On Thu, Sep 16, 2010 at 06:05:59PM -0700, Henry Ptasinski wrote: > On Thu, Sep 16, 2010 at 01:35:15PM -0700, Jason wrote: > > Greg, > > > > Sorry, forgot to include the list. > > > > Please forgive the simple question, but I want to make sure I'm doing this right. In the adventure that has become typedef removal for brcm80211 I found this in or around drivers/staging/brcm80211/include/typedefs.h:73 > > > > #ifndef TYPEDEF_UINT32 > > typedef unsigned int uint32; > > #endif > > > > By itself, this should break 64bit builds, right? As I carefully sed out uint32, should I do their literal substitution, 'unsigned int', or fix it with 'unsigned long'/'uint32_t'? > > > > Up to this point, I've been strictly doing literal substitutions... > > > > Jason, > > Thanks for embarking on this adventure! > > When replacing explicitly-sized (i.e. uint32, int16) types, the substitution > should be with the native type of the same explicit size. I.e. uint32 should > be replaced with uint32_t, int16 should be replaced with int16_t. But if you note, the code above does not really do that for 64bit kernels, right? Are you sure this is correct? And never use the _t versions in the kernel please, use 'u32' instead, the _t types are for userspace programming only and make no sense in the kernel. thanks, greg k-h _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/devel
