Simon Marchi [Thu, 22 Mar 2018 04:27:19 +0000 (00:27 -0400)]
Get rid of VEC(static_tracepoint_marker_p)
This patch replaces VEC(static_tracepoint_marker_p) with std::vector,
and does some c++ification around that. I thought a new overload of
hex2str was useful, so I added it as well as corresponding unit tests.
I also added an overload of ui_out::field_string that takes an
std::string directly.
gdb/ChangeLog:
* tracepoint.h (struct static_tracepoint_marker): Initialize
fields, define default constructor, move constructor and move
assignment, disable the rest.
<str_id, extra>: Make std::string.
(release_static_tracepoint_marker): Remove.
(free_current_marker): Remove.
* tracepoint.c (free_current_marker): Remove.
(parse_static_tracepoint_marker_definition): Adjust to
std::string, use new hex2str overload.
(release_static_tracepoint_marker): Remove.
(print_one_static_tracepoint_marker): Get marker by reference
and adjust to std::string.
(info_static_tracepoint_markers_command): Adjust to std::vector
changes
* target.h (static_tracepoint_marker_p): Remove typedef.
(DEF_VEC_P(static_tracepoint_marker_p)): Remove.
(struct target_ops) <to_static_tracepoint_marker_at>: Return
bool.
<to_static_tracepoint_markers_by_strid>: Return std::vector.
* target-debug.h
(target_debug_print_VEC_static_tracepoint_marker_p_p): Remove.
(target_debug_print_std_vector_static_tracepoint_marker): New.
(target_debug_print_struct_static_tracepoint_marker_p): Rename
to...
(target_debug_print_static_tracepoint_marker_p): ... this.
* target-delegates.c: Re-generate.
* breakpoint.h (struct tracepoint) <static_trace_marker_id>:
Make std::string.
* breakpoint.c (init_breakpoint_sal): Adjust to std::string.
(decode_static_tracepoint_spec): Adjust to std::vector.
(tracepoint_print_one_detail): Adjust to std::string.
(strace_marker_decode_location): Adjust to std::string.
(update_static_tracepoint): Adjust to std::string, remove call
to release_static_tracepoint_marker.
* linux-nat.c (linux_child_static_tracepoint_markers_by_strid):
Adjust to std::vector.
* remote.c (remote_static_tracepoint_marker_at): Return bool.
(remote_static_tracepoint_markers_by_strid): Adjust to
std::vector.
* common/rsp-low.h (hex2str): New overload with explicit count
of bytes.
* common/rsp-low.c (hex2str): New overload with explicit count
of bytes.
* unittests/rsp-low-selftests.c (test_hex2str): New function.
(_initialize_rsp_low_selftests): Add test_hex2str test.
* unittests/tracepoint-selftests.c
(test_parse_static_tracepoint_marker_definition): Adjust to
std::string.
Simon Marchi [Thu, 22 Mar 2018 04:26:37 +0000 (00:26 -0400)]
Make parse_static_tracepoint_marker_definition work with multiple static tracepoint definitions
Since I modify the parse_static_tracepoint_marker_definition function in
the next patch, I wanted to write a unit test for it. Doing so showed
that it doesn't handle multiple consecutive static tracepoint
definitions separated by commas. However, the RSP documentation [1]
states that servers may return multiple definitions, like:
The problem is that the function uses strlen to compute the length of
the last field (the extra field). If there are additional definitions
in addition to the one we are currently parsing, the returned length
will include those definitions, and we'll try to hex-decode past the
extra field.
This patch changes parse_static_tracepoint_marker_definition to consider
the case where the current definition is followed by a comma and more
definitions. It also adds the unit test that found the issue in the
first place.
I don't think this causes any backwards compatibility issues, because
the previous code only handled single static tracepoint definitions, and
the new code handles that correctly.
gdb/ChangeLog:
* tracepoint.c (parse_static_tracepoint_marker_definition):
Consider case where the definition is followed by more
definitions.
* Makefile.in (SUBDIR_UNITTESTS_SRCS): Add
tracepoint-selftests.c.
* unittests/tracepoint-selftests.c: New.
