This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] Enable Intel RDPID instruction.
- From: "H.J. Lu" <hjl dot tools at gmail dot com>
- To: Alexander Fomin <afomin dot mailbox at gmail dot com>
- Cc: Binutils <binutils at sourceware dot org>
- Date: Tue, 10 May 2016 05:08:13 -0700
- Subject: Re: [PATCH] Enable Intel RDPID instruction.
- Authentication-results: sourceware.org; auth=none
- References: <20160503102646 dot GA26920 at msticlxl57 dot ims dot intel dot com> <CAMe9rOo4sCy3b3ttNi+K2Nrz6+4OxiHKaMhOCnZ9ejPxi3PDTg at mail dot gmail dot com> <CAMe9rOquSZXPCEcAJqnwJEcwP7qJ9u+AV0qgFxHBoWaMwZOJXg at mail dot gmail dot com> <20160510120254 dot GA50622 at msticlxl57 dot ims dot intel dot com>
On Tue, May 10, 2016 at 5:02 AM, Alexander Fomin
<afomin.mailbox@gmail.com> wrote:
> On Tue, May 03, 2016 at 07:54:51AM -0700, H.J. Lu wrote:
>> On Tue, May 3, 2016 at 5:05 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> > On Tue, May 3, 2016 at 3:26 AM, Alexander Fomin
>> > <afomin.mailbox@gmail.com> wrote:
>> >> Hi,
>> >> This patch enables Intel RDPID instruction.
>> >>
>> >> Please refer to Intel SDM Vol. 2 (r58) for details about the instruction
>> >> and corresponding CPUID.
>> >>
>> >> I've eliminated i386-init.h and i386-tbl.h from the patch as they were
>> >> just regenerated.
>> >>
>> >> Any objectionis?
>> >>
>> >> Alexander
>> >> ---
>> >> gas/
>> >>
>> >> * doc/c-i386.texi: Document RDPID.
>> >>
>> >> gas/testsuite/
>> >>
>> >> * gas/i386/i386.exp: Add RDPID tests.
>> >> * gas/i386/prefix.d: Adjust.
>> >> * gas/i386/rdpid.s: New test.
>> >> * gas/i386/rdpid.d: Ditto.
>> >> * gas/i386/rdpid-intel.d: Ditto.
>> >> * gas/i386/x86-64-rdpid.s: Ditto.
>> >> * gas/i386/x86-64-rdpid.d: Ditto.
>> >> * gas/i386/x86-64-rdpid-intel.d: Ditto.
>> >>
>> >> opcodes/
>> >>
>> >> * i386-dis.c (prefix_table): Add RDPID instruction.
>> >> * i386-gen.c (cpu_flag_init): Add RDPID flag.
>> >> (cpu_flags): Add RDPID bitfield.
>> >> * i386-opc.h (enum): Add RDPID element.
>> >> (i386_cpu_flags): Add RDPID field.
>> >> * i386-opc.tbl: Add RDPID instruction.
>> >> * i386-init.h: Regenerate.
>> >> * i386-tbl.h: Regenerate.
