gas/ChangeLog:
2016-11-23 Jose E. Marchesi <jose.marchesi@oracle.com>
* testsuite/gas/sparc/sparc.exp (gas_64_check): Make sure the
hwcaps-bump test is run with 64-bit objects.
When the assembler finds an instruction which is part of a higher
opcode architecture it bumps the current opcode architecture. For
example:
$ echo "mwait" | as -bump
{standard input}: Assembler messages:
{standard input}:1: Warning: architecture bumped from "v6" to "v9m" on "mwait"
However, when two instructions pertaining to the same opcode
architecture but associated to different SPARC hardware capabilities
are found in the input stream, and no GAS architecture is specified in
the command line, the assembler bangs:
$ echo "mwait; wr %g0,%g1,%mcdper" | as -bump
{standard input}: Assembler messages:
{standard input}:1: Warning: architecture bumped from "v6" to "v9m" on "mwait"
{standard input}:1: Error: Hardware capability "sparc5" not enabled for "wr".
... and it should'nt, as WRMCDPER pertains to the same architecture
level than MWAIT.
This patch fixes this by extending the definition of sparc opcode
architectures to contain a set of hardware capabilities and making the
assembler to take these capabilities into account when updating the
set of allowed hwcaps when an architecture bump is triggered by some
instruction.
This way, hwcaps associated to architecture levels are maintained in
opcodes, while the assembler keeps the flexibiity of defining GAS
architectures including additional hwcaps (like -Asparcfmaf or the
v8plus* variants).
A test covering this failure case is included.
gas/ChangeLog:
2016-11-22 Jose E. Marchesi <jose.marchesi@oracle.com>
* config/tc-sparc.c: Move HWS_* and HWS2_* definitions to
opcodes/sparc-opc.c.
(sparc_arch): Clarify the new role of the hwcap_allowed and
hwcap2_allowed fields.
(sparc_arch_table): Remove HWS_* and HWS2_* instances from
hwcap_allowed and hwcap2_allowed respectively.
(md_parse_option): Include the opcode arch hwcaps when processing
-A.
(sparc_ip): Use the current opcode arch hwcaps to update
hwcap_allowed, as well of the hwcaps of the instruction triggering
the bump.
* testsuite/gas/sparc/hwcaps-bump.s: New file.
* testsuite/gas/sparc/hwcaps-bump.l: Likewise.
* testsuite/gas/sparc/sparc.exp (gas_64_check): Run tests in
hwcaps-bump.
include/ChangeLog:
2016-11-22 Jose E. Marchesi <jose.marchesi@oracle.com>
* opcode/sparc.h (sparc_opcode_arch): New fields hwcaps and
hwcaps2.
opcodes/ChangeLog:
2016-11-22 Jose E. Marchesi <jose.marchesi@oracle.com>
* sparc-opc.c (HWS_V8): Definition moved from
gas/config/tc-sparc.c.
(HWS_V9): Likewise.
(HWS_VA): Likewise.
(HWS_VB): Likewise.
(HWS_VC): Likewise.
(HWS_VD): Likewise.
(HWS_VE): Likewise.
(HWS_VV): Likewise.
(HWS_VM): Likewise.
(HWS2_VM): Likewise.
(sparc_opcode_archs): Initialize hwcaps and hwcaps2 fields of
existing entries.
gas/
2016-11-18 Claudiu Zissulescu <claziss@synopsys.com>
* testsuite/gas/arc/cl-warn.s: New file.
* testsuite/gas/arc/cpu-pseudop-1.d: Likewise.
* testsuite/gas/arc/cpu-pseudop-1.s: Likewise.
* testsuite/gas/arc/cpu-pseudop-2.d: Likewise.
* testsuite/gas/arc/cpu-pseudop-2.s: Likewise.
* testsuite/gas/arc/cpu-warn2.s: Likewise.
* config/tc-arc.c (selected_cpu): Initialize.
(feature_type): New struct.
(feature_list): New variable.
(arc_check_feature): New function.
(arc_select_cpu): Check for .cpu duplicates. Don't overwrite the
current cpu features. Check if a feature is available for a given
cpu.
(md_parse_option): Test if features are available for a given cpu.
Add support for FCMLA and FCADD complex arithmetic SIMD instructions.
FCMLA has an indexed element variant where the index range has to be
treated specially because a complex number takes two elements and the
indexed vector size depends on the other operands.
These complex number SIMD instructions are part of ARMv8.3
https://community.arm.com/groups/processors/blog/2016/10/27/armv8-a-architecture-2016-additions
include/
2016-11-18 Szabolcs Nagy <szabolcs.nagy@arm.com>
* opcode/aarch64.h (enum aarch64_opnd): Add AARCH64_OPND_IMM_ROT1,
AARCH64_OPND_IMM_ROT2, AARCH64_OPND_IMM_ROT3.
(enum aarch64_op): Add OP_FCMLA_ELEM.
opcodes/
2016-11-18 Szabolcs Nagy <szabolcs.nagy@arm.com>
* aarch64-tbl.h (QL_V3SAMEHSD_ROT, QL_ELEMENT_ROT): Define.
(aarch64_feature_simd_v8_3, SIMD_V8_3): Define.
(aarch64_opcode_table): Add fcmla and fcadd.
(AARCH64_OPERANDS): Add IMM_ROT{1,2,3}.
* aarch64-asm.h (aarch64_ins_imm_rotate): Declare.
* aarch64-asm.c (aarch64_ins_imm_rotate): Define.
* aarch64-dis.h (aarch64_ext_imm_rotate): Declare.
* aarch64-dis.c (aarch64_ext_imm_rotate): Define.
* aarch64-opc.h (enum aarch64_field_kind): Add FLD_rotate{1,2,3}.
* aarch64-opc.c (fields): Add FLD_rotate{1,2,3}.
(operand_general_constraint_met_p): Rotate and index range check.
(aarch64_print_operand): Handle rotate operand.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis-2.c: Likewise.
* aarch64-opc-2.c: Likewise.
gas/
2016-11-18 Szabolcs Nagy <szabolcs.nagy@arm.com>
* config/tc-aarch64.c (parse_operands): Handle AARCH64_OPND_IMM_ROT*.
* testsuite/gas/aarch64/advsimd-armv8_3.d: New.
* testsuite/gas/aarch64/advsimd-armv8_3.s: New.
* testsuite/gas/aarch64/illegal-fcmla.s: New.
* testsuite/gas/aarch64/illegal-fcmla.l: New.
* testsuite/gas/aarch64/illegal-fcmla.d: New.
PR gas/20732
* expr.c (integer_constant): If tc_allow_L_suffix is defined and
non-zero then accept a L or LL suffix.
* testsuite/gas/sparc/pr20732.d: New test source file.
* testsuite/gas/sparc/pr20732.d: New test output file.
* testsuite/gas/sparc/sparc.exp: Run new test.
Add support for system registers introduced in ARMv8.3
for pointer authentication.
opcodes/
2016-11-11 Szabolcs Nagy <szabolcs.nagy@arm.com>
* aarch64-opc.c (aarch64_sys_regs): Add apiakeylo_el1, apiakeyhi_el1,
apibkeylo_el1, apibkeyhi_el1, apdakeylo_el1, apdakeyhi_el1,
apdbkeylo_el1, apdbkeyhi_el1, apgakeylo_el1 and apgakeyhi_el1.
(aarch64_sys_reg_supported_p): Add feature test for new registers.
gas/
2016-11-11 Szabolcs Nagy <szabolcs.nagy@arm.com>
* testsuite/gas/aarch64/sysreg-3.s: New.
* testsuite/gas/aarch64/sysreg-3.d: New.
* testsuite/gas/aarch64/illegal-sysreg-3.l: New.
* testsuite/gas/aarch64/illegal-sysreg-3.d: New.
This patch adds support for a subset of the ARMv8.3 pointer authentication
instructions: XPACLRI, PACIA1716, PACIB1716, AUTIA1716, AUTIA1716, PACIAZ,
PACIASP, PACIBZ, PACISP, AUTIAZ, AUTIASP, AUTIBZ, AUTIBSP.
These are aliases to HINT #0x7, HINT #0x8, HINT #0xa, HINT #0xc, HINT #0xe,
HINT #0x18, HINT #0x19, ..., HINT #0x1f respectively.
For more details about pointer authentication in ARMv8.3 see
https://community.arm.com/groups/processors/blog/2016/10/27/armv8-a-architecture-2016-additions
opcodes/
2016-11-11 Szabolcs Nagy <szabolcs.nagy@arm.com>
* aarch64-tbl.h (aarch64_feature_v8_3, ARMV8_3, V8_3_INSN): New.
(arch64_opcode_table): Add xpaclri, pacia1716, pacib1716, autia1716,
autib1716, paciaz, paciasp, pacibz, pacibsp, autiaz, autiasp, autibz,
autibsp.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis-2.c: Regenerate.
gas/
2016-11-11 Szabolcs Nagy <szabolcs.nagy@arm.com>
* testsuite/gas/aarch64/system-3.s: New.
* testsuite/gas/aarch64/system-3.d: New.
