On 3/23/2018 9:22 PM, Benjamin Herrenschmidt wrote:

quoted

Yes, I'm %100 sure. The answer is both writel() and barrier() generates
a write barrier instruction. I found this by searching the kernel disassembly
for back to back "dsb st" instruction.

I'm not sure you are correct here. As I wrote below, the implementatoin
of writel() contains an *explicit" memory barrier which is completely
different to a barrier() instruction:

OK. I did some directed tests and I'm taking it back. 
barrier() is a compiler reordering statement only.

What got me confused was this sequence:

wmb()
barrier()
writel()

I thought that the second barrier instruction was coming from barrier() but
it was actually coming from writel().

-- 
Sinan Kaya
Qualcomm Datacenter Technologies, Inc. as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.