The following is the end result of a small conversation i had with Sam Leffler regarding the recent ath_hal commit breaking buildkernel on amd64 ... i figured i should probably send the lists about this until a fix gets committed. -------- Forwarded Message -------- From: Pascal Hofstee <caelian_at_gmail.com> To: Sam Leffler <sam_at_errno.com> Subject: Re: [head tinderbox] failure on amd64/amd64 Date: Sun, 07 May 2006 18:32:36 -0700 On Sun, 2006-05-07 at 17:59 -0700, Sam Leffler wrote: > Don't know what to say; I routinely build custom kernels w/ and w/o > ath_hal in the config file and/or the MODULES_OVERRIDE list w/o problems. I had a closer look at the warnings .. and i have at least figured out what's causing the error. The build is complaining it doesn't know how to make the file /usr/src/sys/modules/ath_hal/../../contrib/dev/ath/ public/amd64-elf.opt_ah.h Upon closer inspection this file indeed doesn't exist .. but x86_64-elf.opt_ah.h does, it looks like somewhere the translation from amd64 to x86_64 doesn't take place properly. At this point i decided to take a closer look at the /usr/src/sys/modules/ath_hal/Makefile As it turns out .. the addition of the line to unbreak the Sparc64 build, results in breaking the translation for AMD64. ATH_MODULE_ARCH=${MACHINE_ARCH:S/amd64/x86_64/} ATH_MODULE_ARCH=${MACHINE_ARCH:S/sparc64/sparc64-be/} if i comment the sparc64 entry, make -V ATH_MODULE_ARCH properly returns x86_64 .. if i leave it uncommented make -V ATH_MODULE_ARCH returns amd64 again. I would assume the proper solution here to be to set the ATH_MODULE_ARCH variable conditionally based on MACHINE_ARCH instead of two consecutive definitions as is currently the case. Please correct me if i am wrong ?Received on Sun May 07 2006 - 23:59:03 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:55 UTC