Re: [v1] mm: access to uninitialized struct page
From: Pavel Tatashin
Date: Tue Apr 24 2018 - 16:01:08 EST
Hi Steven,
Thank you for your review:
>> https://lkml.org/lkml/2018/4/18/797
>
> #2, Do not use "lkml.org" it is a very unreliable source.
>
OK
> I'm fine with this change, but what happens if mm_init() traps?
>
> But that is probably not a case we really care about, as it is in the
> very early boot stage.
Yes, the assumption is that we do not trap in mm_init(), which I think
is the case because of early boot, and also I did not see this happen
during testing.
>
>>
>> ftrace_init();
>>
>
> One thing I could add is to move ftrace_init() before trap_init(). But
> that may require some work, because it may still depend on trap_init()
> as well. But making ftrace_init() not depend on trap_init() is easier
> than making it not depend on ftrace_init(). Although it may require
> more arch updates.
>
> I'm not saying that you should move it, it's something that can be
> added later after this change is implemented.
This makes, sense, but should be done outside of this bug fix.
>
> Reviewed-by: Steven Rostedt (VMware) <rostedt@xxxxxxxxxxx>
>
Thank you. I will send out an updated patch.
Pavel