public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: "Michal Jankovič" <michal.jankovic59@gmail.com>
To: Iain Sandoe <iain@sandoe.co.uk>
Cc: GCC Patches <gcc-patches@gcc.gnu.org>
Subject: Re: [PATCH] c++: coroutines - Overlap variables in frame [PR105989]
Date: Tue, 12 Jul 2022 17:14:36 +0200	[thread overview]
Message-ID: <0334F8DB-C49F-4EB2-98C5-91A44A851F70@getmailspring.com> (raw)
In-Reply-To: <236E416B-F744-42C1-806C-4701DB1BAB55@sandoe.co.uk>

Hi Iain,

Thanks for the reply, this is my first time contributing and I am
looking forward to your input.

One other related thing I would like to investigate is reducing the
number of compiler generated variables in the frame, particularly
_Coro_destroy_fn and _Coro_self_handle.  

As I understand it, _Coro_destroy_fn just sets a flag in
_Coro_resume_index and calls _Coro_resume_fn; it should be possible to
move this logic to __builtin_coro_destroy, so that only _Coro_resume_fn
is stored in the frame; this would however change the coroutine ABI - I
don't know if that's a problem.

The _Coro_self_handle should be constructible on-demand from the frame address.

Do you have any advice / opinions on this before I try to implement it?

Michal

On Jul 12 2022, at 4:08 pm, Iain Sandoe <iain@sandoe.co.uk> wrote:

> Hi Michal,
>  
>> On 12 Jul 2022, at 14:35, Michal Jankovič via Gcc-patches
>> <gcc-patches@gcc.gnu.org> wrote:
>>  
>> Currently, coroutine frames store all variables of a coroutine separately,
>> even if their lifetime does not overlap (they are in distinct
>> scopes). This
>> patch implements overlapping distinct variable scopes in the
>> coroutine frame,
>> by storing the frame fields in nested unions of structs. This lowers
>> the size
>> of the frame for larger coroutines significantly, and makes them more usable
>> on systems with limited memory.
>  
> not a review (I will try to take a look at the weekend).
>  
> but … this is one of the two main optimisations on my TODO - so cool
> for doing it.
>  
> (the other related optimisation is to eliminate frame entries for
> scopes without any suspend
> points - which has the potential to save even more space for code with
> sparse use of co_xxxx)
>  
> Iain
>  
>> Bootstrapped and regression tested on x86_64-pc-linux-gnu; new test fails
>> before the patch and succeeds after with no regressions.
>>  
>> 	PR c++/105989
>>  
>> gcc/cp/ChangeLog:
>>  
>> 	* coroutines.cc (struct local_var_info): Add field_access_path.
>> 	(build_local_var_frame_access_expr): New.
>> 	(transform_local_var_uses): Use build_local_var_frame_access_expr.
>> 	(coro_make_frame_entry_id): New.
>> 	(coro_make_frame_entry): Delegate to coro_make_frame_entry_id.
>> 	(struct local_vars_frame_data): Add orig, field_access_path.
>> 	(register_local_var_uses): Generate new frame layout. Create access
>> 	paths to vars.
>> 	(morph_fn_to_coro): Set new fields in local_vars_frame_data.  
>>  
>> gcc/testsuite/ChangeLog:
>>  
>> 	* g++.dg/coroutines/pr105989.C: New test.
>>  
>> <pr105989.patch>
>  
>

  reply	other threads:[~2022-07-12 15:14 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-07-12 13:35 Michal Jankovič
2022-07-12 14:08 ` Iain Sandoe
2022-07-12 15:14   ` Michal Jankovič [this message]
2022-07-12 17:11     ` Iain Sandoe
2022-07-13 12:54       ` Michal Jankovic
2023-05-14 15:36         ` Michal Jankovič
2023-05-14 16:07           ` Iain Sandoe
2023-05-14 16:31             ` Michal Jankovič
2023-05-14 18:51               ` Iain Sandoe

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=0334F8DB-C49F-4EB2-98C5-91A44A851F70@getmailspring.com \
    --to=michal.jankovic59@gmail.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=iain@sandoe.co.uk \
    /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).