Heltec Crash....need help!? #195
-
|
Ok, I've been running the below effects just fine on a single channel, running them all day and all night without an issue. If you've followed my other posts, I finally got multichannel to work (4 51 pixel led strips) running on pins 5, 17, 18, 19 on the heltec esp32. Running the same effect cycle with multi-channels introduced issues....first, the StarryNightEffect seems to only run on a single Channel....i'll dig into that time permitting cause i really like those popping pixels. What happens is, around the 6th or 7th effect, the chip reboots and pukes out a mem dump, recycles and starts all over (back to effect no 1). I have been taking the suspected "violator effect" and systematically moved that effect to the top of the list (in effects.cpp) and then uploading to specifically test that effect. As I monitor the effect changes, it seems to take place at a certain "time" into the light show, ie blows up on effect number 10 in the cycle (i'm still evaluating that), reason i say that is the effects work when i start the cycle with suspect effect. Looks like to me memory is getting corrupted, but this is getting over my head so want to solicit some help PLEASE! Here is the memory dump, and the effects are below that... (I) (loop)(C1) Mem: 92032, LargestBlk: 64500, PSRAM Free: 0/0, Buffer: 0/144, LED FPS: 59, LED Watt: 4, LED Brite: 100%, Audio FPS: 38, Serial FPS: 0, PeakVU: 0.00, MinVU: 0.00, VURatio: 0.00, CPU: 07%, 16%, FreeDraw: 0.014376 Core 1 register dump: Backtrace: 0x4008ff70:0x3ffb2650 0x40090bd3:0x3ffb2670 0x40090e2f:0x3ffb2690 0x400fe7d6:0x3ffb26b0 0x400fe825:0x3ffb26f0 0x400dee16:0x3ffb2730 0x400d8d71:0x3ffb2750 0x400e1015:0x3ffb2820 #0 0x4008ff70:0x3ffb2650 in block_size at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf_block_functions.h:75 (discriminator 4) ELF file SHA256: 10fe6a27684c90dc E (6630) esp_core_dump_flash: Core dump flash config is corrupted! CRC=0x7bd5c66f instead of 0x0 rst:0xc (SW_CPU_RESET),boot:0x17 (SPI_FAST_FLASH_BOOT) |
Beta Was this translation helpful? Give feedback.
Replies: 17 comments 15 replies
-
|
It’s possible it’s memory corruption, but more likely perhaps low memory. Remember it has to reserved buffers for every channel if you have incoming wifi enabled, so that’s can bite you too. How much RAM is free at the time this happens? If it’s under 60-70K the system gets dicey.
However, if its a simply access violation in one of the effects for some reason, you’ll have to find that one and fix it :-)
Can I ask where your stack traces come from? Normally I don’t get a full trace into the ESP code, but did you perhaps build it yourself?
- Dave
… On Jan 7, 2023, at 9:32 AM, stretch911 ***@***.***> wrote:
Ok, I've been running the below effects just fine on a single channel, running them all day and all night without an issue. If you've followed my other posts, I finally got multichannel to work (4 51 pixel led strips) running on pins 5, 17, 18, 19 on the heltec esp32. Running the same effect cycle with multi-channels introduced issues....first, the StarryNightEffect seems to only run on a single Channel....i'll dig into that time permitting cause i really like those popping pixels.
What happens is, around the 6th or 7th effect, the chip reboots and pukes out a mem dump. I have been taking the suspected "violator effect" and systematically moved that effect to the top of the list (in effects.cpp) and then uploading to specifically test that effect. As I monitor the effect changes, it seems to take place at a certain "time" into the light show, ie blows up on effect number 10 in the cycle (i'm still evaluating that), reason i say that is the effects work when i start the cycle with suspect effect. Looks like to me memory is getting corrupted, but this is getting over my head so want to solicit some help PLEASE!
Here is the memory dump, and the effects are below that...
(I) (loop)(C1) Mem: 92032, LargestBlk: 64500, PSRAM Free: 0/0, Buffer: 0/144, LED FPS: 59, LED Watt: 4, LED Brite: 100%, Audio FPS: 38, Serial FPS: 0, PeakVU: 0.00, MinVU: 0.00, VURatio: 0.00, CPU: 07%, 16%, FreeDraw: 0.014376
(I) (loop)(C1) Mem: 92032, LargestBlk: 536870900, PSRAM Free: 0/0, Buffer: 0/144, LED FPS: 58, LED Watt: 3, LED Brite: 100%, Audio FPS: 38, Serial FPS: 0, PeakVU: 0.00, MinVU: 0.00, VURatio: 0.00, CPU: 07%, 17%, FreeDraw: 0.014241
Guru Meditation Error: Core 1 panic'ed (LoadProhibited). Exception was unhandled.
