Re: [PATCH] arm: fix page faults in do_alignment

From: Jing Xiangfeng
Date: Sat Aug 31 2019 - 05:17:07 EST

On 2019/8/31 15:55, Russell King - ARM Linux admin wrote:
> On Sat, Aug 31, 2019 at 09:49:45AM +0800, Jing Xiangfeng wrote:
>> On 2019/8/30 21:35, Russell King - ARM Linux admin wrote:
>>> On Fri, Aug 30, 2019 at 09:31:17PM +0800, Jing Xiangfeng wrote:
>>>> The function do_alignment can handle misaligned address for user and
>>>> kernel space. If it is a userspace access, do_alignment may fail on
>>>> a low-memory situation, because page faults are disabled in
>>>> probe_kernel_address.
>>>> Fix this by using __copy_from_user stead of probe_kernel_address.
>>>> Fixes: b255188 ("ARM: fix scheduling while atomic warning in alignment handling code")
>>>> Signed-off-by: Jing Xiangfeng <jingxiangfeng@xxxxxxxxxx>
>>> NAK.
>>> The "scheduling while atomic warning in alignment handling code" is
>>> caused by fixing up the page fault while trying to handle the
>>> mis-alignment fault generated from an instruction in atomic context.
>> __might_sleep is called in the function __get_user which lead to that bug.
>> And that bug is triggered in a kernel space. Page fault can not be generated.
>> Right?
> Your email is now fixed?

Yeah, I just checked the mailbox, it is normal now.

> All of get_user(), __get_user(), copy_from_user() and __copy_from_user()
> _can_ cause a page fault, which might need to fetch the page from disk.
> All these four functions are equivalent as far as that goes - and indeed
> as are their versions that write as well.
> If the page needs to come from disk, all of these functions _will_
> sleep. If they are called from an atomic context, and the page fault
> handler needs to fetch data from disk, they will attempt to sleep,
> which will issue a warning.
I understand.