Hi Jean, On Tuesday 21 October 2008 20:53:00 Jean Delvare wrote: [snip] > So I don't think that the implementation above is safe, unless the > entries in abituguru3_dmi_detect are sorted specifically to make the > short string comparison always correct. But that's easy to screw this > up later. > > An alternative would be to make a slightly more customized comparison > function: instead of passing strlen(dmi_name) as the length parameter > of strncmp(), you would compute the length by looking for the last > non-space character before the opening parenthesis in board_name. This > would require some more code, but would be more robust. Whether it's > worth it, I'll leave up to you: either do that or leave the code as it > is now. Or drop the patch. Dropping patch 3/3 also works (at least on the boards it was tested on) and the patch only plugs a theoretical problem, and ultimately might pose more problems than it solves, as you outlined. I don't think it's worth adding code which massages the length any further. For example, there's nothing stopping Abit from putting out a future board which uses parens in the model name too. I'd currently vote to drop the patch, and, if it is discovered that the problem we identified is more than theoretical, we can apply this patch as-is, or rework it. -- Cheers, Alistair.