Core 1 register dump:
PC : 0x4008ff73 PS : 0x00060d33 A0 : 0x80090bd6 A1 : 0x3ffb2650
A2 : 0x00d9476c A3 : 0x40090c24 A4 : 0x3ffb26b4 A5 : 0x40d43f70
A6 : 0xfffffffc A7 : 0x00000001 A8 : 0x8008ff8d A9 : 0x3ffb2630
A10 : 0x40d43f70 A11 : 0x20c85f70 A12 : 0x00000000 A13 : 0x3ffb26b4
A14 : 0x00060d20 A15 : 0x00000001 SAR : 0x0000000a EXCCAUSE: 0x0000001c
EXCVADDR: 0x00d94770 LBEG : 0x40087b18 LEND : 0x40087b23 LCOUNT : 0x00000000
Backtrace: 0x4008ff70:0x3ffb2650 0x40090bd3:0x3ffb2670 0x40090e2f:0x3ffb2690 0x400fe7d6:0x3ffb26b0 0x400fe825:0x3ffb26f0 0x400dee16:0x3ffb2730 0x400d8d71:0x3ffb2750 0x400e1015:0x3ffb2820
#0 0x4008ff70:0x3ffb2650 in block_size at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf_block_functions.h:75 (discriminator 4)
(inlined by) block_is_last at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf_block_functions.h:86 (discriminator 4)
(inlined by) tlsf_walk_pool at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf.c:603 (discriminator 4)
#1 <#1> 0x40090bd3:0x3ffb2670 in multi_heap_get_info_impl at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/multi_heap.c:412
#2 <#2> 0x40090e2f:0x3ffb2690 in multi_heap_get_info at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/multi_heap_poisoning.c:364
#3 <#3> 0x400fe7d6:0x3ffb26b0 in heap_caps_get_info at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_caps.c:535
#4 <#4> 0x400fe825:0x3ffb26f0 in heap_caps_get_largest_free_block at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_caps.c:523
#5 <#5> 0x400dee16:0x3ffb2730 in EspClass::getMaxAllocHeap() at C:/Users/sever/.platformio/packages/framework-arduinoespressif32/cores/esp32/Esp.cpp:151
#6 <#6> 0x400d8d71:0x3ffb2750 in loop() at src/main.cpp:866
#7 <#7> 0x400e1015:0x3ffb2820 in loopTask(void*) at C:/Users/sever/.platformio/packages/framework-arduinoespressif32/cores/esp32/main.cpp:50
ELF file SHA256: 10fe6a27684c90dc
E (6630) esp_core_dump_flash: Core dump flash config is corrupted! CRC=0x7bd5c66f instead of 0x0
Rebooting...
ets Jun 8 2016 00:22:57
rst:0xc (SW_CPU_RESET),boot:0x17 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0xee
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:2
load:0x3fff0030,len:1184
load:0x40078000,len:13104
load:0x40080400,len:3036
entry 0x400805e4
E (351) esp_core_dump_flash: No core du���partition found!
E (351) esp_core_dump_flash: No core dump partition found!
(I) (PrintOutputHeader)(C1) NightDriverStrip
new BouncingBallEffect(3, true, true, 1),
new PaletteEffect(RainbowColors_p, 4, 0.1, 0.0, 1.0, 0.0),
new ClassicFireEffect(false, false, 20),
new PaletteEffect(OceanColors_p, 4, 0.1, 0.0, 1.0, 0.0),
new TwinkleStarEffect(),
//new StarryNightEffect<QuietStar>("Ocean Stars", OceanColors_p, STARRYNIGHT_PROBABILITY, 1, LINEARBLEND, 2.0, 0.0, STARRYNIGHT_MUSICFACTOR),
new MeteorEffect(1, 1, 5, .15, .25),
//new StarryNightEffect<BubblyStar>("Bubbly Stars", USAColors_p, STARRYNIGHT_PROBABILITY, 1, LINEARBLEND, 2.0, 0.0, STARRYNIGHT_MUSICFACTOR),
new FireFanEffect(NUM_LEDS, 1, 10, 80, 2, 7, Reverse, false, false, true),
new ColorFillEffect(CRGB::DarkCyan, 1),
//new StarryNightEffect<QuietStar>("Rainbow Twinkle Stars", RainbowColors_p, STARRYNIGHT_PROBABILITY, 1, LINEARBLEND, 2.0, 0.0, STARRYNIGHT_MUSICFACTOR),
new ColorCycleEffect(Sequential),
//new StarryNightEffect<BubblyStar>("Little Blooming Rainbow Stars", BlueColors_p, 1.0, 2, LINEARBLEND, 2.0, 0.0, 1), // Blooming Little Rainbow Stars
//new PaletteEffect(RainbowColors_p, 4, 0.1, 0.0, 1.0, 0.0),
new MeteorEffect(6, 1, 25, .15, .12),
new FireFanEffect(NUM_LEDS, 1, 10, 80, 2, 7, Reverse, false, false),
new MeteorEffect(20, 1, 25, .15, .05),
new RainbowFillEffect(6, 2),
new ClassicFireEffect(false, false, 20),
new MeteorEffect(12, 1, 25, .15, .08),
new RainbowTwinkleEffect(2.0, 14),
new MeteorEffect(),
—
Reply to this email directly, view it on GitHub <#195>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA4HCFZRNJFDD5RDQFIDQZ3WRGSDZANCNFSM6AAAAAATUCAK6E>.
