search

espressif / arduino-esp32

Arduino core for the ESP32
GNU Lesser General Public License v2.1
13.42k stars 7.37k forks source link

Windows 10 PC can't access AP Mode. #2910

Closed GeorgeFlorian closed 4 years ago

GeorgeFlorian commented 5 years ago

Board: ESP32 DevKit V4 with Wrover-B OS: Windows 10 platform = https://github.com/platformio/platform-espressif32.git

I'm having a peculiar issue here. I can't access the WebServer when I am connected to the ESP32 AP Mode using Windows 10. But I can access the WebServer when I am connected to the AP using my Galaxy Note 9.

and

After the PC fails to access the WebServer , when trying to connect with the phone to the AP it doesn't work. It seems like I am limited to one connection to the AP. It just doesn't work. No debug message, no nothing. Not even the phone returns an error. Which is rather strange I'd say...

This is how I set up AP Mode.

//------------------------- logOutput(String)
void logOutput(String string1) {
    circle.push(string1);   
    Serial.println(string1);    
}

logOutput((String)"Starting AP ... ");
  logOutput(WiFi.softAP(ssidAP, passwordAP) ? "Ready" : "WiFi.softAP failed ! (Password must be at least 8 characters long )");
  delay(500);
  logOutput((String)"Setting AP configuration ... ");
  logOutput(WiFi.softAPConfig(local_IP_AP, local_IP_AP, subnetAP) ? "Ready" : "Failed!");
  delay(500);
  logOutput((String)"Soft-AP IP address: ");
  logOutput(WiFi.softAPIP().toString());

Nothing strange until now. This is the -DCORE_DEBUG_LEVEL=5 aka Verbose :

[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 15 - AP_STACONNECTED
dhcps: send_offer>>udp_sendto result 0
[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 17 - AP_STAIPASSIGNED

See the dhcps: send_offer>>udp_sendto result 0 ? I don't get that when I use the phone first and try to access the AP.

And that is all. No other verbose message.

EDIT: I have tried with another Windows 10 Laptop PC and it still does not work.

Nathan-ma commented 5 years ago

Same here, Ap mode seems to be broken!!

dhcps: send_nak>>udp_sendto result 0

Ap mode seems to broken, I already tried normal webserver and AsyncWebServer, both are not working properly, they just stop working from about 90 seconds after you connect with AP.

I can connect with my phone no problem, but after 90 seconds, AP just breaks and I have to disconnect and reconnect the WiFI.

GeorgeFlorian commented 5 years ago

That's just great ! Is there any other way to start/use AP ?

EDIT: I have tried with another Windows 10 Laptop PC and it still does not work.

atanisoft commented 5 years ago

Could this be related to #2899?

Nathan-ma commented 5 years ago

@atanisoft , well I did the test with heap memory and I can confirm the free heap memory size drops when the problem starts like mentioned on the issue you commented!

So Yes I do believe its related...

So the solution might be just change the sdkconfigs on esp-idf , build and replace the libesp32.a to make it work?

atanisoft commented 5 years ago

@DarvosBR yes, that should work to test. I suspect there may be a few other LwIP settings which would be useful to adjust slightly to increase stability in the arduino-esp32 builds.

GeorgeFlorian commented 5 years ago

Could this be related to #2899?

Probably.

So you are saying that this should be a work-around:

Actually, you don't need to re-compile the arduino-esp32 core to change the buffer values.

All you need to do is:

  • Edit \tools\sdk\include\config\sdkconfig.h and adjust the values to match sdkconfig.default from the iperf example folder.
  • Force a full rebuild when compiling your project (not sure how to do this directly from the UI, but changing the flash frequency or upload speed is usually enough to force it).

This applies to the arduino environment. I have no experience with PiO, so can't help you there.

You should see a 13-14kB drop in free heap (as compared to arduino-esp32 defaults). If you don't see this drop, then the changes are not active, and you need to check what you did wrong

On a side note: The buffer values from the iperf config are rather extreme, so later you should probably have a go at it and adjust them to best match your specific application needs.

Also, I've seen you mentioning a possible solution to the AsyncTCP freeze. That would be something extraordinary indeed.

atanisoft commented 5 years ago

@GeorgeFlorian I haven't tested the proposed rebuild but I don't think the steps outlined above will work since this is all pre-compiled into libesp32.a. The only solution I know of is to setup an esp-idf build (cmake or similar) with the arduino-esp32 sdkconfig, build and copy the libs over to the tools/sdk tree where they are today.

I'm not certain it will fix the async code but there may be a small chance that they are related.

GeorgeFlorian commented 5 years ago

@GeorgeFlorian I haven't tested the proposed rebuild but I don't think the steps outlined above will work since this is all pre-compiled into libesp32.a. The only solution I know of is to setup an esp-idf build (cmake or similar) with the arduino-esp32 sdkconfig, build and copy the libs over to the tools/sdk tree where they are today.

I'm not certain it will fix the async code but there may be a small chance that they are related.

That sounds like nothing I have done. I've never used esp-idf and it sounds tricky to use.

As I am now, my best bet is to wait for a stable arduino-core 1.03 in which hopefully somebody will fix have fixed this issue.

Nathan-ma commented 5 years ago

@GeorgeFlorin Don't worry, I'm gonna try both solutions and have a return for you, independent of which solution works (if any do), I will tell you how and event provide file if needed!

I got your back George!

r1dd1ck commented 5 years ago

@atanisoft

I haven't tested the proposed rebuild but I don't think the steps outlined above will work

so you have nothing to back your opinion up with, but you "think" it's garbage.. OK :grin:

I can only say that it worked for me - fixing the stack freeze issue #2899 (with empirical evidence to back that up).

Those buffer parameters are not a fixed part of the pre-compiled core. You can easily check it by increasing the static buffer values and watch your free heap drop as a result.

And to also say something on-topic of #2910 - I don't see how it could be directly related, but who knows.

Easy enough to try, right? :stuck_out_tongue_winking_eye:

atanisoft commented 5 years ago

so you have nothing to back your opinion up with, but you "think" it's garbage.. OK

Just experience from how the build process works in the environments I've used. Anything that is part of IDF is precompiled before any of the arduino-esp32 stack is compiled. Changes to sdkconfig (or sdkconfig.h) will usually require a recompilation of the IDF pieces for the changes to take effect. This is not an issue if you are using arduino as an IDF component though.

Those buffer parameters are not a fixed part of the pre-compiled core. You can easily check it by increasing the static buffer values and watch your free heap drop as a result.

How is this possible, these are used only on the IDF side and IDF code is not recompiled inside Arduino IDE or in PIO when using the arduino framework option.

GeorgeFlorian commented 5 years ago

@GeorgeFlorin Don't worry, I'm gonna try both solutions and have a return for you, independent of which solution works (if any do), I will tell you how and event provide file if needed!

I got your back George!

That is great to hear, man ! Thank you @DarvosBR !

But I don't think that https://github.com/espressif/arduino-esp32/issues/2899 is related. You and me both experience issues when trying to connect to the ESP while it's in AP Mode and that is all. We aren't sending large enough UDP Packets, if we send any at all, so that the Network Stack should crash.

Also, my ESP can stay in AP Mode indefinitely if I made a stable connection or no connection at all. I can connect and disconnect using my phone all day long and it will still work. It will "block" connections only when a failed connection happens. So if I try to connect using my PC and it doesn't work, this will make the ESP not able to accept further connection. The strange thing is that the PC says that it is connected to the respective AP Network, but the server is inaccessible.

r1dd1ck commented 5 years ago

@atanisoft

How is this possible, these are used only on the IDF side and IDF code is not recompiled inside Arduino IDE or in PIO when using the arduino framework option.

:roll_eyes: Too many words, too little effort to find out yourself..

@GeorgeFlorian

The strange thing is that the PC says that it is connected to the respective AP Network, but the server is inaccessible.

Have you checked whether the DHCP gives you a correct IP address/subnet etc. when that happens?

Nathan-ma commented 5 years ago

@r1dd1ck unfortunately the sdkconfig.h solution did not work. I'm gonna try the esp-idf, but I do agree this method should work.

@GeorgeFlorian The whole code works just fine when I access the board by the STA IP. But on AP the code just freezes after a while (the board keeps working, the STA keeps working, but AP just frezzes.

atanisoft commented 5 years ago

Too many words, too little effort to find out yourself..

There is no need to be rude about this. In all other cases of IDF sdkconfig it will require a recompilation of the IDF code for the config change to take effect.