>> >> ---
>> >> gas/doc/c-i386.texi | 3 +-
>> >> gas/testsuite/gas/i386/i386.exp | 4 +
>> >> gas/testsuite/gas/i386/prefix.d | 3 +-
>> >> gas/testsuite/gas/i386/rdpid-intel.d | 11 +
>> >> gas/testsuite/gas/i386/rdpid.d | 11 +
>> >> gas/testsuite/gas/i386/rdpid.s | 5 +
>> >> gas/testsuite/gas/i386/x86-64-rdpid-intel.d | 11 +
>> >> gas/testsuite/gas/i386/x86-64-rdpid.d | 11 +
>> >> gas/testsuite/gas/i386/x86-64-rdpid.s | 5 +
>> >> opcodes/i386-dis.c | 2 +-
>> >> opcodes/i386-gen.c | 3 +
>> >> opcodes/i386-init.h | 223 +-
>> >> opcodes/i386-opc.h | 3 +
>> >> opcodes/i386-opc.tbl | 7 +
>> >> opcodes/i386-tbl.h | 10791 +++++++++++++-------------
>> >> 15 files changed, 5656 insertions(+), 5437 deletions(-)
>> >> create mode 100644 gas/testsuite/gas/i386/rdpid-intel.d
>> >> create mode 100644 gas/testsuite/gas/i386/rdpid.d
>> >> create mode 100644 gas/testsuite/gas/i386/rdpid.s
>> >> create mode 100644 gas/testsuite/gas/i386/x86-64-rdpid-intel.d
>> >> create mode 100644 gas/testsuite/gas/i386/x86-64-rdpid.d
>> >> create mode 100644 gas/testsuite/gas/i386/x86-64-rdpid.s
>> >>
>> >> diff --git a/gas/doc/c-i386.texi b/gas/doc/c-i386.texi
>> >> index adbcb99..edd187b 100644
>> >> --- a/gas/doc/c-i386.texi
>> >> +++ b/gas/doc/c-i386.texi
>> >> @@ -153,6 +153,7 @@ accept various extension mnemonics. For example,
>> >> @code{smap},
>> >> @code{mpx},
>> >> @code{sha},
>> >> +@code{rdpid},
>> >> @code{prefetchwt1},
>> >> @code{clflushopt},
>> >> @code{se1},
>> >> @@ -1173,7 +1174,7 @@ supported on the CPU specified. The choices for @var{cpu_type} are:
>> >> @item @samp{.3dnow} @tab @samp{.3dnowa} @tab @samp{.sse4a} @tab @samp{.sse5}
>> >> @item @samp{.syscall} @tab @samp{.rdtscp} @tab @samp{.svme} @tab @samp{.abm}
>> >> @item @samp{.lwp} @tab @samp{.fma4} @tab @samp{.xop} @tab @samp{.cx16}
>> >> -@item @samp{.padlock} @tab @samp{.clzero} @tab @samp{.mwaitx}
>> >> +@item @samp{.padlock} @tab @samp{.clzero} @tab @samp{.mwaitx} @tab @samp{.rdpid}
>> >> @end multitable
>> >>
>> >> Apart from the warning, there are only two other effects on
>> >> diff --git a/gas/testsuite/gas/i386/i386.exp b/gas/testsuite/gas/i386/i386.exp
>> >> index 9c4fdea..df1518b 100644
>> >> --- a/gas/testsuite/gas/i386/i386.exp
>> >> +++ b/gas/testsuite/gas/i386/i386.exp
>> >> @@ -346,6 +346,8 @@ if [expr ([istarget "i*86-*-*"] || [istarget "x86_64-*-*"]) && [gas_32_check]]
>> >> run_dump_test "mwaitx-bdver4"
>> >> run_list_test "mwaitx-reg"
>> >> run_dump_test "ospke"
>> >> + run_dump_test "rdpid"
>> >> + run_dump_test "rdpid-intel"
>> >>
>> >> # These tests require support for 8 and 16 bit relocs,
>> >> # so we only run them for ELF and COFF targets.
