Re: [PATCH v2 bpf-next 0/6] BPF verifier rotating log

From: Alexei Starovoitov <alexei.starovoitov@gmail.com>
To: Andrii Nakryiko <andrii.nakryiko@gmail.com>
Cc: Andrii Nakryiko <andrii@kernel.org>, bpf <bpf@vger.kernel.org>,
	Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Martin KaFai Lau <martin.lau@kernel.org>,
	lmb@isovalent.com, Timo Beckers <timo@incline.eu>,
	robin.goegge@isovalent.com, Kernel Team <kernel-team@meta.com>
Subject: Re: [PATCH v2 bpf-next 0/6] BPF verifier rotating log
Date: Wed, 29 Mar 2023 22:38:53 -0700	[thread overview]
Message-ID: <CAADnVQKT7Hifb=vV1yu8orgPMkRynNLZykCPKanRqDigE8xVEA@mail.gmail.com> (raw)
In-Reply-To: <CAEf4BzaAcD0HEgJzQH4NTWAzkTXHLS7T-eGGxxhHm2ADROTRrg@mail.gmail.com>

On Tue, Mar 28, 2023 at 4:59 PM Andrii Nakryiko
<andrii.nakryiko@gmail.com> wrote:
>
> On Tue, Mar 28, 2023 at 4:56 PM Andrii Nakryiko <andrii@kernel.org> wrote:
> >
> > This patch set changes BPF verifier log behavior to behave as a rotating log,
> > by default. If user-supplied log buffer is big enough to contain entire
> > verifier log output, there is no effective difference. But where previously
> > user supplied too small log buffer and would get -ENOSPC error result and the
> > beginning part of the verifier log, now there will be no error and user will
> > get ending part of verifier log filling up user-supplied log buffer.  Which
> > is, in absolute majority of cases, is exactly what's useful, relevant, and
> > what users want and need, as the ending of the verifier log is containing
> > details of verifier failure and relevant state that got us to that failure. So
> > this rotating mode is made default, but for some niche advanced debugging
> > scenarios it's possible to request old behavior by specifying additional
> > BPF_LOG_FIXED (8) flag.
> >
> > This patch set adjusts libbpf to allow specifying flags beyond 1 | 2 | 4. We
> > also add --log-size and --log-fixed options to veristat to be able to both
> > test this functionality manually, but also to be used in various debugging
> > scenarios. We also add selftests that tries many variants of log buffer size
> > to stress-test correctness of internal verifier log bookkeeping code.
> >
> > v1->v2:
> >   - return -ENOSPC even in rotating log mode for preserving backwards
> >     compatibility (Lorenz);
>
> I haven't implemented the feature we discussed, where the
> BPF_PROG_LOAD (and BPF_BTF_LOAD) command will return back the full
> size of the buffer that's necessary to contain the complete log
> buffer. I'm building it on top of this patch set and would like to
> send it separately as a follow up, as it touches UAPI some more, and I
> feel like we'll have few revisions just for this. So I didn't want to
> delay these changes. Plus, I think to add this even for BPF_LOG_FIXED
> mode, so it's provided consistently. So I need a bit more time to
> implement this. Hopefully this version will work for everyone and can
> go in sooner.

Lorenz, ping.