Benjamin Herrenschmidt <b...@kernel.crashing.org> writes:
> On Wed, 2014-07-23 at 00:23 +0530, Aneesh Kumar K.V wrote:
>> > A better place for this would be right before the last write to the PMD
>> > (that's also clearing BUSY) in __hash_page_thp(). Basically, it's the
>> > normal lock ordering that's missing here, nothing specific to
>> > mark_hpte_slot_valid() but instead, any state relative to the BUSY bit
>> > in the PMD (including the actual hash writes in update_pp etc...)
>> >
>>
>> IIUC updatepp already have required barriers. ie in updatepp we do tlbie
>> which should take care of the ordering right ?
>
> Only if it succeeds but that doesn't matter, I'd rather we get the
> semantics right. The clearing of the busy bit is an unlock, it should
> have the appropriate barriers like it does in other variants of hash
> page.
ok
>>
>> Now the reason i moved that spm_wmb() to mark_hpte_slot_valid was to
>> pair it with smb_rmb() in get_hpte_slot_array().
>
> Which is also probably in the wrong place. Care to explain to me the
> exact relationship ?
We want to make sure for usage like below we don't reorder the load.
if (pmd_trans_huge(*pmdp)){
get_hpte_slot_array(pmdp)
}
-aneesh
_______________________________________________
Linuxppc-dev mailing list
Linuxppc-dev@lists.ozlabs.org
https://lists.ozlabs.org/listinfo/linuxppc-dev