>> >> @@ -737,6 +739,8 @@ if [expr ([istarget "i*86-*-*"] || [istarget "x86_64-*-*"]) && [gas_64_check]] t
>> >> run_dump_test "x86-64-mwaitx-bdver4"
>> >> run_list_test "x86-64-mwaitx-reg"
>> >> run_dump_test "x86-64-ospke"
>> >> + run_dump_test "x86-64-rdpid"
>> >> + run_dump_test "x86-64-rdpid-intel"
>> >> run_dump_test "x86-64-fence-as-lock-add-yes"
>> >> run_dump_test "x86-64-fence-as-lock-add-no"
>> >>
>> >> diff --git a/gas/testsuite/gas/i386/prefix.d b/gas/testsuite/gas/i386/prefix.d
>> >> index 7afb0c1..8dd200b 100644
>> >> --- a/gas/testsuite/gas/i386/prefix.d
>> >> +++ b/gas/testsuite/gas/i386/prefix.d
>> >> @@ -63,8 +63,7 @@ Disassembly of section .text:
>> >> [ ]*[a-f0-9]+: f3 f0 66 3e 36 90 lock data16 ds ss pause
>> >> [ ]*[a-f0-9]+: f3 f2 67 3e 36 90 repz repnz addr16 ds ss nop
>> >> [ ]*[a-f0-9]+: f3 67 f2 66 3e 36 90 repz addr16 repnz ds ss xchg %ax,%ax
>> >> -[ ]*[a-f0-9]+: f3 0f c7 \(bad\)
>> >> -[ ]*[a-f0-9]+: f8 clc
>> >> +[ ]*[a-f0-9]+: f3 0f c7 f8 rdpid %eax
>> >> [ ]*[a-f0-9]+: 90 nop
>> >> [ ]*[a-f0-9]+: f3 0f c7 \(bad\)
>> >> [ ]*[a-f0-9]+: f0 90 lock nop
>> >> diff --git a/gas/testsuite/gas/i386/rdpid-intel.d b/gas/testsuite/gas/i386/rdpid-intel.d
>> >> new file mode 100644
>> >> index 0000000..f733b17
>> >> --- /dev/null
>> >> +++ b/gas/testsuite/gas/i386/rdpid-intel.d
>> >> @@ -0,0 +1,11 @@
>> >> +#objdump: -dwMintel
>> >> +#name: i386 RDPID (Intel disassembly)
>> >> +#source: rdpid.s
>> >> +
>> >> +.*: +file format .*
>> >> +
>> >> +Disassembly of section .text:
>> >> +
>> >> +0+ <_start>:
>> >> +[ ]*[a-f0-9]+:[ ]*f3 0f c7 f8[ ]*rdpid eax
>> >> +#pass
>> >> diff --git a/gas/testsuite/gas/i386/rdpid.d b/gas/testsuite/gas/i386/rdpid.d
>> >> new file mode 100644
>> >> index 0000000..ef67c84
>> >> --- /dev/null
>> >> +++ b/gas/testsuite/gas/i386/rdpid.d
>> >> @@ -0,0 +1,11 @@
>> >> +#objdump: -dw
>> >> +#name: i386 RDPID insns
>> >> +
>> >> +.*: +file format .*
>> >> +
>> >> +
>> >> +Disassembly of section .text:
>> >> +
>> >> +0+ <_start>:
>> >> +[ ]*[a-f0-9]+:[ ]*f3 0f c7 f8[ ]*rdpid %eax
>> >> +#pass
>> >> diff --git a/gas/testsuite/gas/i386/rdpid.s b/gas/testsuite/gas/i386/rdpid.s
>> >> new file mode 100644
>> >> index 0000000..fe6e401
>> >> --- /dev/null
>> >> +++ b/gas/testsuite/gas/i386/rdpid.s
>> >> @@ -0,0 +1,5 @@
>> >> +# Check 32bit RDPID instructions.
>> >> +
>> >> + .text
>> >> +_start:
>> >> + rdpid %eax
>> >> diff --git a/gas/testsuite/gas/i386/x86-64-rdpid-intel.d b/gas/testsuite/gas/i386/x86-64-rdpid-intel.d
>> >> new file mode 100644
>> >> index 0000000..df695ca
>> >> --- /dev/null
>> >> +++ b/gas/testsuite/gas/i386/x86-64-rdpid-intel.d
>> >> @@ -0,0 +1,11 @@
>> >> +#objdump: -dwMintel
>> >> +#name: x86_64 RDPID (Intel disassembly)
>> >> +#source: x86-64-rdpid.s
>> >> +
>> >> +.*: +file format .*
>> >> +
>> >> +Disassembly of section .text:
>> >> +
>> >> +0+ <_start>:
>> >> +[ ]*[a-f0-9]+:[ ]*f3 48 0f c7 f8[ ]*rdpid rax
>> > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>> >
>> > rdpid doesn't need the REX prefix.
>>
>> Please add a testcase for %r10.
>>
>>
>> --
>> H.J.
>
> Here is an updated version now that we've clarified REX.W behavior.
It is OK. Please check it in.
> Do we need any testcases to make sure REX.W is ignored?
>
It isn't necessary.
Thanks.
--
H.J.