On Fri, 30 Sep 2011 15:32:55 +1000 Stephen Rothwell [off-list ref] wrote:

Hi Andrew,

After merging the akpm tree, today's linux-next build (x86_64 allmodconfig)
failed like this:

In file included from arch/x86/include/asm/uaccess.h:575:0,
                 from include/net/checksum.h:25,
                 from include/linux/skbuff.h:28,
                 from include/linux/icmpv6.h:82,
                 from net/compat.c:19:
In function 'copy_from_user',
    inlined from 'compat_sys_socketcall' at net/compat.c:793:20:
arch/x86/include/asm/uaccess_64.h:64:26: error: call to 'copy_from_user_overflow' declared with attribute error: copy_from_user() buffer size is not provably correct

Caused by commit 15e19cbbbf2a ("x86: implement strict user copy checks
for x86_64") when built with gcc 4.6.0.  This does not fail when built
with 4.5.2.

This:

extern void copy_from_user_overflow(void)
#ifdef CONFIG_DEBUG_STRICT_USER_COPY_CHECKS
	__compiletime_error("copy_from_user() buffer size is not provably correct")
#else
	__compiletime_warning("copy_from_user() buffer size is not provably correct")
#endif

presumably CONFIG_DEBUG_STRICT_USER_COPY_CHECKS=y, so it's doing what
we asked it to do.

The problem here is that the length parameter to copy_from_user() is
obtained by indexing into an array of sizes.  Making the array const
does not help (obviously, since the index is not known at compile time
anyway).

Maybe I need a newer compiler.  For today I have gone back to my 4.5.2
compiler.

That patch is a PITA.  I've been waiting for some saviour to come along
and fix all the warnings it emits before proceeding with it.  As I am
apparently saviourless I shall hide that patch from the mm->linux-next
drop, so only I get to suffer its effects.