On Fri, Oct 21, 2022 at 10:38 PM Jakub Kicinski <kuba@xxxxxxxxxx> wrote: > > On Fri, 21 Oct 2022 23:34:13 +0200 Jiri Olsa wrote: > > > You are right, they should be identical once PTR is deduplicated > > > properly. Sorry, was too quick to jump to conclusions. I was thinking > > > about situations explained by Alan. > > > > > > So, is this still an issue or this was fixed by [0]? > > > > > > [0] https://lore.kernel.org/bpf/1666364523-9648-1-git-send-email-alan.maguire@xxxxxxxxxx/ > > > > yes, it seems to be fixed by that > > > > Jakub, > > could you check with pahole fix [1]? > > If you mean the warning from the subject then those do seem to be gone. > But if I'm completely honest I don't remember how I triggered them in > the first place :S There weren't there on every build for me. > > The objtool warning is still here: > > $ make PAHOLE=~/pahole O=build_allmodconfig/ -j 60 >/tmp/stdout 2>/tmp/stderr; \ > cat /tmp/stderr > > vmlinux.o: warning: objtool: ___ksymtab+bpf_dispatcher_xdp_func+0x0: data relocation to !ENDBR: bpf_dispatcher_xdp_func+0x0 > vmlinux.o: warning: objtool: bpf_dispatcher_xdp+0xa0: data relocation to !ENDBR: bpf_dispatcher_xdp_func+0x0 The effect of the compiler bug was addressed by this fix: https://lore.kernel.org/all/20221018075934.574415-1-jolsa@xxxxxxxxxx/ It's in the bpf tree, but the warning will stay. While the compiler is broken the objtool should keep complaining.
