espsta0 interface does not appear #191
Comments
|
I have moved on, the device is detected, but another error occurs: |
|
I've capture data exchange between Host-side and ESP32: May view in DSView util |
|
|
|
Some text errors from Host side |
|
ESP32 logs: |
|
@mantriyogesh Hello, could you help me please? |
|
Apologies for the delay, will check it today and let you know.. |
|
Can you please:
|
SPI_MODE0: SPI_MODE3: |
|
SPI_MODE1, log from logical analyzer:
ESP Log: |
|
Without changing spi mode and only commenting |
I've try this, see previous comment at 2) |
SPI_MODE2 (by default), ESP Log: Logic analyzer log: |
|
Thank you for investigating at your side & the analyser dump. Observations
Git CommitRelease you have used is ESP-Hosted-NG release 1.0.2 GPIOsNext thing to confirm is GPIOs. Below steps will help verify if the GPIOs are working or not: Three GPIOs needed -
Intention is to trigger next spi_sync_transfer() at esp-hosted/esp_hosted_ng/host/spi/esp_spi.c Line 439 in 29208b6 |
|
@mantriyogesh I Add KERN_ERR: ESP log: Logical analyzer:
|
|
Also i found a bug. If I call rmmod, kernel crashed: |
|
@mantriyogesh Logical analyzer log: |
|
Okay. So interrupts are coming. Are you using jumper wires? If yes, how long are they? |
|
The capture looks not consistent with the general expectations. First thing is MISO on ESP reset (ESP needs to send data and will set DR high. On host's first transaction, MISO will be the receiver. Then further with next message exchanges, MOSI would have some data to send. Without any transaction DR is going down? Can you please verify if DataReady, Handshake and reset pin GPIOs are not used by other peripherals or alternatively change them to ones not in use? |
Originally I used jump wires, but then reduce wire size. I can take a picture, if need, but everything is very confusing there, because of the connected logic analyzer. |
|
Also worth cross checking if wire connections are not touching each other (in case manually cut and joined by hand) & connections are as per |
|
For SPI you can have comfortable wire size like 5-6 cm. it is better if e have all wires at (almost) same length.. As you are anyway using ESP32, Just for trial, can you use NG release 1.0.1? For that, please use the source code from Source (zip). Note: NG release 1.0.2 is not backward compatible. Use firmware binaries and 'Source Code(zip)' exactly from same release. |
On Release 1.0.2 I reduce wires, see picture: All the same, from launch to launch, the interface either appears or does not appear. Once I even managed to scan the network and get two access points on the output of the iw (although in fact there are many more). The next step is to upload version 1.0.1 to the ESP32 and build the kernel module version 1.0.1? |
|
1.0.1 to 1.0.2 there is design change. There is one more possibility of using FG if in case cfg80211 is not must for you. NG is comparatively newer than FG. The FG solution has been present since very long and you should get it up easily. The hardware setup should be same. In case of FG you also can get access to the ESP source code. For FG, if want to build and flash, can use master at both ESP & host. |
|
Have you tried any of these two? |
1.0.1 not working too: |
On *_FG version all works fine: But do I understand correctly that in the FG version I can't use wpa_supplicant and/or iw? |
|
That is correct. The FG is oriented to more of API based approach. It is not connected with wpa_supplicant or iw. It is solution where your user space application can create and maintain the wifi connection just by simple function calls. In contrast to wpa_supplicant, we have open sourced custom control path. We created demo app which should be usable directly. In that you can use python based or c based applicantion, whichever is more suitable for your end application. For the documentation, please follow https://github.com/espressif/esp-hosted/tree/master/esp_hosted_fg. This is simple to use python app demo : |
In this case, the FG version does not suit me. Can we return to fixing the problem with the NG version? |
|
Positive about it. That is something with would let us know to focus if new code is to be suspected or not. |
|
@renyeguo , |
|
not yet，i'm still modify spi
controller of the kernel，problem still exist.
…---- Replied Message ----
| From | ***@***.***> |
| Date | 04/24/2023 18:38 |
| To | ***@***.***> |
| Cc | ***@***.***>***@***.***> |
| Subject | Re: [espressif/esp-hosted] espsta0 interface does not appear (Issue #191) |
@renyeguo ,
I hope the original issues are solved for you now?