You are receiving this because you are subscribed to this thread.
|
Beta Was this translation helpful? Give feedback.
-
|
mem dump shows plenty of ram .... I'm running barebones esp32 WROOM...should be PLENTY of room. I'm not running wifi, or webserver, or sound...or IR... The mem dump came from monitoring the serial port on a HiLetgo ESP32 OLED WiFi Kit V3 Type-C ESP-32 0.96 Inch Blue OLED Display WiFi+Bluetooth CP2012 Internet Development Board for Arduino ESP8266 NodeMCU purchased from Amazon. A little disappointed after spending all this time to get things going that it doesn't run more than a couple hours, all this instability. Lamp will be cool in the background for my team meetings, but that's about it... |
Beta Was this translation helpful? Give feedback.
-
|
one thing i did notice was this value in the logs... what causes LargestBlk to go bananas? (not even sure what it means to be perfectly honest) |
Beta Was this translation helpful? Give feedback.
-
|
Well, if there’s bug, that’s why the code is provided! It’d be great if you could isolate and locate the issue before others run into it.
FWIW, the Atomic Fire Lamp has only 4 channels but works well with a large set of effects. Or at least it worked last time I made one, and others have built it too and send photos, so I know the code works in general.
- Dave
… On Jan 9, 2023, at 7:11 AM, stretch911 ***@***.***> wrote:
mem dump shows plenty of ram ....
i cut back the number of effects to 9, cause seemed to be puking after the 10th effect started up and things run good for an hour or hour and a half, then all of sudden things get frozen. After learning some lessons, the chip is protected w/ capacitors now, so reset gets it going again...but this doesn't seem like proper behavior. LEDs have PLENTY of power too. I don't know why others aren't having the same problem unless everyone else is running on different (older) libraries, or even whatever the chip is flashed with.
I'm running barebones esp32 WROOM...should be PLENTY of room. I'm not running wifi, or webserver, or sound...or IR...
The mem dump came from monitoring the serial port on a HiLetgo ESP32 OLED WiFi Kit V3 Type-C ESP-32 0.96 Inch Blue OLED Display WiFi+Bluetooth CP2012 Internet Development Board for Arduino ESP8266 NodeMCU purchased from Amazon.
A little disappointed after spending all this time to get things going that it doesn't run more than a couple hours, all this instability. Lamp will be cool in the background for my team meetings, but that's about it...
—
Reply to this email directly, view it on GitHub <#195 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA4HCF2LYVT4LWOCQIRE2C3WRQTB5ANCNFSM6AAAAAATUCAK6E>.
You are receiving this because you commented.
|
Beta Was this translation helpful? Give feedback.
-
|
i'll try because these are some sexy effects. i mean i'm only running 204 leds...shouldn't be this hard :) |
Beta Was this translation helpful? Give feedback.
-
|
Here’s my demo (w crash) and photo |
Beta Was this translation helpful? Give feedback.
-
|
i'll be playing w/ it some more because I'd like to get IR and WIFI WebServer and sound running. had to put it aside cause i was getting frustrated. |
Beta Was this translation helpful? Give feedback.
-
|
It looks as if the power is all going through the power rails of the protoboard. Yikes. I would have used direct wiring to the LEDs. Other than splitting the power and feeding both the uC and the LEDs directly (best option in your case), with other firmware like WLED, I sometimes power the LEDs directly and then feed from there to the uC. For example, I have 350 bullet pixels on arches (in one sequence) over the front walkway, to which I feed 12V to in the middle. I put the controller at one end of the string, where it gets power from the string rather than vice versa. This way I can feed power from the middle without needing to inject (or supply to the ESP) at either end of the string. |
Beta Was this translation helpful? Give feedback.
