From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <SRS0=41R+=7Q=deneb.enyo.de=fw@sourceware.org>
Received: from albireo.enyo.de (albireo.enyo.de [37.24.231.21])
	by sourceware.org (Postfix) with ESMTPS id 931303858CDA
	for <libc-alpha@sourceware.org>; Fri, 24 Mar 2023 10:45:54 +0000 (GMT)
DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 931303858CDA
Authentication-Results: sourceware.org; dmarc=none (p=none dis=none) header.from=deneb.enyo.de
Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=deneb.enyo.de
Received: from [172.17.203.2] (port=43409 helo=deneb.enyo.de)
	by albireo.enyo.de ([172.17.140.2]) with esmtps (TLS1.3:ECDHE_SECP256R1__RSA_PSS_RSAE_SHA256__AES_256_GCM:256)
	id 1pfevf-00EzPR-Iv; Fri, 24 Mar 2023 10:45:51 +0000
Received: from fw by deneb.enyo.de with local (Exim 4.96)
	(envelope-from <fw@deneb.enyo.de>)
	id 1pfevf-0012zt-1J;
	Fri, 24 Mar 2023 11:45:51 +0100
From: Florian Weimer <fw@deneb.enyo.de>
To: Xi Ruoyao <xry111@xry111.site>
Cc: Xi Ruoyao via Libc-alpha <libc-alpha@sourceware.org>,  caiyinyu
 <caiyinyu@loongson.cn>,  Wang Xuerui <i@xen0n.name>,  Adhemerval Zanella
 Netto <adhemerval.zanella@linaro.org>,  Andreas Schwab <schwab@suse.de>
Subject: Re: [PATCH] linux: Allow avoiding va_list for generic syscall and
 use it for LoongArch
References: <20230324072745.4138-1-xry111@xry111.site>
	<87mt424h0y.fsf@mid.deneb.enyo.de>
	<30790befdca73a5224b5a9a4491821e33cccc771.camel@xry111.site>
Date: Fri, 24 Mar 2023 11:45:51 +0100
In-Reply-To: <30790befdca73a5224b5a9a4491821e33cccc771.camel@xry111.site> (Xi
	Ruoyao's message of "Fri, 24 Mar 2023 18:36:32 +0800")
Message-ID: <87mt422zs0.fsf@mid.deneb.enyo.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable
X-Spam-Status: No, score=-5.8 required=5.0 tests=BAYES_00,KAM_DMARC_STATUS,SPF_HELO_NONE,SPF_PASS,TXREP autolearn=ham autolearn_force=no version=3.4.6
X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org
List-Id: <libc-alpha.sourceware.org>

* Xi Ruoyao:

> On Fri, 2023-03-24 at 10:47 +0100, Florian Weimer wrote:
>> * Xi Ruoyao via Libc-alpha:
>>=20
>> > Currently GCC generates highly sub-optimal code on architectures
>> > where
>> > the calling convention prefers registers for arugment passing.=A0 This
>> > is
>>=20
>> Typo: ar[gu]ment
>>=20
>> > LoongArch is benefited from this (saving about 430 CPU cycles per
>> > syscall, though I won't call it a significant improvement because
>> > syscall is "slow" in nature).=A0 And in the future we may switch more
>> > ports to use the generic syscall without a performance regression,
>> > reducing the number of target-specific syscall.{c,S} files we need
>> > to
>> > maintain.
>>=20
>> Does this impact the open* and fcntl* wrappers as well?
>
> Yes, on LoongArch all GARs are saved :(.  I'll see if it's possible to
> avoid using va_list for them too.

If we had overread the argument list in open*, that certainly would
have avoided all the grief we had with O_TMPFILE (where the original
glibc implementation did not pass the mode argument to the kernel).  I
think fcntl* already overreads the argument list, so it should be
safe.