From: David Miller <davem@xxxxxxxxxxxxx> Date: Wed, 29 Nov 2017 14:26:44 -0500 (EST) >> mator@ttip:~/sparc$ git bisect good >> 46ad8d2d22c17e2b577adec55ae87161666a3267 is the first bad commit >> commit 46ad8d2d22c17e2b577adec55ae87161666a3267 >> Author: Vijay Kumar <vijay.ac.kumar@xxxxxxxxxx> >> Date: Wed Oct 11 12:50:06 2017 -0600 >> >> sparc64: Use sparc optimized fls and __fls for T4 and above > > Thanks for bisecting, I'll take a look. The problem is that the distance between the generic sparc64 fls/__fls symbols and the NG4 optimized ones are too large for the branch patch. It needs to fit in a 19-bit displacement field. When there are library routines and patching, you have to group them near eachother. This means if the generic routine is "lib-y" you have to make the patch code "lib-y" too. In this case, the NG4fls.o object is placed into "obj-y" which is too far from the "lib-y" code to fit into the disp19 relocation of the patch branch. Vijay, why did you put the NG4fls.o into obj-y? Anatoly, please try this patch: diff --git a/arch/sparc/lib/Makefile b/arch/sparc/lib/Makefile index 0f0f76b..063556f 100644 --- a/arch/sparc/lib/Makefile +++ b/arch/sparc/lib/Makefile @@ -19,7 +19,7 @@ lib-$(CONFIG_SPARC32) += muldi3.o bitext.o cmpdi2.o lib-$(CONFIG_SPARC64) += multi3.o lib-$(CONFIG_SPARC64) += fls.o lib-$(CONFIG_SPARC64) += fls64.o -obj-$(CONFIG_SPARC64) += NG4fls.o +lib-$(CONFIG_SPARC64) += NG4fls.o lib-$(CONFIG_SPARC64) += copy_page.o clear_page.o bzero.o lib-$(CONFIG_SPARC64) += csum_copy.o csum_copy_from_user.o csum_copy_to_user.o -- To unsubscribe from this list: send the line "unsubscribe sparclinux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