-
|
wanted to give a quick update. as you can see in these photos, the wiring is still in "dev mode". Thinking I'd exhausted my ability to debug the source code (which was looking like deep in some included library), I moved the "incomplete" lamp (chip still on breadboard) to a cool spot behind me for my Monday meetings. Side note, it was the first thing the tech guys on my call noticed and immediately asked "WHATS THAT!", so we went down a 15 minute rabbit hole on microprocessors and leds and sensors and stuff. Not sure what happened, but moving to a more stable location? this thing ran 9 effects (as seen above) for 5 mins an effect for like 2 days? then i was disappointed to see it froze again in mid flame...but was encouraged that this surely seemed like a power issue. And in my continued research the SLIGHTEST short or spike will flip these chips on their head. Fortunately I've learned to protect the chip w/ capacitors (and permanent install w/ have fuse) AND don't pull power thru the dev board, a simple reset and was good to go. I was still curious so I moved it back to my work bench and hooked up the serial port. again, the lamp ran for a solid 24 hrs before I moved it back to its display spot. I am out of esp32s (muwahaha) so have some ordered. my intent is to not mess up the ESP32 that is currently working, and on a separate ESP32, add extra effects, sound, IR, web, and wifi and then swap out the boards and make the wiring tidy and permanent. The point of all this is, is I don't think there is a bug. The bug is me. Couple lessons, when compiling this code, make sure you are using the correct board. PlatformIO has a huge list that you can parse. This is an overwhelming list, but keep in mind a heltec is NOT a heltec v2. Second lesson is SOLID WIRING...these lights can draw some current and the little bitty wires that came with your arduino kit can jiggle in the breadboard and stuff. Coming out of the power supply to the lights i am now using 18gauge stranded so there is plenty of bandwidth. Lesson 3, wire the dev board in parallel w/ the lights...not in serial. Pulling power thru that board will fry stuff. Amazon loves me now. :) I'll keep you posted on this thread as my son wants one now, so I see a 2nd build coming soon. my microphones (max4466) and IR sensors(KY022) arrive tomorrow. Stay tuned! |
Beta Was this translation helpful? Give feedback.
-
|
Dave....on the vid where you are doing the Lamp Build...you sorta rush the last few minutes where you install the microphone and the ir...i had to backup, pause...watch as you pan the camera by. looks like you also added a latch switch...would like more info. looks like you drilled a hole in that base and poked the ir sensor out...do you have two ir sensors? wired to the same board? hot glued into the base?....where or how did you mount that microphone? so many questions. thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Great vid on power regulation w/ solution: https://youtu.be/J66_8P043ko |
Beta Was this translation helpful? Give feedback.
-
Sorry for the late reply on this! The one I built for my son, I put a pushbutton on/off switch between the power jack and the rest of the circuit. Same as the microphone, I drilled a hole and stuck it in the base. I used hotglue to mount the mic :-) You can likely do better, or find a mountable mic! |
Beta Was this translation helpful? Give feedback.
-
|
update: amazon delivered 4 new esp32 minis: AITRIP 4pcs D1 Mini NodeMCU ESP32 ESP-WROOM-32 WLAN WiFi Bluetooth Development Board I'm going to use one for testing effects, adding music, wifi, ir, webserver and then once i have all that prototyped, will swap out esp32 in my working lamp. First thing I did after getting 4 strands of leds working was start adding effects...right around effect number 10, the board crashed and started over... :( (I) (loop)(C1) Mem: 88904, LargestBlk: 57332, PSRAM Free: 0/0, Buffer: 0/182, LED FPS: 62, LED Watt: 3, LED Brite: 7%, Audio FPS: 37, Serial FPS: 0, PeakVU: 0.00, MinVU: 0.00, VURatio: 0.00, CPU: 00%, 14%, FreeDraw: 0.015011 Core 1 register dump: Backtrace: 0x4008f1f8:0x3ffb2650 0x4008fe5b:0x3ffb2670 0x400900b7:0x3ffb2690 0x400fc60a:0x3ffb26b0 0x400fc659:0x3ffb26f0 0x400dcde2:0x3ffb2730 0x400d8a8d:0x3ffb2750 0x400deea5:0x3ffb2820 #0 0x4008f1f8:0x3ffb2650 in block_size at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/heap/heap_tlsf_block_functions.h:75 (discriminator 4) ELF file SHA256: c04731ebff3bfc38 E (14511) esp_core_dump_flash: Core dump flash config is corrupted! CRC=0x7bd5c66f instead of 0x0 rst:0xc (SW_CPU_RESET),boot:0x13 (SPI_FAST_FLASH_BOOT) |
Beta Was this translation helpful? Give feedback.
