From: Roland McGrath <roland@hack.frob.com>
To: libc-ports@sourceware.org
Subject: [PATCH roland/arm-elf_machine_dynamic] ARM: Rewrite elf_machine_dynamic in pure C.
Date: Mon, 06 May 2013 21:14:00 -0000 [thread overview]
Message-ID: <20130506211358.0B96D2C061@topped-with-meat.com> (raw)
I don't know the history of why it was ever anything but this.
Today, the compiler produces code for both -marm and -mthumb that is
just about identical to what we had hand-written here. Perhaps it
dates from a time when attribute_hidden wasn't available or wasn't
reliable?
For new variants like arm-nacl, using C leverages the compiler's
knowledge of special requirements and avoids having to get another
variant defined by hand. Is there any reason not to use C everywhere?
Thanks,
Roland
ports/ChangeLog.arm
* sysdeps/arm/dl-machine.h (elf_machine_dynamic): Use a plain C
reference to _GLOBAL_OFFSET_TABLE_ with an STV_HIDDEN declaration.
--- a/ports/sysdeps/arm/dl-machine.h
+++ b/ports/sysdeps/arm/dl-machine.h
@@ -39,30 +39,13 @@ elf_machine_matches_host (const Elf32_Ehdr *ehdr)
/* Return the link-time address of _DYNAMIC. Conveniently, this is the
- first element of the GOT. We used to use the PIC register to do this
- without a constant pool reference, but GCC 4.2 will use a pseudo-register
- for the PIC base, so it may not be in r10. */
+ first element of the GOT. */
static inline Elf32_Addr __attribute__ ((unused))
elf_machine_dynamic (void)
{
- Elf32_Addr dynamic;
-#ifdef __thumb2__
- long tmp;
- asm ("ldr\t%0, 1f\n\t"
- "adr\t%1, 1f\n\t"
- "ldr\t%0, [%0, %1]\n\t"
- "b 2f\n"
- ".align 2\n"
- "1: .word _GLOBAL_OFFSET_TABLE_ - 1b\n"
- "2:" : "=r" (dynamic), "=r"(tmp));
-#else
- asm ("ldr %0, 2f\n"
- "1: ldr %0, [pc, %0]\n"
- "b 3f\n"
- "2: .word _GLOBAL_OFFSET_TABLE_ - (1b+8)\n"
- "3:" : "=r" (dynamic));
-#endif
- return dynamic;
+ /* Declaring this hidden ensures that a PC-relative reference is used. */
+ extern const Elf32_Addr _GLOBAL_OFFSET_TABLE_[] attribute_hidden;
+ return _GLOBAL_OFFSET_TABLE_[0];
}
next reply other threads:[~2013-05-06 21:14 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-05-06 21:14 Roland McGrath [this message]
2013-05-07 15:30 ` Joseph S. Myers
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=20130506211358.0B96D2C061@topped-with-meat.com \
--to=roland@hack.frob.com \
--cc=libc-ports@sourceware.org \
/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).