Re: [PATCH v9 0/5] Re-introduce TX FIFO resize for larger EP bursting

From: Ferry Toth
Date: Thu Jun 17 2021 - 15:54:47 EST


Op 17-06-2021 om 10:30 schreef Wesley Cheng:

On 6/17/2021 12:47 AM, Ferry Toth wrote:

Op 17-06-2021 om 06:25 schreef Wesley Cheng:
On 6/15/2021 12:53 PM, Ferry Toth wrote:

Op 15-06-2021 om 06:22 schreef Wesley Cheng:
On 6/14/2021 12:30 PM, Ferry Toth wrote:
Op 14-06-2021 om 20:58 schreef Wesley Cheng:
On 6/12/2021 2:27 PM, Ferry Toth wrote:

Op 11-06-2021 om 15:21 schreef Andy Shevchenko:
On Fri, Jun 11, 2021 at 4:14 PM Heikki Krogerus
<heikki.krogerus@xxxxxxxxxxxxxxx> wrote:
On Fri, Jun 11, 2021 at 04:00:38PM +0300, Felipe Balbi wrote:

Wesley Cheng <wcheng@xxxxxxxxxxxxxx> writes:
to be honest, I don't think these should go in (apart from
the build
failure) because it's likely to break instantiations of the
core with
differing FIFO sizes. Some instantiations even have some
endpoints with
dedicated functionality that requires the default FIFO size
during coreConsultant instantiation. I know of at OMAP5 and
some Intel
implementations which have dedicated endpoints for processor

With OMAP5, these endpoints are configured at the top of the
endpoints, which means that if a gadget driver gets loaded
and takes
over most of the FIFO space because of this resizing,
processor tracing
will have a hard time running. That being said, processor
tracing isn't
supported in upstream at this moment.

I agree that the application of this logic may differ between
hence why I wanted to keep this controllable by the DT
property, so that
for those which do not support this use case can leave it
disabled.  The
logic is there to ensure that for a given USB configuration,
for each EP
it would have at least 1 TX FIFO.  For USB configurations
utilize all available IN EPs, it would allow re-allocation of
memory to EPs which will actually be in use.
The feature ends up being all-or-nothing, then :-) It sounds
like we can
be a little nicer in this regard.

Don't get me wrong, I think once those features become available
upstream, we can improve the logic.  From what I remember when
sure, I support that. But I want to make sure the first cut isn't
to break things left and right :)

Hence, let's at least get more testing.

Sure, I'd hope that the other users of DWC3 will also see some
big improvements on the TX path with this.
fingers crossed

at Andy Shevchenko's Github, the Intel tracer downstream changes
just to remove physical EP1 and 2 from the DWC3 endpoint list.
If that
right, that's the reason why we introduced the endpoint feature
flags. The end goal was that the UDC would be able to have custom
feature flags paired with ->validate_endpoint() or whatever
allowing it to be enabled. Then the UDC driver could tell UDC
core to
skip that endpoint on that particular platform without
interefering with
everything else.

Of course, we still need to figure out a way to abstract the
dwc3 instantiations.

was the change which ended up upstream for the Intel tracer
then we
could improve the logic to avoid re-sizing those particular EPs.
The problem then, just as I mentioned in the previous paragraph,
will be
coming up with a solution that's elegant and works for all
instantiations of dwc3 (or musb, cdns3, etc).

Well, at least for the TX FIFO resizing logic, we'd only be
needing to
focus on the DWC3 implementation.

You bring up another good topic that I'll eventually needing to be
taking a look at, which is a nice way we can handle vendor
endpoints and how they can co-exist with other "normal"
endpoints.  We
have a few special HW eps as well, which we try to maintain
in our DWC3 vendor driver, but it isn't the most convenient, or
pretty method :).
Awesome, as mentioned, the endpoint feature flags were added
exactly to
allow for these vendor-specific features :-)

I'm more than happy to help testing now that I finally got our
Surface Duo device tree accepted by Bjorn ;-)

However, I'm not sure how the changes would look like in the
so I
would like to wait later down the line to include that :).
Fair enough, I agree. Can we get some more testing of $subject,
Did you test $subject with upstream too? Which gadget drivers
did you
use? How did you test

The results that I included in the cover page was tested with the
upstream kernel on our device.  Below was using the ConfigFS
w/ a
mass storage only composition.

Test Parameters:
    - Platform: Qualcomm SM8150
    - bMaxBurst = 6
    - USB req size = 256kB
    - Num of USB reqs = 16
do you mind testing with the regular request size (16KiB) and 250
requests? I think we can even do 15 bursts in that case.