* testsuite/gas/aarch64/system.d: Update expected output.
ARMv8.3 can be selected with -march=armv8.3-a command line option.
An overview of the ARMv8.3 architecture extension is at
https://community.arm.com/groups/processors/blog/2016/10/27/armv8-a-architecture-2016-additions
gas/
2016-11-11 Szabolcs Nagy <szabolcs.nagy@arm.com>
* config/tc-aarch64.c (aarch64_archs): Add "armv8.3-a".
* doc/c-aarch64.texi (-march): Likewise.
include/
2016-11-11 Szabolcs Nagy <szabolcs.nagy@arm.com>
* opcode/aarch64.h (AARCH64_FEATURE_V8_3): Define.
(AARCH64_ARCH_V8_3): Define.
(AARCH64_ARCH_V8_1, AARCH64_ARCH_V8_2): Simplify.
Since Bad_Opcode and FGRPd9_2 were the same in i386-dis.c, all
Bad_Opcode entries in float_reg were displaced as FGRPd9_2. This
patch adds an entry for Bad_Opcode in fgrps to avoid treating it
as FGRPd9_2.
gas/
PR binutils/20775
* testsuite/gas/i386/i386.exp: Run fpu-bad.
* testsuite/gas/i386/fpu-bad.d: New file.
* testsuite/gas/i386/fpu-bad.s: Likewise.
opcodes/
PR binutils/20775
* i386-dis.c (FGRPd9_2): Replace 0 with 1.
(FGRPd9_4): Replace 1 with 2.
(FGRPd9_5): Replace 2 with 3.
(FGRPd9_6): Replace 3 with 4.
(FGRPd9_7): Replace 4 with 5.
(FGRPda_5): Replace 5 with 6.
(FGRPdb_4): Replace 6 with 7.
(FGRPde_3): Replace 7 with 8.
(FGRPdf_4): Replace 8 with 9.
(fgrps): Add an entry for Bad_Opcode.
When we match against an address type operand within an instruction it
is important that we match exactly the right address type operand early
on, during the opcode selection phase. If we wait until the operand
insertion phase to check that we have the correct address operand, then
it is too late to select an alternative opcode. This becomes important
only when we have multiple opcodes with the same mnemonic, and operand
lists that differ only in the type of the address operands.
This commit fixes this issue, and adds some example instructions that
require this issue to be fixed (the instructions are identical except
for the address type operand).
gas/ChangeLog:
* config/tc-arc.c (find_opcode_match): Use insert function to
validate matching address type operands.
* testsuite/gas/arc/nps400-10.d: New file.
* testsuite/gas/arc/nps400-10.s: New file.
opcodes/ChangeLog:
* arc-opc.c (arc_flag_operands): Add F_DI14.
(arc_flag_classes): Add C_DI14.
* arc-nps400-tbl.h: Add new exc instructions.
Since the bpp instruction has been added the 16 bit wide pc relative
relocs might occur at offset 2 as well at offset 4 in an instruction.
With this patch the different adjustment is passed from
md_gather_operand to md_apply_fix via fx_pcrel_adjust field in the fix
data structure.
No regressions on s390x.
gas/ChangeLog:
2016-11-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
* config/tc-s390.c (md_gather_operands): Set fx_pcrel_adjust.
(md_apply_fix): Use/Set fx_pcrel_adjust.
* testsuite/gas/s390/zarch-zEC12.d: Add bpp reloc test pattern.
* testsuite/gas/s390/zarch-zEC12.s: Add bpp reloc test.
2016-11-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
gas/
* config/tc-arm.c (cortex-m33): Declare new processor.
* doc/c-arm.texi (-mcpu ARM command line option): Document new
Cortex-M33 processor.
* NEWS: Mention ARM Cortex-M33 support.
2016-11-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
gas/
* config/tc-arm.c (cortex-m23): Declare new processor.
* doc/c-arm.texi (-mcpu ARM command line option): Document new
Cortex-M23 processor.
* NEWS: Mention ARM Cortex-M23 support.
The long immediate operand chosen for one of the ldbit tests is
equivalent to a small negative value that would fit inside an s9
operand, leading to the assembler to choose an unexpected (but
legitimate) encoding of the instruction on 32-bit systems, and
therefore causing the test to fail. This commit fixes the test by
changing the offending limm value so that it can no longer be
interpreted as an s9 operand.
gas/ChangeLog:
* testsuite/gas/arc/nps400-6.s: Change ldbit tests so that
limm operands are out of the range of an s9, in order to fix
the test.
* testsuite/gas/arc/nps400-6.d: Updated to match new expected
output.
The current handling for arc instructions longer than 32-bits is all
handled as a special case in both the assembler and disassembler.
The problem with this approach is that it leads to code duplication,
selecting a long instruction is exactly the same process as selecting a
short instruction, except over more bits, in both cases we select based
on bit comparison, and initial operand insertion and extraction.
This commit unifies both the long and short instruction worlds,
converting the core opcodes library from being largely 32-bit focused,
to being largely 64-bit focused.
The changes are, on the whole, not too much. There's obviously a lot of
type changes but otherwise the bulk of the code just works. Most of the
actual functional changes are to code that previously handled the longer
48 or 64 bit instructions. The insert/extract handlers for these have
now been brought into line with the short instruction insert/extract
handlers.
All of the special case handling code that was previously added has now
been removed again. Overall, this commit reduces the amount of code in
the arc assembler and disassembler.
gas/ChangeLog:
* config/tc-arc.c (struct arc_insn): Change type of insn field.
(md_number_to_chars_midend): Support 6- and 8-byte values.
(emit_insn0): Update debug output.
(find_opcode_match): Likewise.
(build_fake_opcode_hash_entry): Delete.
(find_special_case_long_opcode): Delete.
(find_special_case): Remove long format special case handling.
(insert_operand): Change instruction type and update debug print
format.
(assemble_insn): Change instruction type, update debug print
formats, and remove unneeded assert.
include/ChangeLog:
* opcode/arc.h (struct arc_opcode): Change type of opcode and mask
fields.
(struct arc_long_opcode): Delete.
(struct arc_operand): Change types for insert and extract
handlers.
opcodes/ChangeLog:
* arc-dis.c (struct arc_operand_iterator): Remove all fields
relating to long instruction processing, add new limm field.
(OPCODE): Rename to...
(OPCODE_32BIT_INSN): ...this.
(OPCODE_AC): Delete.
(skip_this_opcode): Handle different instruction lengths, update
macro name.
(special_flag_p): Update parameter type.
(find_format_from_table): Update for more instruction lengths.
(find_format_long_instructions): Delete.
(find_format): Update for more instruction lengths.
(arc_insn_length): Likewise.
(extract_operand_value): Update for more instruction lengths.
(operand_iterator_next): Remove code relating to long
instructions.
(arc_opcode_to_insn_type): New function.
(print_insn_arc):Update for more instructions lengths.
* arc-ext.c (extInstruction_t): Change argument type.
* arc-ext.h (extInstruction_t): Change argument type.
* arc-fxi.h: Change type unsigned to unsigned long long
extensively throughout.
* arc-nps400-tbl.h: Add long instructions taken from
arc_long_opcodes table in arc-opc.c.
* arc-opc.c: Update parameter types on insert/extract handlers.
(arc_long_opcodes): Delete.
(arc_num_long_opcodes): Delete.
(arc_opcode_len): Update for more instruction lengths.
In preparation for moving to a world where arc instructions can be 2, 4,
6, or 8 bytes in length, replace the ARC_SHORT macro (which is either
true of false) with an arc_opcode_len function that returns a length in
bytes.
There should be no functional change after this commit.
gas/ChangeLog:
* config/tc-arc.c (assemble_insn): Replace use of ARC_SHORT with
arc_opcode_len.
include/ChangeLog:
* opcode/arc.h (arc_opcode_len): Declare.
(ARC_SHORT): Delete.
opcodes/ChangeLog:
* arc-dis.c (find_format_from_table): Replace use of ARC_SHORT
with arc_opcode_len.
(find_format_long_instructions): Likewise.
* arc-opc.c (arc_opcode_len): New function.
When assembling an instruction replace the short_insn boolean flag with
an integer field for holding the instruction length. This is in
preparation for moving to a world where instructions can be 2, 4, 6, or
8 bytes in length.
gas/ChangeLog:
* config/tc-arc.c (struct arc_insn): Replace short_insn flag with
len field.
(apply_fixups): Update to use len field.
(emit_insn0): Simplify code, making use of len field.
(md_convert_frag): Update to use len field.
(assemble_insn): Update to use len field.
This adds an option for the Qualcomm falkor core, the corresponding
gcc patch is here:
https://gcc.gnu.org/ml/gcc-patches/2016-11/msg00262.html
This was tested with aarch64 and armhf builds and make check and also
by building and running SPEC2006.
* config/tc-aarch64.c (aarch64_cpus): Add falkor.
* config/tc-arm.c (arm_cpus): Likewise.
* doc/c-aarch64.texi: Likewise.
* doc/c-arm.texi: Likewise.
In the ARC assembler, when a cpu type is specified using the .cpu
directive, we rely on the bfd list of arc machine types in order to
validate the cpu name passed in.
