linux-kernel - Re: Change of set_fs() in uaccess.h breaks this header being self contained for x86 (4.14-rc1)

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20170930183124.GF21978@ZenIV.linux.org.uk>
Date:   Sat, 30 Sep 2017 19:31:24 +0100
From:   Al Viro <viro@...IV.linux.org.uk>
To:     Stefan Mätje <Stefan.Maetje@....eu>
Cc:     "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: Change of set_fs() in uaccess.h breaks this header being self
 contained for x86 (4.14-rc1)

On Mon, Sep 18, 2017 at 08:32:19PM +0200, Stefan Mätje wrote:
> Hi,
> 
> there is a change in Linus' mainline kernel
> (https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/arch/x86/include/asm/uaccess.h?id=6c51e67b64d169419fb13318035bb442f9176612)
> that changes the set_fs() macro into an inline function.
> 
> This breaks the build of an out of kernel driver I maintain with this
> message:
> 
> ./arch/x86/include/asm/uaccess.h: In function ‘set_fs’:
> ./arch/x86/include/asm/uaccess.h:31:9: error: dereferencing pointer to
> incomplete type
> 
> set_fs() is referencing the "current" pointer which is of type "struct
> task_struct" declared in include/linux/sched.h.
> 
> Before the change set_fs() was a macro. Because I don't use it it was of
> no concern for me. But now the compiler needs to compile the inline
> function set_fs() and fails.
> 
> Should I add the needed <linux/sched.h> myself in my code or will that
> be changed / fixed in the <asm/uaccess.h> header by including the
> missing header.

Use linux/uaccess.h.  That's the only file that has any business pulling
asm/uaccess.h.

Again, any code that includes asm/uaccess.h directly is wrong and will be
hurt more and more as the time goes and more and more things get moved
from arch-dependent into generic parts.