Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:3 as permitted sender) client-ip=2620:137:e000::3:3;
Message-ID: <79793717-95f8-2bd7-407a-ae547131809d@redhat.com>
Date:   Fri, 22 Sep 2023 20:29:18 -0400
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.14.0
Subject: Re: [PATCH v2] locking/semaphore: Use wake_q to wake up processes
 outside lock critical section
Content-Language: en-US
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     Ingo Molnar <mingo@redhat.com>, Will Deacon <will.deacon@arm.com>,
        Boqun Feng <boqun.feng@gmail.com>, linux-kernel@vger.kernel.org
References: <20220909192848.963982-1-longman@redhat.com>
 <20230921074247.GC14803@noisy.programming.kicks-ass.net>
 <06163c13-03b2-bfe0-90b4-5267a039a02c@redhat.com>
 <20230922194704.GA40277@noisy.programming.kicks-ass.net>
From:   Waiman Long <longman@redhat.com>
In-Reply-To: <20230922194704.GA40277@noisy.programming.kicks-ass.net>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 9/22/23 15:47, Peter Zijlstra wrote:
> On Fri, Sep 22, 2023 at 02:45:04PM -0400, Waiman Long wrote:
>
>> I believe early_printk should only be used in the non-SMP boot process. The
>> use of printk() is frequently used for debugging purpose and the insertion
>> of printk at some lock critical section can cause the lockdep splat to come
>> out obscuring the debugging process.
> By default early_printk is disabled somewhere early, but it has a keep
> argument to keep it around.
>
> Anyway, printk() as it exists today is wholly unsuited for debugging.
> There are too many contexts where it will flat out not work. When you
> use early_print with keep then you can use early_printk() instead of
> printk() to debug.
>
> Also, see the patches I pointed John at. Perf would not be what it is
> without those patches.
>
> Serial lines and early printk are not optional. That is, I flat out
> refuse to develop on machines without them.

Thanks for the debugging tip.

BTW, it is not just printk() that can be problematic in some contexts. I 
believe the various WARN*() calls cause the same kind of lockdep problem 
even though these WARN() calls shouldn't be triggered that often

Cheers,
Longman