From: Jarkko Sakkinen <jarkko.sakkinen@iki.fi>
To: Nick Desaulniers <ndesaulniers@google.com>
Cc: Jarkko Sakkinen <jarkko@kernel.org>,
Andrew Morton <akpm@linux-foundation.org>,
Kees Cook <keescook@chromium.org>,
Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>,
Nathan Chancellor <natechancellor@gmail.com>,
Sedat Dilek <sedat.dilek@gmail.com>,
Marco Elver <elver@google.com>,
Andrey Konovalov <andreyknvl@google.com>,
Masahiro Yamada <masahiroy@kernel.org>,
clang-built-linux <clang-built-linux@googlegroups.com>,
Daniel Borkmann <daniel@iogearbox.net>,
Alexei Starovoitov <ast@kernel.org>,
Will Deacon <will@kernel.org>,
Vincenzo Frascino <vincenzo.frascino@arm.com>,
LKML <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v3 1/7] compiler-clang: add build check for clang 10.0.1
Date: Wed, 4 Nov 2020 03:34:47 +0200 [thread overview]
Message-ID: <20201104013447.GA21728@kapsi.fi> (raw)
In-Reply-To: <CAKwvOdnZSLP_YF3iFDLTHFE=ORxsrCR06s-B2Hk7khSxdC0+5A@mail.gmail.com>
On Tue, Nov 03, 2020 at 05:18:38PM -0800, Nick Desaulniers wrote:
> On Tue, Nov 3, 2020 at 4:38 PM Jarkko Sakkinen <jarkko@kernel.org> wrote:
> >
> > On Tue, Nov 03, 2020 at 10:48:27AM -0800, Nick Desaulniers wrote:
> > > On Mon, Nov 2, 2020 at 8:55 PM Jarkko Sakkinen <jarkko@kernel.org> wrote:
> > > >
> > > > On Wed, Sep 02, 2020 at 03:59:05PM -0700, Nick Desaulniers wrote:
> > > > > +#define CLANG_VERSION (__clang_major__ * 10000 \
> > > > > + + __clang_minor__ * 100 \
> > > > > + + __clang_patchlevel__)
> > > > > +
> > > > > +#if CLANG_VERSION < 100001
> > > > > +# error Sorry, your version of Clang is too old - please use 10.0.1 or newer.
> > > > > +#endif
> > > >
> > > >
> > > > I'm trying to compile a BPF enabled test kernel for a live system and I
> > > > get this error even though I have much newer clang:
> > > >
> > > > ➜ ~ (master) ✔ clang --version
> > > > Ubuntu clang version 11.0.0-2
> > > > Target: x86_64-pc-linux-gnu
> > > > Thread model: posix
> > > > InstalledDir: /usr/bin
> > > >
> > > > Tried to Google for troubleshooter tips but this patch is basically the
> > > > only hit I get :-)
> > >
> > > To check the values of the above preprocessor defines, please run:
> > > $ clang -dM -E - < /dev/null | grep -e __clang_m -e __clang_p
> > >
> > > If you have multiple versions of clang installed, you might not be
> > > running the version you think you are. Particularly, if you're using
> > > bcc, idk if it includes a copy of clang? If that's the case, we may
> > > have to work out how we can support older versions of clang for the
> > > express purposes of bpf.
> >
> > ➜ ~ (master) ✔ clang -dM -E - < /dev/null | grep -e __clang_m -e __clang_p
> > #define __clang_major__ 11
> > #define __clang_minor__ 0
> > #define __clang_patchlevel__ 0
> >
> > I'm compiling the kernel itself with GCC.
> >
> > Here's an example BPF script that fails on me:
> >
> > struct sgx_enclave_add_pages {
> > unsigned long src;
> > unsigned long offset;
> > unsigned long length;
> > unsigned long secinfo;
> > unsigned long flags;
> > unsigned long count;
> > };
> >
> > kprobe:sgx_ioctl
> > {
> > if (arg1 == 0xc030a401) {
> > printf("sgx_ioctl: %d, %lu\n", pid, ((struct sgx_enclave_add_pages *)(arg2))->offset);
> > }
> >
> > }
> > Note that it relies on code not yet in the mainline.
> >
> > If I don't declare structs, things work just fine. E.g. the following
> > works:
> >
> > kprobe:sgx_encl_get_backing
> > {
> > printf("%s\n", func)
> > }
> >
> > BTW, I don't really understand how scripts/clang-version.sh is even
> > supposed to work, if you compile the kernel itself with GCC. In that
> > case there would be no output, right? And thus version gets set to
> > zero...
