samples: zperf: Add initialization for the new device support to the sample main() #71589
Comments
|
It is not broken, I use it regularly. Not initializing it in the main sample was intentional at the time. The shell is enabled in the zperf sample, what is missing in your build command is -DCONFIG_USBD_SHELL=y, see |
jfischer-no
added
Enhancement
jfischer-no
changed the title
|
Thanks for your help! Added some code to main.c: #if defined(CONFIG_USB_DEVICE_STACK_NEXT)
#include <zephyr/usb/usbd.h>
USBD_CONFIGURATION_DEFINE(config_foo, USB_SCD_SELF_POWERED, 200);
USBD_DEVICE_DEFINE(sh_uds_ctx, DEVICE_DT_GET(DT_NODELABEL(zephyr_udc0)),
0x2fe3, 0xffff);
USBD_DESC_LANG_DEFINE(lang);
USBD_DESC_MANUFACTURER_DEFINE(mfr, "ZEPHYR");
USBD_DESC_PRODUCT_DEFINE(product, "Zephyr USBD foobaz");
USBD_DESC_SERIAL_NUMBER_DEFINE(sn, "0123456789ABCDEF");
static struct usbd_contex *my_uds_ctx = &sh_uds_ctx;
#endif
...
int main(void)
{
...
#if defined(CONFIG_USB_DEVICE_STACK_NEXT)
{
int err;
err = usbd_add_configuration(my_uds_ctx, &config_foo);
printk("usbd_add_configuration: %d\n", err);
err = usbd_register_class(my_uds_ctx, "cdc_ecm_0", 1);
printk("usbd_register_class: %d\n", err);
err = usbd_add_descriptor(my_uds_ctx, &lang);
err |= usbd_add_descriptor(my_uds_ctx, &mfr);
err |= usbd_add_descriptor(my_uds_ctx, &product);
err |= usbd_add_descriptor(my_uds_ctx, &sn);
printk("usbd_add_descriptor: %d\n", err);
err = usbd_init(my_uds_ctx);
printk("usbd_init: %d\n", err);
err = usbd_enable(my_uds_ctx);
printk("usbd_enable: %d\n", err);
}
#endifUSB is now starting up (with a delay of 30s because it is still waiting for the IF to come up), the network interface also shows up after that and pinging works. Have to look tomorrow deeper into this. |
|
gosh... I think I'm too blonde today... Forgot to insert my other code to start the server automatically. Looks like: {
static const struct zperf_download_params param = { .port = 5001 };
zperf_tcp_download( ¶m, my_zperf_callback, NULL);
zperf_udp_download( ¶m, my_zperf_callback, NULL); // just with IPv4? And no callback?
}So it seems to work as expected. Thanks for support and if you want to, I can create a PR. |
|
That is a bit to much #71589 (comment). We have common initialization code for the new USB samples, take a look at the samples/bluetooth/hci_usb. It should look something like this: diff --git a/samples/net/zperf/CMakeLists.txt b/samples/net/zperf/CMakeLists.txt
index c4552f97141..dd1a09bbb22 100644
--- a/samples/net/zperf/CMakeLists.txt
+++ b/samples/net/zperf/CMakeLists.txt
@@ -22,3 +22,7 @@ if (CONFIG_NET_L2_ETHERNET)
LOCATION "${CONFIG_NET_SAMPLE_CODE_RAM_NAME}_TEXT" NOKEEP)
endif()
endif()
+
+if (CONFIG_USB_DEVICE_STACK_NEXT)
+ include(${ZEPHYR_BASE}/samples/subsys/usb/common/common.cmake)
+endif()
diff --git a/samples/net/zperf/Kconfig b/samples/net/zperf/Kconfig
index a56cd6b9416..54ab59ca312 100644
--- a/samples/net/zperf/Kconfig
+++ b/samples/net/zperf/Kconfig
@@ -20,3 +20,5 @@ config NET_SAMPLE_CODE_RAM_NAME
Region to relocate networking code to
endif # NET_SAMPLE_CODE_RELOCATE
+
+source "samples/subsys/usb/common/Kconfig.sample_usbd"
diff --git a/samples/net/zperf/src/main.c b/samples/net/zperf/src/main.c
index 8c9d6a5c64c..e81a490ae66 100644
--- a/samples/net/zperf/src/main.c
+++ b/samples/net/zperf/src/main.c
@@ -9,17 +9,40 @@
* @brief Zperf sample.