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
|
You are using ESP32 - SPI |
|
#191 (comment) Did you try to switch spi mode? |
|
@mantriyogesh |
|
@mantriyogesh I have solved this problem. As communicated before, this problem was caused by spi block transmission fragmentation. Now the spi clock can correctly run to 30Mhz.However, the esp side should also need to handle this situation. Ensure the integrity of data reception. |
|
Integrity of data is only dependent upon the spi protocol exchange below. We have data integrity checks with checksum. Also please be assured that timing graph is correct. Cs line is stable, clock and all transitions are correct. We term this as porting the solution. porting, need to be done for host integration one time by user. |
|
Your software should not need fragmentation at all in general for working. Optimisations are all open but basic transport should really not be dependent on fragmentation in place. |
|
@mantriyogesh Some time ago, I was delayed in the development of esp32 due to other things. Now I find that there are several problems:
|
|
|
|
Yes, I agree with the phenomenon of #204. Please help me look at the second question ”I did an experiment. When the ap was successfully connected with wifi, the ap was turned off. The WiFi connection information of eps32 was always connected, and the ap that was just closed could still be scanned after scan operation“ |
|
You might have to check if disconnection is identified immediately by wpa_supplicant or not. wpa_spplicant can be run with extra debugs to understand the scenario. |
|
also, please verify if multiple APs with same SSID in vicinity. |
|
1.wpa_cli -i espsta0
2.udhcpc -i espsta0 -B 3."mate40pro+" is the hotspot with my phone ,then turn off the hotspot. 3.wpa_cli -i espsta0 status (Repeat the process) wpa_state is always COMPLETED, and no matter how many times I get the state, it will be completed after a few minutes |
|
@renyeguo Ideally wpa_cli should behave same way, but can you please check once with wpa_supplicant logs on and test with |
|
@kapilkedawat this modify fix it |
|
@mantriyogesh How to adjust the transmitting power of wifi? I use iwconfig to set this up, but an error occurs,like this: |
|
"iwconfig espsta0 txpower 30” |
|
Hello @renyeguo, That API needs to be added. |
|
@renyeguo Request to create new github issues for non related items. This will help us to track those in better way. |
|
What is the current default transmission power？ @kapilkedawat I have created a new issue。 |
Default Tx power is maximum tx power allowed based on channel/region. ESP devices supports upto 20dbm. |
|
Could you please separately modify a version with the maximum transmitting power to me? I need to do tests to verify the signal problem @mantriyogesh |
|
@renyeguo , esp chips works on max power by default. What do you mean by |
|
sorry, maybe there is something wrong with my expression. What I mean is that the connection between esp and ap is not stable at 8-9 meters when there is a distance between esp and AP, because I am not quite sure how much tx power is by default in the ESP-Hosted scheme at present. We hope to get a debug image of esp-hosted with maximum tx power to confirm whether the problem is caused by our design or insufficient tx power |
|
Hi @renyeguo , We are working on setting of tx_power using Can you share the logs when connection is not stable, I can take a look. Please enable FW logs and set the log levels to debug. |
Hello! I've try to porting to the Allwinner F1C200s SoC device.
I've successfully compile kernel module and load it on target system, but i can't see
espsta0interface inifconfig -aoutput.Kernel logs:
In output of
ps -eI've see this process:160 root [ESP_SPI_WORK_QU]full ps -e output
``` PID USER COMMAND 1 root init 2 root [kthreadd] 4 root [kworker/0:0H-kb] 6 root [mm_percpu_wq] 7 root [ksoftirqd/0] 8 root [kdevtmpfs] 9 root [netns] 10 root [kworker/0:1-pm] 11 root [oom_reaper] 12 root [writeback] 13 root [kcompactd0] 38 root [kblockd] 39 root [watchdogd] 40 root [cfg80211] 41 root [kswapd0] 43 root [spi0] 44 root [kworker/u2:1] 46 root [spi1] 47 root [irq/21-sunxi-mm] 53 root [kworker/0:1H-kb] 58 root [jffs2_gcd_mtd3] 78 root /sbin/syslogd -n 82 root /sbin/klogd -n 93 root /sbin/udevd -d 112 root [kworker/u2:2-ev] 124 root /usr/sbin/haveged -w 1024 -r 0 145 root /usr/sbin/umtprd 150 root -sh 160 root [ESP_SPI_WORK_QU] 173 root [kworker/0:2-eve] 179 root [kworker/0:0-eve] 181 root ps -e ```I found line of code with same name of task:
esp-hosted/esp_hosted_ng/host/main.c
Line 813 in fe0b105
Where I may wrong?
The text was updated successfully, but these errors were encountered: