Message-ID: <533D06F2.7080702@freescale.com>
Date: Thu, 3 Apr 2014 15:00:02 +0800
From: Hongbo Zhang <hongbo.zhang@freescale.com>
User-Agent: Mozilla/5.0 (X11; Linux i686; rv:24.0) Gecko/20100101 Thunderbird/24.1.1
MIME-Version: 1.0
To: Vinod Koul <vinod.koul@intel.com>
CC: Vinod Koul <vkoul@infradead.org>, <dan.j.williams@intel.com>,
        <dmaengine@vger.kernel.org>, <scottwood@freescale.com>,
        <LeoLi@freescale.com>, <linuxppc-dev@lists.ozlabs.org>,
        <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 6/7] DMA: Freescale: use spin_lock_bh instead of spin_lock_irqsave
References: <1389851246-8564-1-git-send-email-hongbo.zhang@freescale.com> <1389851246-8564-7-git-send-email-hongbo.zhang@freescale.com> <1395817294.6569.2.camel@vkoul-udesk3> <533517C1.8080208@freescale.com> <20140329134528.GQ1976@intel.com> <5338EA57.1000509@freescale.com> <20140402163519.GU1976@intel.com>
In-Reply-To: <20140402163519.GU1976@intel.com>
Content-Type: text/plain; charset="ISO-8859-1"; format=flowed
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org


On 04/03/2014 12:35 AM, Vinod Koul wrote:
> On Mon, Mar 31, 2014 at 12:08:55PM +0800, Hongbo Zhang wrote:
>> On 03/29/2014 09:45 PM, Vinod Koul wrote:
>>> On Fri, Mar 28, 2014 at 02:33:37PM +0800, Hongbo Zhang wrote:
>>>> On 03/26/2014 03:01 PM, Vinod Koul wrote:
>>>>> On Thu, 2014-01-16 at 13:47 +0800, hongbo.zhang@freescale.com wrote:
>>>>>> From: Hongbo Zhang <hongbo.zhang@freescale.com>
>>>>>>
>>>>>> The usage of spin_lock_irqsave() is a stronger locking mechanism than is
>>>>>> required throughout the driver. The minimum locking required should be used
>>>>>> instead. Interrupts will be turned off and context will be saved, it is
>>>>>> unnecessary to use irqsave.
>>>>>>
>>>>>> This patch changes all instances of spin_lock_irqsave() to spin_lock_bh(). All
>>>>>> manipulation of protected fields is done using tasklet context or weaker, which
>>>>>> makes spin_lock_bh() the correct choice.
>>>>>>
>>>>>>   /**
>>>>>> @@ -1124,11 +1120,10 @@ static irqreturn_t fsldma_chan_irq(int irq, void *data)
>>>>>>   static void dma_do_tasklet(unsigned long data)
>>>>>>   {
>>>>>>   	struct fsldma_chan *chan = (struct fsldma_chan *)data;
>>>>>> -	unsigned long flags;
>>>>>>   	chan_dbg(chan, "tasklet entry\n");
>>>>>> -	spin_lock_irqsave(&chan->desc_lock, flags);
>>>>>> +	spin_lock_bh(&chan->desc_lock);
>>>>> okay here is the problem :(
>>>>>
>>>>> You moved to _bh variant. So if you grab the lock in rest of the code
>>>>> and irq gets triggered then here we will be spinning to grab the lock.
>>>>> So effectively you made right locking solution into deadlock situation!
>>>> If the rest code grabs lock by spin_lock_bh(), and if irq raised,
>>>> the tasklet could not be executed because it has been disabled by
>>>> the _bh variant function.
>>> yes if you are accessing resources only in tasklet and rest of the code, then
>>> _bh variant works well. The problem here is usage in irq handler
>>>
>> The name dma_do_tasklet may mislead, it is tasklet handler, not irq
>> handler, not a trigger to load tasklet.
>> the irq handler is fsldma_chan_irq, and I don't use lock in it.
> sorry my bad, i misread this as code in fsldma_chan_irq() insteadof
> dma_do_tasklet(). In that case patch is doing the right thing.
>

OK, so I will send a v2 series with only updating 3/7 soon.


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/