Re: [syzbot] WARNING in bpf_test_run

From: Steven Rostedt <rostedt@goodmis.org>
To: Dmitry Vyukov <dvyukov@google.com>
Cc: Yonghong Song <yhs@fb.com>,
	syzbot <syzbot+774c590240616eaa3423@syzkaller.appspotmail.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	andrii@kernel.org, Alexei Starovoitov <ast@kernel.org>,
	Borislav Petkov <bp@alien8.de>, bpf <bpf@vger.kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	David Miller <davem@davemloft.net>,
	Jesper Dangaard Brouer <hawk@kernel.org>,
	"H. Peter Anvin" <hpa@zytor.com>,
	Jim Mattson <jmattson@google.com>,
	John Fastabend <john.fastabend@gmail.com>,
	Joerg Roedel <joro@8bytes.org>, Martin KaFai Lau <kafai@fb.com>,
	kpsingh@kernel.org, Jakub Kicinski <kuba@kernel.org>,
	KVM list <kvm@vger.kernel.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Mark Rutland <mark.rutland@arm.com>,
	masahiroy@kernel.org, Ingo Molnar <mingo@redhat.com>,
	netdev <netdev@vger.kernel.org>,
	Paolo Bonzini <pbonzini@redhat.com>,
	Peter Zijlstra <peterz@infradead.org>,
	rafael.j.wysocki@intel.com,
	Sean Christopherson <seanjc@google.com>,
	Song Liu <songliubraving@fb.com>,
	syzkaller-bugs <syzkaller-bugs@googlegroups.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	vkuznets@redhat.com, wanpengli@tencent.com, will@kernel.org,
	x86@kernel.org
Subject: Re: [syzbot] WARNING in bpf_test_run
Date: Tue, 13 Apr 2021 08:55:22 -0400	[thread overview]
Message-ID: <20210413085522.2caee809@gandalf.local.home> (raw)
In-Reply-To: <CACT4Y+ZYEVsycyzDW9+tXYw-5feZS8otgMWGGZRUCLR=czWtqQ@mail.gmail.com>

On Tue, 13 Apr 2021 09:56:40 +0200
Dmitry Vyukov <dvyukov@google.com> wrote:

> Thanks for looking into this.
> If this is not a kernel bug, then it must not use WARN_ON[_ONCE]. It
> makes the kernel untestable for both automated systems and humans:
> 
> https://lwn.net/Articles/769365/
> 
> <quote>
> Greg Kroah-Hartman raised the problem of core kernel API code that
> will use WARN_ON_ONCE() to complain about bad usage; that will not
> generate the desired result if WARN_ON_ONCE() is configured to crash
> the machine. He was told that the code should just call pr_warn()
> instead, and that the called function should return an error in such
> situations. It was generally agreed that any WARN_ON() or
> WARN_ON_ONCE() calls that can be triggered from user space need to be
> fixed.
> </quote>

I agree. WARN_ON(_ONCE) should be reserved for anomalies that should not
happen ever. Anything that the user could trigger, should not trigger a
WARN_ON.

A WARN_ON is perfectly fine for detecting an accounting error inside the
kernel. I have them scattered all over my code, but they should never be
hit, even if something in user space tries to hit it. (with an exception of
an interface I want to deprecate, where I want to know if it's still being
used ;-) Of course, that wouldn't help bots testing the code. And I haven't
done that in years)

Any anomaly that can be triggered by user space doing something it should
not be doing really needs a pr_warn().

Thanks,

-- Steve