This validation is only used in order to check that the cpu type passed
to the .cpu directive matches any machine type selected earlier on the
command line. Once that initial check has passed a full check is
performed using the assemblers internal list of know cpu types.
The problem is that the assembler knows about more cpu types than bfd,
some cpu types known by the assembler are actually aliases for a base
cpu type plus a specific set of assembler extensions. One such example
is NPS400, though more could be added later.
This commit removes the need for the assembler to use the bfd list of
machine types for validation. Instead the error checking, to ensure
that any value passed to a '.cpu' directive matches any earlier command
line selection, is moved into the function arc_select_cpu.
I have taken the opportunity to bundle the 4 separate static globals
that describe the currently selected machine type into a single
structure (called selected_cpu).
gas/ChangeLog:
* config/tc-arc.c (arc_target): Delete.
(arc_target_name): Delete.
(arc_features): Delete.
(arc_mach_type): Delete.
(mach_type_specified_p): Delete.
(enum mach_selection_type): New enum.
(mach_selection_mode): New static global.
(selected_cpu): New static global.
(arc_eflag): Rename to ...
(arc_initial_eflag): ...this, and make const.
(arc_select_cpu): Update comment, new parameter, check how
previous machine type selection was made, and record this
selection. Use selected_cpu instead of old globals.
(arc_option): Remove use of arc_get_mach, instead use
arc_select_cpu to validate machine type selection. Use
selected_cpu over old globals.
(allocate_tok): Use selected_cpu over old globals.
(find_opcode_match): Likewise.
(assemble_tokens): Likewise.
(arc_cons_fix_new): Likewise.
(arc_extinsn): Likewise.
(arc_extcorereg): Likewise.
(md_begin): Update default machine type selection, use
selected_cpu over old globals.
(md_parse_option): Update machine type selection option handling,
use selected_cpu over old globals.
* testsuite/gas/arc/nps400-0.s: Add .cpu directive.
bfd/ChangeLog:
* cpu-arc.c (arc_get_mach): Delete.
This reverts commit 95e61695c1. People
still want to use older versions of bison, apparently.
Revert 2016-10-06 Alan Modra <amodra@gmail.com>
* config/rl78-parse.y: Do use old %name-prefix syntax.
* config/rx-parse.y: Likewise.
In 32-bit, the REX_B bit in the 3-byte VEX prefix is ignored and the
the highest bit in VEX.vvvv is either 1 or ignored. In 64-bit, we
need to check invalid mask registers.
gas/
PR binutis/20705
* testsuite/gas/i386/i386.exp: Run x86-64-opcode-bad.
* testsuite/gas/i386/x86-64-opcode-bad.d: New file.
* testsuite/gas/i386/x86-64-opcode-bad.s: Likewise.
opcodes/
PR binutis/20705
* i386-dis.c (get_valid_dis386): Ignore the REX_B bit and
the highest bit in VEX.vvvv for the 3-byte VEX prefix in
32-bit mode. Don't check vex.register_specifier in 32-bit
mode.
(OP_E_register): Check invalid mask registers.
(OP_G): Likewise.
(OP_VEX): Likewise.
The disassember was generating invXXX instructions for cases when in reality we
had llockd or scondd instrutions.
opcodes/ChangeLog:
Cupertino Miranda <cmiranda@synopsys.com>
arc-tbl.h: Removed any "inv.+" instructions from the table.
gas/ChangeLog:
Cupertino Miranda <cmiranda@synopsys.com>
testsuite/arc/dis-inv.s: Test to validate patch.
testsuite/arc/dis-inv.d: Likewise.
The ARC (short) instructions are using a special register number to
indicate is the instruction uses a long immediate (LIMM). In the case
of short instruction, this LIMM indicator depends on the ISA version
used. Thus, for ARCv1 processors, the LIMM indicator is 0x3E, the same
value used in "long" instructions. However, for the ARCv2 processors,
this LIMM indicator is 0x1E.
This patch fixes the LIMM detection for ARCv1 ISA and adds two tests.
gas/
2016-10-13 Claudiu Zissulescu <claziss@synopsys.com>
* testsuite/gas/arc/shortlimm_a7.d: New file.
* testsuite/gas/arc/shortlimm_a7.s: Likewise.
* testsuite/gas/arc/shortlimm_hs.d: Likewise.
* testsuite/gas/arc/shortlimm_hs.s: Likewise.
include/
2016-10-13 Claudiu Zissulescu <claziss@synopsys.com>
* opcode/arc.h (ARC_OPCODE_ARCV2): New define.
opcodes/
2016-10-13 Claudiu Zissulescu <claziss@synopsys.com>
* arc-dis.c (find_format_from_table): Discriminate LIMM indicator
usage on ISA basis.
The CFI* testcases fail on MIPS64 because the augmentation string does
not match the regexp. This is because MIPS64 doesn't use the default of
4 for DWARF2_FDE_RELOC_SIZE which ends up as "b" in the augmentation
string. MIPS64 uses the address size which is 8 resulting in "c".
Adding c to the regexp fixes a couple of them. Others also need
adjustments in the FDE header lines due to different
sizes/offsets.
gas/ChangeLog:
2016-10-10 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
* testsuite/gas/cfi/cfi-common-1.d: Adjust regexps for mips64.
* testsuite/gas/cfi/cfi-common-2.d: Likewise.
* testsuite/gas/cfi/cfi-common-3.d: Likewise.
* testsuite/gas/cfi/cfi-common-4.d: Likewise.
* testsuite/gas/cfi/cfi-common-5.d: Likewise.
* testsuite/gas/cfi/cfi-common-7.d: Likewise.
* testsuite/gas/cfi/cfi-common-8.d: Likewise.
* testsuite/gas/cfi/cfi-common-9.d: Likewise.
* testsuite/gas/cfi/cfi-mips-1.d: Likewise.
I noticed a while ago that the rx-elf gas gprel test regressed for no
apparent reason. It turns out that the problem was rx-parse.y using
BFD_RELOC_RX_* values, which may change when other targets add new
relocs. If rx-parse.o doesn't depend on bfd.h, it won't be recompiled.
* Makefile.am (EXTRA_as_new_SOURCES): Add config/rl78-parse.y and
config/rx-parse.y. Move config/bfin-parse.y.
(bfin-parse.@OBJEXT@, rl78-parse.@OBJEXT@, rx-parse.@OBJEXT@): Delete.
($(srcdir)/config/rl78-defs.h): New rule.
* Makefile.in: Regenerate.
The comment logically belongs inside the preprocessor conditional,
but gcc's -Wimplicit-fallthrough loses track of it. Revert when/if
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77817 is fixed.
* app.c (do_scrub_chars): Move fall through comment.
* expr.c (operand): Likewise.
The .cfi_sections directive can be safely used multiple times
with different sections named at any time unless the compact form
of exception handling is requested after CFI information has
been emitted. Only the compact form of CFI information changes
the way in which CFI is generated and therefore cannot be
retrospectively requested after generating CFI information.
gas/
PR gas/20648
* dw2gencfi.c (dot_cfi_sections): Refine the check for
inconsistent .cfi_sections to only consider compact vs non
compact forms.
* testsuite/gas/cfi/cfi-common-9.d: New file.
* testsuite/gas/cfi/cfi-common-9.s: New file.
* testsuite/gas/cfi/cfi.exp: Run new test.
opcode/
PR target/20553
* aarch64-tbl.h (fmla, fmls, fmul, fmulx): Fix opcode mask field.
gas/
* testsuite/gas/aarch64/advsimd-fp16.s (indexed_elem): New high index
testcases for H and S variants. New low index testcases for D variant.
* testsuite/gas/aarch64/advsimd-fp16.d: Update expected results.
This patch adds support for .cfi_val_offset GAS pseudo command which
maps to DW_CFA_val_offset and DW_CFA_val_offset_sf.
gas/ChangeLog:
2016-09-29 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
* doc/as.texinfo: Add docu for .cfi_val_offset.
* dw2gencfi.c (cfi_add_CFA_val_offset): New function.
(dot_cfi): Add case for DW_CFA_val_offset.
(output_cfi_insn): Likewise.
(cfi_pseudo_table): Add entry for cfi_val_offset.
* dw2gencfi.h: Add prototype for cfi_add_CFA_val_offset.
* testsuite/gas/cfi/cfi-common-8.d: New test.
* testsuite/gas/cfi/cfi-common-8.s: New test.
* testsuite/gas/cfi/cfi.exp: Run cfi-common-8 testcase.
binutils/ChangeLog:
2016-09-29 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
* dwarf.c (display_debug_frames): Adjust output line.
cmp[l][o] get an optional L field only when generating 32-bit code.
dcbf, tlbie and tlbiel keep their optional L field, ditto for R field
of tbegin. cmprb, tsr., wlcr[all] and mtsle all change to a
compulsory L field.
L field of dcbf and wclr is 2 bits.
PR 20641
include/
* opcode/ppc.h (PPC_OPERAND_OPTIONAL32): Define.
opcodes/
* ppc-opc.c (L): Make compulsory.
(LOPT): New, optional form of L.
(HTM_R): Define as LOPT.
(L0, L1): Delete.
(L32OPT): New, optional for 32-bit L.
(L2OPT): New, 2-bit L for dcbf.
(SVC_LEC): Update.
(L2): Define.
(insert_l0, extract_l0, insert_l1, extract_l2): Delete.
(powerpc_opcodes <cmpli, cmpi, cmpl, cmp>): Use L32OPT.
<dcbf>: Use L2OPT.
<tlbiel, tlbie>: Use LOPT.
<wclr, wclrall>: Use L2.
gas/
* config/tc-ppc.c (md_assemble): Handle PPC_OPERAND_OPTIONAL32.
* testsuite/gas/ppc/power8.s: Provide tbegin. operand.
* testsuite/gas/ppc/power9.d: Update cmprb disassembly.
This patch extends Tag_GNU_Power_ABI_FP to cover long double ABIs,
makes the assembler warn about undefined tag values, and removes
similar warnings from the linker. I think it is better to not
warn in the linker about undefined tag values as future extensions to
the tags then won't result in likely bogus warnings. This is
consistent with the fact that an older linker won't warn on an
entirely new tag.
include/
* elf/ppc.h (Tag_GNU_Power_ABI_FP): Comment.
bfd/
* elf-bfd.h (_bfd_elf_ppc_merge_fp_attributes): Declare.
* elf32-ppc.c (_bfd_elf_ppc_merge_fp_attributes): New function.
(ppc_elf_merge_obj_attributes): Use it. Don't copy first file
attributes, merge them. Don't warn about undefined tag bits,
or copy unknown values to output.
* elf64-ppc.c (ppc64_elf_merge_private_bfd_data): Call
_bfd_elf_ppc_merge_fp_attributes.
binutils/
* readelf.c (display_power_gnu_attribute): Catch truncated section
for all powerpc attributes. Display long double ABI. Don't
capitalize words, except for names. Show known bits of tag values
when some unknown bits are present. Whitespace fixes.
gas/
* config/tc-ppc.c (ppc_elf_gnu_attribute): New function.
(md_pseudo_table <ELF>): Handle "gnu_attribute".
ld/
* testsuite/ld-powerpc/attr-gnu-4-4.s: Delete.
* testsuite/ld-powerpc/attr-gnu-4-14.d: Delete.
* testsuite/ld-powerpc/attr-gnu-4-24.d: Delete.
* testsuite/ld-powerpc/attr-gnu-4-34.d: Delete.
* testsuite/ld-powerpc/attr-gnu-4-41.d: Delete.
* testsuite/ld-powerpc/attr-gnu-4-32.d: Adjust expected warning.
* testsuite/ld-powerpc/attr-gnu-8-23.d: Likewise.
* testsuite/ld-powerpc/attr-gnu-4-01.d: Adjust expected output.
* testsuite/ld-powerpc/attr-gnu-4-02.d: Likewise.
* testsuite/ld-powerpc/attr-gnu-4-03.d: Likewise.
* testsuite/ld-powerpc/attr-gnu-4-10.d: Likewise.
* testsuite/ld-powerpc/attr-gnu-4-11.d: Likewise.
* testsuite/ld-powerpc/attr-gnu-4-20.d: Likewise.
* testsuite/ld-powerpc/attr-gnu-4-22.d: Likewise.
* testsuite/ld-powerpc/attr-gnu-4-33.d: Likewise.
* testsuite/ld-powerpc/attr-gnu-8-11.d: Likewise.
* testsuite/ld-powerpc/powerpc.exp: Don't run deleted tests.
One of the review comments from the SVE series was that it would
be better to use "must" rather than "should" in error messages.
I think this patch fixes all cases in the AArch64 code.
It also uses "must be" instead of "expected to be".
opcodes/
* aarch64-opc.c (operand_general_constraint_met_p): Use "must be"
rather than "should be" or "expected to be" in error messages.
gas/
* config/tc-aarch64.c (output_operand_error_record): Use "must be"
rather than "should be" or "expected to be" in error messages.
(parse_operands): Likewise.
* testsuite/gas/aarch64/diagnostic.l: Likewise.
* testsuite/gas/aarch64/legacy_reg_names.l: Likewise.
* testsuite/gas/aarch64/sve-invalid.l: Likewise.
* testsuite/gas/aarch64/sve-reg-diagnostic.l: Likewise.
SVE defines new names for existing NZCV conditions, to reflect the
result of instructions like PTEST. This patch adds support for these
names.
The patch also adds comments to the disassembly output to show the
alternative names of a condition code. For example:
cinv x0, x1, cc
becomes:
cinv x0, x1, cc // cc = lo, ul, last
and:
b.cc f0 <...>
becomes:
b.cc f0 <...> // b.lo, b.ul, b.last
Doing this for the SVE names follows the practice recommended by the
SVE specification and is definitely useful when reading SVE code.
If the feeling is that it's too distracting elsewhere, we could add
an option to turn it off.
include/
* opcode/aarch64.h (aarch64_cond): Bump array size to 4.
opcodes/
* aarch64-dis.c (remove_dot_suffix): New function, split out from...
(print_mnemonic_name): ...here.
(print_comment): New function.
(print_aarch64_insn): Call it.
* aarch64-opc.c (aarch64_conds): Add SVE names.
(aarch64_print_operand): Print alternative condition names in
a comment.
gas/
* config/tc-aarch64.c (opcode_lookup): Search for the end of
a condition name, rather than assuming that it will have exactly
2 characters.
(parse_operands): Likewise.
* testsuite/gas/aarch64/alias.d: Add new condition-code comments
to the expected output.
* testsuite/gas/aarch64/beq_1.d: Likewise.
* testsuite/gas/aarch64/float-fp16.d: Likewise.
* testsuite/gas/aarch64/int-insns.d: Likewise.
* testsuite/gas/aarch64/no-aliases.d: Likewise.
* testsuite/gas/aarch64/programmer-friendly.d: Likewise.
* testsuite/gas/aarch64/reloc-insn.d: Likewise.
* testsuite/gas/aarch64/b_c_1.d, testsuite/gas/aarch64/b_c_1.s:
New test.
ld/
* testsuite/ld-aarch64/emit-relocs-280.d: Match branch comments.
* testsuite/ld-aarch64/weak-undefined.d: Likewise.
This patch adds new tests for SVE. It also extends diagnostic.[sl] with
checks for some inappropriate uses of MUL and MUL VL in base AArch64
instructions.
gas/testsuite/
* gas/aarch64/diagnostic.s, gas/aarch64/diagnostic.l: Add tests for
invalid uses of MUL VL and MUL in base AArch64 instructions.
* gas/aarch64/sve-add.s, gas/aarch64/sve-add.d, gas/aarch64/sve-dup.s,
gas/aarch64/sve-dup.d, gas/aarch64/sve-invalid.s,
gas/aarch64/sve-invalid.d, gas/aarch64/sve-invalid.l,
gas/aarch64/sve-reg-diagnostic.s, gas/aarch64/sve-reg-diagnostic.d,
gas/aarch64/sve-reg-diagnostic.l, gas/aarch64/sve.s,
gas/aarch64/sve.d: New tests.
SVE uses some new fields to store W, X and scalar FP registers.
This patch adds corresponding operands.
include/
* opcode/aarch64.h (AARCH64_OPND_SVE_Rm): New aarch64_opnd.
(AARCH64_OPND_SVE_Rn_SP, AARCH64_OPND_SVE_VZn, AARCH64_OPND_SVE_Vd)
(AARCH64_OPND_SVE_Vm, AARCH64_OPND_SVE_Vn): Likewise.
opcodes/
* aarch64-tbl.h (AARCH64_OPERANDS): Add entries for the new SVE core
and FP register operands.
* aarch64-opc.h (FLD_SVE_Rm, FLD_SVE_Rn, FLD_SVE_Vd, FLD_SVE_Vm)
(FLD_SVE_Vn): New aarch64_field_kinds.
* aarch64-opc.c (fields): Add corresponding entries.
(aarch64_print_operand): Handle the new SVE core and FP register
operands.
* aarch64-opc-2.c: Regenerate.
* aarch64-asm-2.c: Likewise.
* aarch64-dis-2.c: Likewise.
gas/
* config/tc-aarch64.c (parse_operands): Handle the new SVE core
and FP register operands.
This patch adds support for the new SVE floating-point immediate
operands. One operand uses the same 8-bit encoding as base AArch64,
but in a different position. The others use a single bit to select
between two values.
One of the single-bit operands is a choice between 0 and 1, where 0
is not a valid 8-bit encoding. I think the cleanest way of handling
these single-bit immediates is therefore to use the IEEE float encoding
itself as the immediate value and select between the two possible values
when encoding and decoding.
As described in the covering note for the patch that added F_STRICT,
we get better error messages by accepting unsuffixed vector registers
and leaving the qualifier matching code to report an error. This means
that we carry on parsing the other operands, and so can try to parse FP
immediates for invalid instructions like:
fcpy z0, #2.5
In this case there is no suffix to tell us whether the immediate should
be treated as single or double precision. Again, we get better error
messages by picking one (arbitrary) immediate size and reporting an error
for the missing suffix later.
include/
* opcode/aarch64.h (AARCH64_OPND_SVE_FPIMM8): New aarch64_opnd.
