From: Alexander Monakov <amonakov@ispras.ru>
To: Martin Jambor <mjambor@suse.cz>
Cc: Artem Klimov <jakmobius@gmail.com>, Jan Hubicka <hubicka@ucw.cz>,
gcc-patches@gcc.gnu.org
Subject: Re: [PATCH] ipa-visibility: Optimize TLS access [PR99619]
Date: Mon, 2 May 2022 19:43:39 +0300 (MSK) [thread overview]
Message-ID: <381c469c-516e-808f-f811-314eddaf2ba9@ispras.ru> (raw)
In-Reply-To: <ri6zgjzj4z8.fsf@suse.cz>
On Mon, 2 May 2022, Martin Jambor wrote:
> > Co-Authored-By: Alexander Monakov <amonakov@gcc.gnu.org>
>
> (Minor nit and I don't care too much, but in GCC we traditionally
> specify co-authors in the ChangeLog entry beginning by providing more
> names, one per line. But perhaps we want to adapt more widely used
> practices.)
I believe this is the recommended way to specify co-authors now (after
Git migration) according to https://gcc.gnu.org/codingconventions.html
(patch discussion below)
> > --- a/gcc/ipa-visibility.cc
> > +++ b/gcc/ipa-visibility.cc
> > @@ -872,6 +872,22 @@ function_and_variable_visibility (bool whole_program)
> > }
> > }
> > }
> > + FOR_EACH_VARIABLE (vnode)
> > + {
> > + tree decl = vnode->decl;
> > +
> > + /* Optimize TLS model based on visibility (taking into account
> > + optimizations done in the preceding loop), unless it was
> > + specified explicitly. */
> > +
> > + if (DECL_THREAD_LOCAL_P (decl)
> > + && !lookup_attribute ("tls_model", DECL_ATTRIBUTES (decl)))
> > + {
> > + enum tls_model new_model = decl_default_tls_model (decl);
> > + gcc_checking_assert (new_model >= decl_tls_model (decl));
> > + set_decl_tls_model (decl, new_model);
> > + }
> > + }
> >
>
> decl_default_tls_model depends on the global optimize flag, which is
> almost always problematic in IPA passes. I was able to make your patch
> ICE using the vis-attr-hidden.c testcase from your patch with:
>
> mjambor@virgil:~/gcc/small/tests/tls$ ~/gcc/small/inst/bin/gcc -O2 -fPIC -flto -c vis-attr-hidden.c
> mjambor@virgil:~/gcc/small/tests/tls$ ~/gcc/small/inst/bin/gcc -fPIC -O0 -shared -flto vis-attr-hidden.o
> during IPA pass: whole-program
> lto1: internal compiler error: in function_and_variable_visibility, at ipa-visibility.cc:888
[snip]
> Note the use of LTO, mismatching -O flags and the -shared flag in the
> link step.
Ah, right. The assert is checking that we don't accidentally downgrade decl's
TLS access model, e.g. from local-dynamic to global-dynamic, and you've shown
how to trigger that. I didn't realize this code can run twice, and with
different 'optimize' levels.
I would suggest to solve this by checking if the new TLS model is stronger,
i.e. instead of this:
gcc_checking_assert (new_model >= decl_tls_model (decl));
set_decl_tls_model (decl, new_model);
do this:
if (new_model >= decl_tls_model (decl))
set_decl_tls_model (decl, new_model);
Does this look reasonable?
> A simple but somewhat lame way to avoid the ICE would be to run your
> loop over variables only from pass_ipa_function_and_variable_visibility
> and not from pass_ipa_whole_program_visibility.
>
> I am afraid a real solution would involve copying relevant entries from
> global_options to the symtab node representing the variable when it is
> created/finalized, properly streaming them for LTO, and modifying
> decl_default_tls_model to rely on those rather than global_options
> itself.
If we agree on the solution above, then this will not be necessary, after all
this transformation looks at optimized whole-program visibility status,
and so initial optimization level should not be relevant.
> But maybe Honza has some other clever idea.
>
> Also, please be careful not to unnecessarily commit trailing blank
> spaces, the empty lines in your patch are not really empty.
Yep, I can take care of whitespace issues.
Thank you!
Alexander
next prev parent reply other threads:[~2022-05-02 16:43 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-17 18:51 Artem Klimov
2022-05-02 8:51 ` Alexander Monakov
2022-05-02 16:04 ` Martin Jambor
2022-05-02 16:43 ` Alexander Monakov [this message]
2022-05-09 16:06 ` Alexander Monakov
2022-05-09 16:47 ` Jan Hubicka
2022-05-09 17:15 ` Alexander Monakov
2022-05-16 15:50 ` Alexander Monakov
2022-05-23 10:56 ` Alexander Monakov
2022-05-25 9:04 ` Jan Hubicka
2022-07-07 15:53 ` [PATCH v2] " Alexander Monakov
2022-07-20 13:04 ` Alexander Monakov
2022-08-05 14:03 ` Alexander Monakov
2022-08-23 15:27 ` Alexander Monakov
2022-08-26 11:32 ` Martin Jambor
2022-08-26 13:35 ` Alexander Monakov
2022-08-30 11:44 ` Martin Jambor
2022-08-30 13:19 ` Alexander Monakov
2022-08-30 14:03 ` Alexander Monakov
2022-09-05 10:39 ` Martin Jambor
2022-05-02 19:28 ` [PATCH] " Martin Liška
2022-05-05 10:50 ` Jan Hubicka
2022-05-05 11:50 ` Alexander Monakov
2022-05-05 11:56 ` Jan Hubicka
2022-05-05 14:41 ` Alexander Monakov
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=381c469c-516e-808f-f811-314eddaf2ba9@ispras.ru \
--to=amonakov@ispras.ru \
--cc=gcc-patches@gcc.gnu.org \
--cc=hubicka@ucw.cz \
--cc=jakmobius@gmail.com \
--cc=mjambor@suse.cz \
/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).