Re: [PATCH v10 0/6] Re-introduce TX FIFO resize for larger EP bursting

From: Pavel Hofman
Date: Mon Aug 23 2021 - 11:01:26 EST

Next message: Soeren Moch: "Re: [Regression 5.14] media: dvb userspace api"
Previous message: Vlastimil Babka: "[PATCH v5 33/35] mm, slub: protect put_cpu_partial() with disabled irqs instead of cmpxchg"
In reply to: Ferry Toth: "Re: [PATCH v10 0/6] Re-introduce TX FIFO resize for larger EP bursting"
Next in thread: Andy Shevchenko: "Re: [PATCH v10 0/6] Re-introduce TX FIFO resize for larger EP bursting"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Dne 22. 08. 21 v 21:43 Ferry Toth napsal(a):

Op 19-08-2021 om 23:04 schreef Pavel Hofman:

Dne 19. 08. 21 v 22:10 Ferry Toth napsal(a):

Hi

Op 19-08-2021 om 09:51 schreef Pavel Hofman:

Hi,

Dne 18. 08. 21 v 21:07 Ferry Toth napsal(a):

Hi,

Op 18-08-2021 om 00:00 schreef Ferry Toth:

Hi,

Op 16-08-2021 om 07:18 schreef Felipe Balbi:

Hi,

Ferry Toth <fntoth@xxxxxxxxx> writes:

Ferry Toth <fntoth@xxxxxxxxx> writes:

Ferry Toth <fntoth@xxxxxxxxx> writes:

Hardware name: Intel Corporation Merrifield/BODEGA BAY, BIOS 542
2015.01.21:18.19.48
RIP: 0010:0x500000000
Code: Unable to access opcode bytes at RIP 0x4ffffffd6.
RSP: 0018:ffffa4d00045fc28 EFLAGS: 00010046
RAX: 0000000500000000 RBX: ffff8cd546aed200 RCX: 0000000000000000
RDX: 0000000000000000 RSI: ffff8cd547bfcae0 RDI: ffff8cd546aed200
RBP: ffff8cd547bfcae0 R08: 0000000000000000 R09: 0000000000000001
R10: ffff8cd541fd28c0 R11: 0000000000000000 R12: ffff8cd547342828
R13: ffff8cd546aed248 R14: 0000000000000000 R15: ffff8cd548b1d000
FS: 0000000000000000(0000) GS:ffff8cd57e200000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000500000000 CR3: 000000000311e000 CR4: 00000000001006f0
Call Trace:
        ? dwc3_remove_requests.constprop.0+0x14d/0x170
        ? __dwc3_gadget_ep_disable+0x7a/0x160
        ? dwc3_gadget_ep_disable+0x3d/0xd0
        ? usb_ep_disable+0x1c/0x
        ? u_audio_stop_capture+0x79/0x120 [u_audio]
        ? afunc_set_alt+0x73/0x80 [usb_f_uac2]

So this is triggered by a SetInterface request...

        ? composite_setup+0x224/0x1b90 [libcomposite]
        ? __dwc3_gadget_kick_transfer+0x160/0x400
        ? dwc3_gadget_ep_queue+0xf3/0x1a0
        ? configfs_composite_setup+0x6b/0x90 [libcomposite]
        ? configfs_composite_setup+0x6b/0x90 [libcomposite]
        ? dwc3_ep0_interrupt+0x459/0xa40
        ? dwc3_thread_interrupt+0x8ee/0xf40
        ? __schedule+0x235/0x6c0
        ? disable_irq_nosync+0x10/0x10
        ? irq_thread_fn+0x1b/0x60
        ? irq_thread+0xc0/0x160
        ? irq_thread_check_affinity+0x70/0x70
        ? irq_forced_thread_fn+0x70/0x70
        ? kthread+0x122/0x140
        ? set_kthread_struct+0x40/0x40
        ? ret_from_fork+0x22/0x30

Do you mind enabling dwc3 traces and collecting them? Trying to figure
out how we got here.

I'll try if I can get the same error by booting with USB in host mode
and then switch to device mode. If so I can enable traces and collect as
you explained me before.

I'll try before monday, as then I fly for a holiday and will not be
available before rc5.

you can enable all of those with kernel cmdline :-)

https://www.kernel.org/doc/html/latest/admin-guide/kernel-parameters.html

you need ftrace_dump_on_oops=1 and also need the correct options on
trace_buf_size and trace_event.

