Re: [PATCH 0/6] mempolicy memory corruption fixlet

[KOSAKI Motohiro]

>>>>>> Yes, that's right direction, I think. Currently, shmem_set_policy()
>>>>>> can't handle
>>>>>> nonlinear mapping.
>>>>>
>>>>> I've been mulling for some time to just remove non linear mappings.
>>>>> AFAIK they were only useful on 32bit and are obsolete and could be
>>>>> emulated with VMAs instead.
>>>>
>>>>
>>>> I agree. It is only userful on 32bit and current enterprise users don't
>>>> use
>>>> 32bit anymore. So, I don't think emulated by vmas cause user visible
>>>> issue.
>>>
>>>
>>> I wish this was true, there are a lot of systems out there still running
>>> 32
>>> bit linux, even on 64 bit capible hardware. This is especially true in
>>> enterprises where they have either homegrown or proprietary software that
>>> isn't 64 bit clean.
>>
>> 32 bit binaries (and entire distros) run fine under 64 bit kernels.
>
> unfortunantly, not quite 100% of the time. It's very good, but the automount
> bug a month or so ago is an example of how you can run into rare problems.
> Many "enterprise" systems are not willing to risk it.

Then we can remove a feature safely. Risk not taker continue to uses old distro
kernel. And some years after, distros discontinue to support 32bit. And then,
problem will vanish automatically.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/