This however is one case where a recompilation is not necessary ONLY because the WiFi lib is calling esp_wifi_init with a locally created wifi_init_config_t (https://github.com/lbernstone/arduino-esp32/blob/master/libraries/WiFi/src/WiFiGeneric.cpp#L114) which the WIFI_INIT_CONFIG_DEFAULT expands to use the sdkconfig.h entries (https://github.com/espressif/esp-idf/blob/e7f85f1987aa9479c2dbab638ca83bcaef99be00/components/esp_wifi/include/esp_wifi.h#L193-L213). This is not typically the case.

@GeorgeFlorian give it a shot and see if it fixes it, if it does then https://github.com/espressif/arduino-esp32/pull/2912 should fix your issues when it has been merged and included in the next release.

Nathan-ma commented 5 years ago

well at least I have something new to share with you guys:

New error msg:

dhcps: send_offer>>udp_sendto result 0 dhcps: send_nak>>udp_sendto result 0

This send_offer msg had never appear before, at least not that I had noticed.

r1dd1ck commented 5 years ago

@atanisoft

@r1dd1ck Too many words, too little effort to find out yourself.. There is no need to be rude about this.

That's not being rude, that's just stating a fact.

I'm sorry, but I go through the effort to provide a working solution, and then someone persistently keeps writing that it's garbage without even bothering to actually try once.. :neutral_face:

This however is one case where a recompilation is not necessary

Glad you have finally checked, and acknowledge now :+1:

Have a nice sunny day :slightly_smiling_face:

Nathan-ma commented 5 years ago

Oh guys this is exciting, we are getting somewhere here!!

CORRUPT HEAP: Bad head at 0x3ffcf4a4. Expected 0xabba1234 got 0x3ffcf574 assertion "head != NULL" failed: file "/Users/ficeto/Desktop/ESP32/ESP32/esp-idf-public/components/heap/multi_heap_poisoning.c", line 214, function: multi_heap_free abort() was called at PC 0x400dd8a7 on core 0

Backtrace: 0x4008cbac:0x3ffb3c60 0x4008cddd:0x3ffb3c80 0x400dd8a7:0x3ffb3ca0 0x4008c821:0x3ffb3cd0 0x40084f16:0x3ffb3cf0 0x400864bd:0x3ffb3d10 0x4000bec7:0x3ffb3d30 0x400f61c3:0x3ffb3d50 0x400f6223:0x3ffb3d70 0x400f626b:0x3ffb3d90 0x400f895a:0x3ffb3db0 0x400f89c2:0x3ffb3dd0 0x400d6ba1:0x3ffb3df0 0x400f5724:0x3ffb3e10 0x40088b49:0x3ffb3e40

Rebooting...

The new error issue I got!

so just a few extra informations, I just did a Fresh 1.0.2 instalation of the arduino-esp32 using the board manager from arduino IDE and the only thing I modify was the this:

WIFI_STATIC_RX_BUFFER_NUM = 8 → 10
WIFI_DYNAMIC_RX_BUFFER_NUM = 10 → 32
WIFI_RX_BA_WIN = 16 → 6

And yes I did try recompiling libesp32.a on idf on the last attempts, but this one was the original comp

r1dd1ck commented 5 years ago

@DarvosBR Oh guys this is exciting, we are getting somewhere here!!

Well, so far it only got you to a point where you will need to do a full manual clean-up & re-install :yum: because that heap corruption is very likely caused by some old files not being properly removed/replaced in the process :hankey:

Or if not, then maybe forcing a full rebuild will help - the caching algorithm sometimes acts silly :roll_eyes:

atanisoft commented 5 years ago

CORRUPT HEAP: Bad head at 0x3ffcf4a4. Expected 0xabba1234 got 0x3ffcf574 assertion "head != NULL" failed: file "/Users/ficeto/Desktop/ESP32/ESP32/esp-idf-public/components/heap/multi_heap_poisoning.c", line 214, function: multi_heap_free abort() was called at PC 0x400dd8a7 on core 0

I agree with @r1dd1ck on the cleanup/rerun. But, it would be good to decode the stacktrace before cleanup/rerun. If this is from AsyncUDP there was this merge on master that is not on 1.0.2: https://github.com/espressif/arduino-esp32/commit/e9389e3122fc47227f961eaaa6541380ef4f2b0c

GeorgeFlorian commented 5 years ago

Coming back with some updates of my own:

Without changing the code I know get dhcps: send_offer>>udp_sendto result 0 even when I am connecting with my phone. It can connect though.