    - USB Speed = Super-Speed
    - Function Driver: Mass Storage (w/ ramdisk)
    - Test Application: CrystalDiskMark


TXFIFO Depth = 3 max packets

Test Case | Data Size | AVG tput (in MB/s)
Sequential|1 GB x     |
Read      |9 loops    | 193.60
             |           | 195.86
             |           | 184.77
             |           | 193.60

TXFIFO Depth = 6 max packets

Test Case | Data Size | AVG tput (in MB/s)
Sequential|1 GB x     |
Read      |9 loops    | 287.35
           |           | 304.94
             |           | 289.64
             |           | 293.61
I remember getting close to 400MiB/sec with Intel platforms without
resizing FIFOs and I'm sure the FIFO size was set to 2x1024,
though my
memory could be failing.

Then again, I never ran with CrystalDiskMark, I was using my own
(it's somewhere in github. If you care, I can look up the URL).

We also have internal numbers which have shown similar
improvements as
well.  Those are over networking/tethering interfaces, so testing
loopback over TCP/UDP.
loopback iperf? That would skip the wire, no?

size of 2 and TX threshold of 1, this would really be not
beneficial to
us, because we can only change the TX threshold to 2 at max,
and at
least in my observations, once we have to go out to system
memory to
fetch the next data packet, that latency takes enough time
for the
controller to end the current burst.
What I noticed with g_mass_storage is that we can amortize the
cost of
fetching data from memory, with a deeper request queue.
Whenever I
test(ed) g_mass_storage, I was doing so with 250 requests. And
that was
enough to give me very good performance. Never had to poke at TX
resizing. Did you try something like this too?

I feel that allocating more requests is a far simpler and more
method that changing FIFO sizes :)

I wish I had a USB bus trace handy to show you, which would
make it
clear how the USB bus is currently utilized with TXFIFO size 2 vs
6.  So
by increasing the number of USB requests, that will help if there
was a
bottleneck at the SW level where the application/function driver
utilizing the DWC3 was submitting data much faster than the HW was
processing them.

So yes, this method of increasing the # of USB reqs will
with situations such as HSUSB or in SSUSB when EP bursting isn't
The TXFIFO resize comes into play for SSUSB, which utilizes
Hmm, that's not what I remember. Perhaps the TRB cache size plays a
here too. I have clear memories of testing this very scenario of
bursting (using g_mass_storage at the time) because I was curious
it. Back then, my tests showed no difference in behavior.

It could be nice if Heikki could test Intel parts with and without
changes on g_mass_storage with 250 requests.
Andy, you have a system at hand that has the DWC3 block enabled,
right? Can you help out here?
I'm not sure if i will have time soon, I Cc'ed to Ferry who has a few
more test cases (I have only one or two) and maybe can help. But I'll
keep this in mind.
I just tested on 5.13.0-rc4 on Intel Edison (x86_64). All 5 patches
apply. Switching between host/gadget works, no connections
dropping, no
errors in dmesg.

In host mode I connect a smsc9504 eth+4p hub. In gadget mode I have
composite device created from configfs with gser / eem /
mass_storage /

Tested with iperf3 performance in host (93.6Mbits/sec) and gadget
(207Mbits/sec) mode. Compared to v5.10.41 without patches host
(93.4Mbits/sec) and gadget (198Mbits/sec).

Gadget seems to be a little faster with the patches, but that might
be caused  by something else, on v5.10.41 I see the bitrate bouncing
between 207 and 199.

I saw a mention to test iperf3 to self (loopback). 3.09 Gbits/sec.
v5.10.41 3.07Gbits/sec. Not bad for a 500MHz device.

With gnome-disks I did a read access benchmark 35.4MB/s, with v5.10.41
34.7MB/s. This might be limited by Edison's internal eMMC speed (when
booting U-Boot reads the kernel with 21.4 MiB/s).

Hi Ferry,

Thanks for the testing.  Just to double check, did you also enable the
property, which enabled the TXFIFO resize feature on the platform?  For
example, for the QCOM SM8150 platform, we're adding the following to
device tree node:


If not, then your results at least confirms that w/o the property
present, the changes won't break anything :).  Thanks again for the
initial testing!
I applied the patch now to 5.13.0-rc5 + the following:

Hi Ferry,

Quick question...there was a compile error with the V9 patch series, as
it was using the dwc3_mwidth() incorrectly. I will update this with the
proper use of the mdwidth, but which patch version did you use?
Hi Ferry,

