Received-SPF: pass (google.com: domain of linux-kernel+bounces-10343-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Message-ID: <61d4f3f6-dc91-449a-9666-67f1eaca414c@quicinc.com>
Date: Sat, 23 Dec 2023 14:02:50 +0530
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH v2] usb: dwc3: gadget: Ignore dwc3 interrupt if GEVNTCOUNT
 reads corrupt value
To: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
CC: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-usb@vger.kernel.org" <linux-usb@vger.kernel.org>,
        "quic_ppratap@quicinc.com" <quic_ppratap@quicinc.com>,
        "quic_jackp@quicinc.com" <quic_jackp@quicinc.com>
References: <20231221151620.16030-1-quic_kriskura@quicinc.com>
 <20231222221907.l7s5olt6uubwfjlh@synopsys.com>
Content-Language: en-US
From: Krishna Kurapati PSSNV <quic_kriskura@quicinc.com>
In-Reply-To: <20231222221907.l7s5olt6uubwfjlh@synopsys.com>
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit


On 12/23/2023 3:49 AM, Thinh Nguyen wrote:
> On Thu, Dec 21, 2023, Krishna Kurapati wrote:
>> In the current implementation, the check_event_buf call reads the
>> GEVNTCOUNT register to determine the amount of event data generated
>> and copies it from ev->buf to ev->cache after masking interrupt.
>>
>> During copy if the amount of data to be copied is more than
>> (length - lpos), we fill the ev->cache till the end of 4096 byte
>> buffer allocated and then start filling from the top (lpos = 0).
>>
>> In one instance of SMMU crash it is observed that GEVNTCOUNT register
>> reads more than 4096 bytes:
>>
>> dwc3_readl   base=0xffffffc0091dc000  offset=50188  value=63488
>>
>> (offset = 50188 -> 0xC40C)  -> reads 63488 bytes
>>
>> As per crash dump:
>> dwc->lpos = 2056
>>
>> and evt->buf is at 0xFFFFFFC009185000 and the crash is at
>> 0xFFFFFFC009186000. The diff which is exactly 0x1000 bytes.
>>
>> We first memcpy 2040 bytes from (buf + lpos) to (buf + 0x1000).
>>
>> And then we copy the rest of the data (64388 - 2040) from beginning
>> of dwc->ev_buf. While doing so we go beyond bounds as we are trying
>> to memcpy 62348 bytes into a 4K buffer resulting in crash.
>>
>> Fix this by ignoring the interrupt when GEVNTCOUNT register reads a
>> value more than the event ring allocated.
>>
>> Signed-off-by: Krishna Kurapati <quic_kriskura@quicinc.com>
>> ---
>> Changes in v2:
>> Instead of fixing amount of data being copied from ring, ignored
>> the interrupt when count is corrupt as per suggestion from Thinh.
>>
>> Link to v1:
>> https://urldefense.com/v3/__https://lore.kernel.org/all/20230521100330.22478-1-quic_kriskura@quicinc.com/__;!!A4F2R9G_pg!ewM3u9Pdk8yD9eU24sOuQqmhm8M2VpGXH8zALqVWGKffGbcJxrtyKKlUPuh8SS2arqO09DjnC9atC3bemEpx-g5UQMllbA$
>>
>>   drivers/usb/dwc3/gadget.c | 10 +++++++++-
>>   1 file changed, 9 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
>> index 858fe4c299b7..e27933fdcce3 100644
>> --- a/drivers/usb/dwc3/gadget.c
>> +++ b/drivers/usb/dwc3/gadget.c
>> @@ -4451,6 +4451,7 @@ static irqreturn_t dwc3_thread_interrupt(int irq, void *_evt)
>>   static irqreturn_t dwc3_check_event_buf(struct dwc3_event_buffer *evt)
>>   {
>>   	struct dwc3 *dwc = evt->dwc;
>> +	int ret = IRQ_WAKE_THREAD;
> 
> irqreturn_t instead of int?
> 
>>   	u32 amount;
>>   	u32 count;
>>   
>> @@ -4480,6 +4481,12 @@ static irqreturn_t dwc3_check_event_buf(struct dwc3_event_buffer *evt)
>>   	if (!count)
>>   		return IRQ_NONE;
>>   
>> +	if (count > evt->length) {
>> +		dev_err(dwc->dev, "GEVTCOUNT corrupt\n");
>> +		ret = IRQ_NONE;
>> +		goto done;
>> +	}
>> +
>>   	evt->count = count;
>>   	evt->flags |= DWC3_EVENT_PENDING;
>>   
>> @@ -4493,9 +4500,10 @@ static irqreturn_t dwc3_check_event_buf(struct dwc3_event_buffer *evt)
>>   	if (amount < count)
>>   		memcpy(evt->cache, evt->buf, count - amount);
>>   
>> +done:
>>   	dwc3_writel(dwc->regs, DWC3_GEVNTCOUNT(0), count);
> 
> Don't update the GEVNTCOUNT if the read is invalid. We're not processing
> any event, so we should not write back the "count" that we did not
> process.
> 
>>   
Thanks for the review Thinh.

If we don't update, won't the register always be non-zero ? It will keep 
triggering the dwc3_interrupt unnecessarily right ?

Regards,
Krishna,