lkml.org 
[lkml]   [2019]   [Aug]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRE: [PATCH v9 5/6] usb:cdns3 Add Cadence USB3 DRD Driver
Date
>
>Hi,
>
>Pawel Laszczak <pawell@cadence.com> writes:
>>>>>>>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.
>
>okay. Just to be sure I understand correctly. If you mask USB_IEN, then
>we would miss a cable disconnect event. Right?
>
>>>>>>>>>> + 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);
>> }
>> ??
>
>right, you can do it right before giving back the request. Or right
>after.
>
>> 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
>
>Did you notice that it doesn't reenable interrupts, though?

I noticed that there is a lack of reenabling interrupts :)

The problem is that If I have disabled interrupt the kernel complains
for using dma_free_coherent function in such place.

Here you have a fragment of complaints:
[ 7420.502863] WARNING: CPU: 0 PID: 10260 at kernel/dma/mapping.c:281 dma_free_attrs+0xa0/0xd0
[ 7420.502866] Modules linked in: usb_f_mass_storage cdns3(OE) cdns3_pci_wrap(OE) libcomposite
...
[ 7420.502965] cdns3_gadget_giveback+0x159/0x2a0 [cdns3]
[ 7420.502975] cdns3_transfer_completed+0xc5/0x3c0 [cdns3]
[ 7420.502986] cdns3_device_thread_irq_handler+0x1b1/0xab0 [cdns3]
[ 7420.502991] ? __schedule+0x333/0x7e0
[ 7420.503001] irq_thread_fn+0x26/0x60
[ 7420.503006] ? irq_thread+0xa8/0x1b0
[ 7420.503011] irq_thread+0x10e/0x1b0
[ 7420.503015] ? irq_forced_thread_fn+0x80/0x80
[ 7420.503021] ? wake_threads_waitq+0x30/0x30
[ 7420.503029] kthread+0x12c/0x150
[ 7420.503034] ? irq_thread_check_affinity+0xe0/0xe0
[ 7420.503038] ? kthread_park+0x90/0x90
[ 7420.503045] ret_from_fork+0x3a/0x50
[ 7420.503061] irq event stamp: 2962
[ 7420.503065] hardirqs last enabled at (2961): [<ffffffffb252672c>] _raw_spin_unlock_irq+0x2c/0x40
[ 7420.503070] hardirqs last disabled at (2962): [<ffffffffb25268f5>] _raw_spin_lock_irqsave+0x25/0x60
[ 7420.503074] softirqs last enabled at (2918): [<ffffffffb2800340>] __do_softirq+0x340/0x451
[ 7420.503079] softirqs last disabled at (2657): [<ffffffffb1aa02b6>] irq_exit+0xc6/0xd0
[ 7420.503082] ---[ end trace d02652af11011c3b ]---

Maybe it's a bug in implementation of this function. I allocate memory with flag GFP_ATOMIC with
disabled interrupt, but I can't free such memory.

--
pawell

\
 
 \ /
  Last update: 2019-08-12 12:28    [W:0.651 / U:0.028 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site