_0,
PREFIX_EVEX_0F3A26, PREFIX_EVEX_0F3A27, PREFIX_EVEX_0F3A56,
PREFIX_EVEX_0F3A57, PREFIX_EVEX_0F3A66, PREFIX_EVEX_0F3A67,
PREFIX_EVEX_0F3AC2, PREFIX_EVEX_MAP5_10, PREFIX_EVEX_MAP5_11,
PREFIX_EVEX_MAP5_1D, PREFIX_EVEX_MAP5_2A, PREFIX_EVEX_MAP5_2C,
PREFIX_EVEX_MAP5_2D, PREFIX_EVEX_MAP5_2E, PREFIX_EVEX_MAP5_2F,
PREFIX_EVEX_MAP5_51, PREFIX_EVEX_MAP5_58, PREFIX_EVEX_MAP5_59,
PREFIX_EVEX_MAP5_5A_W_0, PREFIX_EVEX_MAP5_5A_W_1,
PREFIX_EVEX_MAP5_5B_W_0, PREFIX_EVEX_MAP5_5B_W_1,
PREFIX_EVEX_MAP5_5C, PREFIX_EVEX_MAP5_5D, PREFIX_EVEX_MAP5_5E,
PREFIX_EVEX_MAP5_5F, PREFIX_EVEX_MAP5_78, PREFIX_EVEX_MAP5_79,
PREFIX_EVEX_MAP5_7A, PREFIX_EVEX_MAP5_7B, PREFIX_EVEX_MAP5_7C,
PREFIX_EVEX_MAP5_7D_W_0, PREFIX_EVEX_MAP6_13, PREFIX_EVEX_MAP6_56,
PREFIX_EVEX_MAP6_57, PREFIX_EVEX_MAP6_D6, PREFIX_EVEX_MAP6_D7
(enum): Add EVEX_MAP5 and EVEX_MAP6.
(enum): Add EVEX_W_MAP5_5A, EVEX_W_MAP5_5B,
EVEX_W_MAP5_78_P_0, EVEX_W_MAP5_78_P_2, EVEX_W_MAP5_79_P_0,
EVEX_W_MAP5_79_P_2, EVEX_W_MAP5_7A_P_2, EVEX_W_MAP5_7A_P_3,
EVEX_W_MAP5_7B_P_2, EVEX_W_MAP5_7C_P_0, EVEX_W_MAP5_7C_P_2,
EVEX_W_MAP5_7D, EVEX_W_MAP6_13_P_0, EVEX_W_MAP6_13_P_2,
(get_valid_dis386): Properly handle new instructions.
(intel_operand_size): Handle new modes.
(OP_E_memory): Ditto.
(OP_EX): Ditto.
* i386-dis-evex.h: Updated for AVX512_FP16.
* i386-dis-evex-mod.h: Updated for AVX512_FP16.
* i386-dis-evex-prefix.h: Updated for AVX512_FP16.
* i386-dis-evex-reg.h : Updated for AVX512_FP16.
* i386-dis-evex-w.h : Updated for AVX512_FP16.
* i386-gen.c (cpu_flag_init): Add CPU_AVX512_FP16_FLAGS,
and CPU_ANY_AVX512_FP16_FLAGS. Update CPU_ANY_AVX512F_FLAGS
and CPU_ANY_AVX512BW_FLAGS.
(cpu_flags): Add CpuAVX512_FP16.
(opcode_modifiers): Add DistinctDest.
* i386-opc.h (enum): (AVX512_FP16): New.
(i386_opcode_modifier): Add reqdistinctreg.
(i386_cpu_flags): Add cpuavx512_fp16.
(EVEXMAP5): Defined as a macro.
(EVEXMAP6): Ditto.
* i386-opc.tbl: Add Intel AVX512_FP16 instructions.
* i386-init.h: Regenerated.
* i386-tbl.h: Ditto.
opcodes/
* s390-opc.c (INSTR_SIY_RD): New instruction format.
(MASK_SIY_RD): New instruction mask.
* s390-opc.txt: Change instruction format of lpswey to SIY_RD.
gas/
* testsuite/gas/s390/zarch-arch14.d: Remove last operand of
lpswey.
* testsuite/gas/s390/zarch-arch14.s: Likewise.
While EVEX.L'L are indeed ignored when EVEX.b stands for just SAE,
EVEX.b itself is not ignored when an insn permits neither rounding
control nor SAE.
While changing this aspect of EVEX.b handling, also alter unduly set
embedded broadcast: Don't call BadOp(), screwing up subsequent
disassembly, but emit "{bad}" instead.