On S390, the guarded storage register set is only valid if guarded storage
is active. Reading/writing the register set yields errors if this is not
the case. Then gdbserver emits warnings like these:
Warning: ptrace(regsets_store_inferior_registers): No data available
Apart from confusing the user, this can also lead to test case failures
due to unexpected output. To suppress this, make the guarded storage
regsets read-only for now.
gdb/gdbserver/ChangeLog:
* linux-s390-low.c (s390_fill_gs): Remove function.
(s390_fill_gsbc): Remove function.
(s390_regsets): Set fill functions for the guarded storage regsets
to NULL.
Andreas Arnez [Wed, 21 Mar 2018 16:41:44 +0000 (17:41 +0100)]
S390: Enable re-attaching with native-extended-gdbserver
On s390x, when running attach.exp with native-extended-gdbserver,
gdbserver crashes in find_regno like this:
.../regcache.c:252: A problem internal to GDBserver has been detected.
Unknown register tdb0 requested
On the GDB side it looks like this:
(gdb) attach 31568
Attaching to process 31568
Remote connection closed
The test case attempts to attach to a new process via the already running
gdbserver. Thus s390_arch_setup is called a second time, and that's where
the problem occurs. In order to determine the word width (32 or 64 bits),
s390_arch_setup reads the pswm register through the regcache. For that it
uses a temporary tdesc which is supposed to work for all s390 targets,
since the actual tdesc has not been determined yet. But in this second
round this doesn't work, because s390_regsets has been updated already and
now contains regsets not described by the temporary tdesc, such as the one
containing tdb0.
This is fixed by rearranging the logic in s390_arch_setup.
gdb/gdbserver/ChangeLog:
* linux-s390-low.c (s390_get_hwcap): Replace tdesc parameter by
the word size. Add comment.
(s390_get_wordsize): New function.
(s390_arch_setup): No longer select a temporary tdesc to fetch the
pswm with it. Instead, use s390_get_wordsize to determine the
word size first and derive the correct tdesc from that directly.
Alan Modra [Mon, 19 Mar 2018 09:44:05 +0000 (20:14 +1030)]
Make tls_mask unsigned in elf32-ppc.c
* elf32-ppc.c (struct ppc_elf_link_hash_entry): Make tls_mask
field unsigned.
(update_local_sym_info): Likewise for local_got_tls_masks.
(ppc_elf_tls_optimize): Likewise for lgot_masks.
Stephen Roberts [Tue, 20 Mar 2018 13:54:54 +0000 (13:54 +0000)]
Replace the linear search in find_pc_sect_line with a binary search.
This patch addresses slowness when setting breakpoints, especially in
heavily templatized code. Profiling showed that find_pc_sect_line in
symtab.c was the performance bottleneck. The original logic performed a
linear search over ordered data. This patch uses a binary search, as
suggested by comments around the function. There are no behavioural
changes, but gdb is now faster at setting breakpoints in template code.
Tested using on make check on an x86 target. The optimisation speeds up
the included template-breakpoints.py performance test by a factor of 7
on my machine.
ChangeLog:
2018-03-20 Stephen Roberts <stephen.roberts@arm.com>
* gdb/symtab.c (find_pc_sect_line): now uses binary search.
gdb/testsuite/
* gdb.perf/template-breakpoints.cc: New file.
* gdb.perf/template-breakpoints.exp: New file.
* gdb.perf/template-breakpoints.py: New file.
gc.exp: Remove extraneous LD flags from the personality test
Complement commit 8988502d7534 ("MIPS/LD/testsuite: Correct dynamic
links with VR4100, VR4300 and VR5000") and commit bf48520113c4 ("gc.exp:
Fix a typo: $LFLAGS -> $LDFLAGS"), and remove extraneous additional LD
flags explicitly passed to the personality test on invocation. This
passing has accidentally leaked from a WIP version of the former change.
With the actual version committed the $LDFLAGS global variable is used
instead for `run_dump_test' to use implicitly, so that `-call_shared'
precedes `-lpersonality' on the linker's command line, as otherwise a
shared `libpersonality.so' library wouldn't be considered for linking,
causing the test to fail.
ld/
* testsuite/ld-gc/gc.exp: Remove extraneous LD flags from the
personality test.
