public inbox for gcc-help@gcc.gnu.org
 help / color / mirror / Atom feed
From: Bob Plantz <rgplantz@outlook.com>
To: Richard Earnshaw <Richard.Earnshaw@foss.arm.com>,
	"gcc-help@gcc.gnu.org" <gcc-help@gcc.gnu.org>
Subject: RE: :got: modifier in assembler
Date: Wed, 23 Nov 2022 16:38:20 +0000	[thread overview]
Message-ID: <BYAPR04MB4661B770BBA28F743AD4F08CD40C9@BYAPR04MB4661.namprd04.prod.outlook.com> (raw)
In-Reply-To: <7e7e65b3-bf34-e8b6-0613-0fc9d0beaf33@foss.arm.com>

Thank you very much, Richard. This is precisely what I've been searching for!

--Bob

-----Original Message-----
From: Richard Earnshaw <Richard.Earnshaw@foss.arm.com> 
Sent: Wednesday, November 23, 2022 8:23 AM
To: Bob Plantz <rgplantz@outlook.com>; gcc-help@gcc.gnu.org
Subject: Re: :got: modifier in assembler



On 22/11/2022 19:12, Bob Plantz via Gcc-help wrote:
> I'm using the Arm A64 instruction set on a Raspberry Pi running Raspberry Pi 64-bit. Given the C code:
> 
> extern int x, y, z;
> 
> void addTwo(void)
> {
> 
>    z = x + y;
> }
> 
> gcc produces the assembly language:
>          .arch armv8-a
>          .file   "addTwoGlobal.c"
>          .text
>          .align  2
>          .global addTwo
>          .type   addTwo, %function
> addTwo:
>          adrp    x0, :got:x
>          ldr     x0, [x0, #:got_lo12:x]
>          ldr     w1, [x0]
>          adrp    x0, :got:y
>          ldr     x0, [x0, #:got_lo12:y]
>          ldr     w0, [x0]
>          add     w1, w1, w0
>          adrp    x0, :got:z
>          ldr     x0, [x0, #:got_lo12:z]
>          str     w1, [x0]
>          nop
>          ret
>          .size   addTwo, .-addTwo
>          .ident  "GCC: (Debian 10.2.1-6) 10.2.1 20210110"
>          .section        .note.GNU-stack,"",@progbits
> 
> 
> 
> I believe that the :got: and :got_lo12: modifiers tell the linker to use the address for x from the global offset table for the offset values in the
>          adrp    x0, :got:x
>          ldr     x0, [x0, #:got_lo12:x]
> 
> 
> Instructions.
> 
> Using gdb, I can see that adrp loads the offset from the pc to the page number of x and that #:got_lo12:x is an immediate value that is the offset of x within that page.
> 
> But I cannot find where :got: and :got_lo12: are documented.
> 
> 
> 
> 
> 

The relocation directives should really be documented in GAS (the assembler), but the text there is pretty short and not very comprehensive [1].  The relocation names are essentially taken from the ELF specification for AArch64
(https://github.com/ARM-software/abi-aa/blob/main/aaelf64/aaelf64.rst)
and the relocation markers used by GAS largely follow the relocation names described in that document.

R.

[1]
https://sourceware.org/binutils/docs-2.39/as.html#AArch64_002dRelocations


      reply	other threads:[~2022-11-23 16:38 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-22 19:12 Bob Plantz
2022-11-23 16:22 ` Richard Earnshaw
2022-11-23 16:38   ` Bob Plantz [this message]

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=BYAPR04MB4661B770BBA28F743AD4F08CD40C9@BYAPR04MB4661.namprd04.prod.outlook.com \
    --to=rgplantz@outlook.com \
    --cc=Richard.Earnshaw@foss.arm.com \
    --cc=gcc-help@gcc.gnu.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).