Their sole use is for {,V}EXTRACTPS / {,V}P{EXT,INS}RB respectively; for
consistency also limit use of dqw_mode to Jdqw. 64-bit disassembly
reflecting REX.W / VEX.W is not in line with the assembler's opcode
table having NoRex64 / VexWIG in all respective templates, i.e. assembly
input isn't being honored there either. Obviously the 0FC5 encodings of
{,V}PEXTRW then also need adjustment for consistency reasons.
It has only a single use and can easily be represented by dq_mode
instead. Plus its handling in intel_operand_size() was duplicating
that of vex_vsib_{d,q}_w_dq_mode anyway.
Unlike the high bit of VEX.vvvv / EVEX.vvvv, EVEX.V' is not ignored
outside of 64-bit mode. Oddly enough there already are tests for these
cases, but their expectations were wrong. (This may have been based on
an old SDM version, where the restriction wasn't properly spelled out.)
What so far was OP_E_register() can be easily reused also for OP_G().
Add suitable parameters to the function and move the invocation of
swap_operand() to OP_E(). Adjust MOVSXD's first operand: There never was
a need to use movsxd_mode there, and its use gets in the way of the code
folding.
With EVEX.W clear the instruction doesn't ignore the rounding mode, but
(like for other insns without rounding semantics) EVEX.b set causes #UD.
Hence the handling of EVEX.W needs to be done when processing
evex_rounding_64_mode, not at the decode stages.
Derive a new 64-bit testcase from the 32-bit one to cover the different
EVEX.W treatment in both cases.
Some extern declarations differ in constnes to their definitions too.
Let's make sure this sort of thing doesn't happen again, but putting
the externs in a header where they belong.
gas/
* config/tc-nds32.c (nds32_keyword_gpr): Don't declare.
(md_begin): Constify k.
opcodes/
* nds32-dis.c (nds32_find_reg_keyword): Constify arg and return.
(nds32_parse_audio_ext, nds32_parse_opcode): Constify psys_reg.
(nds32_field_table, nds32_opcode_table, nds32_keyword_table),
(nds32_opcodes, nds32_operand_fields, nds32_keywords),
(nds32_keyword_gpr): Move declarations to..
* nds32-asm.h: ..here, constifying to match definitions.
This makes it easier to override to point to an older version of guile.
The current cgen code doesn't work with guile-2, so need to point to an
older guile-1.8.
For the variables that don't need to be exported, mark them static.
For the ones shared between modules, add a "nds32_" prefix to avoid
collisions with these common variable names.
This is exporting the variable "opcodes" as a large writable blob.
This is not a namespace friendly name, so add a "microblaze" prefix,
and then sprinkle const over its definition & use.
This commit provides a small performance improvement when starting up
CGEN based disassemblers by making use of __builtin_popcount.
The #if check used in this commit was copied from bfd/elf32-arm.c
where __builtin_popcount is also used.
I ran into this code while investigating some GDB tests that would
occasionally timeout. One of the reason these tests were having
problems is that the m16c and m32c disassemblers take so long to
initialise themselves. Speeding up count_decodable_bits helps, but is
not a total solution. Still, this felt like an easy win which added
minimal extra complexity, so I figure its worth doing.
opcodes/ChangeLog:
* cgen-dis.c (count_decodable_bits): Use __builtin_popcount when
available.
Commit 54758c3e39 made changes to the picojava support based on
https://sourceware.org/pipermail/binutils/2005-November/045136.html
An update from picojava to picojava II, I think. Unfortunately the
patch neglected any changes to the gas testsuite, resulting in
"FAIL: pj" since that date. This patch makes a few relatively simple
changes to cure the regression.
gas/
* config/tc-pj.c (md_apply_fix): Apply PJ_CODE_REL32 relocs.
* testsuite/gas/pj/ops.s: Update jsr, ret, getstatic,
putstatic, getfield, putfield, invokevirtual, invokespecial,
invokestatic, invokeinterface, goto_w, jsr_w assembly. Delete
version 1 picojava opcodes.
* testsuite/gas/pj/ops.d: Match expected output.
opcodes/
* pj-dis.c (print_insn_pj): Don't print trailing tab. Do
print separator for pcrel insns.
Fixes a 16 year old bug report, which even came with a patch.
opcodes/
PR 1202
* mcore-dis.c (print_insn_mcore): Correct loopt disassembly.
Use unsigned int for inst.
gas/
PR 1202
* testsuite/gas/mcore/allinsn.d: Correct loopt expected output.
The idea of this change is simple: Populate a data structure, namely
"disasm_option_and_arg_t" from "include/dis-asm.h", to encompass the
disassembly options and their possible arguments.