Tom Tromey [Sun, 2 Oct 2016 16:50:20 +0000 (10:50 -0600)]
Convert observers to C++
This converts observers from using a special source-generating script
to be plain C++. This version of the patch takes advantage of C++11
by using std::function and variadic templates; incorporates Pedro's
patches; and renames the header file to "observable.h" (this change
eliminates the need for a clean rebuild).
Note that Pedro's patches used a template lambda in tui-hooks.c, but
this failed to compile on some buildbot instances (presumably due to
differing C++ versions); I replaced this with an ordinary template
function.
Regression tested on the buildbot.
gdb/ChangeLog
2018-03-19 Pedro Alves <palves@redhat.com>
Tom Tromey <tom@tromey.com>
Andreas Arnez [Mon, 19 Mar 2018 12:13:09 +0000 (13:13 +0100)]
Testsuite: Fix ambiguous "break" due to libinproctrace
Some of GDB's trace test cases define a function end() and place a
breakpoint there with "break end". However, when libinproctrace is linked
to the binary, there are multiple methods named "end", such as
std::string::end() from the C++ library or format_pieces::end() from
common/format.h. GDB then creates multiple breakpoints instead of just a
single one, and some FAILs result, such as these:
FAIL: gdb.trace/trace-mt.exp: ftrace on: break end
FAIL: gdb.trace/trace-mt.exp: ftrace off: break end
Fix this by adding the "-qualified" option to the break commands. For
consistency, change all occurrences of "break end" (and similar) in all
trace test cases, even if the current behavior does not cause problems.
Also, consequently use the gdb_breakpoint convenience proc.
Tom Tromey [Fri, 16 Mar 2018 23:02:11 +0000 (17:02 -0600)]
Change auto_load_objfile_script_1 to use std::string
This replaces some manual string manipulation in
auto_load_objfile_script_1 with std::string, simplifying the code and
allowing the removal of some cleanups.
Tested by the buildbot.
2018-03-17 Tom Tromey <tom@tromey.com>
* auto-load.c (auto_load_objfile_script_1): Use std::string.
Tom Tromey [Fri, 23 Feb 2018 21:42:26 +0000 (14:42 -0700)]
Remove target_fileio_close_cleanup
This removes target_fileio_close_cleanup in favor of a new RAII class.
The new class is similar to scoped_fd but calls
target_fileio_close_cleanup rather than close.
Regression tested by the buildbot.
gdb/ChangeLog
2018-03-17 Tom Tromey <tom@tromey.com>
* target.c (class scoped_target_fd): New.
(target_fileio_close_cleanup): Remove.
(target_fileio_read_alloc_1): Use scoped_target_fd.
Simon Marchi [Fri, 16 Mar 2018 20:06:23 +0000 (16:06 -0400)]
Add silent Makefile rules
Many projects (e.g. the Linux kernel) and build systems use "silent"
rules, which means that they'll only print a summary of what's being
done instead of printing all the detailed command lines. While chatting
on the #gdb IRC channel, I realized a few people (including me) thought
it would be nice to have it in GDB too.
The idea is that too much text is not useful, the important information
gets lost. If there's only the essential information, it's more likely
to be useful. Most of the time, when I look at the build output, it's
to see how it's progressing. By just printing a brief summary of each
operation, I can easily spot what's currently being compiled and
therefore how the build progresses (with time you know the order in
which files are compiled almost by heart).
As with other projects (Linux, automake-based things, probably others),
it's possible to print the complete command lines by passing V=1 to make
(or any other non-zero value).
I had one hesitation about this: when people report build failures, we
are more likely to miss the full compile command line. We'll probably
sometimes need to ask people to include the build log with "make V=1".
I don't think it's a big downside, if other projects the size of the
Linux kernel can live with it, I'm sure we can too.
