Hi Arnd,

>>
>>If you need the barrier after the write, it probably was already faulty
>>before, because writel only implies a barrier before the store, not
>>after. Of course all the barriers likely made the whole process so
>>slow that you never hit that race in the end.
>
>ya, it could have worked in this way and i never saw a race issue before this.
>The only reason for changing this was to optimise out the additonal barriers
>that were happening. I do not see any issue now as well, only that the writes 
>would
>be faster.
>
 I reposted a patch here [1] with comments, i had to delete the
old accessors though as it became unused now.

http://www.spinics.net/lists/arm-kernel/msg505448.html

Regards,
 Sricharan

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu

Reply via email to