-
|
....continued... now that I have several chips running NightDriverStrip, I've been able to test many effects, very systematically. The reboot issue is not, because its trying to cycle thru too many effects...I'm running 23 effects at the moment without a hitch. As I was testing each effect, I found that several effects freaked the chip out and it would just reboot. those guys above, sorta make sense to me, because they are most likely (haven't deep dove) because I don't have a fan matrix...I don't have fans defined and these are Fan Array effects (i think). The one that was tripping me up on previous builds etc turns out to be: //new ColorCycleEffect(Sequential), constant reboot I used this as my first effect and the chip just went into constant reboot cycle. I commented it out, and my 23 effects have been running trouble free for a few days now. Thought this may help others. I haven't dug into why the ColorCycleEffect(Sequential) blows stuff up, but it does and I'm fine just commenting it out for now :) |
Beta Was this translation helpful? Give feedback.
-
|
Make sure your NUM_FANS and FAN_SIZE are set appropiately. If either was zero, it might wind up doing div by zero conceivably.
This only affects those effects (nice!) which use radial drawing, like fans and reels.
- Dave
… On Jan 17, 2023, at 6:05 AM, stretch911 ***@***.***> wrote:
....continued...
now that I have several chips running NightDriverStrip, I've been able to test many effects, very systematically.
The reboot issue is not, because its trying to cycle thru too many effects...I'm running 23 effects at the moment without a hitch.
As I was testing each effect, I found that several effects freaked the chip out and it would just reboot.
//new FanBeatEffect("FanBeat"), // constant reboot
//new TapeReelEffect("TapeReelEffect"), // constant reboot
//new PaletteReelEffect("PaletteReelEffect"), // constant reboot
those guys above, sorta make sense to me, because they are most likely (haven't deep dove) because I don't have a fan matrix...I don't have fans defined and these are Fan Array effects (i think).
The one that was tripping me up on previous builds etc turns out to be:
//new ColorCycleEffect(Sequential), constant reboot
I used this as my first effect and the chip just went into constant reboot cycle. I commented it out, and my 23 effects have been running trouble free for a few days now.
Thought this may help others. I haven't dug into why the ColorCycleEffect(Sequential) blows stuff up, but it does and I'm fine just commenting it out for now :)
—
Reply to this email directly, view it on GitHub <#195 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA4HCF4AI2IHQ55BF6RXCR3WS2RJPANCNFSM6AAAAAATUCAK6E>.
You are receiving this because you commented.
|
Beta Was this translation helpful? Give feedback.
-
|
So I picked up a Heltec V3, but of course it’s not yet supported by the PlatformIO boards list. Can you tell me what you did to get it to work with the V3, and I’ll see if I can narrow down the bug!
Thanks
Dave
… On Jan 9, 2023, at 7:31 AM, Dave Plummer ***@***.***> wrote:
Well, if there’s bug, that’s why the code is provided! It’d be great if you could isolate and locate the issue before others run into it.
FWIW, the Atomic Fire Lamp has only 4 channels but works well with a large set of effects. Or at least it worked last time I made one, and others have built it too and send photos, so I know the code works in general.
- Dave
> On Jan 9, 2023, at 7:11 AM, stretch911 ***@***.***> wrote:
>
>
> mem dump shows plenty of ram ....
> i cut back the number of effects to 9, cause seemed to be puking after the 10th effect started up and things run good for an hour or hour and a half, then all of sudden things get frozen. After learning some lessons, the chip is protected w/ capacitors now, so reset gets it going again...but this doesn't seem like proper behavior. LEDs have PLENTY of power too. I don't know why others aren't having the same problem unless everyone else is running on different (older) libraries, or even whatever the chip is flashed with.
>
> I'm running barebones esp32 WROOM...should be PLENTY of room. I'm not running wifi, or webserver, or sound...or IR...
>
> The mem dump came from monitoring the serial port on a HiLetgo ESP32 OLED WiFi Kit V3 Type-C ESP-32 0.96 Inch Blue OLED Display WiFi+Bluetooth CP2012 Internet Development Board for Arduino ESP8266 NodeMCU purchased from Amazon.
>
> A little disappointed after spending all this time to get things going that it doesn't run more than a couple hours, all this instability. Lamp will be cool in the background for my team meetings, but that's about it...
>
> —
> Reply to this email directly, view it on GitHub <#195 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA4HCF2LYVT4LWOCQIRE2C3WRQTB5ANCNFSM6AAAAAATUCAK6E>.
> You are receiving this because you commented.