Max Filippov [Thu, 15 Mar 2018 17:35:59 +0000 (10:35 -0700)]
xtensa: bfd: fix assertion in xlate_offset_with_removed_text
Linking objects containing jumps targeting the end of a section triggers
assertion in the xlate_offset_with_removed_text. Such jumps may be
generated by a compiler as a dead code and not removed at -O0. Allow
such jumps.
While at it make bsearch argument match comparison function expectations
and use bfd_vma for address fields in the struct xlate_map_entry.
bfd/
2018-03-15 Max Filippov <jcmvbkbc@gmail.com>
* elf32-xtensa.c (xlate_map_entry): Change types of address
fields from 'unsigned' to 'bfd_vma'.
(xlate_offset_with_removed_text): Use struct xlate_map_entry as
the key argument to bsearch. Allow offsets past the end of a
section, use the last map entry for translation of such offsets.
Jim Wilson [Wed, 14 Mar 2018 23:04:03 +0000 (16:04 -0700)]
RISC-V: Add .insn support.
gas/ChangeLog
2018-03-07 Kito Cheng <kito.cheng@gmail.com>
* config/tc-riscv.c (opcode_name_list): New.
(opcode_names_hash): Likewise.
(init_opcode_names_hash): Likewise.
(opcode_name_lookup): Likewise.
(validate_riscv_insn): New argument length, and add new format
which used in .insn directive.
(md_begin): Refine hash table initialization logic into
init_opcode_hash.
(init_opcode_hash): New.
(my_getOpcodeExpression): Parse opcode name for .insn.
(riscv_ip): New argument hash, able to handle .insn directive.
(s_riscv_insn): Handler for .insn directive.
(riscv_pseudo_table): New entry for .insn.
* doc/c-riscv.texi: Add documentation for .insn directive.
* testsuite/gas/riscv/insn.d: Add testcase for .insn directive.
* testsuite/gas/riscv/insn.s: Likewise.
Tom Tromey [Thu, 15 Feb 2018 03:11:16 +0000 (20:11 -0700)]
Special case NULL when using printf's %s format
This changes the printf command's %s and %ls formats to special-case
NULL, and print "(null)" for these. This is PR cli/14977. This
behavior seems a bit friendlier; I was undecided on whether other
invalid pointers should be handled specially somehow, so for the time
being I've left those out.
gdb/ChangeLog
2018-03-14 Tom Tromey <tom@tromey.com>
PR cli/14977:
* printcmd.c (printf_c_string, printf_wide_c_string): Special case
for NULL.
gdb/gdbserver/ChangeLog
2018-03-14 Tom Tromey <tom@tromey.com>
PR cli/14977:
* ax.c (ax_printf): Special case for NULL.
gdb/testsuite/ChangeLog
2018-03-14 Tom Tromey <tom@tromey.com>
PR cli/14977:
* gdb.base/printcmds.exp (test_printf): Add printf test of %s with
a null pointer.
* gdb.base/wchar.exp: Likewise.
Tom Tromey [Wed, 14 Feb 2018 18:12:17 +0000 (11:12 -0700)]
Allow - in %p for printf
PR cli/19918 points out that a printf format like "%-5p" will cause a
gdb crash. The bug is problem is that printf_pointer doesn't take the
"-" flag into account.
gdb/ChangeLog
2018-03-14 Tom Tromey <tom@tromey.com>
PR cli/19918:
* printcmd.c (printf_pointer): Allow "-" in format.
gdb/testsuite/ChangeLog
2018-03-14 Tom Tromey <tom@tromey.com>
PR cli/19918:
* gdb.base/printcmds.exp (test_printf): Add printf test using '-'
flag.
to adjust from the original out-of-line save/restore function address
in sfpr to a copy at the end of stub_sec goes wrong when stub_sec is
padded, because the copy is no longer at the end of stub_sec. The
solution is to pad before copying sfpr, so the copy is always at the
end of stub_sec.
* elf64-ppc.c (sfpr_define): Adjust for stub_sec size having
sfpr size added before defining alias symbols.
(ppc64_elf_build_stubs): Add stub section padding before
copying sfpr contents and defining save/restore alias symbols.
