Re: [PATCH 4/6] improve sys_personality for compat architectures

["H. Peter Anvin" <hpa@zytor.com>]

On 02/03/2010 09:13 AM, David Miller wrote:
> From: Arnd Bergmann <arnd@arndb.de>
> Date: Wed, 3 Feb 2010 18:06:27 +0100
> 
>>> But if the consensus is that we should fix this properly I can
>>> replace the patch with one introducing a compat_sys_personality
>>> which only gets used for compat tasks.
>>
>> Right now, sparc64 and powerpc64 use sys32_personality for both native
>> and compat tasks, x86 never uses it and all others use it only for
>> compat tasks. That seems more sensible if we keep this function at
>> all.
> 
> If it only gets used for compat tasks, you can only switch in
> one direction.  I think it needs to be handled for both compat
> and non-compat tasks, in order to allow for that.

That seems odd... with the wrapper you can ever only switch in one
direction, and if you use it for non-compat tasks you wouldn't be ever
to switch back, period.

> That's why powerpc64 and sparc64 do things the way they do,
> I am pretty sure.

As far as I can tell, the ppc64 and sparc64 implementations would seem
to be trapdoors from which no return is possible.  That's a pretty
defensible position in some ways -- it mimics the 32-bit machine even
down to the personality() syscall -- but it definitely has disadvantages.

The x86 method of simply not bothering doesn't seem to have caused
problems -- our compat (and noncompat) tasks happily return PER_LINUX32
if that is the mode and we don't seem to have had complaints with it.
If userspace ever had an issue with it -- and they might have, at one
point in history libc used to call personality() during startup, which
it doesn't seem to anymore -- they presumably have worked through it.

As such, I'm more than a little reluctant to change the current behavior.

	-hpa