On Edison-Arduino I have a switch to go to device mode, after which
udev triggers a script configure gadgets through configfs.

I tried to log following these instructions:

https://www.kernel.org/doc/html/latest/driver-api/usb/dwc3.html#reporting-bugs <https://www.kernel.org/doc/html/latest/driver-api/usb/dwc3.html#reporting-bugs>

Unfortunately the kernel crashes so badly I can not get to the ` cp
/t/trace /root/trace.txt` line (after a while the watchdog kicks).

What to do next?

Pass ftrace_dump_on_oops to kernel cmdline.

No sure if I did this right, on oops everything is pushed to console
(115k2 serial), I hope nothing essential is lost.

I copied the screen buffer to file see attached.

Thank you, I bet it took quite a some time :-) Anyway, looking at
the logs around Set Interface requests, we can track every endpoint
that's disabled. I'll take a guess and assume we're failing at the last
Set Interface, that means we should have something odd with ep6in, but
everything looks fine in the trace output:

[   75.823107] irq/14-d-596       0d... 42789194us : dwc3_gadget_ep_enable: ep6in: mps 192/346 streams 16 burst 0 ring 0/0 flags E:swbp:<
[   75.835472] irq/14-d-596       0d... 42789198us : dwc3_alloc_request: ep6in: req 0000000002c71409 length 0/0 zsI ==> 0
[   75.846416] irq/14-d-596       0d... 42789202us : dwc3_ep_queue: ep6in: req 0000000002c71409 length 0/192 zsI ==> -115
[   75.857360] irq/14-d-596       0d... 42789204us : dwc3_alloc_request: ep6in: req 00000000a324f5d0 length 0/0 zsI ==> 0
[   75.868301] irq/14-d-596       0d... 42789206us : dwc3_ep_queue: ep6in: req 00000000a324f5d0 length 0/192 zsI ==> -115
[   75.879244] irq/14-d-596       0d... 42789209us : dwc3_event: event (000020c2): ep0in: Transfer Not Ready [0] (Not Active) [Status Phase]
[   75.891880] irq/14-d-596       0d... 42789211us : dwc3_prepare_trb: ep0in: trb 000000004c0ae319 (E0:D0) buf 000000001bded000 size 0 ctrl 00000c33 (HLcs:SC:status2)
[   75.989131] irq/14-d-596       0d... 42789224us : dwc3_gadget_ep_cmd: ep0in: cmd 'Start Transfer' [406] params 00000000 1bded000 00000000 --> status: Successful
[   76.096261] irq/14-d-596       0d... 42789272us : dwc3_event: event (0000c042): ep0in: Transfer Complete (sIL) [Status Phase]
[   76.107834] irq/14-d-596       0d... 42789275us : dwc3_complete_trb: ep0out: trb 000000004c0ae319 (E0:D0) buf 000000001bded000 size 0 ctrl 00000c32 (hLcs:SC:status2)
[   76.122944] irq/14-d-596       0d... 42789277us : dwc3_gadget_giveback: ep0out: req 00000000cb1bd3cd length 0/0 zsI ==> 0
[   76.134160] irq/14-d-596       0d... 42789280us : dwc3_prepare_trb: ep0out: trb 000000004c0ae319 (E0:D0) buf 000000001bded000 size 8 ctrl 00000c23 (HLcs:SC:setup)
[   76.231322] irq/14-d-596       0d... 42789292us : dwc3_gadget_ep_cmd: ep0out: cmd 'Start Transfer' [406] params 00000000 1bded000 00000000 --> status: Successful
[   76.297418] kworker/-23        0d... 42789670us : dwc3_ep_queue: ep3in: req 0000000029586135 length 0/96 ZsI ==> -115
[   76.308278] kworker/-23        0d... 42789695us : dwc3_prepare_trb: ep3in: trb 00000000b81213d6 (E1:D0) buf 0000000003b7a800 size 96 ctrl 00000811 (Hlcs:sC:normal)
[   76.395294] kworker/-23        0d... 42789707us : dwc3_gadget_ep_cmd: ep3in: cmd 'Update Transfer' [60007] params 00000000 00000000 00000000 --> status: Successful
[   76.471900] irq/14-d-596       0d... 42789842us : dwc3_event: event (0000c040): ep0out: Transfer Complete (sIL) [Setup Phase]
[   76.489308] irq/14-d-596       0d... 42789845us : dwc3_ctrl_req: Set Interface(Intf = 5, Alt.Setting = 0)
[   76.505650] irq/14-d-596       0d... 42789851us : dwc3_ep_dequeue: ep6in: req 0000000002c71409 length 0/192 zsI ==> -115
