Carl-Daniel Hailfinger wrote:
All unknown SPI chips claim to have status UNTESTED for probe/read/erase/write. That's incorrect. Since the chips are unknown, read/erase/write are unavailable for them. And if probe worked, they wouldn't have needed the generic vendor match in the first place. Mark those chips as BAD for probe/read/erase/write.
That change revealed another bug: If a chip has any TEST_BAD_* flag set, we don't even list the unsupported functions, giving the user the impression that the unsupported functions are tested. Fix that bug as well.
Signed-off-by: Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net
Thanks to Uwe for testing and finding a bug.
Nak.
Unknown chips can not be tested and known to be bad, which is what TEST_BAD_ means.
If anything, please remove the unknown chips.
The TEST_BAD_ logic improvement is good, but I would like it to explicitly test each flag instead of assuming that TEST_OK_ and TEST_BAD_ flags will overlap. That may not always be the case. Also the _SHIFT define isn't so nice imo. Fix those and that part is:
Acked-by: Peter Stuge peter@stuge.se