Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Thu, 1 Oct 2020 07:47:31 +0200
From:   Petr Vorel <petr.vorel@gmail.com>
To:     Rich Felker <dalias@libc.org>
Cc:     linux-kernel@vger.kernel.org, musl@lists.openwall.com
Subject: Re: [musl] [PATCH 1/1] linux/sysinfo.h: Add guarder for struct
 sysinfo
Message-ID: <20201001054731.GA193908@x230>
Reply-To: Petr Vorel <petr.vorel@gmail.com>
References: <20200930214636.186132-1-petr.vorel@gmail.com>
 <20200930234212.GK17637@brightrain.aerifal.cx>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200930234212.GK17637@brightrain.aerifal.cx>
Precedence: bulk

Hi Rich,

...
> I don't think this is the right way to do it. It prevents getting
> access to the kernel uapi structure (which may be wanted) if you're
> not using glibc or if you include kernel headers before any libc
> headers. Rather, <linux/kernel.h>, whose only real purpose is
> providing this structure to legacy applications that don't know the
> renamed name for it, should not be implicitly included by other
> headers. There's an existing thread on the topic but I don't have the
> link handy. IIRC I proposed moving the alignment macros or whatever
> other useful stuff is in <linux/kernel.h> to a separate header and
> getting rid of all the indirect inclusions of <linux/kernel.h>.

Thanks for a review and tip!

Kind regards,
Petr