Re: [PATCH v4 bpf-next 3/5] kbuild: build kernel module BTFs if BTF is enabled and pahole supports it

[Andrii Nakryiko <andrii.nakryiko@gmail.com>]

On Mon, Nov 16, 2020 at 1:24 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Mon, 16 Nov 2020 12:34:17 -0800 Andrii Nakryiko wrote:
> > > This change, commit 5f9ae91f7c0d ("kbuild: Build kernel module BTFs if BTF is enabled and pahole
> > > supports it") currently in net-next, linux-next, etc. breaks the use-case of compiling only a specific
> > > kernel module (both in-tree and out-of-tree, e.g. 'make M=drivers/net/ethernet/intel/ice') after
> > > first doing a 'make modules_prepare'.  Previously, that use-case would result in a warning noting
> > > "Symbol info of vmlinux is missing. Unresolved symbol check will be entirely skipped" but now it
> > > errors out after noting "No rule to make target 'vmlinux', needed by '<...>.ko'.  Stop."
> > >
> > > Is that intentional?
> >
> > I wasn't aware of such a use pattern, so definitely not intentional.
> > But vmlinux is absolutely necessary to generate the module BTF. So I'm
> > wondering what's the proper fix here? Leave it as is (that error
> > message is actually surprisingly descriptive, btw)? Force vmlinux
> > build? Or skip BTF generation for that module?
>
> For an external out-of-tree module there is no guarantee that vmlinux
> will even be on the system, no? So only the last option can work in
> that case.


Ok, how about something like the patch below. With that I seem to be
getting the desired behavior:

$ make clean
$ touch drivers/acpi/button.c
$ make M=drivers/acpi
make[1]: Entering directory `/data/users/andriin/linux-build/default-x86_64'
  CC [M]  drivers/acpi/button.o
  MODPOST drivers/acpi/Module.symvers
  LD [M]  drivers/acpi/button.ko
  BTF [M] drivers/acpi/button.ko
Skipping BTF generation for drivers/acpi/button.ko due to
unavailability of vmlinux
make[1]: Leaving directory `/data/users/andriin/linux-build/default-x86_64'
$ readelf -S ~/linux-build/default/drivers/acpi/button.ko | grep BTF -A1
... empty ...

Now with normal build:

$ make all
...
LD [M]  drivers/acpi/button.ko
BTF [M] drivers/acpi/button.ko
...
$ readelf -S ~/linux-build/default/drivers/acpi/button.ko | grep BTF -A1
  [60] .BTF              PROGBITS         0000000000000000  00029310
       000000000000ab3f  0000000000000000           0     0     1



diff --git a/scripts/Makefile.modfinal b/scripts/Makefile.modfinal
index 02b892421f7a..d49ec001825d 100644
--- a/scripts/Makefile.modfinal
+++ b/scripts/Makefile.modfinal
@@ -38,7 +38,12 @@ quiet_cmd_ld_ko_o = LD [M]  $@
     $(if $(ARCH_POSTLINK), $(MAKE) -f $(ARCH_POSTLINK) $@, true)

 quiet_cmd_btf_ko = BTF [M] $@
-      cmd_btf_ko = LLVM_OBJCOPY=$(OBJCOPY) $(PAHOLE) -J --btf_base vmlinux $@
+      cmd_btf_ko =                             \
+    if [ -f vmlinux ]; then                        \
+        LLVM_OBJCOPY=$(OBJCOPY) $(PAHOLE) -J --btf_base vmlinux $@; \
+    else                                \
+        printf "Skipping BTF generation for %s due to unavailability
of vmlinux\n" $@ 1>&2; \
+    fi;

 # Same as newer-prereqs, but allows to exclude specified extra dependencies
 newer_prereqs_except = $(filter-out $(PHONY) $(1),$?)
@@ -49,7 +54,7 @@ if_changed_except = $(if $(call
newer_prereqs_except,$(2))$(cmd-check),      \
     printf '%s\n' 'cmd_$@ := $(make-cmd)' > $(dot-target).cmd, @:)

 # Re-generate module BTFs if either module's .ko or vmlinux changed
-$(modules): %.ko: %.o %.mod.o scripts/module.lds vmlinux FORCE
+$(modules): %.ko: %.o %.mod.o scripts/module.lds $(if
$(KBUILD_BUILTIN),vmlinux) FORCE
     +$(call if_changed_except,ld_ko_o,vmlinux)
 ifdef CONFIG_DEBUG_INFO_BTF_MODULES
     +$(if $(newer-prereqs),$(call cmd,btf_ko))