On Sat, 2008-03-08 at 07:57 +0100, Johannes Berg wrote: > On Fri, 2008-03-07 at 19:17 -0800, Harvey Harrison wrote: > > Also remove Hi16/Lo16 and open code them in the one place they > > are used. > > > int i, j; > > + __le16 *ptr = (__le16 *)ta; /* address is held in LE byteorder */ > > + > > + p1k[0] = tsc_IV32 & 0xffff; > > + p1k[1] = tsc_IV32 >> 16; > > + p1k[2] = __le16_to_cpup(ptr++); > > + p1k[3] = __le16_to_cpup(ptr++); > > + p1k[4] = __le16_to_cpup(ptr); > > > > - p1k[0] = Lo16(tsc_IV32); > > - p1k[1] = Hi16(tsc_IV32); > > - p1k[2] = Mk16(ta[1], ta[0]); > > - p1k[3] = Mk16(ta[3], ta[2]); > > - p1k[4] = Mk16(ta[5], ta[4]); > > NACK! Now the code assumes natural alignment of the u16s which is, as > far as I can tell, completely bogus. Please review > Documentation/unaligned-memory-access.txt. I didn't think about the alignment constraints, true. Maybe common _unaligned versions should be added to the kernel infrastructure instead of hiding these hand-rolled versions in tkip.c? Harvey -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
