From: "H.J. Lu" <hjl.tools@gmail.com>
To: Michael Matz <matz@suse.de>
Cc: Sriraman Tallam <tmsriram@google.com>,
GCC Patches <gcc-patches@gcc.gnu.org>,
David Li <davidxl@google.com>
Subject: Re: [RFC][PATCH][X86_64] Eliminate PLT stubs for specified external functions via -fno-plt=
Date: Sun, 10 May 2015 15:19:00 -0000 [thread overview]
Message-ID: <CAMe9rOpF996wsqCjCUtApvSh8tNcf--=+OnfnNuSyKm0VPoUKQ@mail.gmail.com> (raw)
On Sat, May 9, 2015 at 9:34 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Mon, May 4, 2015 at 7:45 AM, Michael Matz <matz@suse.de> wrote:
>> Hi,
>>
>> On Thu, 30 Apr 2015, Sriraman Tallam wrote:
>>
>>> We noticed that one of our benchmarks sped-up by ~1% when we eliminated
>>> PLT stubs for some of the hot external library functions like memcmp,
>>> pow. The win was from better icache and itlb performance. The main
>>> reason was that the PLT stubs had no spatial locality with the
>>> call-sites. I have started looking at ways to tell the compiler to
>>> eliminate PLT stubs (in-effect inline them) for specified external
>>> functions, for x86_64. I have a proposal and a patch and I would like to
>>> hear what you think.
>>>
>>> This comes with caveats. This cannot be generally done for all
>>> functions marked extern as it is impossible for the compiler to say if a
>>> function is "truly extern" (defined in a shared library). If a function
>>> is not truly extern(ends up defined in the final executable), then
>>> calling it indirectly is a performance penalty as it could have been a
>>> direct call.
>>
>> This can be fixed by Alans idea.
>>
>>> Further, the newly created GOT entries are fixed up at
>>> start-up and do not get lazily bound.
>>
>> And this can be fixed by some enhancements in the linker and dynamic
>> linker. The idea is to still generate a PLT stub and make its GOT entry
>> point to it initially (like a normal got.plt slot). Then the first
>> indirect call will use the address of PLT entry (starting lazy resolution)
>> and update the GOT slot with the real address, so further indirect calls
>> will directly go to the function.
>>
>> This requires a new asm marker (and hence new reloc) as normally if
>> there's a GOT slot it's filled by the real symbols address, unlike if
>> there's only a got.plt slot. E.g. a
>>
>> call *foo@GOTPLT(%rip)
>>
>> would generate a GOT slot (and fill its address into above call insn), but
>> generate a JUMP_SLOT reloc in the final executable, not a GLOB_DAT one.
>>
>
> I added the "relax" prefix support to x86 assembler on users/hjl/relax
> branch
>
> at
>
> https://sourceware.org/git/?p=binutils-gdb.git;a=summary
>
> [hjl@gnu-tools-1 relax-3]$ cat r.S
> .text
> relax jmp foo
> relax call foo
> relax jmp foo@plt
> relax call foo@plt
> [hjl@gnu-tools-1 relax-3]$ ./as -o r.o r.S
> [hjl@gnu-tools-1 relax-3]$ ./objdump -drw r.o
>
> r.o: file format elf64-x86-64
>
>
> Disassembly of section .text:
>
> 0000000000000000 <.text>:
> 0: 66 e9 00 00 00 00 data16 jmpq 0x6 2: R_X86_64_RELAX_PC32 foo-0x4
> 6: 66 e8 00 00 00 00 data16 callq 0xc 8: R_X86_64_RELAX_PC32 foo-0x4
> c: 66 e9 00 00 00 00 data16 jmpq 0x12 e: R_X86_64_RELAX_PLT32foo-0x4
> 12: 66 e8 00 00 00 00 data16 callq 0x18 14: R_X86_64_RELAX_PLT32foo-0x4
> [hjl@gnu-tools-1 relax-3]$
>
> Right now, the relax relocations are treated as PC32/PLT32 relocations.
> I am working on linker support.
>
I implemented the linker support for x86-64:
00000000 <main>:
0: 48 83 ec 08 sub $0x8,%rsp
4: e8 00 00 00 00 callq 9 <main+0x9> 5: R_X86_64_PC32 plt-0x4
9: e8 00 00 00 00 callq e <main+0xe> a: R_X86_64_PLT32 plt-0x4
e: e8 00 00 00 00 callq 13 <main+0x13> f: R_X86_64_PC32 bar-0x4
13: 66 e8 00 00 00 00 data16 callq 19 <main+0x19> 15:
R_X86_64_RELAX_PC32 bar-0x4
19: 66 e8 00 00 00 00 data16 callq 1f <main+0x1f> 1b:
R_X86_64_RELAX_PLT32 bar-0x4
1f: 66 e8 00 00 00 00 data16 callq 25 <main+0x25> 21:
R_X86_64_RELAX_PC32 foo-0x4
25: 66 e8 00 00 00 00 data16 callq 2b <main+0x2b> 27:
R_X86_64_RELAX_PLT32 foo-0x4
2b: 31 c0 xor %eax,%eax
2d: 48 83 c4 08 add $0x8,%rsp
31: c3 retq
00400460 <main>:
400460: 48 83 ec 08 sub $0x8,%rsp
400464: e8 d7 ff ff ff callq 400440 <plt@plt>
400469: e8 d2 ff ff ff callq 400440 <plt@plt>
40046e: e8 ad ff ff ff callq 400420 <bar@plt>
400473: ff 15 ff 03 20 00 callq *0x2003ff(%rip) # 600878
<_DYNAMIC+0xf8>
400479: ff 15 f9 03 20 00 callq *0x2003f9(%rip) # 600878
<_DYNAMIC+0xf8>
40047f: 66 e8 f3 00 00 00 data16 callq 400578 <foo>
400485: 66 e8 ed 00 00 00 data16 callq 400578 <foo>
40048b: 31 c0 xor %eax,%eax
40048d: 48 83 c4 08 add $0x8,%rsp
400491: c3 retq
Sriraman, can you give it a try?
