Side note:
Am 19.12.2016 um 16:22 schrieb Icenowy Zheng:
[...]
According to some facts:
- The register based access to SID is weird: it needs ~5 register operations per word of SID.
[...]
My experiments seem to indicate that Allwinner's implementation might be overly complicated. I have used an alternative approach and did not notice any issues so far.
They start by reading SID_PRCTL and use bit mask operations to get in the key index value (PG_INDEX), then use a second write to set OP_LOCK and READ_START. As far as I can tell, it's sufficient to construct (and write) a single value from these three components. Allwinner finally clears out the used bits by another mask operation, simply writing zero seems to get that job done equally well.
My implementation can be seen in https://github.com/n1tehawk/sunxi-tools/blob/20161220_sid-fix/uart0-hellowor... or as assembler code in https://github.com/n1tehawk/sunxi-tools/blob/20161220_sid-fix/sid_read_root....
A corresponding PR is at https://github.com/linux-sunxi/sunxi-tools/pull/91 , or you may clone my branch directly ("git clone https://github.com/n1tehawk/sunxi-tools.git -b 20161220_sid-fix").
Testers would be very welcome.
Regards, B. Nortmann