Re: make tinderbox failures

On Fri, Oct 26, 2012 at 3:24 PM, Warner Losh <imp_at_bsdimp.com> wrote:
>
> On Oct 26, 2012, at 1:08 PM, Garrett Cooper wrote:
>
>> On Fri, Oct 26, 2012 at 10:26 AM,  <mdf_at_freebsd.org> wrote:
>>> Running make tinderbox locally I see failures that aren't being
>>> reported by the automated tinderbox builds.  I'm curious what's
>>> different about the environment.  Failures are in the following:
>>>
>>> arm ARMADAXP kernel failed, check _.arm.ARMADAXP for details
>>> mips SENTRY5 kernel failed, check _.mips.SENTRY5 for details
>>> mips XLPN32 kernel failed, check _.mips.XLPN32 for details
>>> mips XLR kernel failed, check _.mips.XLR for details
>>> mips XLRN32 kernel failed, check _.mips.XLRN32 for details
>>>
>>> _.arm.ARMADAXP:
>>> /data/sb/bsd.git/sys/arm/mv/armadaxp/armadaxp_mp.c: In function
>>> 'platform_mp_start_ap':
>>> /data/sb/bsd.git/sys/arm/mv/armadaxp/armadaxp_mp.c:181: warning:
>>> passing argument 1 of 'pmap_kextract' makes integer from pointer
>>> without a cast
>
> This still needs some love...
>
>>> _.mips.SENTRY5:
>>> config: Error: device "siba" is unknown
>
> OK.  Traced this down and fixed it.  I'd planned on moving siba to files, and it looks like I never finished that.  My bad.
>
>>> _.mips.XLPN32 and _.mips.XLRN32:
>>> linking kernel.debug
>>> stack_machdep.o: In function `stack_capture':
>>> /data/sb/bsd.git/sys/mips/mips/stack_machdep.c:(.text+0x34): undefined
>>> reference to `stack_zero'
>>> /data/sb/bsd.git/sys/mips/mips/stack_machdep.c:(.text+0x34):
>>> relocation truncated to fit: R_MIPS_26 against `stack_zero'
>>> /data/sb/bsd.git/sys/mips/mips/stack_machdep.c:(.text+0x13c):
>>> undefined reference to `stack_put'
>>> /data/sb/bsd.git/sys/mips/mips/stack_machdep.c:(.text+0x13c):
>>> relocation truncated to fit: R_MIPS_26 against `stack_put'
>
> stack_machdep needs to depend on options stack or ddb.  Fixed.  Unsure why other kernels didn't uncover this.
>
>>> _.mips.XLR:
>>> linking kernel.debug
>>> board.o: In function `xlr_board_info_setup':
>>> /data/sb/bsd.git/sys/mips/rmi/board.c:(.text+0x3a4): undefined
>>> reference to `__ucmpdi2'
>>> /data/sb/bsd.git/sys/mips/rmi/board.c:(.text+0x3a4): relocation
>>> truncated to fit: R_MIPS_26 against `__ucmpdi2'
>>> /data/sb/bsd.git/sys/mips/rmi/board.c:(.text+0x3e4): undefined
>>> reference to `__ucmpdi2'
>>> /data/sb/bsd.git/sys/mips/rmi/board.c:(.text+0x3e4): relocation
>>> truncated to fit: R_MIPS_26 against `__ucmpdi2'
>
> Fixed earlier this week.
>
>>> Since I don't work on arm or mips I generally ignore these.  But it
>>> makes it harder to have confidence in a global change when make
>>> tinderbox throws errors locally.
>>
>> These are probably due to the changes that Warner made recently to the
>> mips conf files.
>
> Some of them...

Thanks for fixes!  Is the reason the public tinderbox didn't see these
because it has explicit lists of arch/CONF kernels to build, while
"make tinderbox" on my machine just builds all the capitalized files
as configuration files?

Thanks,
matthew