Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <9075b67b-9290-8d31-abe5-3755d4c93d97@linux.intel.com>
Date:   Wed, 15 Jun 2022 17:32:39 +0300
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Firefox/91.0 Thunderbird/91.8.1
Content-Language: en-US
To:     Hongyu Xie <xy521521@gmail.com>, mathias.nyman@intel.com,
        gregkh@linuxfoundation.org
Cc:     linux-usb@vger.kernel.org, linux-kernel@vger.kernel.org,
        125707942@qq.com, wangqi@kylinos.cn, xiongxin@kylinos.cn,
        Hongyu Xie <xiehongyu1@kylinos.cn>
References: <20220615090703.1812560-1-xiehongyu1@kylinos.cn>
From:   Mathias Nyman <mathias.nyman@linux.intel.com>
Subject: Re: [PATCH v2 -next] usb: xhci: disable irq during initialization
In-Reply-To: <20220615090703.1812560-1-xiehongyu1@kylinos.cn>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 15.6.2022 12.07, Hongyu Xie wrote:
> irq is disabled in xhci_quiesce(called by xhci_halt, with bit:2 cleared
> in USBCMD register), but xhci_run(called by usb_add_hcd) re-enable it.
> It's possible that you will receive thousands of interrupt requests
> after initialization for 2.0 roothub. And you will get a lot of
> warning like, "xHCI dying, ignoring interrupt. Shouldn't IRQs be
> disabled?". This amount of interrupt requests will cause the entire
> system to freeze.
> This problem was first found on a device with ASM2142 host controller
> on it
I see, makes sense to enable interrupts as close as possible to where we
start the host.

For some unknown reason xHCI specification recommends enabling interrupts
(INTE and IE bits) before setting the Run/Stop bit, see xhci 4.2
Host Controller Initialization, "Note" part

So the order here needs to be swapped to comply with spec.
This unfortunately also creates a small window where xHC can interrupt
before xHC is running, but it's better than current situation.

This small window could be solved by protecting this part with a
spin_lock_irqsave().

Thanks
-Mathias