Here are the serial debug verbose when I hit Connect (to ESP's AP) on my phone:

Soft-AP IP address: 
109.108.112.114
[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 15 - AP_STACONNECTED
[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 17 - AP_STAIPASSIGNED
[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 17 - AP_STAIPASSIGNED
[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 17 - AP_STAIPASSIGNED
dhcps: send_offer>>udp_sendto result 0
dhcps: send_offer>>udp_sendto result 0
dhcps: send_offer>>udp_sendto result 0
[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 17 - AP_STAIPASSIGNED

As you can see: [D][WiFiGeneric.cpp:336] _eventCallback(): Event: 17 - AP_STAIPASSIGNED and dhcps: send_offer>>udp_sendto result 0 show up 3 times just for one connection.

What is happening ?

I am using Platformio IDE and here is how the platformio.ini looks like:

[env:esp32doit-devkit-v1]
platform = https://github.com/platformio/platform-espressif32.git
board = esp32dev
framework = arduino
board_build.flash_mode = qio
upload_port = /dev/ttyUSB0
monitor_port = /dev/ttyUSB0
monitor_speed = 115200

build_flags = -DCORE_DEBUG_LEVEL=5

#libraries
lib_deps =  https://github.com/me-no-dev/ESPAsyncWebServer.git
            https://github.com/me-no-dev/AsyncTCP.git
            https://github.com/marcmerlin/SmartMatrix_GFX.git
            https://github.com/marcmerlin/Framebuffer_GFX.git
            https://github.com/adafruit/Adafruit-GFX-Library.git
            https://github.com/FastLED/FastLED.git

As you can see from platform = https://github.com/platformio/platform-espressif32.git I am using the develop branch of platformio which gets updated pretty often. I don't know if updated regarding this issue were pushed though.

Nathan-ma commented 5 years ago

@GeorgeFlorian yep same issue here, let me ask you something about this George (and I want to to know currently working to solve this problem my own, as soon as I get success I will let you know).

My question is: when you do connect with your phone, you have a stable connection even when that thing appears on your serial monitor (the DHCPS thing)? for me, no matter what I do, after 30 seconds connection freezes and I can refresh the page for example, until I quit the Wifi connection and reconnect.

Can you please try this for me ? connect it to your phone, refresh like 3 times with a 20 seconds interval and see if you can successfully do a fourth refresh!

Thank you very much

GeorgeFlorian commented 5 years ago

refresh like 3 times with a 20 seconds interval and see if you can successfully do a fourth refresh!

Refresh what ?

If I connect to the ESP's AP using the phone, the connection is stable for an undefined period of time.

Nathan-ma commented 5 years ago

Refresh I meant if you are using a web server, but just now it hit me, maybe that's not your case!

GeorgeFlorian commented 5 years ago

Refresh I meant if you are using a web server, but just now it hit me, maybe that's not your case!

Actually it is. I am using ESPAsyncWebServer. And it does not crash in AP, even after 4 refreshes.

Nathan-ma commented 5 years ago

than I must recheck my hardware, i'm using ESPAsyncwebserver myself to read the value of a simple sensor and after a while AP freezes. Thank you George, this piece of information is valuable

GeorgeFlorian commented 5 years ago

than I must recheck my hardware, i'm using ESPAsyncwebserver myself to read the value of a simple sensor and after a while AP freezes. Thank you George, this piece of information is valuable

I haven't changed the code. Here is a little update:

I am now able to connect to the ESP32's AP using an Asus G771JM Laptop, that has Intel Dual Band Wireless-AC 7260. It seems pretty stable. What I don't like is the fact that it takes more than 30 seconds to connect to it. Using a phone takes under 5 seconds. Using a PC with a ROG Strix Z270E Gaming Motherboard that has Qualcomm Atheros Wireless I was able to connect just twice in many, many tries.

Also, regarding the ESPAsyncWebServer in AP, I still get some hangings and "infinite loading" after some time. The first 2-3 refreshes work well. While it does NOT crash it sends me warnings:

[W][AsyncTCP.cpp:903] _poll(): rx timeout 4
[W][AsyncTCP.cpp:903] _poll(): ack timeout 4

Closing the browers and trying again seems to fix it.

While in STATION mode, it works fairly fine for an undefined period of time, even after multiple refreshes.

