On Tue, May 04, 2021 at 09:41:14AM +0200, Arnd Bergmann wrote: > On Tue, May 4, 2021 at 6:43 AM Greg Kroah-Hartman > <gregkh@xxxxxxxxxxxxxxxxxxx> wrote: > > On Mon, May 03, 2021 at 09:16:42PM +0200, Arnd Bergmann wrote: > > > On Mon, May 3, 2021 at 7:00 PM 'Nick Desaulniers' via Clang Built > > > Linux <clang-built-linux@xxxxxxxxxxxxxxxx> wrote: > > > > > > >> ERROR: "__memcat_p" [drivers/hwtracing/stm/stm_core.ko] undefined! > > > > > > > > > > I'm fairly sure this is unrelated to my patch, but I don't see what > > > > > happened here. > > > > > > > > It's unrelated to your patch. It was fixed in 5.7 by > > > > 7273ad2b08f8ac9563579d16a3cf528857b26f49 and a few other dependencies > > > > according to https://github.com/ClangBuiltLinux/linux/issues/515. > > > > > > > > > > Ah right, the big hammer. > > > > > > Greg, not sure what we want to do here. Backporting > > > > > > 7273ad2b08f8 ("kbuild: link lib-y objects to vmlinux forcibly when > > > CONFIG_MODULES=y") > > > > > > to v5.4 and earlier would be an easy workaround, but it has the potential > > > of adding extra bloat to the kernel image since it links in all other > > > library objects as well. > > > > I've lost the thread here, but what _real_ problem is happening here > > that doing the above is required? > > Randconfig builds fail if drivers/hwtracing/stm/stm_core.ko is a loadable > modules and nothing inside vmlinux forces lib/memcat_p.o to be > linked in. A simpler workaround for v5.4 would be: > > diff --git a/lib/Makefile b/lib/Makefile > index a5c5f342ade0..745d1207e9e2 100644 > --- a/lib/Makefile > +++ b/lib/Makefile > @@ -33,7 +33,7 @@ lib-y := ctype.o string.o vsprintf.o cmdline.o \ > flex_proportions.o ratelimit.o show_mem.o \ > is_single_threaded.o plist.o decompress.o kobject_uevent.o \ > earlycpio.o seq_buf.o siphash.o dec_and_lock.o \ > - nmi_backtrace.o nodemask.o win_minmax.o memcat_p.o \ > + nmi_backtrace.o nodemask.o win_minmax.o \ > buildid.o > > lib-$(CONFIG_PRINTK) += dump_stack.o > @@ -48,7 +48,7 @@ obj-y += bcd.o sort.o parser.o debug_locks.o random32.o \ > bsearch.o find_bit.o llist.o memweight.o kfifo.o \ > percpu-refcount.o rhashtable.o \ > once.o refcount.o usercopy.o errseq.o bucket_locks.o \ > - generic-radix-tree.o > + generic-radix-tree.o memcat_p.o > obj-$(CONFIG_STRING_SELFTEST) += test_string.o > obj-y += string_helpers.o > obj-$(CONFIG_TEST_STRING_HELPERS) += test-string_helpers.o > > which is the same as what 7273ad2b08f8 does, but only for this one file > instead of all of lib/*.o. I think a "one off" change would be best here. Can you submit it? thanks, greg k-h
