On 15.01.20 13:29, Carl-Daniel Hailfinger wrote:
VIA VL805 support, first draft. No idea if it actually works. It is highly likely that SPI accesses with a readcnt not being a multiple of 4 will return incorrect data due to the unknown encoding of the register containing SPI responses of the chip. That will be obvious from any verbose log, though, and once I have such data, adjusting the code is trivial.
Reverse engineered based on PCI traces created by cleverca22 on a Raspberry Pi 4 Model B.
RFC, with some unrelated (automatic programmer driver writer) and some related changes (buggy PCI patch reverted).
In case someone is feeling less brave, I also have a version which aborts early after the init sequence.
Signed-off-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net
My apologies for the whitespace-mangled patch. I checked that the patch was sent correctly to the mail server of my provider, but after the round trip it was mangled. No idea where that happened, but I'll try to debug the problem before I send the next version.
Regards, Carl-Daniel