The only thing that I do not like in this situation is the fact that it's slow and after long-ish period of times in which a web-page was left open in the browser, upon refreshing the page or clicking a submit button, it sends either one of the following warnings or both:

[W][AsyncTCP.cpp:883] _poll(): pcb is NULL
[W][AsyncTCP.cpp:903] _poll(): rx timeout 4

After an hour of letting the ESP32 in STA Mode and a web-page open in the browser I got a Guru Meditation Error:

Guru Meditation Error: Core  0 panic'ed (LoadProhibited). Exception was unhandled.
Core 0 register dump:
PC      : 0x4010ff5e  PS      : 0x00060630  A0      : 0x8010bb26  A1      : 0x3ffd50b0  
A2      : 0x3ffdfa84  A3      : 0x00000020  A4      : 0x00000002  A5      : 0x3ffb9c40  
A6      : 0x00000000  A7      : 0x3ffd45e0  A8      : 0x3741a32b  A9      : 0xabba13d7  
A10     : 0x74786574  A11     : 0x000001a3  A12     : 0xabba1234  A13     : 0x00000b38  
A14     : 0x00000020  A15     : 0x00000000  SAR     : 0x00000019  EXCCAUSE: 0x0000001c  
EXCVADDR: 0x00000030  LBEG    : 0x4000c2e0  LEND    : 0x4000c2f6  LCOUNT  : 0xffffffff  

Backtrace: 0x4010ff5e:0x3ffd50b0 0x4010bb23:0x3ffd50e0 0x401377d9:0x3ffd5100 0x401084d8:0x3ffd5120 0x400887fd:0x3ffd5150
0x400887fd: rtc_clk_slow_freq_set at /Users/ficeto/Desktop/ESP32/ESP32/esp-idf-public/components/soc/esp32/rtc_clk.c line 256

The issue with this is that after the reboot, the code that's on the ESP32 does NOT work. The serial monitor doesn't say a thing. I have to manually reset the ESP by pressing the on-board reset button.

Here is an issue that I've just opened on ESPAsyncWebServer's github detailing more of the slow-downs I experience : https://github.com/me-no-dev/ESPAsyncWebServer/issues/535

GeorgeFlorian commented 5 years ago

Any luck with this ?

I still can't connect all the time using a laptop or a PC. So the IP gets assigned but I can't connect.

[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 17 - AP_STAIPASSIGNED

but there is no 

[D][WiFiGeneric.cpp:336] _eventCallback(): Event: 15 - AP_STACONNECTED
MkLHX commented 5 years ago

there is something not work actually on AsyncTCP i have same kind of troubles.

Server not accessible in AP mode, tcp communication crash, werbserver crash sometimes...

https://github.com/espressif/arduino-esp32/issues/2995 https://github.com/espressif/arduino-esp32/issues/2979

GeorgeFlorian commented 5 years ago

there is something not work actually on AsyncTCP i have same kind of troubles.

Server not accessible in AP mode, tcp communication crash, werbserver crash sometimes...

2995

2979

I don't think that the AsyncTCP or the ESPAsyncWebServer are at fault here. I think that it has something to do with WiFi . I can't connect using my Home PC to the ESP32's AP, but I can use my office Laptop to connect to it and the web-server works perfectly fine. They both run under Windows 10.

Also, check this: https://github.com/espressif/arduino-esp32/issues/3037

MkLHX commented 5 years ago

there is something not work actually on AsyncTCP i have same kind of troubles. Server not accessible in AP mode, tcp communication crash, werbserver crash sometimes...

2995

2979

I don't think that the AsyncTCP or the ESPAsyncWebServer are at fault here. I think that it has something to do with WiFi . I can't connect using my Home PC to the ESP32's AP, but I can use my office Laptop to connect to it and the web-server works perfectly fine. They both run under Windows 10.

Also, check this: #3037

Thanks, made the same, i cross my fingers it's seems to work...

MkLHX commented 5 years ago

Hi!

troubles back again https://github.com/espressif/arduino-esp32/issues/3037#issuecomment-530931276

stale[bot] commented 4 years ago

[STALE_SET] This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 14 days if no further activity occurs. Thank you for your contributions.

stale[bot] commented 4 years ago

[STALE_DEL] This stale issue has been automatically closed. Thank you for your contributions.

melvinpmathew commented 4 years ago

@GeorgeFlorian did you get this issue resolved? Win10 + AP mode?

johnnytolengo commented 1 year ago

has anyone fix that issue?