#10994 NORM 1.75-so: xo-1.75 kernel rpmbuild erroneously attempts to build with "nommu"
Zarro Boogs per Child
bugtracker at laptop.org
Mon Jun 13 18:11:17 EDT 2011
#10994: xo-1.75 kernel rpmbuild erroneously attempts to build with "nommu"
-----------------------------+----------------------------------------------
Reporter: martin.langhoff | Owner: martin.langhoff
Type: defect | Status: new
Priority: normal | Milestone: 1.75-software
Component: kernel | Version: not specified
Keywords: | Next_action: never set
Verified: 0 | Deployment_affected:
Blockedby: | Blocking:
-----------------------------+----------------------------------------------
Our current 2.6.29 tree, when built under rpmbuild fails with errors:
{{{
arch/arm/kernel/built-in.o: In function `module_finalize':
/home/martin/src/rpm-
build/kernel-2.6.39_xo1.75/linux-2.6.39.armv7l/arch/arm/kernel/module.c:344:
undefined reference to `fixup_smp'
arch/arm/mm/built-in.o: In function `arm_mm_memblock_reserve':
/home/martin/src/rpm-
build/kernel-2.6.39_xo1.75/linux-2.6.39.armv7l/arch/arm/mm/mmu.c:902:
undefined reference to `swapper_pg_dir'
mm/built-in.o:(.data+0x5d0): undefined reference to `swapper_pg_dir'
}}}
For some reason, under rpmbuild, CONFIG_MMU=y is getting ignored, and the
build fails because "there is no MMU configured".
If we look into the builddir,
* the .config file is correct
* no defconfig files are present that are likely to be overriding our
config
* executing the shellscript created by rpm completes the build correctly,
picking head.S instead of head-nommu.S
* otherwise, a manual call to make completes the build correctly
Unfortunately, attempting to call make [opts] zImage twice in %build does
not succeed.
The usual makefile debug strategies (--debug and V=1) don't show anything
interesting. Diffing the output of a good and a bad build (with --debug
V=1 does not show anything interesting. Make is not telling us why it
decides to build head-nommu.S
The file that controls this seems to be arch/arm/Makefile - see CONFIG_MMU
and MMUEXT.
For the time being, we have an ugly workaround in place. We should however
get to the bottom of it, as it is worrying that something is overriding or
mangling our options.
--
Ticket URL: <http://dev.laptop.org/ticket/10994>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system
More information about the Bugs
mailing list