On Wed, 01 Dec 2010 17:28:28 -0600 Timur Tabi timur@freescale.com wrote:
Scott Wood wrote:
You told me that since I'm doing a read following a write to uncached memory,
that I don't need a sync.
No, I was talking about a write followed by a read, to the same location.
That's what I said. Read following write == write followed by read.
Sorry, misread.
The sync I was concerned about wasn't between that write and the following read, but between the read and whatever comes after the read.
out_8 should be fixed to behave like the other accessors.
Ok, but I'm not using any of the I/O accessors, so this doesn't affect me.
Yes you are, in set_mux_to_diu(). But it's actually setbits_8(), which will do an in_8() first, which has synchronization, so it should be OK.
I just need to make sure that the read is executed after the write, and that the read completes before I continue.
Right. It was that last bit I was talking about.
So I should I put an isync between the write and the read,
Not necessary since they're the same address, and wouldn't help if they weren't (you'd want sync or mbar 1 in that case, not isync).
and a sync after the read?
If you were to immediately follow it with out_8 as currently defined, yes. But setbits_8 should be OK.
-Scott