>
|
Beta Was this translation helpful? Give feedback.
-
|
sorry just seeing this..i'm pretty sure mine is a v1. |
Beta Was this translation helpful? Give feedback.
-
|
The v3 chip discussion had made some progress in the issues tab. |
Beta Was this translation helpful? Give feedback.
-
|
Hey everyone, loving this library and set of effects! It doesn't seem GPIO-specific, I can move pins to other pins and behavior is the same. What I don't see is a crash dump, how can I configure more debugging stuff than the usual "Monitor"? I will move to a different room in my house to test that fun theory 😆 |
Beta Was this translation helpful? Give feedback.
-
|
Does this board have psram? We're dealing with a number of issues right now
with low memory configurations and more LEDs requires more memory.
If you scroll backwards in the serial monitor, what are the last few things
you see before it reboots?
…On Mon, Nov 13, 2023, 1:19 PM rdecarreau ***@***.***> wrote:
Hey everyone, loving this library and set of effects!
I am running a heltecv3wifi and setup NUM_CHANNELS to 3, each of just 6
LEDs, and I see a crash-loop on the Demo. 😞
2 channels of 6 works fine all day.
It doesn't seem GPIO-specific, I can move pins to other pins and behavior
is the same.
It doesn't seem power-specific; I only have 6 pins and even if I don't
wire anything the controller alone boot-loops.
What I don't see is a crash dump, how can I configure more debugging stuff
than the usual "Monitor"?
I will move to a different room in my house to test that fun theory 😆
—
Reply to this email directly, view it on GitHub
<#195 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCSD3YEOWKKG27CEI224X3YEJXC7AVCNFSM6AAAAAATUCAK6GVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TKNJXG44TM>
.
You are receiving this because you are subscribed to this thread.Message
ID:
<PlummersSoftwareLLC/NightDriverStrip/repo-discussions/195/comments/7557796
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
I used the Heltec link from the readme, this is the page from the manufacturer: https://heltec.org/project/wifi-kit-32-v3/. Looking at the spec, I see this list of memory: That doesn't seem to indicate that there is PSRAM, but there are two sets of SRAM. EDIT: I will update with serial output soon. |
Beta Was this translation helpful? Give feedback.
-
|
Figuring out esp32 hardware is probably one of the worst things about this
platform. I actually have a change in my tree that thoroughly identifies it
in startup and adds a command to the telnet debugger that tells you what's
actually present.
Both
https://docs.platformio.org/en/latest/boards/espressif32/heltec_wifi_kit_32_V3.html
and the schematic shows no mention of psram.
You didn't show the last few things printed before it crashed, but I'll bet
that message every five seconds shows you don't have enough memory.
Are you able to run the failing combination in a debugger? We may be able
to help you get a. Breakinpoint on a failed assertion,. For example.
Like Rutger did in a patch this weekend for spectrum, you may be able to
buy breathing room by disabling some of the other on board services.
I don't know if we've made a. Change in recent weeks that drove up memory
use or we're just facing the death of a thousand paper cuts (the trend in a
project is always to add code, rarely shrinking it) but we seem to be
facing a rash of these sorts of problems lately...well, assuming this IS
the problem, but I'm betting on it.
…On Mon, Nov 13, 2023, 2:54 PM rdecarreau ***@***.***> wrote:
I used the Heltec link from the readme, this is the page from the
manufacturer: https://heltec.org/project/wifi-kit-32-v3/.
Looking at the spec, I see this list of memory:
384KB ROM; 512KB SRAM; 16KB RTC SRAM; 8MB SiP Flash
That doesn't seem to indicate that there is PSRAM, but there are two sets
of SRAM.
—
Reply to this email directly, view it on GitHub
<#195 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCSD3ZA2LCLOR5NJ3YJH63YEKCHBAVCNFSM6AAAAAATUCAK6GVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TKNJYGUZTK>
.