The V9 set gets applied to 5.13.0-rc5 by Yocto, if it doesn't apply it
stops the build. I didn't notice any compile errors, they stop the whole
build process too. But warnings are ignored. I'll check the logs to be sure.
Ah, ok. I think the incorrect usage of the API will result as a warning
as seen in Greg's compile log:

drivers/usb/dwc3/gadget.c: In function ‘dwc3_gadget_calc_tx_fifo_size’:
drivers/usb/dwc3/gadget.c:653:45: warning: passing argument 1 of
‘dwc3_mdwidth’ makes pointer from integer without a cast [-Wint-conversion]
653 | mdwidth = dwc3_mdwidth(dwc->hwparams.hwparams0);
I found exactly this warning in my logs. Sorry for this noise, I was watching for an error.
This is probably why the page fault occurs, as we're not passing in the
DWC3 struct. I will send out a V10 shortly after testing it on my device.

Wesley Cheng

Wesley Cheng

--- a/drivers/usb/dwc3/dwc3-pci.c
+++ b/drivers/usb/dwc3/dwc3-pci.c
@@ -124,6 +124,7 @@ static const struct property_entry
dwc3_pci_mrfld_properties[] = {
+    PROPERTY_ENTRY_BOOL("tx-fifo-resize"),

 and when switching to gadget mode unfortunately received the following

BUG: unable to handle page fault for address: 00000000202043f2
#PF: supervisor read access in kernel mode
#PF: error_code(0x0000) - not-present page
PGD 0 P4D 0
Oops: 0000 [#1] SMP PTI
CPU: 0 PID: 617 Comm: Not tainted
5.13.0-rc5-edison-acpi-standard #1
Hardware name: Intel Corporation Merrifield/BODEGA BAY, BIOS 542
RIP: 0010:dwc3_gadget_check_config+0x33/0x80
Code: 59 04 00 00 04 74 61 48 c1 ee 10 48 89 f7 f3 48 0f b8 c7 48 89 c7
39 81 60 04 00 00 7d 4a 89 81 60 04 00 00 8b 81 08 04 00 00 <81> b8 e8
03 00 00 32 33 00 00 0f b6 b0 09 04 00 00 75 0d 8b 80 20
RSP: 0018:ffffb5550038fda0 EFLAGS: 00010297
RAX: 000000002020400a RBX: ffffa04502627348 RCX: ffffa04507354028
RDX: 0000000000000000 RSI: 000000000000003c RDI: 0000000000000004
RBP: ffffa04508ac0550 R08: ffffa04503a75b2c R09: 0000000000000000
R10: 0000000000000216 R11: 000000000002eba0 R12: ffffa04508ac0550
R13: dead000000000100 R14: ffffa04508ac0600 R15: ffffa04508ac0520
FS:  00007f7471e2f740(0000) GS:ffffa0453e200000(0000)
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000202043f2 CR3: 0000000003f38000 CR4: 00000000001006f0
Call Trace:
 configfs_composite_bind+0x2f4/0x430 [libcomposite]
 gadget_dev_desc_UDC_store+0xbc/0x130 [libcomposite]
RIP: 0033:0x7f7471f1ff53
Code: 8b 15 21 cf 0c 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f
00 64 8b 04 25 18 00 00 00 85 c0 75 14 b8 01 00 00 00 0f 05 <48> 3d 00
f0 ff ff 77 55 c3 0f 1f 40 00 48 83 ec 28 48 89 54 24 18
RSP: 002b:00007fffa3dcd328 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
RAX: ffffffffffffffda RBX: 000000000000000c RCX: 00007f7471f1ff53
RDX: 000000000000000c RSI: 00005614d615a770 RDI: 0000000000000001
RBP: 00005614d615a770 R08: 000000000000000a R09: 00007f7471fb20c0
R10: 0000000000000000 R11: 0000000000000246 R12: 000000000000000c
R13: 00007f7471fee520 R14: 000000000000000c R15: 00007f7471fee720
Modules linked in: usb_f_uac2 u_audio usb_f_mass_storage usb_f_eem
u_ether usb_f_serial u_serial libcomposite rfcomm iptable_nat bnep
snd_sof_nocodec spi_pxa2xx_platform dw_dmac smsc snd_sof_pci_intel_tng
snd_sof_pci snd_sof_acpi_intel_byt snd_sof_intel_ipc snd_sof_acpi
smsc95xx snd_sof pwm_lpss_pci pwm_lpss snd_sof_xtensa_dsp
snd_intel_dspcfg snd_soc_acpi_intel_match snd_soc_acpi dw_dmac_pci
intel_mrfld_pwrbtn intel_mrfld_adc dw_dmac_core spi_pxa2xx_pci brcmfmac
brcmutil leds_gpio hci_uart btbcm ti_ads7950
industrialio_triggered_buffer kfifo_buf ledtrig_timer ledtrig_heartbeat
mmc_block extcon_intel_mrfld sdhci_pci cqhci sdhci led_class
intel_soc_pmic_mrfld mmc_core btrfs libcrc32c xor zstd_compress
zlib_deflate raid6_pq
CR2: 00000000202043f2
---[ end trace 5c11fe50dca92ad4 ]---

No I didn't. Afaik we don't have a devicetree property to set.