>
> That script is only used by KBUILD. include/linux/compiler-clang.h is
> what's included into include/linux/compiler_types.h and causes the
> error. The eBFP tools must be including kernel headers and defining
> `__clang__`. Forgive my complete ignorance of eBPF, but how do you
> build that script? I assume the tool is using Clang, since eBPF
Thanks a lot for helping with this :-)
I'm using bpftrace as the frontend.
> relies on the LLVM backend (not sure if the GCC eBPF backend is good
> to go quite yet), and that version of clang is older.
>
> I wonder if we should guard the version check with __BPF_TRACING__
> similar to arch/x86/include/asm/cpufeature.h? Care to test:
Before I received this response, I did git revert for this commit
and things started working again.
> ```
> diff --git a/include/linux/compiler-clang.h
> b/include/linux/compiler-clang.h
> index dd7233c48bf3..98cff1b4b088 100644
> --- a/include/linux/compiler-clang.h
> +++ b/include/linux/compiler-clang.h
> @@ -8,8 +8,10 @@
> + __clang_patchlevel__)
>
> #if CLANG_VERSION < 100001
> +#ifndef __BPF_TRACING__
> # error Sorry, your version of Clang is too old - please use 10.0.1 or newer.
> #endif
> +#endif
>
> /* Compiler specific definitions for Clang compiler */
> ```
> --
Shouldn't "#ifndef" be before the whole version check? Otherwise,
LGTM. Please CC me once there is a properly formed patch to try out.
> Thanks,
> ~Nick Desaulniers
/Jarkko
next prev parent reply other threads:[~2020-11-04 1:35 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-09-02 22:59 [PATCH v3 0/7] set clang minimum version to 10.0.1 Nick Desaulniers
2020-09-02 22:59 ` [PATCH v3 1/7] compiler-clang: add build check for clang 10.0.1 Nick Desaulniers
2020-11-03 4:55 ` Jarkko Sakkinen
2020-11-03 6:38 ` Nathan Chancellor
2020-11-03 9:00 ` Jarkko Sakkinen
2020-11-03 18:48 ` Nick Desaulniers
2020-11-04 0:38 ` Jarkko Sakkinen
2020-11-04 1:18 ` Nick Desaulniers
2020-11-04 1:34 ` Jarkko Sakkinen [this message]
2020-11-04 1:36 ` Jarkko Sakkinen
2020-11-17 3:04 ` John Hubbard
2020-11-17 18:46 ` Nick Desaulniers
2020-11-18 2:31 ` Nathan Chancellor
2020-11-18 2:37 ` Andrew Morton
2020-09-02 22:59 ` [PATCH v3 2/7] Revert "kbuild: disable clang's default use of -fmerge-all-constants" Nick Desaulniers
2020-09-02 22:59 ` [PATCH v3 3/7] Revert "arm64: bti: Require clang >= 10.0.1 for in-kernel BTI support" Nick Desaulniers
2020-09-02 22:59 ` [PATCH v3 4/7] Revert "arm64: vdso: Fix compilation with clang older than 8" Nick Desaulniers
2020-09-02 22:59 ` [PATCH v3 5/7] Partially revert "ARM: 8905/1: Emit __gnu_mcount_nc when using Clang 10.0.0 or newer" Nick Desaulniers
2020-09-02 22:59 ` [PATCH v3 6/7] kasan: Remove mentions of unsupported Clang versions Nick Desaulniers
2020-09-02 22:59 ` [PATCH v3 7/7] compiler-gcc: improve version error Nick Desaulniers
2020-09-07 21:25 ` Will Deacon
2020-09-03 14:06 ` [PATCH v3 0/7] set clang minimum version to 10.0.1 Sedat Dilek
2020-09-03 17:28 ` Nathan Chancellor
2020-09-04 5:42 ` Miguel Ojeda
2020-09-07 16:12 ` Arvind Sankar
2020-09-08 4:42 ` Nathan Chancellor
2020-09-07 21:26 ` Will Deacon
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20201104013447.GA21728@kapsi.fi \
--to=jarkko.sakkinen@iki.fi \
--cc=akpm@linux-foundation.org \
--cc=andreyknvl@google.com \
--cc=ast@kernel.org \
--cc=clang-built-linux@googlegroups.com \
--cc=daniel@iogearbox.net \
--cc=elver@google.com \
--cc=jarkko@kernel.org \
--cc=keescook@chromium.org \
--cc=linux-kernel@vger.kernel.org \
--cc=masahiroy@kernel.org \
--cc=miguel.ojeda.sandonis@gmail.com \
--cc=natechancellor@gmail.com \
--cc=ndesaulniers@google.com \
--cc=sedat.dilek@gmail.com \
--cc=vincenzo.frascino@arm.com \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.