You are receiving this because you commented.Message ID:
<PlummersSoftwareLLC/NightDriverStrip/repo-discussions/195/comments/7558535
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
--- Terminal on COM3 | 115200 8-N-1
Core 1 register dump: Backtrace: 0x7ad00baa:0x3fc97230 0x403781e5:0x3fc97280 0x40377b74:0x3fc972a0 0x00040022:0x3fc97360 |<-CORRUPTED #0 0x7ad00baa:0x3fc97230 in ?? ??:0 ELF file SHA256: 4cff054c421c8f69 E (4063) esp_core_dump_flash: Core dump flash config is corrupted! CRC=0x7bd5c66f instead of 0x0
Core 1 register dump: Backtrace: 0x40382c58:0x3fc97100 0x4202c335:0x3fc97120 0x4202c724:0x3fc97150 0x403781f9:0x3fc971a0 0x403792b8:0x3fc971c0 0x4202c6bf:0x3fc97280 0x40377b74:0x3fc972a0 0x00040022:0x3fc97360 |<-CORRUPTED #0 0x40382c58:0x3fc97100 in mwdt_ll_write_protect_disable at /Users/ficeto/Desktop/ESP32/ESP32S2/esp-idf-public/components/hal/esp32s3/include/hal/mwdt_ll.h:259 ELF file SHA256: 4cff054c421c8f69 E (2957) esp_core_dump_flash: Core dump flash config is corrupted! CRC=0x7bd5c66f instead of 0x0 |
Beta Was this translation helpful? Give feedback.
-
|
This is a couple boot loops. What's interesting is eventually it stays alive, but the effects are not running. I guess that is some feature? |
Beta Was this translation helpful? Give feedback.
-
|
It's interesting that the two present differently.
In both of them, you're dying while starting the network; well before the
first draw.
You're low on memory, but in both cases there's like 60+kb free in the
heap, so even though we're still creating threads and such, we're not into
what I'd expect to be code red.
To test if it's an issue from being low on memory, Disable the use_oled or
even use_wifi.and see if the breaking point moves.
Another, similar test would be to make the third pin the only one and/or
male the first two strips one px long. Either can change the memory
footprint.
I have esp32-s3 boards. I can artificially disable psram . Is your
configuration otherwise simple to reproduce or are there effects and
changes not checked in? (Or that I could get from another repo.)
Remember that for testing, you don't actually. Even need to connect
strips.. want to test 8 strips with 2000 lights? Configure it and run.
(That's actually. Almost sure to fail without psram. I'm just. Exaggerating
for effect that you don't even need to connect anything to the pins.
Oh.. there's. An idea.. are the pins you're using actually free on this
board or are they being used by that. Display? I can imagine chaos if we
try to hijack a. Pin used by other hardware.
…On Mon, Nov 13, 2023, 5:45 PM rdecarreau ***@***.***> wrote:
This is a couple boot loops. What's interesting is eventually it stays
alive, but the effects are not running. I guess that is some feature?
—
Reply to this email directly, view it on GitHub
<#195 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCSD364D3D5JZQVIFVXUE3YEKWJ5AVCNFSM6AAAAAATUCAK6GVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TKNJZGY2DI>
.
You are receiving this because you commented.Message ID:
<PlummersSoftwareLLC/NightDriverStrip/repo-discussions/195/comments/7559644
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Yeah so this is just the Demo for the heltecv3wifi; I configured 3 channels and that was all I had to do to replicate this. Also yeah good call on the pins to use, I do that wrong all the time (in my defense most of these datasheets are a bit ambiguous). Good ideas on disabling things! I went on a bit of a spree finding and 0-ing #defines after thinking this was a memory issue; so far still no changes have kept it from rebooting. Each time reporting a different EXCCAUSE, so far (I started reading up on these, quite interesting). I will probably buy a couple other variants of the esp32, maybe a seeed studio one as well because that form-factor is just super cool! My actual setup doesn't need many channels so I am refactoring for now to use a single channel of ~50px. |
Beta Was this translation helpful? Give feedback.
-
|
An interesting other note: |
Beta Was this translation helpful? Give feedback.
-
|
Good news! I can reproduce this.
Well, it's not the best news because it still crashes, but it means that
there's someone else on the case that's able to help you.
I still see the log "Creating SSD1306 Screen". Tried full clean, still
there? I must be doing something else wrong.
You're doing nothing wrong. Our build system is [ censored ]. This is
actually configured in three different places. (Rutger and I are arguing
about that in another PR right now, actually...) You're probably turning it
off in globals.h. It' being turned ON in platform.ini. :-/
It also catches my eye that you said you turned networking off, yet the
logs are clearly spawning a network thread and connecting to your WiFi. Is
this just a mixup where I'm confusing things or mixing discussions?
Anyway, I'm on the case!
…On Mon, Nov 13, 2023 at 9:33 PM rdecarreau ***@***.***> wrote:
An interesting other note:
Even though I have flagged USE_OLED to 0, I still see the log "Creating
SSD1306 Screen". Tried full clean, still there? I must be doing something
else wrong.
—
Reply to this email directly, view it on GitHub
<#195 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ACCSD33ILE6POXF42Q4K2IDYELRAFAVCNFSM6AAAAAATUCAK6GVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TKNRQHE2TI>
.
