On Mon, May 24, 2010 at 8:16 PM, Felipe Balbi <felipe.balbi@xxxxxxxxx> wrote: > personally I don't like this patch since it makes the C-sources rely on > indirect header inclusion by another header. That's really error prone and > will cause build failures if someone decides to remove the include > <linux/device.h> from plat/mailbox.h. First of all, you cannot expect each and every source file to include each and every header it's using directly. It would be nice, but it's not happening now, and probably will not happen ever. Now, with a few exceptions, I didn't assume somebody else would be loading anything, in fact, the patch is mostly _adding_ headers that were missing. If you take a look at the modified files regarding kernel.h and module.h: * arch/arm/mach-omap1/mailbox.c: Not really using kernel.h, missing module.h * arch/arm/mach-omap2/mailbox.c: Missing module.h * arch/arm/plat-omap/mailbox.c Missing kernel.h So it was clear to me that these were not carefully added, and probably were redundant, so I choose to remove them completely. If you have something against that, I can add them where it makes sense but the current patch is not really making the situation much worst. Now, regarding device.h: * arch/arm/mach-omap2/mailbox.c Using it, but missing * arch/arm/plat-omap/include/plat/mailbox.h Was using it, but a 'struct device' would suffice * arch/arm/plat-omap/mailbox.c Using it correctly Now, I am pretty sure that platform_device.h will always depend on device.h that's why I removed them, but I can add them again if somebody wants to, but again, the situation is already bad. -- Felipe Contreras -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
