netdev - Re: [PATCH net] scripts/pahole-flags.sh: Make sure pahole --version works

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Wed, 5 Jan 2022 19:45:53 -0800
From:   Andrii Nakryiko <andrii.nakryiko@...il.com>
To:     Saeed Mahameed <saeed@...nel.org>
Cc:     Daniel Borkmann <daniel@...earbox.net>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        Networking <netdev@...r.kernel.org>, bpf <bpf@...r.kernel.org>,
        Saeed Mahameed <saeedm@...dia.com>,
        Andrii Nakryiko <andrii@...nel.org>,
        Jiri Olsa <jolsa@...hat.com>
Subject: Re: [PATCH net] scripts/pahole-flags.sh: Make sure pahole --version works

On Wed, Jan 5, 2022 at 2:14 PM Saeed Mahameed <saeed@...nel.org> wrote:
>
> On Wed, Jan 05, 2022 at 02:42:01PM +0100, Daniel Borkmann wrote:
> >On 12/31/21 8:56 AM, Saeed Mahameed wrote:
> >>From: Saeed Mahameed <saeedm@...dia.com>
> >>
> >>I had a broken pahole and it's been driving me crazy to see tons of the
> >>following error messages on every build.
> >>
> >>pahole: symbol lookup error: pahole: undefined symbol: btf_gen_floats
> >>scripts/pahole-flags.sh: line 12: [: : integer expression expected
> >>scripts/pahole-flags.sh: line 16: [: : integer expression expected
> >>
> >>Address this by redirecting pahole --version stderr to devnull,
> >>and validate stdout has a non empty string, otherwise exit silently.
> >
> >I'll leave this up to Andrii, but broken pahole version sounds like it would
> >have been better to fix the local pahole installation instead [rather than the
> >kernel having to guard against it, especially if it's driving you crazy]?
> >
>
> Already did :)
>
> >I could image that silent exit on empty version string due to broken pahole
> >deployment might rather waste developer's time to then go and debug why btf
> >wasn't generated..
> >
>
> Good point, I was mainly thinking about developers who are not familiar with btf
> and who have no time debugging irrelevant build issues, but up to you, I
> personally like silent build scripts.
>

Sorry, trying to understand. If you didn't use BTF (and thus
CONFIG_DEBUG_INFO_BTF is not set), is pahole still being called? If
yes, we might want to address that, I suppose.

But if you have a broken pahole that emits something to stderr
(undefined symbol in shared library), then I agree with Daniel that we
shouldn't be working around that in Linux build script.

> >>Fixes: 9741e07ece7c ("kbuild: Unify options for BTF generation for vmlinux and modules")
> >>CC: Andrii Nakryiko <andrii@...nel.org>
> >>CC: Jiri Olsa <jolsa@...hat.com>
> >>Signed-off-by: Saeed Mahameed <saeedm@...dia.com>
> >>---
> >>  scripts/pahole-flags.sh | 3 ++-
> >>  1 file changed, 2 insertions(+), 1 deletion(-)
> >>
> >>diff --git a/scripts/pahole-flags.sh b/scripts/pahole-flags.sh
> >>index e6093adf4c06..b3b53f890d40 100755
> >>--- a/scripts/pahole-flags.sh
> >>+++ b/scripts/pahole-flags.sh
> >>@@ -7,7 +7,8 @@ if ! [ -x "$(command -v ${PAHOLE})" ]; then
> >>      exit 0
> >>  fi
> >>-pahole_ver=$(${PAHOLE} --version | sed -E 's/v([0-9]+)\.([0-9]+)/\1\2/')
> >>+pahole_ver=$(${PAHOLE} --version 2>/dev/null | sed -E 's/v([0-9]+)\.([0-9]+)/\1\2/')
> >>+[ -z "${pahole_ver}" ] && exit 0
> >>  if [ "${pahole_ver}" -ge "118" ] && [ "${pahole_ver}" -le "121" ]; then
> >>      # pahole 1.18 through 1.21 can't handle zero-sized per-CPU vars
> >>
> >