You are receiving this because you commented.Message ID:
<PlummersSoftwareLLC/NightDriverStrip/repo-discussions/195/comments/7560954
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Holy cow, you wouldn't believe this journey! Before I type a wall of words,
can you please add one line in platformio.ini, rebuild, and confirm success
or failure?
My working theory is that it's not out of memory at all (I can actually
reproduce this on my device with 16MB!) but is suffering from a bad
lock/IPL interaction when the LED completion interrupt runs on one
processor while the other processor is scribbling those .2Hz messages to
flash.
Please jam a
-D FASTLED_ESP32_FLASH_LOCK=1
right after the -DDEMO in your platformio.ini that entry for
[env:heltecv3demo] and report back.
Bonus points for quantifying any impact on the stability (flickering,
jankiness) of the LED strips as network and "disk" traffic is generated.
For network traffic, telnet to the board and run 'v' to set the debug level
to verbose where it'll generate a few lines of jibberish per frame. Type
'w' to bring the system back to printing things of severity 'w'arning or
higher. In case you've forgotten, control and ] will bring you back to the
telnet command prompt where you can 'quit' or send a standard EOF with
control and D.
This isn't the way I'll want to officially fix this (it's neither Heltec
nor DEMO specific) but this will be enough to see if this train of thought
pays off enough to keep pursuing it.
…On Mon, Nov 13, 2023 at 11:45 PM Robert Lipe ***@***.***> wrote:
Good news! I can reproduce this.
Well, it's not the best news because it still crashes, but it means that
there's someone else on the case that's able to help you.
> I still see the log "Creating SSD1306 Screen". Tried full clean, still
there? I must be doing something else wrong.
You're doing nothing wrong. Our build system is [ censored ]. This is
actually configured in three different places. (Rutger and I are arguing
about that in another PR right now, actually...) You're probably turning it
off in globals.h. It' being turned ON in platform.ini. :-/
It also catches my eye that you said you turned networking off, yet the
logs are clearly spawning a network thread and connecting to your WiFi. Is
this just a mixup where I'm confusing things or mixing discussions?
Anyway, I'm on the case!
On Mon, Nov 13, 2023 at 9:33 PM rdecarreau ***@***.***>
wrote:
> An interesting other note:
> Even though I have flagged USE_OLED to 0, I still see the log "Creating
> SSD1306 Screen". Tried full clean, still there? I must be doing something
> else wrong.
>
> —
> Reply to this email directly, view it on GitHub
> <#195 (reply in thread)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/ACCSD33ILE6POXF42Q4K2IDYELRAFAVCNFSM6AAAAAATUCAK6GVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TKNRQHE2TI>
> .
> You are receiving this because you commented.Message ID:
> <PlummersSoftwareLLC/NightDriverStrip/repo-discussions/195/comments/7560954
> @github.com>
>
|
Beta Was this translation helpful? Give feedback.
-
|
To your previous comment about the network config in my logs, yeah that was just me posting logs from a run where it was still enabled. I can successfully disable wifi. *** Remote debug - over telnet - for ESP32 - version 3.0.5
(I) (loop)(C1) WiFi: WL_CONNECTED, IP: 192.168.1.93, Mem: 59024, LargestBlk: 49140, PSRAM Free: 0/0, LED FPS: 27 LED Bright: 100%, LED Watts: 0, CPU: 0(I) (loop)(C1) WiFi: WL_CONNECTED, IP: 192.168.1.93, Mem: 59024, LargestBlk: 49140, PSRAM Free: 0/0, LED FPS: 27 LED Bright: 100%, LED Watts: 0, CPU: 043%, 012%, FreeDraw: 0.023
(V) (WiFiDraw)(C1) WifIDraw claims to have drawn 0 pixels
|
Beta Was this translation helpful? Give feedback.
-
|
There is not a lot there, but, what is interesting is that after I sent "w" at the end, telnet appears to disconnect. I have to reboot and connect again even though the effect is running fine. |
Beta Was this translation helpful? Give feedback.
-
|
Moving followups to #515 |
Beta Was this translation helpful? Give feedback.
wanted to give a quick update. as you can see in these photos, the wiring is still in "dev mode". Thinking I'd exhausted my ability to debug the source code (which was looking like deep in some included library), I moved the "incomplete" lamp (chip still on breadboard) to a cool spot behind me for my Monday meetings. Side note, it was the first thing the tech guys on my call noticed and immediately asked "WHATS THAT!", so we went down a 15 minute rabbit hole on microprocessors and leds and sensors and stuff.
Not sure what happened, but moving to a more stable location? this thing ran 9 effects (as seen above) for 5 mins an effect for like 2 days? then i was disappointed to see it froze ag…