Received-SPF: pass (google.com: domain of linux-kernel+bounces-185449-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Message-ID: <93ae2776-a661-40fc-86da-cfb9a39d9fce@quicinc.com>
Date: Tue, 21 May 2024 14:08:58 -0700
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH 6/7] remoteproc: qcom_q6v5_pas: Add hwspinlock bust on
 stop
To: Konrad Dybcio <konrad.dybcio@linaro.org>,
        Bjorn Andersson
	<andersson@kernel.org>,
        Baolin Wang <baolin.wang@linux.alibaba.com>,
        Peter
 Zijlstra <peterz@infradead.org>, Ingo Molnar <mingo@redhat.com>,
        Will Deacon
	<will@kernel.org>, Waiman Long <longman@redhat.com>,
        Boqun Feng
	<boqun.feng@gmail.com>, Jonathan Corbet <corbet@lwn.net>,
        Mathieu Poirier
	<mathieu.poirier@linaro.org>,
        Rob Herring <robh@kernel.org>,
        Krzysztof
 Kozlowski <krzk+dt@kernel.org>,
        Conor Dooley <conor+dt@kernel.org>,
        Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
CC: <linux-remoteproc@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
        <linux-doc@vger.kernel.org>, <linux-arm-msm@vger.kernel.org>,
        <devicetree@vger.kernel.org>, Richard Maina <quic_rmaina@quicinc.com>
References: <20240516-hwspinlock-bust-v1-0-47a90a859238@quicinc.com>
 <20240516-hwspinlock-bust-v1-6-47a90a859238@quicinc.com>
 <722c57dc-98cf-4ec8-93c5-aef55e6084c7@linaro.org>
Content-Language: en-US
From: Chris Lew <quic_clew@quicinc.com>
In-Reply-To: <722c57dc-98cf-4ec8-93c5-aef55e6084c7@linaro.org>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 8bit


On 5/21/2024 10:38 AM, Konrad Dybcio wrote:
> 
> 
> On 5/17/24 00:58, Chris Lew wrote:
>> From: Richard Maina <quic_rmaina@quicinc.com>
>>
>> When remoteproc goes down unexpectedly this results in a state where any
>> acquired hwspinlocks will remain locked possibly resulting in deadlock.
>> In order to ensure all locks are freed we include a call to
>> hwspin_lock_bust() during remoteproc shutdown.
>>
>> For qcom_q6v5_pas remoteprocs, each remoteproc has an assigned id that
>> is used to take the hwspinlock. Remoteproc should use this id to try and
>> bust the lock on remoteproc stop.
>>
>> This edge case only occurs with q6v5_pas watchdog crashes. The error
>> fatal case has handling to clear the hwspinlock before the error fatal
>> interrupt is triggered.
>>
>> Signed-off-by: Richard Maina <quic_rmaina@quicinc.com>
>> Signed-off-by: Chris Lew <quic_clew@quicinc.com>
>> ---
> 
> 
>>   drivers/remoteproc/qcom_q6v5_pas.c | 28 ++++++++++++++++++++++++++++
>>   1 file changed, 28 insertions(+)
>>
>> diff --git a/drivers/remoteproc/qcom_q6v5_pas.c 
>> b/drivers/remoteproc/qcom_q6v5_pas.c
>> index 54d8005d40a3..57178fcb9aa3 100644
>> --- a/drivers/remoteproc/qcom_q6v5_pas.c
>> +++ b/drivers/remoteproc/qcom_q6v5_pas.c
>> @@ -10,6 +10,7 @@
>>   #include <linux/clk.h>
>>   #include <linux/delay.h>
>>   #include <linux/firmware.h>
>> +#include <linux/hwspinlock.h>
>>   #include <linux/interrupt.h>
>>   #include <linux/kernel.h>
>>   #include <linux/module.h>
>> @@ -52,6 +53,7 @@ struct adsp_data {
>>       const char *ssr_name;
>>       const char *sysmon_name;
>>       int ssctl_id;
>> +    int hwlock_id;
>>       int region_assign_idx;
>>       int region_assign_count;
>> @@ -84,6 +86,9 @@ struct qcom_adsp {
>>       bool decrypt_shutdown;
>>       const char *info_name;
>> +    struct hwspinlock *hwlock;
>> +    int hwlock_id;
> 
> IIRC, this is the same one that is passed in the DT.
> 
> Can we get it dynamically from there?
> 

The argument passed in DT is the index of the hwlock in the TCSR mutex 
region. The index determines use of hwlock[0..n]

This id is supposed to be the identifier that is passed into 
hwspin_lock_bust(). The actual value that we would read from 
hwlock[0..n] to see if we need to bust the lock.

Maybe the naming of this variable is confusing. Do you have any 
suggestions to make it clearer? Could call it hwlock_bust_id.

We could also try increasing the #hwlock-cells to 2 and have something 
like <&phandle index bust_id>. To me this seemed odd for clients that 
weren't planning on using the bust_id.

> Konrad