[   76.523315] irq/14-d-596       0d... 42789854us : dwc3_gadget_giveback: ep6in: req 0000000002c71409 length 0/192 zsI ==> -104
[   76.541427] irq/14-d-596       0d... 42789857us : dwc3_free_request: ep6in: req 0000000002c71409 length 0/192 zsI ==> -104
[   76.559267] irq/14-d-596       0d... 42789859us : dwc3_ep_dequeue: ep6in: req 00000000a324f5d0 length 0/192 zsI ==> -115
[   76.576937] irq/14-d-596       0d... 42789861us : dwc3_gadget_giveback: ep6in: req 00000000a324f5d0 length 0/192 zsI ==> -104
[   76.595046] irq/14-d-596       0d... 42789862us : dwc3_free_request: ep6in: req 00000000a324f5d0 length 0/192 zsI ==> -104
[   76.612892] irq/14-d-596       0d... 42789865us : dwc3_gadget_ep_disable: ep6in: mps 192/346 streams 16 burst 0 ring 0/0 flags E:swbp:<
[   76.665535] irq/14-d-596       0d... 42789873us : dwc3_event: event (000020c2): ep0in: Transfer Not Ready [0] (Not Active) [Status Phase]
[   76.684716] irq/14-d-596       0d... 42789875us : dwc3_prepare_trb: ep0in: trb 000000004c0ae319 (E0:D0) buf 000000001bded000 size 0 ctrl 00000c33 (HLcs:SC:status2)
[   76.819195] irq/14-d-596       0d... 42789886us : dwc3_gadget_ep_cmd: ep0in: cmd 'Start Transfer' [406] params 00000000 1bded000 00000000 --> status: Successful
[   76.926324] irq/14-d-596       0d... 42789930us : dwc3_event: event (0000c042): ep0in: Transfer Complete (sIL) [Status Phase]
[   76.937892] irq/14-d-596       0d... 42789933us : dwc3_complete_trb: ep0out: trb 000000004c0ae319 (E0:D0) buf 000000001bded000 size 0 ctrl 00000c32 (hLcs:SC:status2)
[   76.953003] irq/14-d-596       0d... 42789935us : dwc3_gadget_giveback: ep0out: req 00000000cb1bd3cd length 0/0 zsI ==> 0
[   76.964217] irq/14-d-596       0d... 42789938us : dwc3_prepare_trb: ep0out: trb 000000004c0ae319 (E0:D0) buf 000000001bded000 size 8 ctrl 00000c23 (HLcs:SC:setup)
[   77.061379] irq/14-d-596       0d... 42789950us : dwc3_gadget_ep_cmd: ep0out: cmd 'Start Transfer' [406] params 00000000 1bded000 00000000 --> status: Successful
[   77.168595] irq/14-d-596       0d... 42790509us : dwc3_event: event (0000c040): ep0out: Transfer Complete (sIL) [Setup Phase]
[   77.180159] irq/14-d-596       0d... 42790512us : dwc3_ctrl_req: Get String Descriptor(Index = 18, Length = 255)
[   77.190578] irq/14-d-596       0d... 42790537us : dwc3_prepare_trb: ep0in: trb 000000004c0ae319 (E0:D0) buf 0000000003b68000 size 36 ctrl 00000c53 (HLcs:SC:data)
[   77.287648] irq/14-d-596       0d... 42790550us : dwc3_gadget_ep_cmd: ep0in: cmd 'Start Transfer' [406] params 00000000 1bded000 00000000 --> status: Successful
[   77.333107] irq/14-d-596       0d... 42790557us : dwc3_event: event (000010c2): ep0in: Transfer Not Ready [0] (Not Active) [Data Phase]
[   77.407223] irq/14-d-596       0d... 42790575us : dwc3_event: event (000090c2): ep0in: Transfer Not Ready [0] (Active) [Data Phase]
[   77.480985] irq/14-d-596       0d... 42790588us : dwc3_event: event (0000c042): ep0in: Transfer Complete (sIL) [Data Phase]
[   77.492376] irq/14-d-596       0d... 42790590us : dwc3_complete_trb: ep0out: trb 000000004c0ae319 (E0:D0) buf 0000000003b68000 size 0 ctrl 00000c52 (hLcs:SC:data)
[   77.507221] irq/14-d-596       0d... 42790595us : dwc3_gadget_giveback: ep0out: req 00000000cb1bd3cd length 36/36 ZsI ==> 0
[   77.518609] irq/14-d-596       0d... 42790597us : dwc3_event: event (000020c0): ep0out: Transfer Not Ready [0] (Not Active) [Status Phase]
[   77.531332] irq/14-d-596       0d... 42790598us : dwc3_prepare_trb: ep0out: trb 000000004c0ae319 (E0:D0) buf 000000001bded000 size 0 ctrl 00000c43 (HLcs:SC:status3)
[   77.628669] irq/14-d-596       0d... 42790609us : dwc3_gadget_ep_cmd: ep0out: cmd 'Start Transfer' [406] params 00000000 1bded000 00000000 --> status: Successful