Nick Clifton [Tue, 13 Mar 2018 16:10:45 +0000 (16:10 +0000)]
Add range changing to STABS parsing functions, in order to prevent buffer overruns.
PR 22955
* stabs.c (parse_number): Add p_end parameter and use it to check
the validity of the pp parameter. Add checks to prevent walking
off the end of the string buffer.
(parse_stab_string): Likewise.
(parse_stab_type): Likewise.
(parse_stab_type_number): Likewise.
(parse_stab_range_type): Likewise.
(parse_stab_sun_builtin_type): Likewise.
(parse_stab_sun_floating_type): Likewise.
(parse_stab_enum_type): Likewise.
(parse_stab_struct_type): Likewise.
(parse_stab_baseclasses): Likewise.
(parse_stab_struct_fields): Likewise.
(parse_stab_cpp_abbrev): Likewise.
(parse_stab_one_struct_field): Likewise.
(parse_stab_members): Likewise.
(parse_stab_tilde_field): Likewise.
(parse_stab_array_type): Likewise.
* parse_stab: Compute the end of the string and then pass it on to
individual parser functions.
bfd * pei-x86_64.c (pex64_get_unwind_info): Change to a boolean
function. Add an end address parameter. Check access of the data
pointer to make sure that they do not extend beyond the end
address. Return FALSE if any check fails. Add the end address
pointer to the ui structure.
(pex64_xdata_print_uwd_codes): Check accesses of the raw unwind
codes to make sure that they do not extend beyond the end address
pointer. Print an error message and return immediately if any
check fails.
* riscv-tdep.c (riscv_sw_breakpoint_from_kind): Add localization
to error message string.
(riscv_register_name): Use xsnprintf instead of sprintf.
(riscv_insn::fetch_instruction): Use gdb_assert instead of
internal_error.
(riscv_print_arg_location): Use gdb_assert_not_reached instead of
error.
(riscv_push_dummy_call): Likewise.
Yao Qi [Mon, 12 Mar 2018 09:15:39 +0000 (09:15 +0000)]
Fix ia64 GDB build
Commit 849d0ba8 breaks GDB build for ia64 with --with-libunwind-ia64=yes.
This patch fixes it.
gdb:
2018-03-12 Yao Qi <yao.qi@linaro.org>
* ia64-libunwind-tdep.c (libunwind_get_reg_special): Change
parameter type to readable_regcache.
* ia64-libunwind-tdep.h (libunwind_get_reg_special): Update
the declaration.
Tom Tromey [Sun, 11 Mar 2018 03:47:06 +0000 (20:47 -0700)]
Use std::vector for field lists in dwarf2read.c
This changes dwarf2read.c to use std::vector rather than a linked list
when managing the fields and base classes to be added to a type. This
removes some bookkeeping types and also allows the removal of some
cleanups.
gdb/ChangeLog
2018-03-11 Tom Tromey <tom@tromey.com>
Nick Clifton [Fri, 9 Mar 2018 14:37:36 +0000 (14:37 +0000)]
Fix Sparc, s390 and AArch64 targets so that they can handle relocs against ifunc symbols found in note sections.
Following on from PR 22929, I have found the same problem exists with
other ifunc supporting targets too. Plus see this link for the bug
being reported against the s390x binutils for Fedora rawhide:
So I am going to check in the patch below which applies the same
change that H.J. made for the x86_64 target to the other affected
targets. (Specifically: S390, AArch64 and Sparc). Plus it adds a new
test to the linker testsuite to make sure that this problem stays
fixed.
bfd * elf64-s390.c (elf_s390_relocate_section): Move check for
relocations against non-allocated sections to before the code that
handles ifunc relocations.
* elf32-s390.c (elf_s390_relocate_section): Likewise.
* elfnn-aarch64.c (elfNN_aarch64_final_link_relocate): Treat
relocs against IFUNC symbols in non-allocated sections as relocs
against FUNC symbols.
* elfxx-sparc.c (_bfd_sparc_elf_relocate_section): Likewise.
ld * testsuite/ld-ifunc/ifuncmod5.s: New test. Checks that targets
that support IFUNC symbols can handle relocations against those
symbols in NOTE sections.
* testsuite/ld-ifunc/ifuncmod5.d: New file: Driver for the new
test.
* testsuite/ld-ifunc/ifunc.exp: Run the new test.
Tom Tromey [Wed, 7 Mar 2018 22:55:01 +0000 (15:55 -0700)]
Change enable_thread_stack_temporaries to an RAII class
This started as a patch to change enable_thread_stack_temporaries to
be an RAII class, but then I noticed that this code used a VEC, so I
went ahead and did a bit more C++-ification, changing
stack_temporaries_enabled to a bool and changing stack_temporaries to
a std::vector.
Regression tested by the buildbot.
gdb/ChangeLog
2018-03-08 Tom Tromey <tom@tromey.com>
* infcall.c (struct call_return_meta_info)
<stack_temporaries_enabled>: Remove.
(get_call_return_value, call_function_by_hand_dummy): Update.
* thread.c (disable_thread_stack_temporaries): Remove.
(enable_thread_stack_temporaries): Remove.
(thread_stack_temporaries_enabled_p): Return bool.
(push_thread_stack_temporary, value_in_thread_stack_temporaries)
(get_last_thread_stack_temporary): Update.
* eval.c (evaluate_subexp): Update.
* gdbthread.h (class enable_thread_stack_temporaries): Now a
class, not a function.
(value_ptr, value_vec): Remove typedefs.
(class thread_info) <stack_temporaries_enabled>: Now bool.
<stack_temporaries>: Now a std::vector.
(thread_stack_temporaries_enabled_p)
(value_in_thread_stack_temporaries): Return bool.
Simon Marchi [Fri, 9 Mar 2018 00:00:03 +0000 (19:00 -0500)]
Fix misreporting of omitted bytes for large remote packets
In remote.c, when the output of "set debug remote" is truncated, the
number of characters reported is incorrect. What is reported is the
number of characters added by the quoting, not the number of characters
that were truncated.
Simon Marchi [Thu, 8 Mar 2018 23:57:53 +0000 (18:57 -0500)]
Use std::string to simplify build_id_to_debug_bfd
Using std::string here makes the string building simpler thank playing
with char*. A stack allocation is replaced with heap allocation, but
I don't think this is really performance-critical code.
gdb/ChangeLog:
* build-id.c (build_id_to_debug_bfd): Use std::string.
Simon Marchi [Thu, 8 Mar 2018 23:04:46 +0000 (18:04 -0500)]
Add xml_escape_text_append and use it
[This patch should go on top of "linux_qxfer_libraries_svr4: Use
std::string", I should have sent them together as a series.]
I noticed that linux_qxfer_libraries_svr4 used xml_escape_text, which
returns an std::string. That string is then copied into a larger
buffer. It would be more efficient if we had a version of
xml_escape_text which appended to an existing string instead of
returning a new one. This is what this patch does.
I manually verified that the output of linux_qxfer_libraries_svr4 didn't
change before/after the patch.
gdb/ChangeLog:
* common/xml-utils.c (xml_escape_text): Move code to...
(xml_escape_text_append): ... this new function.
* common/xml-utils.h (xml_escape_text_append): New declaration.
* unittests/xml-utils-selftests.c (test_xml_escape_text_append):
New function.
(_initialize_xml_utils): register test_xml_escape_text_append as
a selftest.
gdb/gdbserver/ChangeLog:
* linux-low.c (linux_qxfer_libraries_svr4): Use
xml_escape_text_append.
there seems to be situations where the remote-stdio-gdbserver board
fails to delete the uploaded binary file. Passing "target" fixes the
issue for Christian who reported the bug.
I did not experience this problem, but passing "target" to remote_exec
still works for me, so I'm fine with changing it.
Any objection?
gdb/testsuite/ChangeLog:
PR gdb/22841
* boards/remote-stdio-gdbserver.exp (${board}_file): Pass
"target" to remote_exec.
the local gdbserver boards (except native-extended-gdbserver...) were
considered as remote by DejaGNU. To avoid DejaGNU trying to use ssh/scp
to download the files to the target (which is actually local), the
gdbserver-base.exp file defined some _download, _upload and _file board
operations to override the default behavior, and instead just use local
operations.
