Hi, On Tue, Jun 14, 2011 at 12:45 PM, Felipe Balbi <balbi@xxxxxx> wrote: > Hi, > > On Tue, Jun 14, 2011 at 11:33:32AM -0400, Arnaud Lacombe wrote: >> On Tue, Jun 14, 2011 at 3:32 AM, Felipe Balbi <balbi@xxxxxx> wrote: >> > On Tue, Jun 14, 2011 at 08:28:05AM +0200, Michal Nazarewicz wrote: >> >> On Mon, 13 Jun 2011 17:35:17 +0200, Felipe Balbi <balbi@xxxxxx> wrote: >> >> >We also need to a find a better way to get rid of the inclusion of C >> >> >source files. I understand why Dave did it, but it's been quite some >> >> >time since those patches were committed - e.g. >> >> >7e75bc0f9006e995a0fa25f0a285addc3d5fd5cb - and no progress has been made >> >> >to fix that up. Probably Clang's link-time optimization would help a lot >> >> >on that, but Clang still needs quite some work to be usable for us, so >> >> >we need something else. Any ideas ? >> >> >> >> To me it always seemed like it's a bug in/lack of feature of Kbuild. >> >> Would making Kbuild allow building module from separate .o files be >> >> a problem? >> > >> > maybe the guys at linux-kbuild@vger can help answering this. >> > >> What is the problem exactly ? >> >> The following patch: >> >> diff --git a/drivers/usb/gadget/Makefile b/drivers/usb/gadget/Makefile >> index 4fe92b1..cc432b8 100644 >> --- a/drivers/usb/gadget/Makefile >> +++ b/drivers/usb/gadget/Makefile >> @@ -34,6 +34,12 @@ obj-$(CONFIG_USB_FUSB300) += fusb300_udc.o >> # USB gadget drivers >> # >> g_zero-y := zero.o >> +g_zero-y += composite.o >> +g_zero-y += usbstring.o >> +g_zero-y += config.o >> +g_zero-y += epautoconf.o >> +g_zero-y += f_sourcesink.o >> +g_zero-y += f_loopback.o > > yes, you can do that. But the problem is the runtime memory footprint > that we will have with that. At least that was the reason why Greg had > asked Dave to change it to how it is done now. > What number are we talking about ? Using the same hack in `f_loopback.c'[0], here is the numbers I get on x86-64, using gcc 4.5.1, with -O2: combined (C): 14017 1440 288 15745 3d81 drivers/usb/gadget/g_zero.o not combined (P): 13835 1472 296 15603 3cf3 drivers/usb/gadget/g_zero.o the result at -Os is slightly worse: (C) 11318 1344 256 12918 3276 drivers/usb/gadget/g_zero.o (P) 11353 1360 264 12977 32b1 drivers/usb/gadget/g_zero.o - Arnaud [0]: btw, this is a nice source of code duplication, as sourcesink_add() and loopback_add() are pretty much identical. To some extend, f_loopback.c and f_sourcesink.c seem to share very similar code structure... > Do we support gcc --combine already ? > not I know of, but my knowledge are limited. - Arnaud > -- > balbi > -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