Do you mind adding a few prints in dwc3_remove_requests to tell us which
endpoint is being processed? Then we'll know for sure which one caused
the crash.

I wouldn't mind but am leaving on a holiday, won't have time until 6 aug.
not a problem, we'll still be here when you're back :-)

Well, let's go then :-)

To get back in the mood I have retested 5.13.0, 5.14.0-rc1, 5.14.0-rc2
and 5.14.0-rc5.

I find that 5.13.0 works fine, and the issue starts from 5.14.0-rc1.
That's great finding. We have a bisection point. There are a total of
13764 commits between v5.13 and v5.14-rc1

    $ git rev-list --count v5.13..v5.14-rc1
    13764

git bisect should find the offending commit in at most 14 tries. That's
not too bad.

I correctly guesstimated that the problem got introduced by the usb merge 79160a60

"Merge tag 'usb-5.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb"

116 commits(7 bisects).

24f779dac8f3efb9629adc0e486914d93dc45517 is the first bad commit

"usb: gadget: f_uac2/u_audio: add feedback endpoint support"

Ruslan's 3 patches are related to each other so I reverted all three 24f779da...e89bb428 and applied the reverts to rc1.

I can confirm this indeed resolves the problem in rc1.

Is late now, tomorrow evening I will apply the reverts to rc6.

With these reverts rc6 works fine as well.

So, where do we go from here?

I know the patches have been tested on dwc2 (by me and others). I do not know if Ruslan or Jerome tested them on dwc3 but probably not. Ruslan has talked about RPi (my case too) and BeagleboneBlack, both with dwc2. Perhaps the dwc2 behaves a bit differently than dwc3?

The patches add a new EP-IN for async feedback. I am sorry I have not followed your long thread (it started as unrelated to uac). Does the problem appear with f_uac1 or f_uac2? Please how have you reached the above problem?

I'm sorry too. I first believed the issue was related to the patch mentioned in the subject line.

The problem appaers with f_uac2. I bost Edison_Arduino board in host mode (there is a switch allowing to select host/device mode). When flipping the switch to device mode udev run a script:
But as I am using configfs (excerpt follows) and just disabling the last 2 line resolves the issue, I'm guessing uac2 is the issue. Or exceeding the available resources.

# Create directory structure
mkdir "${GADGET_BASE_DIR}"
cd "${GADGET_BASE_DIR}"
mkdir -p configs/c.1/strings/0x409
mkdir -p strings/0x409

# Serial device
mkdir functions/gser.usb0
ln -s functions/gser.usb0 configs/c.1/
###

# Ethernet device
mkdir functions/eem.usb0
echo "${DEV_ETH_ADDR}" > functions/eem.usb0/dev_addr
echo "${HOST_ETH_ADDR}" > functions/eem.usb0/host_addr
ln -s functions/eem.usb0 configs/c.1/

# Mass Storage device
mkdir functions/mass_storage.usb0
echo 1 > functions/mass_storage.usb0/stall
echo 0 > functions/mass_storage.usb0/lun.0/cdrom
echo 0 > functions/mass_storage.usb0/lun.0/ro
echo 0 > functions/mass_storage.usb0/lun.0/nofua
echo "${USBDISK}" > functions/mass_storage.usb0/lun.0/file
ln -s functions/mass_storage.usb0 configs/c.1/

# UAC2 device
mkdir functions/uac2.usb0
ln -s functions/uac2.usb0 configs/c.1
....

As you say, could perhaps the reason be that the extra EP-IN added in those patches (previously 1, now 2 with the default config you use) exceeds your EP-IN max count or available fifos somehow? You have a number of functions initialized. If you change the load order of the functions, do you get the error later with a different function? Just guessing...