(AARCH64_OPND_SVE_I1_HALF_ONE, AARCH64_OPND_SVE_I1_HALF_TWO)
(AARCH64_OPND_SVE_I1_ZERO_ONE): Likewise.
opcodes/
* aarch64-tbl.h (AARCH64_OPERANDS): Add entries for the new SVE FP
immediate operands.
* aarch64-opc.h (FLD_SVE_i1): New aarch64_field_kind.
* aarch64-opc.c (fields): Add corresponding entry.
(operand_general_constraint_met_p): Handle the new SVE FP immediate
operands.
(aarch64_print_operand): Likewise.
* aarch64-opc-2.c: Regenerate.
* aarch64-asm.h (ins_sve_float_half_one, ins_sve_float_half_two)
(ins_sve_float_zero_one): New inserters.
* aarch64-asm.c (aarch64_ins_sve_float_half_one): New function.
(aarch64_ins_sve_float_half_two): Likewise.
(aarch64_ins_sve_float_zero_one): Likewise.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis.h (ext_sve_float_half_one, ext_sve_float_half_two)
(ext_sve_float_zero_one): New extractors.
* aarch64-dis.c (aarch64_ext_sve_float_half_one): New function.
(aarch64_ext_sve_float_half_two): Likewise.
(aarch64_ext_sve_float_zero_one): Likewise.
* aarch64-dis-2.c: Regenerate.
gas/
* config/tc-aarch64.c (double_precision_operand_p): New function.
(parse_operands): Use it to calculate the dp_p input to
parse_aarch64_imm_float. Handle the new SVE FP immediate operands.
This patch adds the new SVE integer immediate operands. There are
three kinds:
- simple signed and unsigned ranges, but with new widths and positions.
- 13-bit logical immediates. These have the same form as in base AArch64,
but at a different bit position.
In the case of the "MOV Zn.<T>, #<limm>" alias of DUPM, the logical
immediate <limm> is not allowed to be a valid DUP immediate, since DUP
is preferred over DUPM for constants that both instructions can handle.
- a new 9-bit arithmetic immediate, of the form "<imm8>{, LSL #8}".
In some contexts the operand is signed and in others it's unsigned.
As an extension, we allow shifted immediates to be written as a single
integer, e.g. "#256" is equivalent to "#1, LSL #8". We also use the
shiftless form as the preferred disassembly, except for the special
case of "#0, LSL #8" (a redundant encoding of 0).
include/
* opcode/aarch64.h (AARCH64_OPND_SIMM5): New aarch64_opnd.
(AARCH64_OPND_SVE_AIMM, AARCH64_OPND_SVE_ASIMM)
(AARCH64_OPND_SVE_INV_LIMM, AARCH64_OPND_SVE_LIMM)
(AARCH64_OPND_SVE_LIMM_MOV, AARCH64_OPND_SVE_SHLIMM_PRED)
(AARCH64_OPND_SVE_SHLIMM_UNPRED, AARCH64_OPND_SVE_SHRIMM_PRED)
(AARCH64_OPND_SVE_SHRIMM_UNPRED, AARCH64_OPND_SVE_SIMM5)
(AARCH64_OPND_SVE_SIMM5B, AARCH64_OPND_SVE_SIMM6)
(AARCH64_OPND_SVE_SIMM8, AARCH64_OPND_SVE_UIMM3)
(AARCH64_OPND_SVE_UIMM7, AARCH64_OPND_SVE_UIMM8)
(AARCH64_OPND_SVE_UIMM8_53): Likewise.
(aarch64_sve_dupm_mov_immediate_p): Declare.
opcodes/
* aarch64-tbl.h (AARCH64_OPERANDS): Add entries for the new SVE
integer immediate operands.
* aarch64-opc.h (FLD_SVE_immN, FLD_SVE_imm3, FLD_SVE_imm5)
(FLD_SVE_imm5b, FLD_SVE_imm7, FLD_SVE_imm8, FLD_SVE_imm9)
(FLD_SVE_immr, FLD_SVE_imms, FLD_SVE_tszh): New aarch64_field_kinds.
* aarch64-opc.c (fields): Add corresponding entries.
(operand_general_constraint_met_p): Handle the new SVE integer
immediate operands.
(aarch64_print_operand): Likewise.
(aarch64_sve_dupm_mov_immediate_p): New function.
* aarch64-opc-2.c: Regenerate.
* aarch64-asm.h (ins_inv_limm, ins_sve_aimm, ins_sve_asimm)
(ins_sve_limm_mov, ins_sve_shlimm, ins_sve_shrimm): New inserters.
* aarch64-asm.c (aarch64_ins_limm_1): New function, split out from...
(aarch64_ins_limm): ...here.
(aarch64_ins_inv_limm): New function.
(aarch64_ins_sve_aimm): Likewise.
(aarch64_ins_sve_asimm): Likewise.
(aarch64_ins_sve_limm_mov): Likewise.
(aarch64_ins_sve_shlimm): Likewise.
(aarch64_ins_sve_shrimm): Likewise.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis.h (ext_inv_limm, ext_sve_aimm, ext_sve_asimm)
(ext_sve_limm_mov, ext_sve_shlimm, ext_sve_shrimm): New extractors.
* aarch64-dis.c (decode_limm): New function, split out from...
(aarch64_ext_limm): ...here.
(aarch64_ext_inv_limm): New function.
(decode_sve_aimm): Likewise.
(aarch64_ext_sve_aimm): Likewise.
(aarch64_ext_sve_asimm): Likewise.
(aarch64_ext_sve_limm_mov): Likewise.
(aarch64_top_bit): Likewise.
(aarch64_ext_sve_shlimm): Likewise.
(aarch64_ext_sve_shrimm): Likewise.
* aarch64-dis-2.c: Regenerate.
gas/
* config/tc-aarch64.c (parse_operands): Handle the new SVE integer
immediate operands.
This patch adds support for addresses of the form:
[<base>, #<offset>, MUL VL]
This involves adding a new AARCH64_MOD_MUL_VL modifier, which is
why I split it out from the other addressing modes.
For LD2, LD3 and LD4, the offset must be a multiple of the structure
size, so for LD3 the possible values are 0, 3, 6, .... The patch
therefore extends value_aligned_p to handle non-power-of-2 alignments.
include/
* opcode/aarch64.h (AARCH64_OPND_SVE_ADDR_RI_S4xVL): New aarch64_opnd.
(AARCH64_OPND_SVE_ADDR_RI_S4x2xVL, AARCH64_OPND_SVE_ADDR_RI_S4x3xVL)
(AARCH64_OPND_SVE_ADDR_RI_S4x4xVL, AARCH64_OPND_SVE_ADDR_RI_S6xVL)
(AARCH64_OPND_SVE_ADDR_RI_S9xVL): Likewise.
(AARCH64_MOD_MUL_VL): New aarch64_modifier_kind.
opcodes/
* aarch64-tbl.h (AARCH64_OPERANDS): Add entries for new MUL VL
operands.
* aarch64-opc.c (aarch64_operand_modifiers): Initialize
the AARCH64_MOD_MUL_VL entry.
(value_aligned_p): Cope with non-power-of-two alignments.
(operand_general_constraint_met_p): Handle the new MUL VL addresses.
(print_immediate_offset_address): Likewise.
(aarch64_print_operand): Likewise.
* aarch64-opc-2.c: Regenerate.
* aarch64-asm.h (ins_sve_addr_ri_s4xvl, ins_sve_addr_ri_s6xvl)
(ins_sve_addr_ri_s9xvl): New inserters.
* aarch64-asm.c (aarch64_ins_sve_addr_ri_s4xvl): New function.
(aarch64_ins_sve_addr_ri_s6xvl): Likewise.
(aarch64_ins_sve_addr_ri_s9xvl): Likewise.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis.h (ext_sve_addr_ri_s4xvl, ext_sve_addr_ri_s6xvl)
(ext_sve_addr_ri_s9xvl): New extractors.
* aarch64-dis.c (aarch64_ext_sve_addr_reg_mul_vl): New function.
(aarch64_ext_sve_addr_ri_s4xvl): Likewise.
(aarch64_ext_sve_addr_ri_s6xvl): Likewise.
(aarch64_ext_sve_addr_ri_s9xvl): Likewise.
* aarch64-dis-2.c: Regenerate.
gas/
* config/tc-aarch64.c (SHIFTED_NONE, SHIFTED_MUL_VL): New
parse_shift_modes.
(parse_shift): Handle SHIFTED_MUL_VL.
(parse_address_main): Add an imm_shift_mode parameter.
(parse_address, parse_sve_address): Update accordingly.
(parse_operands): Handle MUL VL addressing modes.
This patch adds most of the new SVE addressing modes and associated
operands. A follow-on patch adds MUL VL, since handling it separately
makes the changes easier to read.
The patch also introduces a new "operand-dependent data" field to the
operand flags, based closely on the existing one for opcode flags.
For SVE this new field needs only 2 bits, but it could be widened
in future if necessary.
include/
* opcode/aarch64.h (AARCH64_OPND_SVE_ADDR_RI_U6): New aarch64_opnd.
(AARCH64_OPND_SVE_ADDR_RI_U6x2, AARCH64_OPND_SVE_ADDR_RI_U6x4)
(AARCH64_OPND_SVE_ADDR_RI_U6x8, AARCH64_OPND_SVE_ADDR_RR)
(AARCH64_OPND_SVE_ADDR_RR_LSL1, AARCH64_OPND_SVE_ADDR_RR_LSL2)
(AARCH64_OPND_SVE_ADDR_RR_LSL3, AARCH64_OPND_SVE_ADDR_RX)
(AARCH64_OPND_SVE_ADDR_RX_LSL1, AARCH64_OPND_SVE_ADDR_RX_LSL2)
(AARCH64_OPND_SVE_ADDR_RX_LSL3, AARCH64_OPND_SVE_ADDR_RZ)
(AARCH64_OPND_SVE_ADDR_RZ_LSL1, AARCH64_OPND_SVE_ADDR_RZ_LSL2)
(AARCH64_OPND_SVE_ADDR_RZ_LSL3, AARCH64_OPND_SVE_ADDR_RZ_XTW_14)
(AARCH64_OPND_SVE_ADDR_RZ_XTW_22, AARCH64_OPND_SVE_ADDR_RZ_XTW1_14)
(AARCH64_OPND_SVE_ADDR_RZ_XTW1_22, AARCH64_OPND_SVE_ADDR_RZ_XTW2_14)
(AARCH64_OPND_SVE_ADDR_RZ_XTW2_22, AARCH64_OPND_SVE_ADDR_RZ_XTW3_14)
(AARCH64_OPND_SVE_ADDR_RZ_XTW3_22, AARCH64_OPND_SVE_ADDR_ZI_U5)
(AARCH64_OPND_SVE_ADDR_ZI_U5x2, AARCH64_OPND_SVE_ADDR_ZI_U5x4)
(AARCH64_OPND_SVE_ADDR_ZI_U5x8, AARCH64_OPND_SVE_ADDR_ZZ_LSL)
(AARCH64_OPND_SVE_ADDR_ZZ_SXTW, AARCH64_OPND_SVE_ADDR_ZZ_UXTW):
Likewise.
opcodes/
* aarch64-tbl.h (AARCH64_OPERANDS): Add entries for the new SVE
address operands.
* aarch64-opc.h (FLD_SVE_imm6, FLD_SVE_msz, FLD_SVE_xs_14)
(FLD_SVE_xs_22): New aarch64_field_kinds.
(OPD_F_OD_MASK, OPD_F_OD_LSB, OPD_F_NO_ZR): New flags.
(get_operand_specific_data): New function.
* aarch64-opc.c (fields): Add entries for FLD_SVE_imm6, FLD_SVE_msz,
FLD_SVE_xs_14 and FLD_SVE_xs_22.
(operand_general_constraint_met_p): Handle the new SVE address
operands.
(sve_reg): New array.
(get_addr_sve_reg_name): New function.
(aarch64_print_operand): Handle the new SVE address operands.
* aarch64-opc-2.c: Regenerate.
* aarch64-asm.h (ins_sve_addr_ri_u6, ins_sve_addr_rr_lsl)
(ins_sve_addr_rz_xtw, ins_sve_addr_zi_u5, ins_sve_addr_zz_lsl)
(ins_sve_addr_zz_sxtw, ins_sve_addr_zz_uxtw): New inserters.
* aarch64-asm.c (aarch64_ins_sve_addr_ri_u6): New function.
(aarch64_ins_sve_addr_rr_lsl): Likewise.
(aarch64_ins_sve_addr_rz_xtw): Likewise.
(aarch64_ins_sve_addr_zi_u5): Likewise.
(aarch64_ins_sve_addr_zz): Likewise.
(aarch64_ins_sve_addr_zz_lsl): Likewise.
(aarch64_ins_sve_addr_zz_sxtw): Likewise.
(aarch64_ins_sve_addr_zz_uxtw): Likewise.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis.h (ext_sve_addr_ri_u6, ext_sve_addr_rr_lsl)
(ext_sve_addr_rz_xtw, ext_sve_addr_zi_u5, ext_sve_addr_zz_lsl)
(ext_sve_addr_zz_sxtw, ext_sve_addr_zz_uxtw): New extractors.
* aarch64-dis.c (aarch64_ext_sve_add_reg_imm): New function.
(aarch64_ext_sve_addr_ri_u6): Likewise.
(aarch64_ext_sve_addr_rr_lsl): Likewise.
(aarch64_ext_sve_addr_rz_xtw): Likewise.
(aarch64_ext_sve_addr_zi_u5): Likewise.
(aarch64_ext_sve_addr_zz): Likewise.
(aarch64_ext_sve_addr_zz_lsl): Likewise.
(aarch64_ext_sve_addr_zz_sxtw): Likewise.
(aarch64_ext_sve_addr_zz_uxtw): Likewise.
* aarch64-dis-2.c: Regenerate.
gas/
* config/tc-aarch64.c (REG_TYPE_SVE_BASE, REG_TYPE_SVE_OFFSET): New
register types.
(get_reg_expected_msg): Handle them.
(aarch64_addr_reg_parse): New function, split out from
aarch64_reg_parse_32_64. Handle Z registers too.
(aarch64_reg_parse_32_64): Call it.
(parse_address_main): Add base_qualifier, offset_qualifier,
base_type and offset_type parameters. Handle SVE base and offset
registers.
(parse_address): Update call to parse_address_main.
(parse_sve_address): New function.
(parse_operands): Parse the new SVE address operands.
Some SVE instructions count the number of elements in a given vector
pattern and allow a scale factor of [1, 16] to be applied to the result.
This scale factor is written ", MUL #n", where "MUL" is a new operator.
E.g.:
UQINCD X0, POW2, MUL #2
This patch adds support for this kind of operand.
All existing operators were shifts of some kind, so there was a natural
range of [0, 63] regardless of context. This was then narrowered further
by later checks (e.g. to [0, 31] when used for 32-bit values).
In contrast, MUL doesn't really have a natural context-independent range.
Rather than pick one arbitrarily, it seemed better to make the "shift"
amount a full 64-bit value and leave the range test to the usual
operand-checking code. I've rearranged the fields of aarch64_opnd_info
so that this doesn't increase the size of the structure (although I don't
think its size is critical anyway).
include/
* opcode/aarch64.h (AARCH64_OPND_SVE_PATTERN_SCALED): New
aarch64_opnd.
(AARCH64_MOD_MUL): New aarch64_modifier_kind.
(aarch64_opnd_info): Make shifter.amount an int64_t and
rearrange the fields.
opcodes/
* aarch64-tbl.h (AARCH64_OPERANDS): Add an entry for
AARCH64_OPND_SVE_PATTERN_SCALED.
* aarch64-opc.h (FLD_SVE_imm4): New aarch64_field_kind.
* aarch64-opc.c (fields): Add a corresponding entry.
(set_multiplier_out_of_range_error): New function.
(aarch64_operand_modifiers): Add entry for AARCH64_MOD_MUL.
(operand_general_constraint_met_p): Handle
AARCH64_OPND_SVE_PATTERN_SCALED.
(print_register_offset_address): Use PRIi64 to print the
shift amount.
(aarch64_print_operand): Likewise. Handle
AARCH64_OPND_SVE_PATTERN_SCALED.
* aarch64-opc-2.c: Regenerate.
* aarch64-asm.h (ins_sve_scale): New inserter.
* aarch64-asm.c (aarch64_ins_sve_scale): New function.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis.h (ext_sve_scale): New inserter.
* aarch64-dis.c (aarch64_ext_sve_scale): New function.
* aarch64-dis-2.c: Regenerate.
gas/
* config/tc-aarch64.c (SHIFTED_MUL): New parse_shift_mode.
(parse_shift): Handle it. Reject AARCH64_MOD_MUL for all other
shift modes. Skip range tests for AARCH64_MOD_MUL.
(process_omitted_operand): Handle AARCH64_OPND_SVE_PATTERN_SCALED.
(parse_operands): Likewise.
The SVE instructions have two enumerated operands: one to select a
vector pattern and another to select a prefetch operation. The latter
is a cut-down version of the base AArch64 prefetch operation.
Both types of operand can also be specified as raw enum values such as #31.
Reserved values can only be specified this way.
If it hadn't been for the pattern operand, I would have been tempted
to use the existing parsing for prefetch operations and add extra
checks for SVE. However, since the patterns needed new enum parsing
code anyway, it seeemed cleaner to reuse it for the prefetches too.
Because of the small number of enum values, I don't think we'd gain
anything by using hash tables.
include/
* opcode/aarch64.h (AARCH64_OPND_SVE_PATTERN): New aarch64_opnd.
(AARCH64_OPND_SVE_PRFOP): Likewise.
(aarch64_sve_pattern_array): Declare.
