[kbuild-all] Re: arch/powerpc/kernel/optprobes.c:34:1: error: unused function 'is_kprobe_ppc_optinsn_slot'

Le 12/05/2021 à 17:06, Masami Hiramatsu a écrit : > On Wed, 12 May 2021 21:50:57 +0900 > Masami Hiramatsu <mhiramat(a)kernel.org&gt; wrote: > >> On Wed, 12 May 2021 11:52:30 +0200 >> Christophe Leroy <christophe.leroy(a)csgroup.eu&gt; wrote: > >>> >>> Le 12/05/2021 à 11:36, Christophe Leroy a écrit : >>>> >>>> >>>> Le 11/05/2021 à 20:46, Nathan Chancellor a écrit : >>>>> On 5/11/2021 9:47 AM, Christophe Leroy wrote: >>>>>> Hi All, >>>>>> >>>>>> Le 11/05/2021 à 18:30, kernel test robot a écrit : >>>>>>> Hi Christophe, >>>>>>> >>>>>>> First bad commit (maybe != root cause): >>>>>>> >>>>>>> tree: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master >>>>>>> head:   1140ab592e2ebf8153d2b322604031a8868ce7a5 >>>>>>> commit: eacf4c0202654adfa94bbb17b5c5c77c0be14af8 powerpc: Enable OPTPROBES on PPC32 >>>> >>>> Ok, the problem appears on PPC32 with that patch, but it must have been there on PPC64 since the >>>> implementation of optprobes with commit 51c9c0843993 ("powerpc/kprobes: Implement Optprobes") >>>> >>>> >>>>>>> date:   3 weeks ago >>>>>>> config: powerpc-randconfig-r033-20210511 (attached as .config) >>>>>>> compiler: clang version 13.0.0 (https://github.com/llvm/llvm-project >>>>>>> a0fed635fe1701470062495a6ffee1c608f3f1bc) >>>>>> >>>>>> A clang issue ? >>>>>> >>>>>> >>>>>>> reproduce (this is a W=1 build): >>>>>>>          wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O >>>>>>> ~/bin/make.cross >>>>>>>          chmod +x ~/bin/make.cross >>>>>>>          # install powerpc cross compiling tool for clang build >>>>>>>          # apt-get install binutils-powerpc-linux-gnu >>>>>>>          # >>>>>>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit... >>>>>>> >>>>>>>          git remote add linus https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git >>>>>>>          git fetch --no-tags linus master >>>>>>>          git checkout eacf4c0202654adfa94bbb17b5c5c77c0be14af8 >>>>>>>          # save the attached .config to linux build tree >>>>>>>          COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 ARCH=powerpc >>>>>>> >>>>>>> If you fix the issue, kindly add following tag as appropriate >>>>>>> Reported-by: kernel test robot <lkp(a)intel.com&gt; >>>>>>> >>>>>>> All errors (new ones prefixed by >>): >>>>>>> >>>>>>>>> arch/powerpc/kernel/optprobes.c:34:1: error: unused function 'is_kprobe_ppc_optinsn_slot' >>>>>>>>> [-Werror,-Wunused-function] >>>>>>>     DEFINE_INSN_CACHE_OPS(ppc_optinsn); >>>>>>>     ^ >>>>>>>     include/linux/kprobes.h:306:20: note: expanded from macro 'DEFINE_INSN_CACHE_OPS' >>>>>>>     static inline bool is_kprobe_##__name##_slot(unsigned long addr)        \ >>>>>> >>>>>> That's a 'static inline', shouldn't generate an 'unused function' warning. >>>>> This is a W=1 build, which means that "inline" does not silence unused function warnings with >>>>> clang because of commit 6863f5643dd7 ("kbuild: allow Clang to find unused static inline functions >>>>> for W=1 build"). >>>> >>>> Ok, I didn't know that. >>> >>> After looking at it in more details, I don't really know what should be done to avoid that. >> >> Ah, thanks for reporting! >> >>> >>> This unused function is defined as part of DEFINE_INSN_CACHE_OPS(ppc_optinsn). >>> >>> The code uses other items defined by macro DEFINE_INSN_CACHE_OPS, so it cannot be removed. >>> >>> Solution could be to hide that in a .h, but is that worth it ? >> >> The best solution is to remove ppc_optinsn, but that is defined in kernel/kprobes.c >> just because page allocation is different. >> I think the easiest fix is to add weak alloc_optinsn_page() and override it >> in arch/powerpc/kernel/optprobe.c. > > Please try this patch. > > Thank you, Yeah, look pretty similar to the two patches I sent 30 minutes ago.

> > From 7782fa154349923ed52baa8843cc1d5eed26e82d Mon Sep 17 00:00:00 2001 > From: Masami Hiramatsu <mhiramat(a)kernel.org&gt; > Date: Wed, 12 May 2021 23:33:10 +0900 > Subject: [PATCH] powerpc/kprobes: Define arch specific optinsn memory > allocation > > @@ -203,14 +193,14 @@ int arch_prepare_optimized_kprobe(struct optimized_kprobe *op, struct kprobe *p) > unsigned long nip, size; > int rc, i; > > - kprobe_ppc_optinsn_slots.insn_size = MAX_OPTINSN_SIZE; > + kprobe_optinsn_slots.insn_size = MAX_OPTINSN_SIZE; Is this still needed ? It looks like it is already done by init_kprobes() in kernel/kprobes.c

> > nip = can_optimize(p); > if (!nip) > return -EILSEQ; > > /* Allocate instruction slot for detour buffer */ > - buff = get_ppc_optinsn_slot(); > + buff = get_optinsn_slot(); > if (!buff) > return -ENOMEM; > Christophe