[Devel] Re: [PATCH] diskquota: 32bit quota tools on 64bit architectures

[Mikael Pettersson]

Andrew Morton writes:
 > On Mon, 18 Jun 2007 12:21:47 +0400
 > Vasily Tarasov <vtaras at openvz.org> wrote:
 > 
 > > From: Vasily Tarasov <vtaras at openvz.org>
 > > 
 > > This patch should be applied after Arnd Bergmann's patch,
 > > that intoduces new compat types: 
 > > http://lkml.org/lkml/2007/6/15/98
 > > 
 > > OpenVZ Linux kernel team has discovered the problem 
 > > with 32bit quota tools working on 64bit architectures.
 > > In 2.6.10 kernel sys32_quotactl() function was replaced by sys_quotactl() with
 > > the comment "sys_quotactl seems to be 32/64bit clean, enable it for 32bit"
 > > However this isn't right. Look at if_dqblk structure:
 > > 
 > > struct if_dqblk {
 > >         __u64 dqb_bhardlimit;
 > >         __u64 dqb_bsoftlimit;
 > >         __u64 dqb_curspace;
 > >         __u64 dqb_ihardlimit;
 > >         __u64 dqb_isoftlimit;
 > >         __u64 dqb_curinodes;
 > >         __u64 dqb_btime;
 > >         __u64 dqb_itime;
 > >         __u32 dqb_valid;
 > > };
 > > 
 > > For 32 bit quota tools sizeof(if_dqblk) == 0x44.
 > > But for 64 bit kernel its size is 0x48, 'cause of alignment!
 > > Thus we got a problem. Attached patch reintroduce sys32_quotactl() function,
 > > that handles this and related situations.
 > > 
 > > Signed-off-by: Vasily Tarasov <vtaras at openvz.org>
 > > 
 > > ---
 > > 
 > > In OpenVZ technology 32 bit Virtual Environments over 
 > > 64 bit OS are common, hence we have customers, that complains on this bad quota
 > > behaviour: 
 > > 
 > > # /usr/bin/quota
 > > quota: error while getting quota from /dev/sda1 for 0: Success
 > > 
 > > The reason is caused above.
 > 
 > Only x86_64 and ia64 are fixed.  Would it be correct to assume that the
 > other CONFIG_COMPAT architectures also need to be fixed?

I complained about this very issue when a previous version of
this patch was submitted last week, and Arnd explained that
non-x86 doesn't have a problem here because alignof(u64) is the
same in 32- and 64-bit modes.

However, the fact that the patch description talks about 32-
and 64-bit machines _in_general_, while the patch clearly only
handles x86-32 on x86-64 and ia64, is itself a bug. A more
precise patch description and a better comment in the code
is in order, I think.

/Mikael