On Mon, 2022-06-06 at 13:40 +0300, Andy Shevchenko wrote: > On Mon, Jun 06, 2022 at 08:48:05AM +0800, kernel test robot wrote: > > Hi Sander, > > > > I love your patch! Yet something to improve: > > > > [auto build test ERROR on akpm-mm/mm-everything] > > [also build test ERROR on linus/master v5.18 next-20220603] > > [If your patch is applied to the wrong git tree, kindly drop us a note. > > And when submitting patch, we suggest to use '--base' as documented in > > https://git-scm.com/docs/git-format-patch] > > > > url: > > https://github.com/intel-lab-lkp/linux/commits/Sander-Vanheule/cpumask-Fix-invalid-uniprocessor-assumptions/20220606-004959 > > base: https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything > > config: i386-randconfig-a009 > > (https://download.01.org/0day-ci/archive/20220606/202206060858.wA0FOzRy-lkp@xxxxxxxxx/config) > > compiler: gcc-11 (Debian 11.3.0-1) 11.3.0 > > reproduce (this is a W=1 build): > > # https://github.com/intel-lab-lkp/linux/commit/37b3f10c4604ea299b454f39ac5ba3cad903ae16 > > git remote add linux-review https://github.com/intel-lab-lkp/linux > > git fetch --no-tags linux-review Sander-Vanheule/cpumask-Fix-invalid-uniprocessor- > > assumptions/20220606-004959 > > git checkout 37b3f10c4604ea299b454f39ac5ba3cad903ae16 > > # save the config file > > mkdir build_dir && cp config build_dir/.config > > make W=1 O=build_dir ARCH=i386 SHELL=/bin/bash > > > > If you fix the issue, kindly add following tag where applicable > > Reported-by: kernel test robot <lkp@xxxxxxxxx> > > > > All errors (new ones prefixed by >>): > > > > ld: arch/x86/kernel/cpu/cacheinfo.o: in function `__cache_amd_cpumap_setup': > > arch/x86/kernel/cpu/cacheinfo.c:890: undefined reference to `cpu_llc_shared_map' > > > > ld: arch/x86/kernel/cpu/cacheinfo.c:895: undefined reference to `cpu_llc_shared_map' > > Seems like somewhere we need stubs for UP builds for those cache related functions. > I think I finally figured out what's going on here. cpu_llc_shared_map is always declared with DECLARE_PER_CPU_READ_MOSTLY, but defined in arch/x86/kernel/smpboot.c which only builds on CONFIG_SMP=y. cpu_llc_shared_map is accessed in a for_each_cpu loop: for_each_cpu(i, cpu_llc_shared_mask(cpu)) The old (wrong) UP implementation would ignore the mask, so cpu_llc_shared_map access was optimised out and the linker would never see that symbol. Adding a stub for the two inline functions in arch/x86/include/smp.h won't be sufficient I'm afraid. But I think we can safely make the following assumptions (nobody complained before): * anything using cpumask_first_zero(), cpumask_next_zero(), and for_each_cpu_not() was expecting an empty mask on UP builds, * anything else would have expected a filled mask on UP builds. I'll think about how to identify all the possible cases, but may not be able to spend a lot of time on this in the two coming weeks. Any suggestions, or alternative solutions, are of course welcome. Best, Sander