(aarch64_sve_prfop_array): Likewise.
opcodes/
* aarch64-tbl.h (AARCH64_OPERANDS): Add entries for
AARCH64_OPND_SVE_PATTERN and AARCH64_OPND_SVE_PRFOP.
* aarch64-opc.h (FLD_SVE_pattern): New aarch64_field_kind.
(FLD_SVE_prfop): Likewise.
* aarch64-opc.c: Include libiberty.h.
(aarch64_sve_pattern_array): New variable.
(aarch64_sve_prfop_array): Likewise.
(fields): Add entries for FLD_SVE_pattern and FLD_SVE_prfop.
(aarch64_print_operand): Handle AARCH64_OPND_SVE_PATTERN and
AARCH64_OPND_SVE_PRFOP.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis-2.c: Likewise.
* aarch64-opc-2.c: Likewise.
gas/
* config/tc-aarch64.c (parse_enum_string): New function.
(po_enum_or_fail): New macro.
(parse_operands): Handle AARCH64_OPND_SVE_PATTERN and
AARCH64_OPND_SVE_PRFOP.
This patch adds qualifiers to represent /z and /m suffixes on
predicate registers.
include/
* opcode/aarch64.h (AARCH64_OPND_QLF_P_Z): New aarch64_opnd_qualifier.
(AARCH64_OPND_QLF_P_M): Likewise.
opcodes/
* aarch64-opc.c (aarch64_opnd_qualifiers): Add entries for
AARCH64_OPND_QLF_P_[ZM].
(aarch64_print_operand): Print /z and /m where appropriate.
gas/
* config/tc-aarch64.c (vector_el_type): Add NT_zero and NT_merge.
(parse_vector_type_for_operand): Assert that the skipped character
is a '.'.
(parse_predication_for_operand): New function.
(parse_typed_reg): Parse /z and /m suffixes for predicate registers.
(vectype_to_qualifier): Handle NT_zero and NT_merge.
This patch adds the Zn and Pn registers, and associated fields and
operands.
include/
* opcode/aarch64.h (AARCH64_OPND_CLASS_SVE_REG): New
aarch64_operand_class.
(AARCH64_OPND_CLASS_PRED_REG): Likewise.
(AARCH64_OPND_SVE_Pd, AARCH64_OPND_SVE_Pg3, AARCH64_OPND_SVE_Pg4_5)
(AARCH64_OPND_SVE_Pg4_10, AARCH64_OPND_SVE_Pg4_16)
(AARCH64_OPND_SVE_Pm, AARCH64_OPND_SVE_Pn, AARCH64_OPND_SVE_Pt)
(AARCH64_OPND_SVE_Za_5, AARCH64_OPND_SVE_Za_16, AARCH64_OPND_SVE_Zd)
(AARCH64_OPND_SVE_Zm_5, AARCH64_OPND_SVE_Zm_16, AARCH64_OPND_SVE_Zn)
(AARCH64_OPND_SVE_Zn_INDEX, AARCH64_OPND_SVE_ZnxN)
(AARCH64_OPND_SVE_Zt, AARCH64_OPND_SVE_ZtxN): New aarch64_opnds.
opcodes/
* aarch64-tbl.h (AARCH64_OPERANDS): Add entries for new SVE operands.
* aarch64-opc.h (FLD_SVE_Pd, FLD_SVE_Pg3, FLD_SVE_Pg4_5)
(FLD_SVE_Pg4_10, FLD_SVE_Pg4_16, FLD_SVE_Pm, FLD_SVE_Pn, FLD_SVE_Pt)
(FLD_SVE_Za_5, FLD_SVE_Za_16, FLD_SVE_Zd, FLD_SVE_Zm_5, FLD_SVE_Zm_16)
(FLD_SVE_Zn, FLD_SVE_Zt, FLD_SVE_tzsh): New aarch64_field_kinds.
* aarch64-opc.c (fields): Add corresponding entries here.
(operand_general_constraint_met_p): Check that SVE register lists
have the correct length. Check the ranges of SVE index registers.
Check for cases where p8-p15 are used in 3-bit predicate fields.
(aarch64_print_operand): Handle the new SVE operands.
* aarch64-opc-2.c: Regenerate.
* aarch64-asm.h (ins_sve_index, ins_sve_reglist): New inserters.
* aarch64-asm.c (aarch64_ins_sve_index): New function.
(aarch64_ins_sve_reglist): Likewise.
* aarch64-asm-2.c: Regenerate.
* aarch64-dis.h (ext_sve_index, ext_sve_reglist): New extractors.
* aarch64-dis.c (aarch64_ext_sve_index): New function.
(aarch64_ext_sve_reglist): Likewise.
* aarch64-dis-2.c: Regenerate.
gas/
* config/tc-aarch64.c (NTA_HASVARWIDTH): New macro.
(AARCH64_REG_TYPES): Add ZN and PN.
(get_reg_expected_msg): Handle them.
(parse_vector_type_for_operand): Add a reg_type parameter.
Skip the width for Zn and Pn registers.
(parse_typed_reg): Extend vector handling to Zn and Pn. Update the
call to parse_vector_type_for_operand. Set HASVARTYPE for Zn and Pn,
expecting the width to be 0.
(parse_vector_reg_list): Restrict error about [BHSD]nn operands to
REG_TYPE_VN.
(vectype_to_qualifier): Use S_[BHSD] qualifiers for NTA_HASVARWIDTH.
(parse_operands): Handle the new Zn and Pn operands.
(REGSET16): New macro, split out from...
(REGSET31): ...here.
(reg_names): Add Zn and Pn entries.
SVE has some instructions in which the same register appears twice
in the assembly string, once as an input and once as an output.
This patch adds a general mechanism for that.
The patch needs to add new information to the instruction entries.
One option would have been to extend the flags field of the opcode
to 64 bits (since we already rely on 64-bit integers being available
on the host). However, the *_INSN macros mean that it's easy to add
new information as top-level fields without affecting the existing
table entries too much. Going for that option seemed to give slightly
neater code.
include/
* opcode/aarch64.h (aarch64_opcode): Add a tied_operand field.
(AARCH64_OPDE_UNTIED_OPERAND): New aarch64_operand_error_kind.
opcodes/
* aarch64-tbl.h (CORE_INSN, __FP_INSN, SIMD_INSN, CRYP_INSN)
(_CRC_INSN, _LSE_INSN, _LOR_INSN, RDMA_INSN, FP16_INSN, SF16_INSN)
(V8_2_INSN, aarch64_opcode_table): Initialize tied_operand field.
* aarch64-opc.c (aarch64_match_operands_constraint): Check for
tied operands.
gas/
* config/tc-aarch64.c (output_operand_error_record): Handle
AARCH64_OPDE_UNTIED_OPERAND.
SVE predicate operands can appear in three forms:
1. unsuffixed: "Pn"
2. with a predication type: "Pn/[ZM]"
3. with a size suffix: "Pn.[BHSD]"
No variation is allowed: unsuffixed operands cannot have a (redundant)
suffix, and the suffixes can never be dropped. Unsuffixed Pn are used
in LDR and STR, but they are also used for Pg operands in cases where
the result is scalar and where there is therefore no choice to be made
between "merging" and "zeroing". This means that some Pg operands have
suffixes and others don't.
It would be possible to use context-sensitive parsing to handle
this difference. The tc-aarch64.c code would then raise an error
if the wrong kind of suffix is used for a particular instruction.
However, we get much more user-friendly error messages if we parse
all three forms for all SVE instructions and record the suffix as a
qualifier. The normal qualifier matching code can then report cases
where the wrong kind of suffix is used. This is a slight extension
of existing usage, which really only checks for the wrong choice of
suffix within a particular kind of suffix.
The only catch is a that a "NIL" entry in the qualifier list
specifically means "no suffix should be present" (case 1 above).
NIL isn't a wildcard here. It also means that an instruction that
requires all-NIL qualifiers can fail to match (because a suffix was
supplied when it shouldn't have been); this requires a slight change
to find_best_match.
This patch adds an F_STRICT flag to select this behaviour.
The flag will be set for all SVE instructions. The behaviour
for other instructions doesn't change.
include/
* opcode/aarch64.h (F_STRICT): New flag.
opcodes/
* aarch64-opc.c (match_operands_qualifier): Handle F_STRICT.
gas/
* config/tc-aarch64.c (find_best_match): Simplify, allowing an
instruction with all-NIL qualifiers to fail to match.
In the review of the original version of this series, Richard didn't
like the use of boolean parameters to parse_address_main. I think we
can just get rid of them and leave the callers to check the addressing
modes. As it happens, the handling of ADDR_SIMM9{,_2} already did this
for relocation operators (i.e. it used parse_address_reloc and then
rejected relocations).
The callers are already set up to reject invalid register post-indexed
addressing, so we can simply remove the accept_reg_post_index parameter
without adding any more checks. This again creates a corner case where:
.equ x2, 1
ldr w0, [x1], x2
was previously an acceptable way of writing "ldr w0, [x1], #1" but
is now rejected.
Removing the "reloc" parameter means that two cases need to check
explicitly for relocation operators.
ADDR_SIMM9_2 appers to be unused. I'll send a separate patch
to remove it.