The same patch also changed remote-stdio-gdbserver.exp to make it
inherit from gdbserver-base.exp. Since then, this board (which is
actually remote) uses the overrides with local file operations. As a
result, files are never actually copied to the target.
I think we can simply remove the overrides from gdbserver-base.exp.
Because all boards should be properly considered local or remote by
DejaGNU, it should by default use the right method for transferring
files.
H.J. Lu [Thu, 8 Mar 2018 14:36:40 +0000 (06:36 -0800)]
x86: Treat relocation against IFUNC symbol as FUNC
When resolving a relocation against IFUNC symbol in a SHT_NOTE section
without SHF_ALLOC, we treat it as relocation against FUNC symbol since
it needs the address of IFUNC symbol, not the address returned by IFUNC
function.
bfd/
PR ld/22929
* elf32-i386.c (elf_i386_relocate_section): Treat relocation
against IFUNC symbol in SHT_NOTE section without SHF_ALLOC as
relocation against FUNC symbol.
* elf64-x86-64.c (elf_x86_64_relocate_section): Likewise.
Jan Beulich [Thu, 8 Mar 2018 07:58:55 +0000 (08:58 +0100)]
x86: fold several AVX512VL templates
The differences between some of the register and memory forms of the
same insn often don't really require the templates to be separate. For
example, Disp8MemShift is simply irrelevant to register forms. Fold
these as far as possible, and also fold register-only forms. Further
folding is possible, but needs other prereq work done first.
A note regarding EVEXDYN: This is intended to be used only when no other
properties of the template would make is_evex_encoding() return true. In
all "normal" cases I think it is preferable to omit this indicator, to
keep the table half way readable.
Jan Beulich [Thu, 8 Mar 2018 07:58:05 +0000 (08:58 +0100)]
x86: fold certain AVX512 rotate and shift templates
Their memory forms were bogusly using VexLWP instead of VexNDD. Adjust
VexNDD handling to cope with these, allowing their register and memory
forms to be folded.
Jan Beulich [Thu, 8 Mar 2018 07:56:47 +0000 (08:56 +0100)]
x86: fold a few AVX512F templates
The differences between some of the register and memory forms of the
same insn often don't really require the templates to be separate. For
example, Disp8MemShift is simply irrelevant to register forms. Fold them
as far as possible. Further folding is possible, but needs other prereq
work done first.
Jan Beulich [Thu, 8 Mar 2018 07:52:27 +0000 (08:52 +0100)]
x86: drop {X,Y,Z}MMWORD_MNEM_SUFFIX
They aren't really useful (anymore?): The conflicting operand size check
isn't applicable to any insn validly using respective memory operand
sizes (and if they're used wrongly, another error would result), and the
logic in process_suffix() can be easily changed to work without them.
While re-structuring conditionals in process_suffix() also drop the
CMPXCHG8B special case in favor of a NoRex64 attribute in the opcode
table.
Jan Beulich [Thu, 8 Mar 2018 07:51:18 +0000 (08:51 +0100)]
x86: correct operand size match checks for BMI/BMI2 insns
Some BMI/BMI2 insns allow their middle operands to be a memory one. In
such a case, matching register types between operands 0 and 1 as well as
1 and 2 won't help - operands 0 and 2 also need to be checked.
Jan Beulich [Thu, 8 Mar 2018 07:48:10 +0000 (08:48 +0100)]
x86: simplify result processing of cpu_flags_match()
Make more obvious what the success and failure paths are, and in
particular that what used to be at the "skip" label can't be reached
by what used to be straight line code.
Jan Beulich [Thu, 8 Mar 2018 07:47:32 +0000 (08:47 +0100)]
x86: add GFNI, VAES, and VPCLMUL checking to cpu_flags_match()
Just like for the AVX/AES and AVX/PCLMUL combinations, AVX/GFN,
AVX512F/GFNI, AVX512F/VAES, and AVX512F/PCLMUL need special handling to
deal with the pair of required checks specified in the templates.