[PATCH 5/6] ARM: re-implement physical address space switching

[santosh.shilimkar@oracle.com (santosh shilimkar)]

On 4/15/2015 5:07 AM, Mark Rutland wrote:
> Hi,
>
[..]

>>>>> +ENTRY(lpae_pgtables_remap_asm)
>>>>> +       stmfd   sp!, {r4-r8, lr}
>>>>> +
>>>>> +       mrc     p15, 0, r8, c1, c0, 0           @ read control reg
>>>>> +       bic     ip, r8, #CR_M                   @ disable caches and MMU
>>>>> +       mcr     p15, 0, ip, c1, c0, 0
>>>>> +       dsb
>>>>> +       isb
>>>>
>>>> Shouldn't the DSB be between the STMFD and the MCR (given the SP doesn't
>>>> point to an idmap/physical address)?
>>>>
>>>> I don't see why we need a DSB after the write to the SCTLR.
>>>>
>> dsb can be moved up after stmfd but leaving as above should be fine
>> as well.
>
> I don't think that it's safe to leave it where it is. Currently the
> STMFD could be reordered w.r.t. the cp15 accesses, and hence the write
> may occur with translation disabled (and would go to the wrong physical
> address).
>
> We need to ensure that the STMFD is executed before the MCR potentially
> changes the execution context. The DSB will ensure that in addition to
> ensuring completion of the write (i.e. it isn't left sat in a write
> buffer).
>
I see your point. Thanks for expanding it.

Regards,
Santosh