netdev
[Top] [All Lists]

Re: [PATCH 2/3] add open iscsi netlink interface to iscsi transport clas

To: herbert@xxxxxxxxxxxxxxxxxxx
Subject: Re: [PATCH 2/3] add open iscsi netlink interface to iscsi transport class
From: YOSHIFUJI Hideaki / 吉藤英明 <yoshfuji@xxxxxxxxxxxxxx>
Date: Thu, 05 May 2005 20:55:30 +0900 (JST)
Cc: michaelc@xxxxxxxxxxx, linux-scsi@xxxxxxxxxxxxxxx, netdev@xxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, yoshfuji@xxxxxxxxxxxxxx
In-reply-to: <E1DTehn-0005gJ-00@gondolin.me.apana.org.au>
Organization: USAGI Project
References: <20050505.185503.78606493.yoshfuji@linux-ipv6.org> <E1DTehn-0005gJ-00@gondolin.me.apana.org.au>
Sender: netdev-bounce@xxxxxxxxxxx
In article <E1DTehn-0005gJ-00@xxxxxxxxxxxxxxxxxxxxxxxx> (at Thu, 05 May 2005 
21:39:47 +1000), Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> says:

> YOSHIFUJI Hideaki / ???? <yoshfuji@xxxxxxxxxxxxxx> wrote:
> > In article <42798AC1.2010308@xxxxxxxxxxx> (at Wed, 04 May 2005 19:53:53 
> > -0700), Mike Christie <michaelc@xxxxxxxxxxx> says:
> > 
> >> +struct iscsi_uevent {
> > :
> >> +} __attribute__ ((aligned (sizeof(unsigned long))));
> > 
> > I think it'd be better to use sizeof(uint64_t) or something.
> 
> Is there a benefit in aligning on 64-bit boundaries for 32-bit platforms?

Well, this if for avoiding inconsistency between kernel and userspace.
If you use unsigned long, sizeof(unsigned long) may be inconssitent
between kernel and userland like this:
  kernel:    sizeof(unsigned long) == 8
  userspace: sizeof(unsigned long) == 4

--yoshfuji

<Prev in Thread] Current Thread [Next in Thread>