You should be able to switch the default async EP-OUT (which configures the new feedback EP-IN ) to adaptive EP-OUT (which requires no feedback EP) with c_sync=8 parameter of f_uac2.

https://elixir.bootlin.com/linux/v5.14-rc6/source/drivers/usb/gadget/function/f_uac2.c#L47

https://elixir.bootlin.com/linux/v5.14-rc6/source/drivers/usb/gadget/function/f_uac2.c#L830

https://elixir.bootlin.com/linux/v5.14-rc6/source/include/uapi/linux/usb/ch9.h#L453

Does that fix the problem?

Not sure how to do that. Do you mean the module should have a parameter called c_sync? `modinfo` list no parameters for usb_f_uac2.

Those are configfs params, not available in modinfo.

I checked and the value is "adaptive" https://elixir.bootlin.com/linux/v5.14-rc7/source/drivers/usb/gadget/function/f_uac2.c#L1312

In your configfs script:

# UAC2 device
mkdir functions/uac2.usb0
ln -s functions/uac2.usb0 configs/c.1

On the USB Host:
cat /proc/asound/Gadget/stream:
Playback:
Status: Stop
Interface 1
Altset 1
Format: S16_LE
Channels: 2
Endpoint: 0x01 (1 OUT) (ASYNC)
Rates: 64000
Data packet interval: 125 us
Bits: 16
Channel map: FL FR
Sync Endpoint: 0x82 (2 IN)
Sync EP Interface: 1
Sync EP Altset: 1
Implicit Feedback Mode: No

lsusb -v -d 1d6b:0104 | | grep EP.*IN
bEndpointAddress 0x81 EP 1 IN
bEndpointAddress 0x82 EP 2 IN
bEndpointAddress 0x83 EP 3 IN

I have additional patches applied which define controls via EP IN interrupt mode, not part of that patchset.

Switching to the adaptive mode:
# UAC2 device
mkdir functions/uac2.usb0
echo "adaptive" > functions/uac2.usb0/c_sync
ln -s functions/uac2.usb0 configs/c.1

On the USB Host:
cat /proc/asound/Gadget/stream:
Playback:
Status: Stop
Interface 1
Altset 1
Format: S16_LE
Channels: 2
Endpoint: 0x01 (1 OUT) (ADAPTIVE)
Rates: 64000
Data packet interval: 125 us
Bits: 16
Channel map: FL FR

lsusb -v -d 1d6b:0104 | grep EP.*IN
bEndpointAddress 0x81 EP 1 IN
bEndpointAddress 0x82 EP 2 IN

The feedback EP-IN is gone because the mode is adaptive now.

If you think the new input endpoints should not be created by default for resource-compatibility reasons, the adaptive mode can be set by default in a fixed patch.

Also the patches defining the new interrupt endpoints can have the controls disabled by default (and not allocate the EP-IN). These patches on top of the async patches are already accepted by Greg for usb-next branch https://kernel.googlesource.com/pub/scm/linux/kernel/git/gregkh/usb/+/eaf6cbe0992052a46d93047dc122fad5126aa3bd . That's why I was trying to sort out the async patches without reverting them as it will call for more reverts.

Thanks,

Pavel.

Next message: Soeren Moch: "Re: [Regression 5.14] media: dvb userspace api"
Previous message: Vlastimil Babka: "[PATCH v5 33/35] mm, slub: protect put_cpu_partial() with disabled irqs instead of cmpxchg"
In reply to: Ferry Toth: "Re: [PATCH v10 0/6] Re-introduce TX FIFO resize for larger EP bursting"
Next in thread: Andy Shevchenko: "Re: [PATCH v10 0/6] Re-introduce TX FIFO resize for larger EP bursting"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]