Re: [PATCH 1 of 3] add phys_addr_t for holding physical addresses

[Jeremy Fitzhardinge <jeremy@goop.org>]

Andrew Morton wrote:
> On Fri, 22 Aug 2008 14:11:16 -0700
> Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>
>   
>> Andrew Morton wrote:
>>     
>>> On Tue, 19 Aug 2008 13:02:50 -0700
>>> Jeremy Fitzhardinge <jeremy@goop.org> wrote:
>>>
>>>   
>>>       
>>>> Add a kernel-wide "phys_addr_t" which is guaranteed to be able to hold
>>>> any physical address.  By default it equals the word size of the
>>>> architecture, but a 32-bit architecture can set ARCH_PHYS_ADDR_T_64BIT
>>>> if it needs a 64-bit phys_addr_t.
>>>>
>>>>     
>>>>         
>>> You say this is a bugfix but you don't describe the bug.  This makes it
>>> rather hard to make the 2.6.2[5678] decisions.
>>>
>>> Ditto on [patch 2/3].
>>>   
>>>       
>> 1/3 is not a bugfix in itself, but a pre-requisite for 2/3.
>>
>> 2/3 replaces an ad-hoc Xen fix with a general fix to prevent address
>> truncation when using PFN_PHYS() on any PFN above the 4G mark.  The Xen
>> crash is the only bug I know of that's directly attributable to this,
>> and it was already addressed in older kernels with the casts in the Xen
>> code that this patch removes.
>>
>> So I don't think there's any strong need to push this to earlier kernels.
>>
>>     
>
> Still confused.  The above implies that 2.6.27 doesn't need fixing
> either, because the typecasts already avoid the crash.
>   

Yes, that's true.  It's a bit more of a comprehensive and correct fix; 
I think it's fairly low risk at this point in the -rc series, but it
could be deferred (or just defer 3/3, which really is cosmetic).

    J