From: "Vladimir 'φ-coder/phcoder' Serbinenko" <phcoder@gmail.com>
To: The development of GNU GRUB <grub-devel@gnu.org>
Subject: Re: NEED_REGISTER_FRAME_INFO can be replaced by -static-libgcc
Date: Sat, 17 Apr 2010 22:52:11 +0200 [thread overview]
Message-ID: <4BCA1F7B.7080601@gmail.com> (raw)
In-Reply-To: <4BC59A74.5080900@t-online.de>
[-- Attachment #1: Type: text/plain, Size: 2128 bytes --]
Christian Franke wrote:
> Vladimir 'φ-coder/phcoder' Serbinenko wrote:
>> Christian Franke wrote:
>>
>>> The *_frame_info symbols are set undefined to force linkage of the
>>> libgcc_s shared library or dll.
>>>
>>> This can be prevented by TARGET_LDFLAGS=-static-libgcc. To build from
>>> grub-1.98 tarball on Cygwin, run configure with this argument.
>>>
>>> May also be necessary on other build platforms when -shared-libgcc is
>>> the default. It may be possible simply set -static-libgcc
>>> unconditionally.
>>>
>>>
>> We already supply -lgcc --static-libgcc. Do you know why it still had
>> issues?
>>
>
> The above is only set when linking the kernel image.
>
> According to 'gcc -dumpspecs' of Cygwin gcc 4.3.4-3, options '-u
> ___[de]register_frame_info' are always passed to the linker if
> -static-libgcc is not specified. This is Cygwin and MinGW specific: In
> the exe startup code these symbols are loaded only if present, so the
> DLL must be forced to load first.
>
> As a consequence, the symbols are also set undefined when the GRUB
> pre-*.o files are generated with 'ld -r'.
>
> The attached patch works for me with the bzr revision preceding the
> NEED_REGISTER_FRAME_INFO fix.
>
Doesn't it risk linking e.g. __bswapsi2 into every module which uses it
during partial link? It would increase module size on RISC
>
>> And I also doubt usefullness of pulling these functions since reference
>> to them is purely dummy: no relocation uses it so it will only increase
>> code size. Another concern is the behviour of these functions in grub
>> environment if they ever get called.
>>
>>
>
> AFAICS from the gcc source, the functions are used in EH stack
> unwinding code in libgcc. The compiler itself does not generate calls
> to these functions.
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Grub-devel mailing list
> Grub-devel@gnu.org
> http://lists.gnu.org/mailman/listinfo/grub-devel
--
Regards
Vladimir 'φ-coder/phcoder' Serbinenko
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 293 bytes --]
next prev parent reply other threads:[~2010-04-17 20:52 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-04-13 19:10 NEED_REGISTER_FRAME_INFO can be replaced by -static-libgcc Christian Franke
2010-04-13 19:24 ` Vladimir 'φ-coder/phcoder' Serbinenko
2010-04-14 10:35 ` Christian Franke
2010-04-17 20:52 ` Vladimir 'φ-coder/phcoder' Serbinenko [this message]
2010-04-19 15:13 ` Christian Franke
2010-04-24 21:19 ` Vladimir 'φ-coder/phcoder' Serbinenko
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=4BCA1F7B.7080601@gmail.com \
--to=phcoder@gmail.com \
--cc=grub-devel@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