--
H.J.
next reply other threads:[~2015-05-10 15:19 UTC|newest]
Thread overview: 65+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-05-10 15:19 H.J. Lu [this message]
[not found] ` <CAAs8HmwWSDY+KjKcB4W=TiYV0Pz7NSvfL_8igp+hPT-LU1utTg@mail.gmail.com>
2015-05-21 21:31 ` Sriraman Tallam
2015-05-21 21:39 ` Sriraman Tallam
2015-05-21 22:02 ` Pedro Alves
2015-05-21 22:02 ` Jakub Jelinek
2015-05-22 1:47 ` H.J. Lu
2015-05-22 3:38 ` Xinliang David Li
2015-05-21 22:34 ` Sriraman Tallam
2015-05-22 9:22 ` Pedro Alves
2015-05-22 15:13 ` Sriraman Tallam
2015-05-28 18:53 ` Sriraman Tallam
2015-05-28 19:05 ` H.J. Lu
2015-05-28 19:48 ` Sriraman Tallam
2015-05-28 20:19 ` H.J. Lu
2015-05-28 21:27 ` Sriraman Tallam
2015-05-28 21:31 ` H.J. Lu
2015-05-28 21:52 ` Sriraman Tallam
2015-05-28 22:48 ` H.J. Lu
2015-05-29 3:51 ` Sriraman Tallam
2015-05-29 5:13 ` H.J. Lu
2015-05-29 7:13 ` Sriraman Tallam
2015-05-29 17:36 ` Sriraman Tallam
2015-05-29 17:52 ` H.J. Lu
2015-05-29 18:33 ` Sriraman Tallam
2015-05-29 20:50 ` Jan Hubicka
2015-05-29 22:56 ` Sriraman Tallam
2015-05-29 23:08 ` Sriraman Tallam
[not found] ` <CAJA7tRYsMiq7rx34c=z6KwRdwYxxaeP6Z6qzA4XEwnJSMT7z=Q@mail.gmail.com>
2015-05-30 4:44 ` Sriraman Tallam
2015-06-01 8:24 ` Ramana Radhakrishnan
2015-06-01 18:01 ` Sriraman Tallam
2015-06-01 18:41 ` Ramana Radhakrishnan
2015-06-01 18:55 ` Sriraman Tallam
2015-06-01 20:33 ` Ramana Radhakrishnan
2015-06-02 18:27 ` Sriraman Tallam
2015-06-02 19:59 ` Bernhard Reutner-Fischer
2015-06-02 20:09 ` Sriraman Tallam
2015-06-02 21:18 ` Bernhard Reutner-Fischer
2015-06-02 21:09 ` Ramana Radhakrishnan
2015-06-02 21:25 ` Xinliang David Li
2015-06-02 21:52 ` Bernhard Reutner-Fischer
2015-06-02 21:40 ` Sriraman Tallam
2015-06-03 14:37 ` Ramana Radhakrishnan
2015-06-03 18:53 ` Sriraman Tallam
2015-06-03 20:16 ` Richard Henderson
2015-06-03 20:59 ` Sriraman Tallam
2015-06-04 16:56 ` Sriraman Tallam
2015-06-04 17:30 ` Richard Henderson
2015-06-04 21:34 ` Sriraman Tallam
2015-07-24 19:02 ` H.J. Lu
2015-06-03 19:57 ` Richard Henderson
-- strict thread matches above, loose matches on Subject: below --
2015-05-01 0:31 Sriraman Tallam
2015-05-01 3:21 ` Alan Modra
2015-05-01 3:26 ` Sriraman Tallam
2015-05-01 15:01 ` Andi Kleen
2015-05-01 16:19 ` Xinliang David Li
2015-05-01 16:23 ` H.J. Lu
2015-05-01 16:26 ` Xinliang David Li
2015-05-01 18:06 ` Sriraman Tallam
2015-05-02 12:12 ` Andi Kleen
2015-05-01 17:50 ` Sriraman Tallam
2015-05-04 14:45 ` Michael Matz
2015-05-04 16:43 ` Xinliang David Li
2015-05-04 16:58 ` Michael Matz
2015-05-04 17:22 ` Xinliang David Li
2015-05-09 16:35 ` H.J. Lu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAMe9rOpF996wsqCjCUtApvSh8tNcf--=+OnfnNuSyKm0VPoUKQ@mail.gmail.com' \
--to=hjl.tools@gmail.com \
--cc=davidxl@google.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=matz@suse.de \
--cc=tmsriram@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).