*/
#include <zephyr/usb/usb_device.h>
+#include <zephyr/usb/usbd.h>
#include <zephyr/net/net_config.h>
#ifdef CONFIG_NET_LOOPBACK_SIMULATE_PACKET_DROP
#include <zephyr/net/loopback.h>
#endif
+
+#if defined(CONFIG_USB_DEVICE_STACK_NEXT)
+#include <sample_usbd.h>
+
+static int enable_usb_device_next(void)
+{
+ struct usbd_contex *sample_usbd = sample_usbd_init_device(NULL);
+
+ if (sample_usbd == NULL) {
+ printk("Failed to initialize USB device");
+ return -ENODEV;
+ }
+
+ return usbd_enable(sample_usbd);
+}
+#endif /* CONFIG_USB_DEVICE_STACK_NEXT */
+
int main(void)
{
-#if defined(CONFIG_USB_DEVICE_STACK)
+#if defined(CONFIG_USB_DEVICE_STACK) || defined(CONFIG_USB_DEVICE_STACK_NEXT)
int ret;
+#if defined(CONFIG_USB_DEVICE_STACK)
ret = usb_enable(NULL);
+#endif
+#if defined(CONFIG_USB_DEVICE_STACK_NEXT)
+ ret = enable_usb_device_next();
+#endif
if (ret != 0) {
printk("usb enable error %d\n", ret);
}Whereby it is too much ifdefery, IS_ENABLED() could also work. |
|
Seems that I have to integrate the current development branch. Is there any doc about this? (I'm actually using nRF Connect SDK) |
nRF Connect SDK development model has the concept of upmerge. Currently the HWMv2 is being upmerged to NCS (See nrfconnect/sdk-nrf#14626 for more information). Considering how complex upmerge is, I believe it is better for you to simply wait until NCS has all the required Zephyr changes integrated. In the meantime, it is probably best to use upstream Zephyr directly (if you can get away without NCS specific features). |
It is much simpler: installed zephyr to ~/zephyrproject according to installation doc & the VSCode plugin automatically detects it and one can select "Zephyr v3.6.99". Now your suggestion from above is compiling & working. Question 1: ok if I extend the example so that a zperf_tcp_download() server is auto started? This makes testing a lot easier. Question 2: how stable is the current ECM driver? Is it work in progress? |
Everything in subsys/usb/device_next is experimental. CDC ECM is not a work in progress. |
Notes
Trying to create an NCM driver for the new USB stack, using zperf with ECM as a starting point, see #71451
Describe the bug
Using the zperf example and selecting the configuration with CONFIG_USB_DEVICE_STACK_NEXT=y. west command line:
After flashing the target gives the following log output:
It seems that the interface is not coming up automatically. BTW: to me the start sequence is not clear at all ;-)
USB is not starting at all, meaning that the host computer does not see a (new) USB device. (before you ask: with the old USB stack everything is working fine).
To Reproduce
see above
Expected behavior
I'ld expect that USB starts up, the interface is becoming available and the target device provides some IP functionality.
Impact
annoying...
Logs and console output
see above, log output with more debug looks like this:
Environment (please complete the following information):
-- west build: generating a build system
Loading Zephyr default modules (Zephyr base).
-- Application: /mnt/d/u/src/ncs/zperf_checko
-- CMake version: 3.21.0
-- Found Python3: /home/hardy/ncs/toolchains/2be090971e/usr/local/bin/python3.9 (found suitable version "3.9.18", minimum required is "3.8") found components: Interpreter
-- Cache files will be written to: /home/hardy/.cache/zephyr
-- Zephyr version: 3.5.99 (/home/hardy/ncs/v2.6.0/zephyr)
-- Found west (found suitable version "1.2.0", minimum required is "0.14.0")
-- Board: nrf52840dk_nrf52840
-- Found host-tools: zephyr 0.16.5 (/home/hardy/ncs/toolchains/2be090971e/opt/zephyr-sdk)
-- Found toolchain: zephyr 0.16.5 (/home/hardy/ncs/toolchains/2be090971e/opt/zephyr-sdk)
Additional context
set
CONFIG_NET_SOCKETS_POLL_MAX=10to make network startup. Otherwise log showsThe text was updated successfully, but these errors were encountered: