Open NicoHood opened 8 years ago
facchinm
commented
8 years ago Hi Nico, the stacktrace looks somehow related with Intel display drivers (as you can see from the timestamps, USB events seems uncorrelated with the kernel oops). Is your sketch doing something graphic-related?
NicoHood
commented
8 years ago My sketch controls HID and reads the serial. The time difference is probably because the error occured (Arduino disconnected), then I noticed it some minutes afterwards and switched to tty1 (where then the error was triggered). Next time i could just turn off the pc and see what the log then says. But the error happens very few times.
NicoHood
commented
8 years ago Good for you, bad for me:
My PC crashed again. (and i instantly tunred it off, no switching to tty1) The kernel message sais absolutely nothing. I have no idea how to debug this. its not causes by any usb device it seems. But when the bug occurs the usb devices will give errors, and the display also. I developed an usb bootloader though, but this should not crash my pc, i've been doing this quite some time now.
Any ideas appreciated, however it is not an arduino bug as i first though.
NicoHood
commented
8 years ago I have to reopen this issue, as I had this error again yesterday.
The bug was caused by the IDE, I am pretty sure. I developed for a very long time and the Serial was used a lot. I do not know if this is a linux driver, java or arduino issue, but is has to be one of those.
I will test again and check my ram usage carefully. It might still be a java crash, as it also has some internal memory management I guess. Maybe one who know the code better can give a statement about that.
The last sketch I used printed a single char ('s') aka Serial.println("s"); very fast (every 1~10ms).
NicoHood
commented
8 years ago Yes I am right with my assumption. It takes quite a long time to fill the ram (thatswhy it happened very few times), but it does fill up. I started with 3gb of ram usage and now i got 4gb. The IDE now uses 1.3gb of ram at the moment, increasing slow, but increasing.
I think no screenshots are required that you believe me.
I think one of the problems with the new IDE is the feature, that the serial monitor can stay open, even after uploading. Thatswhy the ram is not freed. On top of that the ram is also not freed if the serial monitor is closed! I have not tested if it continues to fill. The ram is only freed when you close the last IDE window.
And that causes the freeze and the VERY slow PC. Because it started to swap and therefor it was not fully unfunctional but extremely slow. This should be fixed for sure. Luckly I have 16gb and do not notice this bug that fast. Others don't, especially for the ARM builds.
Test sketch:
void setup() {
// put your setup code here, to run once:
Serial.begin(115200);
}
void loop() {
// put your main code here, to run repeatedly:
Serial.println("t");
}I've updated the title. I have not tested this issue again with the latest version, not on any other OS. I assume it still got not fixed yet.
I've been using my Arduino for quite some time and I got a weird error every week or so: Randomly my PC (elementary os, based on ubuntu 14.04 x64) does not recognize any keyboard/mouse inputs anymore. The Mouse is still moving though. I cannot do anything, well sometimes it reacts a very short timespan after waiting minutes.
So I switched to tty1 and tried to debug it. Switching takes about 2~3 minutes to do so. In the log below you can see the CPU errors that occur when switching to tty1 and back to tty7. This CPU errors happen when trying to switch, they are not the main reason it seems.
And the only thing you can see is, that my Arduino suddenly disconnected. (No I did not plug it off). I then switched to tty1, check dmesg, plugged the arduino out and back in (to veryfy the device number) and switched back to tty7 where it fully crashed. Afterwards I saved the log:
I am using th latest nightly IDE (selfcompiled via git). I also worked on the USB core as you remember, but I have no clue what is happening here. I also cannot post my sketch, it is quite complex. Also I think this was not the problem, since it only presses USB HID keys when I do so with my remote. I also did not use the Serial port, so the device was in idle. It happen with and without usb hub, with different cables.
Also I cannot debug this further because it happens so few times. But I remember that my Arduino sketch crashed somehow, since it did not react to my IR remote anymore. No HID was working and no leds (ir remote can trigger leds). But the watchdog also did not restart the MCU. So I guess the usb core hangs at some function in idle mode. Any Ideas?