This will make it easier to manage or extend those options by adapting
entries in a data structure, "arc_options". There will be lesser need
to hard-code the options in the code itself. Moreover, ARC GDB will
use this population function, "disassembler_options_arc ()", to enable
the "set disassembler-option" for ARC targets. The gdb change will be
in a separate patch though.
The changes in this patch can be divided into:
1) Introduction of "disassembler_options_arc ()" that will return a
"disasm_option_and_arg_t" structure representing the disassembly
options and their likely arguments.
2) New data type "arc_options_arg_t" and new data "arc_options".
These are the internals for keeping track of options and arguments
entries that can easily be extended.
3) To print the options, the "print_arc_disassembler_options ()" has
been adjusted to use this dynamically built structure instead of having
them hard-coded inside.
To see this in effect, one can look into the output of:
$ ./binutils/objdump --help
...
The following ARC specific disassembler options are...
...
include/ChangeLog:
* dis-asm.h (disassembler_options_arc): New prototype.
opcodes/ChangeLog:
* arc-dis.c (arc_option_arg_t): New enumeration.
(arc_options): New variable.
(disassembler_options_arc): New function.
(print_arc_disassembler_options): Reimplement in terms of
"disassembler_options_arc".
opcodes/
* ppc-dis.c (lookup_powerpc): Test deprecated field when -Many.
Don't special case PPC_OPCODE_RAW.
(lookup_prefix): Likewise.
(lookup_vle, lookup_spe2): Similarly. Add dialect parameter and..
(print_insn_powerpc): ..update caller.
* ppc-opc.c (EXT): Define.
(powerpc_opcodes): Mark extended mnemonics with EXT.
(prefix_opcodes, vle_opcodes): Likewise.
(XISEL, XISEL_MASK): Add cr field and simplify.
(powerpc_opcodes): Use XISEL with extended isel mnemonics and sort
all isel variants to where the base mnemonic belongs. Sort dstt,
dststt and dssall.
gas/
* testsuite/gas/ppc/raw.s,
* testsuite/gas/ppc/raw.d: New test.
* testsuite/gas/ppc/ppc.exp: Run it.
Group legacy instructions using the COP0, COP2, COP3 opcodes together
and by their coprocessor number, and move them towards the end of the
opcode table. No functional change.
With the addition of explicit ISA exclusions this is maybe not strictly
necessary anymore as the individual legacy instructions are not supposed
to match ISA levels or CPU implementations that have discarded them or
replaced with a new instruction each, but let's not have them scattered
randomly across blocks of unrelated instruction sets where someone chose
to put them previously. Perhaps they could be put back in alphabetical
order in the main instruction block, but let's leave it for another
occasion.
opcodes/
* mips-opc.c (mips_builtin_opcodes): Reorder legacy COP0, COP2,
COP3 opcode instructions.
Adjust opcode table entries for coprocessor instructions that have been
removed from certain ISA levels or CPU implementations as follows:
- remove CP0 memory access instructions from MIPS II up as the LWC0 and
SWC0 opcodes have been reused for the LL and SC instructions
respectively[1]; strictly speaking LWC0 and SWC0 have never really
been defined in the first place[2], but let's keep them for now in
case an odd implementation did,
- remove CP0 branch instructions from MIPS IV[3] and MIPS32[4] up, as
they have been removed as from those ISAs,
- remove CP0 control register move instructions from MIPS32 up, as they
have been removed as from that ISA[5],
- remove the RFE instruction from MIPS III[6] and MIPS32[7] up, as it
has been removed as from those ISAs in favour to ERET,
- remove CP2 instructions from Vr5400 CPUs as their encodings have been
reused for the multimedia instruction set extensions[8] and no CP2
registers exist[9],
- remove CP3 memory access instructions from MIPS III up as coprocessor
3 has been removed as from that ISA[10][11] and from MIPS32 up as the
LWC3 opcode has been reused for the PREF instruction and consequently
all the four memory access instructions removed from the ISA (though
the COP3 opcode has been retained)[12].
Update the testsuite accordingly.
References:
[1] Charles Price, "MIPS IV Instruction Set", MIPS Technologies, Inc.,
Revision 3.2, September, 1995, Table A-38 "CPU Instruction Encoding
- MIPS II Architecture", p. A-178
[2] same, Section A.2.5.1 "Coprocessor Load and Store", p. A-12
[3] "MIPS R10000 Microprocessor User's Manual", Version 2.0, MIPS
Technologies, Inc., January 29, 1997, Section 14.25 "CP0
Instructions", Subsection "Branch on Coprocessor 0", p. 285
[4] "MIPS32 Architecture For Programmers, Volume II: The MIPS32
Instruction Set", MIPS Technologies, Inc., Document Number:
MD00086, Revision 1.00, June 9, 2003, Table A-9 "MIPS32 COP0
Encoding of rs Field", p. 242
[5] same
[6] Joe Heinrich, "MIPS R4000 Microprocessor User's Manual", Second
Edition, MIPS Technologies, Inc., April 1, 1994, Figure A-2 "R4000
Opcode Bit Encoding", p. A-182
[8] "Vr5432 64-bit MIPS RISC Microprocessor User's Manual, Volume 1",
NEC Electronics Inc., Document No. U13751EU5V0UM00, May 2000,
Section 1.2.3 "CPU Instruction Set Overview", p. 9
[9] "Vr5432 64-bit MIPS RISC Microprocessor User's Manual, Volume 2",
NEC Electronics Inc., Document No. U13751EU5V0UM00, May 2000,
Section 19.2 "Multimedia Instruction Format", p. 681
[10] Charles Price, "MIPS IV Instruction Set", MIPS Technologies, Inc.,
Revision 3.2, September, 1995, Section A 8.3.4 "Coprocessor 3 -
COP3 and CP3 load/store", p. A-176
[11] same, Table A-39 "CPU Instruction Encoding - MIPS III
Architecture", p. A-179
[12] "MIPS32 Architecture For Programmers, Volume II: The MIPS32
Instruction Set", MIPS Technologies, Inc., Document Number:
MD00086, Revision 1.00, August 29, 2002, Table A-2 "MIPS32 Encoding
of the Opcode Field", p. 241
opcodes/
* mips-opc.c (mips_builtin_opcodes): Update exclusion list for
"ldc2", "ldc3", "lwc0", "lwc2", "lwc3", "sdc2", "sdc3", "swc0",
"swc2", "swc3", "cfc0", "ctc0", "bc2f", "bc2fl", "bc2t",
"bc2tl", "cfc2", "ctc2", "dmfc2", "dmtc2", "mfc2", "mtc2",
"bc3f", "bc3fl", "bc3t", "bc3tl", "cfc3", "ctc3", "mfc3",
"mtc3", "bc0f", "bc0fl", "bc0t", "bc0tl", "rfe", "c2", "c3",
"cop2", and "cop3" entries.
gas/
* testsuite/gas/mips/mips32@isa-override-1.d: Update for LDC3
instruction removal.
* testsuite/gas/mips/mips32r2@isa-override-1.d: Likewise.
Coprocessor 3 has been removed from the MIPS ISA as from MIPS III[1][2]
with the LDC3 and SDC3 instructions having been replaced with LD and SD
instructions respectively and therefore the doubleword move instructions
from and to that coprocessor have never materialized (for 32-bit ISAs
coprocessor 3 has likewise been removed as from MIPS32r2[3]). Remove
the DMFC3 and DMTC3 instructions from the opcode table then to avoid
confusion.
References:
[1] Charles Price, "MIPS IV Instruction Set", MIPS Technologies, Inc.,
Revision 3.2, September, 1995, Section A 8.3.4 "Coprocessor 3 - COP3
and CP3 load/store", p. A-176
[2] same, Table A-39 "CPU Instruction Encoding - MIPS III Architecture",
p. A-179
[3] "MIPS32 Architecture For Programmers, Volume II: The MIPS32
Instruction Set", MIPS Technologies, Inc., Document Number: MD00086,
Revision 2.00, June 9, 2003, Table A-2 "MIPS32 Encoding of the
Opcode Field", p. 317
opcodes/
* mips-opc.c (mips_builtin_opcodes): Remove "dmfc3" and "dmtc3"
entries and associated comments.
Fix a commit b015e599c7 ("[MIPS] Add new virtualization instructions"),
<https://sourceware.org/ml/binutils/2013-05/msg00118.html>, regression
and bring the disassembly of the RFE instruction back for the relevant
ISA levels.
It is because the "rfe" opcode table entry was incorrectly moved behind
the catch-all generic "c0" entry for CP0 instructions, causing output
like:
00: 42000010 c0 0x10
to be produced rather than:
00: 42000010 rfe
even for ISA levels that do include the RFE instruction.
Move the "rfe" entry ahead of "c0" then, correcting the problem. Add a
suitable test case.
opcodes/
* mips-opc.c (mips_builtin_opcodes): Move the "rfe" entry ahead
of "c0".
gas/
* testsuite/gas/mips/rfe.d: New test.
* testsuite/gas/mips/rfe.s: New test source.
* testsuite/gas/mips/mips.exp: Run the new test.
The two CP1 control registers defined by legacy ISAs used to be referred
to by various names, such as FCR0, FCR31, FSR, however their documented
full names have always been the Implementation and Revision, and Control
and Status respectively, so the FIR and FCSR acronyms coming from modern
ISA revisions will be just as unambiguous while improving the clarity of
disassembly. Do not update the TX39 though as it did not have an FPU.
opcodes/
* mips-dis.c (mips_cp1_names_mips): New variable.
(mips_arch_choices): Use it rather than `mips_cp1_names_numeric'
for "r3000", "r4000", "r4010", "vr4100", "vr4111", "vr4120",
"r4300", "r4400", "r4600", "r4650", "r5000", "vr5400", "vr5500",
"r5900", "r6000", "rm7000", "rm9000", "r8000", "r10000",
"r12000", "r14000", "r16000", "mips5", "loongson2e", and
"loongson2f".
gas/
* testsuite/gas/mips/cp1-names-r3900.d: New test.
* testsuite/gas/mips/mips.exp: Run the new test.
* testsuite/gas/mips/branch-misc-3.d: Update disassembly
according to changes to opcodes.
* testsuite/gas/mips/cp1-names-r3000.d: Likewise.
* testsuite/gas/mips/cp1-names-r4000.d: Likewise.
* testsuite/gas/mips/relax-swap1-mips1.d: Likewise.
* testsuite/gas/mips/relax-swap1-mips2.d: Likewise.
* testsuite/gas/mips/trunc.d: Likewise.
The CP0 control register set has never been defined, however encodings
for the CFC0 and CTC0 instructions remained available for implementers
up until the MIPS32 ISA declared them invalid and causing the Reserved
Instruction exception[1]. Therefore we handle them for both assembly
and disassembly, however in the latter case the names of CP0 registers
from the regular set are incorrectly printed if named registers are
requested. This is because we do not define separate operand classes
for coprocessor regular and control registers respectively, which means
the disassembler has no way to tell the two cases apart. Consequently
nonsensical disassembly is produced like:
cfc0 v0,c0_random
Later the MIPSr5 ISA reused the encodings for XPA ASE MFHC0 and MTHC0
instructions[2] although it failed to document them in the relevant
opcode table until MIPSr6 only.
Correct the issue then by defining a new register class, OP_REG_CONTROL,
and corresponding operand codes, `g' and `y' for the two positions in
the machine instruction a control register operand can take. Adjust the
test cases affected accordingly.
While at it swap the regular MIPS opcode table "cfc0" and "ctc0" entries
with each other so that they come in the alphabetical order.
References:
[1] "MIPS32 Architecture For Programmers, Volume II: The MIPS32
Instruction Set", MIPS Technologies, Inc., Document Number: MD00086,
Revision 1.00, August 29, 2002, Table A-9 "MIPS32 COP0 Encoding of
rs Field", p. 242
[2] "MIPS Architecture For Programmers, Volume II-A: The MIPS32
Instruction Set", MIPS Technologies, Inc., Document Number: MD00086,
Revision 5.04, December 11, 2013, Section 3.2 "Alphabetical List of
Instructions", pp. 195, 216
include/
* opcode/mips.h: Document `g' and `y' operand codes.
(mips_reg_operand_type): Add OP_REG_CONTROL enumeration
constant.
gas/
* tc-mips.c (convert_reg_type) <OP_REG_CONTROL>: New case.
(macro) <M_TRUNCWS, M_TRUNCWD>: Use the `g' rather than `G'
operand code.
opcodes/
* mips-dis.c (print_reg) <OP_REG_COPRO>: Move control register
handling code over to...
<OP_REG_CONTROL>: ... this new case.
* mips-opc.c (decode_mips_operand) <'g', 'y'>: New cases.
(mips_builtin_opcodes): Update "cfc1", "ctc1", "cttc1", "cttc2",
"cfc0", "ctc0", "cfc2", "ctc2", "cfc3", and "ctc3" entries
replacing the `G' operand code with `g'. Update "cftc1" and
"cftc2" entries replacing the `E' operand code with `y'.
* micromips-opc.c (decode_micromips_operand) <'g'>: New case.
(micromips_opcodes): Update "cfc1", "cfc2", "ctc1", and "ctc2"
entries replacing the `G' operand code with `g'.
binutils/
* testsuite/binutils-all/mips/mips-xpa-virt-1.d: Correct CFC0
operand disassembly.
* testsuite/binutils-all/mips/mips-xpa-virt-3.d: Likewise.
