From: Pawel Laszczak <pawell@cadence.com>
To: Felipe Balbi <felipe.balbi@linux.intel.com>,
"devicetree@vger.kernel.org" <devicetree@vger.kernel.org>
Cc: "gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>,
"linux-usb@vger.kernel.org" <linux-usb@vger.kernel.org>,
"hdegoede@redhat.com" <hdegoede@redhat.com>,
"heikki.krogerus@linux.intel.com"
<heikki.krogerus@linux.intel.com>,
"robh+dt@kernel.org" <robh+dt@kernel.org>,
"rogerq@ti.com" <rogerq@ti.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"jbergsagel@ti.com" <jbergsagel@ti.com>,
"nsekhar@ti.com" <nsekhar@ti.com>, "nm@ti.com" <nm@ti.com>,
Suresh Punnoose <sureshp@cadence.com>,
"peter.chen@nxp.com" <peter.chen@nxp.com>,
Jayshri Dajiram Pawar <jpawar@cadence.com>,
Rahul Kumar <kurahul@cadence.com>
Subject: RE: [PATCH v9 5/6] usb:cdns3 Add Cadence USB3 DRD Driver
Date: Mon, 12 Aug 2019 09:13:56 +0000 [thread overview]
Message-ID: <BYAPR07MB4709C07ED94C952886858F14DDD30@BYAPR07MB4709.namprd07.prod.outlook.com> (raw)
In-Reply-To: <87imr2u77c.fsf@gmail.com>
Hi,
>
>Pawel Laszczak <pawell@cadence.com> writes:
>>>> I have such situation in which one interrupt line is shared with ehci and cdns3 driver.
>>>> In such case this function returns error code.
>>>
>>>which function returns error code?
>>
>> devm_request_threaded_irq, of course if I set IRQF_SHARED | IRQF_ONESHOT.
>> As I remember it was EBUSY error.
>
>oh, right. That's probably because the handlers must agree on IRQ flags.
>
>>>> So probably I will need to mask only the reported interrupts.
>>>
>>>you should mask all interrupts from your device, otherwise you top-halt
>>>may still end up reentrant.
>>>
>>>> I can't mask all interrupt using controller register because I can miss some of them.
>>>
>>>why would you miss them? They would be left in the register until you
>>>unmask them and the line is raised again.
>>
>> I consult this with author of controller.
>> We have:
>> USB_IEN and USB_ISTS for generic interrupts
>> EP_IEN and EP_ISTS for endpoint interrupts
>>
>> Both these group works different.
>> For endpoint I can disable all interrupt and I don't miss any of them.
>> So it's normal behavior.
>>
>> But USB_ISTS work little different. If we mask all interrupt in USB_IEN
>> then when new event occurs the EP_ISTS will not be updated.
>
>wait a minute. When you mask USB_ISTS, then EP_ISTS isn't updated? Is
>this a quirk on the controller or a design choice?
>
>> It's not standard and not expected behavior but it works in this way.
>
>Yeah, sounds rather odd.
>
Oh no. My mistake. Of course I mean USB_ISTS.
If we mask all interrupt in USB_IEN
then when new event occurs the USB_ISTS will not be updated.
>>>>>>>> + /* check USB device interrupt */
>>>>>>>> + reg = readl(&priv_dev->regs->usb_ists);
>>>>>>>> +
>>>>>>>> + if (reg) {
>>>>>>>> + writel(reg, &priv_dev->regs->usb_ists);
>>>>>>>> + cdns3_check_usb_interrupt_proceed(priv_dev, reg);
>>>>>>>> + ret = IRQ_HANDLED;
>>>>>>>
>>>>>>>now, because you _don't_ mask this interrupt, you're gonna have
>>>>>>>issues. Say we actually get both device and endpoint interrupts while
>>>>>>>the thread is already running with previous endpoint interrupts. Now
>>>>>>>we're gonna reenter the top half, because device interrupts are *not*
>>>>>>>masked, which will read usb_ists and handle it here.
>>>>>>
>>>>>> Endpoint interrupts are masked in cdns3_device_irq_handler and stay masked
>>>>>> until they are not handled in threaded handler.
>>>>>
>>>>>Quick question, then: these ISTS registers, are they masked interrupt
>>>>>status or raw interrupt status?
>>>>
>>>> Yes it's masked, but after masking them the new interrupts will not be reported
>>>> In ISTS registers. Form this reason I can mask only reported interrupt.
>>>
>>>and what happens when you unmask the registers? Do they get reported?
>>
>> No they are not reported in case of USB_ISTS register.
>> They should be reported in case EP_ISTS, but I need to test it.
>
>okay, please _do_ test and verify the behavior. The description above
>sounds really surprising to me. Does it really mean that if you mask all
>USB_ISTS and then disconnect the cable while interrupt is masked, you
>won't know cable was disconnected?
Yes, exactly.
Initially I've tested it and it's work correct.
I can even simply write 0 to EP_IEN in hard irq and ~0 in thread handler.
It's simplest and sufficient way.
>
>>>>>>>> + struct cdns3_aligned_buf *buf, *tmp;
>>>>>>>> +
>>>>>>>> + list_for_each_entry_safe(buf, tmp, &priv_dev->aligned_buf_list,
>>>>>>>> + list) {
>>>>>>>> + if (!buf->in_use) {
>>>>>>>> + list_del(&buf->list);
>>>>>>>> +
>>>>>>>> + spin_unlock_irqrestore(&priv_dev->lock, flags);
>>>>>>>
>>>>>>>creates the possibility of a race condition
>>>>>> Why? In this place the buf can't be used.
>>>>>
>>>>>but you're reenabling interrupts, right?
>>>>
>>>> Yes, driver frees not used buffers here.
