On Wed, 14 Jan 2015 23:52:42 +0100 (CET) Jernej Škrabec jernej.skrabec@planet.si wrote:
Hello Stefan,
here you have report without -c parameter. I also removed my local changes, so it should be correct.
Thanks!
You once said, that maybe another probe function should be written for this chip. Can you elaborate a bit? I take a look at datasheet and found out that actually almost every field in chip definition structure can be filled from SFDP data. Is this a way to go?
I have been working on a complete probing rewrite. Some vendors share the 16b model RDID IDs between incompatible models, but provide other means to distinguish them. In the worst case we would have to issue a second probe opcode and derive the model from the second reply. Or we would have to look at the SFDP responses... (I don't remember what the solution for this series of Spansion chips is from the top of my head, but something similar). Relying completely on SFDP is a bad idea. While it is standardized in a way that flashrom can somewhat work with it (we do use it, if the chip is unknown but supports SFDP), the data in there is often not completely correct... and sometimes even completely bogus. Also, not too many chips support it.
So my plan for this model and all others is to probe not a single function but a number of them and compare the results. This is also important for systems where some opcodes are blocked by the system (e.g. many Intel-based boards).