But I'd be happy to test that as well. But where to set the property?

dwc3_pci_mrfld_properties[] in dwc3-pci?

Hi Ferry,

Not too sure which DWC3 driver is used for the Intel platform, but I
believe that should be the one. (if that's what is normally used)  We'd
just need to add an entry w/ the below:


Wesley Cheng

Wesley Cheng

Now with endpoint bursting, if the function notifies the host that
bursting is supported, when the host sends the ACK for the Data
it should have a NumP value equal to the bMaxBurst reported in
the EP
Yes and no. Looking back at the history, we used to configure NUMP
on bMaxBurst, but it was changed later in commit
4e99472bc10bda9906526d725ff6d5f27b4ddca1 by yours truly because
of a
problem reported by John Youn.

And now we've come full circle. Because even if I believe more
are enough for bursting, NUMP is limited by the RxFIFO size. This
up supporting your claim that we need RxFIFO resizing if we want to
squeeze more throughput out of the controller.

However, note that this is about RxFIFO size, not TxFIFO size. In
looking at Table 8-13 of USB 3.1 r1.0, we read the following about
(emphasis is mine):

         "Number of Packets (NumP). This field is used to
indicate the
         number of Data Packet buffers that the **receiver** can
         accept. The value in this field shall be less than or
equal to
         the maximum burst size supported by the endpoint as
         by the value in the bMaxBurst field in the Endpoint
         Descriptor (refer to Section 9.6.7)."

So, NumP is for the receiver, not the transmitter. Could you
what you mean here?

/me keeps reading

Hmm, table 8-15 tries to clarify:

         "Number of Packets (NumP).

         For an OUT endpoint, refer to Table 8-13 for the
description of
         this field.

         For an IN endpoint this field is set by the endpoint to
         number of packets it can transmit when the host resumes
         transactions to it. This field shall not have a value
         than the maximum burst size supported by the endpoint as
         indicated by the value in the bMaxBurst field in the
         Companion Descriptor. Note that the value reported in this
         may be treated by the host as informative only."

However, if I remember correctly (please verify dwc3 databook),
DCFG was only for receive buffers. Thin, John, how does dwc3
NumP for TX/IN endpoints? Is that computed as a function of
TxFIFO size?

desc.  If we have a TXFIFO size of 2, then normally what I have
seen is
that after 2 data packets, the device issues a NRDY.  So then we'd
to send an ERDY once data is available within the FIFO, and the
sequence happens until the USB request is complete.  With this
NRDY/ERDY handshake going on, you actually see that the bus is
utilized.  When we increase an EP's FIFO size, then you'll see
bursts for a request, until the request is done, or if the host
runs out
of RXFIFO. (ie no interruption [on the USB protocol level] during
request data transfer)
Unfortunately I don't have access to a USB sniffer anymore :-(

Good points.

Wesley, what kind of testing have you done on this on
different devices?

As mentioned above, these changes are currently present on end
devices for the past few years, so its been through a lot of
testing :).
all with the same gadget driver. Also, who uses USB on android
these days? Most of the data transfer goes via WiFi or
Bluetooth, anyway

I guess only developers are using USB during development to
flash dev
images heh.

I used to be a customer facing engineer, so honestly I did see
really interesting and crazy designs.  Again, we do have
products that use the same code, and it has been working in
for a
few years as well.  The TXFIFO sizing really has helped with
use cases, which use isoc endpoints, since esp. in those lower
end CPU
chips where latencies across the system are much larger, and a
ISOC interval leads to a pop in your ear.
This is good background information. Thanks for bringing this
up. Admitedly, we still have ISOC issues with dwc3. I'm
interested in
knowing if a deeper request queue would also help here.

Remember dwc3 can accomodate 255 requests + link for each
endpoint. If
our gadget driver uses a low number of requests, we're never
using the TRB ring in our benefit.

We're actually using both a deeper USB request queue + TX fifo
resizing. :).
okay, great. Let's see what John and/or Thinh respond WRT dwc3 TX