MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII;
 format=flowed
Content-Transfer-Encoding: 7bit
Date: Wed, 29 Mar 2017 21:12:23 -0700
From: Vikram Mulukutla <markivx@codeaurora.org>
To: Sudeep Holla <sudeep.holla@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>,
        Will Deacon <will.deacon@arm.com>, linux-kernel@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org,
        linux-kernel-owner@vger.kernel.org
Subject: Re: spin_lock behavior with ARM64 big.Little/HMP
In-Reply-To: <f17207e7-e58c-2e42-04a8-50ed83a9616f@arm.com>
References: <400ab4b8b2354c5b9283f6ed657363a0@codeaurora.org>
 <8d9d6333-0ebe-65c4-c6f1-3e3475e3e535@arm.com>
 <f9fc9549d801fece7422d97d5d89df8b@codeaurora.org>
 <f17207e7-e58c-2e42-04a8-50ed83a9616f@arm.com>
Message-ID: <537888bcf5dd5ea3c3418fc9e9e3ff79@codeaurora.org>
User-Agent: Roundcube Webmail/1.2.1
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 702
Lines: 21


Hi Sudeep,

> 
> Interesting. Just curious if this is r0p0/p1 A53 ? If so, is the errata
> 819472 enabled ?

Sorry for bringing this up after the loo-ong delay, but I've been 
assured that the A53 involved is > r0p1. I've also confirmed this 
problem on multiple internal platforms, and I'm pretty sure that it 
occurs on any b.L out there today. Also, we found the same problematic 
lock design used in the workqueue code in the kernel, causing the same 
livelock. It's very very rare and requires a perfect set of 
circumstances.

If it would help I can provide a unit test if you folks would be 
generous enough to test it on the latest Juno or something b.L that's 
also upstream.

Thanks,
Vikram