On Sun, 25 Mar 2007, Randy Dunlap wrote: > On Sun, 25 Mar 2007 08:40:49 -0400 (EDT) Robert P. J. Day wrote: > > > > > i'd never noticed this before, but a small number of Kconfig files > > have some variation of "depends on m": > > > > $ grep "depends on m" $(find . -name Kconfig) > > ... > > ./drivers/scsi/pcmcia/Kconfig: depends on m && !64BIT > > ./drivers/scsi/pcmcia/Kconfig: depends on m > > ./drivers/scsi/pcmcia/Kconfig: depends on m && !64BIT > > ./drivers/scsi/pcmcia/Kconfig: depends on m > > ./drivers/scsi/pcmcia/Kconfig: depends on m > > ./drivers/isdn/hysdn/Kconfig: depends on m && PROC_FS && PCI && BROKEN_ON_SMP > > ./crypto/Kconfig: depends on m > > > > what's the rationale here? one consequence of this is that, even > > when you "make allyesconfig", these features will be built as modules. > > that is, here's the tail end of the build output after "make > > allyesconfig": > > > > CC crypto/tcrypt.mod.o > > LD [M] crypto/tcrypt.ko > > CC drivers/i2c/busses/i2c-stub.mod.o > > LD [M] drivers/i2c/busses/i2c-stub.ko > > CC drivers/isdn/hardware/eicon/diva_mnt.mod.o > > LD [M] drivers/isdn/hardware/eicon/diva_mnt.ko > > CC drivers/net/wan/cosa.mod.o > > LD [M] drivers/net/wan/cosa.ko > > CC drivers/net/wan/dscc4.mod.o > > LD [M] drivers/net/wan/dscc4.ko > > CC drivers/net/wan/hostess_sv11.mod.o > > LD [M] drivers/net/wan/hostess_sv11.ko > > CC drivers/net/wan/sealevel.mod.o > > LD [M] drivers/net/wan/sealevel.ko > > CC drivers/net/wan/z85230.mod.o > > LD [M] drivers/net/wan/z85230.ko > > CC drivers/scsi/pcmcia/aha152x_cs.mod.o > > LD [M] drivers/scsi/pcmcia/aha152x_cs.ko > > CC drivers/scsi/pcmcia/fdomain_cs.mod.o > > LD [M] drivers/scsi/pcmcia/fdomain_cs.ko > > CC drivers/scsi/pcmcia/nsp_cs.mod.o > > LD [M] drivers/scsi/pcmcia/nsp_cs.ko > > CC drivers/scsi/pcmcia/qlogic_cs.mod.o > > LD [M] drivers/scsi/pcmcia/qlogic_cs.ko > > CC drivers/scsi/pcmcia/sym53c500_cs.mod.o > > LD [M] drivers/scsi/pcmcia/sym53c500_cs.ko > > CC sound/oss/msnd.mod.o > > LD [M] sound/oss/msnd.ko > > CC sound/oss/msnd_classic.mod.o > > LD [M] sound/oss/msnd_classic.ko > > CC sound/oss/msnd_pinnacle.mod.o > > LD [M] sound/oss/msnd_pinnacle.ko > > > > what's so special about those features that they're built as > > modules even with "make allyesconfig"? just curious. > > This is all fairly old stuff. I think that someone decided long ago > that using these drivers only made sense as loadable modules. This > was probably before hotplug was working, and much much before udev > was born. So before hotplug worked, this was a reasonable thing to > do. > > (Well, the crypto module isn't old, but it's just a testing > module, so why build it into the kernel binary file?) oh, it's fairly clear that *some* of that stuff makes sense only as modules (such as that crypto module). and there's certainly no point wasting any cleanup time on anything OSS-related. but the rest of it certainly deserves a closer look -- i posted to LKML on the weirdness behind the PCMCIA stuff. rday -- ======================================================================== Robert P. J. Day Linux Consulting, Training and Annoying Kernel Pedantry Waterloo, Ontario, CANADA http://fsdev.net/wiki/index.php?title=Main_Page ======================================================================== -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
