From: "Chen, Kenneth W" <kenneth.w.chen@intel.com>
To: "'Pete Zaitcev'" <zaitcev@redhat.com>,
       "Matt Helsley" <matthltc@us.ibm.com>
Cc: "Erik Jacobson" <erikj@sgi.com>, <guillaume.thouvenin@bull.net>,
       <linux-kernel@vger.kernel.org>
Subject: RE: [PATCH] connector: Some fixes for ia64 unaligned access errors
Date: Mon, 11 Dec 2006 18:07:08 -0800
Message-ID: <000001c71d92$3aa76c70$e434030a@amr.corp.intel.com>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: 7bit
Thread-Index: Accdjo4zfo9eXwARTN+j6Pv5mjRxNAAAl8BQ
In-Reply-To: <20061211172907.305473cf.zaitcev@redhat.com>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1586
Lines: 38

Pete Zaitcev wrote on Monday, December 11, 2006 5:29 PM
> On Mon, 11 Dec 2006 15:52:47 -0800, Matt Helsley <matthltc@us.ibm.com> wrote:
> 
> > 	I'm shocked memcpy() introduces 8-byte stores that violate architecture
> > alignment rules. Is there any chance this a bug in ia64's memcpy()
> > implementation? I've tried to read it but since I'm not familiar with
> > ia64 asm I can't make out significant parts of it in
> > arch/ia64/lib/memcpy.S.
> 
> The arch/ia64/lib/memcpy.S is probably fine, it must be gcc doing
> an inline substitution of a well-known function.

arch/ia64/lib/memcpy.S is fine because it does alignment check at the very
beginning of the function and depends on the alignment of dst/src alignment,
it does different thing.  The unaligned access is coming from gcc inlined
version of memcpy.

But looking deeply, memory allocation for proc_event in proc_for_connector
doesn't looked correct at all:

In drivers/connector/cn_proc.c:
#define CN_PROC_MSG_SIZE (sizeof(struct cn_msg) + sizeof(struct proc_event))

void proc_fork_connector(struct task_struct *task)
{
        struct cn_msg *msg;
        struct proc_event *ev;
        __u8 buffer[CN_PROC_MSG_SIZE];

You can't do that because gcc assumes struct proc_event aligns on certain
boundary.  Doing fancy hand crafting like that breaks code generated by gcc.

- Ken
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/