>>>> I think that it's the safest place for this purpose.
>>>
>>>I guess you missed the point a little. Since you reenable interrupts
>>>just to free the buffer, you end up creating the possibility for a race
>>>condition. Specially since you don't mask all interrupt events. The
>>>moment you reenable interrupts, one of your not-unmasked interrupt
>>>sources could trigger, then top-half gets scheduled which tries to wake
>>>up the IRQ thread again and things go boom.
>>
>> Ok, I think I understand. So I have 3 options:
>> 1. Mask the USB_IEN and EP_IEN interrupts, but then I can lost some USB_ISTS
>> events. It's dangerous options.
>
>sure sounds dangerous, but also sounds quite "peculiar" :-)
>
>> 2. Remove implementation of handling unaligned buffers and assume that
>> upper layer will worry about this. What with vendor specific drivers that
>> can be used by companies and not upstreamed ?
>> It could be good to have such safety mechanism even if it is not currently used.
>
>dunno. It may become dead code that's NEVER used :-)
>
>> 3. Delegate this part of code for instance to separate thread that will be called
>> In free time.
>
>Yet another thread? Can't you just run this right before giving back the
>USB request? So, don't do it from IRQ handler, but from giveback path?
Do you mean in:
if (request->complete) {
spin_unlock(&priv_dev->lock);
if (priv_dev->run_garbage_collector) {
....
}
usb_gadget_giveback_request(&priv_ep->endpoint,
request);
spin_lock(&priv_dev->lock);
}
??
I ask because this is finally also called from IRQ handler:
cdns3_device_thread_irq_handler
-> cdns3_check_ep_interrupt_proceed
-> cdns3_transfer_completed
-> cdns3_gadget_giveback
-> usb_gadget_giveback_request
--
Pawell
next prev parent reply other threads:[~2019-08-12 9:16 UTC|newest]
Thread overview: 44+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-07-05 10:57 [PATCH v9 0/6] Introduced new Cadence USBSS DRD Driver Pawel Laszczak
2019-07-05 10:57 ` [PATCH v9 1/6] dt-bindings: add binding for USBSS-DRD controller Pawel Laszczak
2019-07-05 10:57 ` [PATCH v9 2/6] usb:gadget Separated decoding functions from dwc3 driver Pawel Laszczak
2019-07-05 11:27 ` Greg KH
2019-07-05 11:39 ` Pawel Laszczak
2019-07-05 11:49 ` Greg KH
2019-07-05 12:03 ` Pawel Laszczak
2019-07-05 11:39 ` Felipe Balbi
2019-07-05 11:44 ` Pawel Laszczak
2019-08-07 10:17 ` Roger Quadros
2019-08-07 10:22 ` Felipe Balbi
2019-08-08 3:07 ` Pawel Laszczak
2019-07-05 10:57 ` [PATCH v9 3/6] usb:gadget Patch simplify usb_decode_set_clear_feature function Pawel Laszczak
2019-07-05 10:57 ` [PATCH v9 4/6] usb:gadget Simplify usb_decode_get_set_descriptor function Pawel Laszczak
2019-07-05 10:57 ` [PATCH v9 5/6] usb:cdns3 Add Cadence USB3 DRD Driver Pawel Laszczak
2019-07-05 11:55 ` Felipe Balbi
2019-07-07 17:56 ` Pawel Laszczak
2019-07-08 6:29 ` Felipe Balbi
2019-07-08 10:59 ` Pawel Laszczak
2019-07-08 11:15 ` Felipe Balbi
2019-07-08 11:45 ` Pawel Laszczak
2019-07-08 11:59 ` Felipe Balbi
2019-07-08 12:39 ` Pawel Laszczak
2019-07-09 4:23 ` Pawel Laszczak
2019-07-09 6:36 ` Felipe Balbi
2019-07-09 7:07 ` Pawel Laszczak
2019-07-09 7:22 ` Felipe Balbi
2019-07-09 7:29 ` Pawel Laszczak
2019-07-10 8:25 ` Pawel Laszczak
2019-07-08 7:11 ` Felipe Balbi
2019-07-15 11:00 ` Pawel Laszczak
2019-08-07 10:34 ` Felipe Balbi
2019-08-10 20:39 ` Pawel Laszczak
2019-08-12 1:55 ` Peter Chen
2019-08-12 4:43 ` Pawel Laszczak
2019-08-12 5:24 ` Felipe Balbi
2019-08-12 7:13 ` Pawel Laszczak
2019-08-12 8:19 ` Felipe Balbi
2019-08-12 9:13 ` Pawel Laszczak [this message]
2019-08-12 9:45 ` Felipe Balbi
2019-08-12 10:26 ` Pawel Laszczak
2019-08-12 10:34 ` Felipe Balbi
2019-08-12 14:20 ` Alan Stern
2019-07-05 10:57 ` [PATCH v9 6/6] usb:cdns3 Fix for stuck packets in on-chip OUT buffer Pawel Laszczak
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=BYAPR07MB4709C07ED94C952886858F14DDD30@BYAPR07MB4709.namprd07.prod.outlook.com \
--to=pawell@cadence.com \
--cc=devicetree@vger.kernel.org \
--cc=felipe.balbi@linux.intel.com \
--cc=gregkh@linuxfoundation.org \
--cc=hdegoede@redhat.com \
--cc=heikki.krogerus@linux.intel.com \
--cc=jbergsagel@ti.com \
--cc=jpawar@cadence.com \
--cc=kurahul@cadence.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=nm@ti.com \
--cc=nsekhar@ti.com \
--cc=peter.chen@nxp.com \
--cc=robh+dt@kernel.org \
--cc=rogerq@ti.com \
--cc=sureshp@cadence.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).