This patch makes parse_address temporarily equivalent to
parse_address_main, but later patches in the series will need
to keep the distinction.
gas/
* config/tc-aarch64.c (parse_address_main): Remove reloc and
accept_reg_post_index parameters. Parse relocations and register
post indexes unconditionally.
(parse_address): Remove accept_reg_post_index parameter.
Update call to parse_address_main.
(parse_address_reloc): Delete.
(parse_operands): Call parse_address instead of parse_address_main.
Update existing callers of parse_address and make them check
inst.reloc.type where appropriate.
* testsuite/gas/aarch64/diagnostic.s: Add tests for relocations
in ADDR_SIMPLE, SIMD_ADDR_SIMPLE, ADDR_SIMM7 and ADDR_SIMM9 addresses.
Also test for invalid uses of post-index register addressing.
* testsuite/gas/aarch64/diagnostic.l: Update accordingly.
aarch64_reg_parse_32_64 is currently used to parse address registers,
among other things. It returns two bits of information about the
register: whether it's W rather than X, and whether it's a zero register.
SVE adds addressing modes in which the base or offset can be a vector
register instead of a scalar, so a choice between W and X is no longer
enough. It's more convenient to pass the type of register around as
a qualifier instead.
As it happens, two callers of aarch64_reg_parse_32_64 already wanted
the information in the form of a qualifier, so the change feels pretty
natural even without SVE.
Also, the function took two parameters to control whether {W}SP
and (W|X)ZR should be accepted. We tend to get slightly better
error messages by accepting them regardless and getting the caller
to do the check, rather than potentially treating "xzr", "sp" etc.
as constants. This is easier to do if the function returns the
reg_entry rather than just the register number.
This does create a corner case where:
.equ sp, 1
ldr w0, [x0, sp]
was previously an acceptable way of writing "ldr w0, [x0, #1]",
but I don't think it's important to continue supporting that.
We already rejected things like:
.equ sp, 1
add x0, x1, sp
To ensure these new error messages "win" when matching against
several candidate instruction entries, we need to use the same
address-parsing code for all addresses, including ADDR_SIMPLE
and SIMD_ADDR_SIMPLE. The next patch also relies on this.
Finally, aarcch64_check_reg_type was written in a pretty
conservative way. It should always be equivalent to a single
bit test.
gas/
* config/tc-aarch64.c (REG_TYPE_R_Z, REG_TYPE_R_SP): New register
types.
(get_reg_expected_msg): Handle them and REG_TYPE_R64_SP.
(aarch64_check_reg_type): Simplify.
(aarch64_reg_parse_32_64): Return the reg_entry instead of the
register number. Return the type as a qualifier rather than an
"isreg32" boolean. Remove reject_sp, reject_rz and isregzero
parameters.
(parse_shifter_operand): Update call to aarch64_parse_32_64_reg.
Use get_reg_expected_msg.
(parse_address_main): Likewise. Use aarch64_check_reg_type.
(po_int_reg_or_fail): Replace reject_sp and reject_rz parameters
with a reg_type parameter. Update call to aarch64_parse_32_64_reg.
Use aarch64_check_reg_type to test the result.
(parse_operands): Update after the above changes. Parse ADDR_SIMPLE
addresses normally before enforcing the syntax restrictions.
* testsuite/gas/aarch64/diagnostic.s: Add tests for a post-index
zero register and for a stack pointer index.
* testsuite/gas/aarch64/diagnostic.l: Update accordingly.
Also update existing diagnostic messages after the above changes.
* testsuite/gas/aarch64/illegal-lse.l: Update the error message
for 32-bit register bases.
Since some SVE constants are no longer explicitly tied to the 8-bit
FP immediate format, it seems better to move the range checks out of
parse_aarch64_imm_float and into the callers.
gas/
* config/tc-aarch64.c (parse_aarch64_imm_float): Remove range check.
(parse_operands): Check the range of 8-bit FP immediates here instead.
Previously:
fmov d0, #2
would give an error:
Operand 2 should be an integer register
whereas the user probably just forgot to add the ".0" to make:
fmov d0, #2.0
This patch reports an invalid floating point constant unless the
operand is obviously a register.
The FPIMM8 handling is only relevant for SVE. Without it:
fmov z0, z1
would try to parse z1 as an integer immediate zero (the res2 path),
whereas it's more likely that the user forgot the predicate. This is
tested by the final patch.
gas/
* config/tc-aarch64.c (parse_aarch64_imm_float): Report a specific
low-severity error for registers.
(parse_operands): Report an invalid floating point constant for
if parsing an FPIMM8 fails, and if no better error has been
recorded.
* testsuite/gas/aarch64/diagnostic.s,
testsuite/gas/aarch64/diagnostic.l: Add tests for integer operands
to FMOV.
SVE has single-bit floating-point constants that don't really
have any relation to the AArch64 8-bit floating-point encoding.
(E.g. one of the constants selects between 0 and 1.) The easiest
way of representing them in the aarch64_opnd_info seemed to be
to use the IEEE float representation directly, rather than invent
some new scheme.
This patch paves the way for that by making the code that converts IEEE
doubles to IEEE floats accept any value in the range of an IEEE float,
not just zero and 8-bit floats. It leaves the range checking to the
caller (which already handles it).
gas/
* config/tc-aarch64.c (aarch64_double_precision_fmovable): Rename
to...
(can_convert_double_to_float): ...this. Accept any double-precision
value that converts to single precision without loss of precision.
(parse_aarch64_imm_float): Update accordingly.
To remove parsing ambiguities and to avoid register names being
accidentally added to the symbol table, the immediate parsing
routines reject things like:
.equ x0, 0
add v0.4s, v0.4s, x0
An explicit '#' must be used instead:
.equ x0, 0
add v0.4s, v0.4s, #x0
Of course, it wasn't possible to predict what other register
names might be added in future, so this behaviour was restricted
to the register names that were defined at the time. For backwards
compatibility, we should continue to allow things like:
.equ p0, 0
add v0.4s, v0.4s, p0
even though p0 is now an SVE register.
However, it seems reasonable to extend the x0 behaviour above to
SVE registers when parsing SVE instructions, especially since none
of the SVE immediate formats are relocatable. Doing so removes the
same parsing ambiguity for SVE instructions as the x0 behaviour removes
for base AArch64 instructions.
As a prerequisite, we then need to be able to tell the parsing routines
which registers to reject. This patch changes the interface to make
that possible, although the set of rejected registers doesn't change
at this stage.
gas/
* config/tc-aarch64.c (parse_immediate_expression): Add a
reg_type parameter.
(parse_constant_immediate): Likewise, and update calls.
(parse_aarch64_imm_float): Likewise.
(parse_big_immediate): Likewise.
(po_imm_nc_or_fail): Update accordingly, passing down a new
imm_reg_type variable.
(po_imm_of_fail): Likewise.
(parse_operands): Likewise.
Rename parse_neon_reg_list to parse_vector_reg_list and take
in the required register type as an argument. Later patches
will reuse the function for SVE registers.
gas/
* config/tc-aarch64.c (parse_neon_reg_list): Rename to...
(parse_vector_reg_list): ...this and take a register type
as input.
(parse_operands): Update accordingly.
Generalise the name of parse_neon_type_for_operand to
parse_vector_type_for_operand. Later patches will add SVEisms to it.
gas/
* config/tc-aarch64.c (parse_neon_type_for_operand): Rename to...
(parse_vector_type_for_operand): ...this.
(parse_typed_reg): Update accordingly.
Similar to the previous patch, but this time for the neon_type_el
structure.
gas/
* config/tc-aarch64.c (neon_type_el): Rename to...
(vector_type_el): ...this.
(parse_neon_type_for_operand): Update accordingly.
(parse_typed_reg): Likewise.
(aarch64_reg_parse): Likewise.
(vectype_to_qualifier): Likewise.
(parse_operands): Likewise.
(eq_neon_type_el): Likewise. Rename to...
(eq_vector_type_el): ...this.
(parse_neon_reg_list): Update accordingly.
Later patches will add SVEisms to neon_el_type, so this patch renames
it to something more generic.
gas/
* config/tc-aarch64.c (neon_el_type: Rename to...
(vector_el_type): ...this.
(neon_type_el): Update accordingly.
(parse_neon_type_for_operand): Likewise.
(vectype_to_qualifier): Likewise.
A false return from parse_neon_operand_type had an overloaded
meaning: either the parsing failed, or there was nothing to parse
(which isn't necessarily an error). The only caller, parse_typed_reg,
would therefore not consume the suffix if it was invalid but instead
(successfully) parse the register without a suffix. It would still
leave inst.parsing_error with an error about the invalid suffix.
It seems wrong for a successful parse to leave an error message,
so this patch makes parse_typed_reg return PARSE_FAIL instead.
The patch doesn't seem to make much difference in practice.
Most possible follow-on errors use set_first_error and so the
error about the suffix tended to win despite the successful parse.
gas/
* config/tc-aarch64.c (parse_neon_operand_type): Delete.
(parse_typed_reg): Call parse_neon_type_for_operand directly.
Jose E. Marchesi