From: Fangrui Song <i@maskray.me>
To: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Cc: Jan Beulich <jbeulich@suse.com>, binutils@sourceware.org
Subject: Re: [PATCH 2/4] gas: Add support for LLVM addrsig and addrsig_sym directives on ELF.
Date: Wed, 25 May 2022 01:07:41 -0700 [thread overview]
Message-ID: <DS7PR12MB5765F60AACCF62A36852452ECBD69@DS7PR12MB5765.namprd12.prod.outlook.com> (raw)
In-Reply-To: <CANqewP3JH81H7YisJ_0zsfp0UMy9vd5=_tnyqz2s64c5wW6OxQ@mail.gmail.com>
On Wed, May 25, 2022 at 12:58 AM Tatsuyuki Ishi via Binutils
<binutils@sourceware.org> wrote:
>
> > Why two directives? IOW what's the point of silently ignoring
> > .addrsig_sym when there's no .addrsig anywhere?
>
> Because addrsig is an opt-in mechanism. A file with only .addrsig means that
> none of the symbols are address significant, while a file without is treated
> as if all symbols were.
>
> The silent ignore behavior is not ideal, but it could be added later. The
> primary target of this patch is compiler generated output.
The idea is that if no symbol is address significant, clang -S just
emits `.addrsig` with no `.addrsig_sym $sym`.
The .llvm_addrsig section is empty but it conveys information to
ld.lld --icf=safe.
In llvm-mc, `.addrsig_sym` accepts exactly one symbol. Technically, if
the directive were designed to support zero or any number of symbols,
`.addrsig` could be removed.
> > And such a toggle-on
> > directive likely would want to allow expressing also (or even only)
> > via a command line option.
>
> How would that work? That leads to two kind of end results, 1. mark *all*
> symbols as address-insignificant (if there's no addrsig_sym directives)
> or maybe 2. toggle off llvm_addrsig support (but why?).
next prev parent reply other threads:[~2022-05-25 8:07 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-05-25 6:42 [PATCH 0/4] gas: Add support for LLVM addrsig and addrsig_sym Tatsuyuki Ishi
2022-05-25 6:42 ` [PATCH 1/4] elf: Add definition for SHT_LLVM_ADDRSIG Tatsuyuki Ishi
2022-05-26 10:03 ` Jan Beulich
2022-05-26 10:13 ` Tatsuyuki Ishi
2022-05-26 10:19 ` Jan Beulich
2022-05-26 10:36 ` Tatsuyuki Ishi
2022-05-27 17:28 ` Peter Collingbourne
2022-05-25 6:42 ` [PATCH 2/4] gas: Add support for LLVM addrsig and addrsig_sym directives on ELF Tatsuyuki Ishi
2022-05-25 7:23 ` Alan Modra
2022-05-25 7:49 ` Jan Beulich
2022-05-25 7:58 ` Tatsuyuki Ishi
2022-05-25 8:07 ` Fangrui Song [this message]
2022-05-25 8:34 ` Jan Beulich
2022-05-25 8:41 ` Tatsuyuki Ishi
2022-05-25 6:42 ` [PATCH 3/4] bfd: Output SH_LINK to .symtab for SHT_LLVM_ADDRSIG Tatsuyuki Ishi
2022-05-25 7:46 ` Jan Beulich
2022-05-25 6:42 ` [PATCH 4/4] gas: Add basic test for addrsig Tatsuyuki Ishi
2022-05-25 7:30 ` Alan Modra
2022-05-25 7:19 ` [PATCH 0/4] gas: Add support for LLVM addrsig and addrsig_sym Fangrui Song
2022-05-25 7:34 ` Alan Modra
2022-05-25 7:59 ` Fangrui Song
2022-05-25 8:01 ` Tatsuyuki Ishi
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=DS7PR12MB5765F60AACCF62A36852452ECBD69@DS7PR12MB5765.namprd12.prod.outlook.com \
--to=i@maskray.me \
--cc=binutils@sourceware.org \
--cc=ishitatsuyuki@gmail.com \
--cc=jbeulich@suse.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).