Feedback-ID: i56a14606:Fastmail
User-Agent: Cyrus-JMAP/3.9.0-alpha0-156-g081acc5ed5-fm-20230206.001-g081acc5e
Mime-Version: 1.0
Message-Id: <17740a63-624f-47da-a701-dd218f95f9da@app.fastmail.com>
In-Reply-To: <Y+MXk9A+gB1W6T/n@1wt.eu>
References: <cover.1675734681.git.chenfeiyang@loongson.cn>
 <1af05c1441e9f96870be1cc20b1162e3f5043b2e.1675734681.git.chenfeiyang@loongson.cn>
 <1e7da4e7-392d-4a9e-aa95-0599a0c84419@app.fastmail.com>
 <CACWXhK==yVZGZbY+3DUAuxL34=gKwQv20Cw3y1+QKir0D2F3EQ@mail.gmail.com>
 <Y+MXk9A+gB1W6T/n@1wt.eu>
Date:   Wed, 08 Feb 2023 08:29:47 +0100
From:   "Arnd Bergmann" <arnd@arndb.de>
To:     "Willy Tarreau" <w@1wt.eu>,
        "chris.chenfeiyang" <chris.chenfeiyang@gmail.com>
Cc:     "Paul E. McKenney" <paulmck@kernel.org>,
        "Feiyang Chen" <chenfeiyang@loongson.cn>,
        "Huacai Chen" <chenhuacai@kernel.org>,
        "Jiaxun Yang" <jiaxun.yang@flygoat.com>, loongarch@lists.linux.dev,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH 1/3] nolibc: Add statx() support to implement sys_stat()
Content-Type: text/plain
Precedence: bulk

On Wed, Feb 8, 2023, at 04:31, Willy Tarreau wrote:
> On Wed, Feb 08, 2023 at 10:09:48AM +0800, Feiyang Chen wrote:
>> On Tue, 7 Feb 2023 at 22:31, Arnd Bergmann <arnd@arndb.de> wrote:
> (...)
>> > Given that all architectures implement statx the same way, I wonder
>> > if we can't just kill off the old function here and always use statx.
>> >
>> > That would also allow removing the architecture specific
>> > sys_stat_struct definitions in all arch-*.h files.
>> >
>> 
>> Hi, Arnd,
>> 
>> I'd really like to make all architectures use sys_statx() instead
>> of sys_stat(). I just fear we might get dragged into a long discussion.
>> Can I send a patch series to do this later?
>
> I generally agree with the Arnd's points overall and I'm fine with the
> rest of your series. On this specific point, I'm fine with your proposal,
> let's just start with sys_statx() only on this arch, please add a comment
> about this possibility in the commit message that brings statx(),
> indicating that other archs are likely to benefit from it as well, and
> let's see after this if we can migrate all archs to statx.

Ok, makes sense. Just change the description then to note that this
will also fix riscv32 (which has only statx) as well as arc, hexagon,
nios2 and openrisc (which have statx and stat64 but not stat or newstat).

   Arnd