On Thu, 08 Jan 2009 20:21:35 +0100, Hans de Goede wrote: > Alistair John Strachan wrote: > > I think the consensus we came to at the time was to switch to strncmp'ing a > > subset of the DMI string, but in this case we'll need strncasecmp. This isn't > > too terrible as false positives are pretty unlikely (we already check the > > board manufacturer too, which does seem to be unchanging). Hans? > > That was my idea too, just make the dmi string stored in the driver: "IP35 Pro" > and do a strncasecmp for the length of the string in the driver, with the one > from the BIOS, if it is a match assume we have found an abituguru3 equipped > motherboard. That's not enough, because some board names match the beginning of other supported board names (for example AW8 and AW8D.) So the algorithm must really compare both strings on a strip-everything-after-parenthesis-and-every-trailing-white-space basis. That's a little more code, but still not rocket science, and that's the only safe way. (Actually, no, there's a second way: put the abituguru3_motherboards[] entries in a clever order. But that could break easily, so I wouldn't recommend it.) -- Jean Delvare
