Re: What kernel to use?
Ethan Benson wrote:
> On Mon, Feb 12, 2001 at 11:14:41AM -0500, Adam C Powell IV wrote:
>
> > #
> > # USB HID
> > #
> > CONFIG_USB_HID=y
> > CONFIG_USB_WACOM=y
> > CONFIG_USB_WMFORCE=m
> > CONFIG_INPUT_KEYBDEV=y
> > CONFIG_INPUT_MOUSEDEV=y
> > CONFIG_INPUT_MOUSEDEV_SCREEN_X=1024
> > CONFIG_INPUT_MOUSEDEV_SCREEN_Y=768
> > CONFIG_INPUT_JOYDEV=m
> > CONFIG_INPUT_EVDEV=y
> >
> > #
> > # Mac device drivers
> > #
> > CONFIG_INPUT_ADBHID=y
> > CONFIG_MAC_HID=y
> > CONFIG_MAC_ADBKEYCODES=y
> > CONFIG_MAC_EMUMOUSEBTN=y
> > CONFIG_PPC_RTC=y
> > CONFIG_MAC_FLOPPY=y
> > CONFIG_MAC_SERIAL=y
> > CONFIG_SERIAL_CONSOLE=y
>
> these parts look ok.
>
> i have the following disabled:
>
> # CONFIG_USB_WACOM is not set
> # CONFIG_USB_WMFORCE is not set
> # CONFIG_MAC_FLOPPY is not set
> # CONFIG_SERIAL_CONSOLE is not set
These shouldn't change anything.
> > "ADB" and "HID" don't exist anywhere else in the config.
> >
> > And here's the build error with stock 2.2.18 from kernel.org:
> >
> > cc -D__KERNEL__ -I/usr/src/kernel-source-2.2.18/include -Wall -Wstrict-prototypes -O2
> > -fomit-frame-pointer -fno-strict-aliasing -D__powerpc__ -fsigned-char -msoft-float -pipe
> > -fno-builtin -ffixed-r2 -Wno-uninitialized -mmultiple -mstring -c -o mac_hid.o
> > mac_hid.c
> > mac_hid.c: In function `mac_hid_sysctl_keycodes':
> > mac_hid.c:296: structure has no member named `ppc_kbd_sysrq_xlate'
> > mac_hid.c:303: structure has no member named `ppc_kbd_sysrq_xlate'
> > make[4]: *** [mac_hid.o] Error 1
>
> i had this when my kernel configuration was bogus. if you would like
> to try my .config let me know, it will compile with stock 2.2.18. i
> use the tree as i do on my x86 boxes.
I'd love to see your .config. I guess you leave out CONFIG_MAGIC_SYSRQ?
There's a trivial patch to mac_hid.c which fixes the above and has been posted to the list a
couple of times (me 2000-12-29, Christian Pernegger 2001-1-25). I don't see how a .config
could do the same, and skipping a feature doesn't count as fixing a bug.
> > Furthermore, if one can generate a config using "make config" which fails to build or
> > link or boot or otherwise function perfectly in any way, is that not a kernel bug? The
> > config system should disallow any improper build options for the (sub)arch selected,
> > right?
>
> the config system is broken and won't be fixed until 2.5. (it can't
> be fixed only replaced, its a big pile of donkey dung)
Sure, the current system is not the most elegant there is, but it seems unwise to just
dismiss the whole thing without trying to see where Config.in errors may be breaking builds,
or to blame bugs in the released 2.2.18 kernel.org source (e.g. mac_hid.c) on the config
system.
Let me rephrase: can you produce a .config using "make config" which does not build, without
a bug somewhere in the source (or a Makefile or Config.in)?
-Adam P.
GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6
Welcome to the best software in the world today cafe!
Reply to: