lmac.c 662 ?

Have been getting this mysterious message on serial debug output before wdt reset:

lmac.c 662

The message "lmac.c 662" is not some debug print from my own sketch. Have seen it more than once now, so it doesn't seem to be a random glitch on serial...

Does anyone have an idea what's it about ?
(and how to prevent would be even better...)

[unfoundbug]

Not got a solution as of yet, but I get this as well using the SDK, V1.2 so it's not specific to the arduino branch
Sometimes this actually pushes out as an exception rather than the lmac line.

Not quite sure yet whether its caused during the amount of time with interrupts disabled or in transmission, will be working on this later.

[sticilface]

I'm getting a lot of these... and with a simplified sketch that is just running neopixelbus and an animation with OTA enabled. I still get them without wifi and OTA, but just the lmac. I don't get the stack trace very often. However, locating 40103bd the function has lots of lmac references in it, so it looks pretty similar.

here is sketch
http://pastebin.com/Pambzjwn

here is the elf dump
https://dl.dropboxusercontent.com/u/12627633/OTAdump.txt

here is the stack

Fatal exception (28):
epc1=0x40103bdb, epc2=0x00000000, epc3=0x00000000, excvaddr=0x00000024, depc=0x00000000

Exception (28):
epc1=0x40103bdb epc2=0x00000000 epc3=0x00000000 excvaddr=0x00000024 depc=0x00000000

ctx: cont 
sp: 3ffea710 end: 3ffeaa10 offset: 01a0

>>>stack>>>
3ffea8b0:  00040000 3ee85010 53002200 3ffec640  
3ffea8c0:  4000050c 3fffc278 4010550c 3fffc200  
3ffea8d0:  00000022 000000ff 3e9c9c9d 402025d8  
3ffea8e0:  40203881 00000030 0000001e ffffffff  
3ffea8f0:  4020387c 00000020 3fff2ea0 00000004  
3ffea900:  399428ea 00000000 00000000 60000308  
3ffea910:  00000080 00000040 000000c7 3fff2af8  
3ffea920:  3ffe9950 3fffdc20 3ffeaa3c 00000030  
3ffea930:  00000007 00000006 402017fa 3ffe99f0  
3ffea940:  3fff2808 00000008 00000000 0000003b  
3ffea950:  003bc300 00000000 3ffeaa34 40202eb0  
3ffea960:  63000000 3e409c00 3ffe8b62 3f800000  
3ffea970:  3e70dc37 3ef0dc37 3fff4368 40202450  
3ffea980:  3f1b5b00 3fff3134 402022bc 402023dc  
3ffea990:  3fff2818 00000000 402022bc 00000001  
3ffea9a0:  00000023 0000008d 3fff2bf8 402037be  
3ffea9b0:  00000009 00000009 3fff2bf8 4020375c  
3ffea9c0:  00101abc 432f0000 3fff39bc 00000000  
3ffea9d0:  00000000 3fffdc20 3ffe9950 3fff2af8  
3ffea9e0:  3fffdc20 00000000 3ffe9650 40202c5a  
3ffea9f0:  3fffdc20 00000000 3ffeaa34 40201876  
3ffeaa00:  00000000 00000000 3ffe99f0 4010046c  
<<<stack<<<

[Duality4Y]

I notice this too, and it shows up more on my network then anywhere.

[sticilface]

Using the latest commits, these have decreased significantly for me, almost completely.

[sticilface]

Thought I'd throw some more info up here in the hope of a solution. I'm still getting this not as frequently but every hour or two... here is the offending function from the elf dump.... with the offending address starred. It is not a function in my sketch!

40103c8c <lmacProcessAckTimeout>:
40103c8c:   f0c112          addi    a1, a1, -16
40103c8f:   0109        s32i.n  a0, a1, 0
40103c91:   feef85          call0   40102b8c <lmacIsActive>
40103c94:   02dc        bnez.n  a2, 40103ca8 <lmacProcessAckTimeout+0x1c>
40103c96:   fbc821          l32r    a2, 40102bb8 <lmacIsIdle+0x18>
40103c99:   fbc831          l32r    a3, 40102bbc <lmacIsIdle+0x1c>
40103c9c:   e2a442          movi    a4, 0x4e2
40103c9f:   f7bd01          l32r    a0, 40101b94 <ets_timer_setfn+0x1c>
40103ca2:   0000c0          callx0  a0
40103ca5:   ffff06          j   40103ca5 <lmacProcessAckTimeout+0x19>
40103ca8:   fbb841          l32r    a4, 40102b88 <call_user_start_local+0x420>
40103cab:   000422          l8ui    a2, a4, 0
40103cae:   a438        l32i.n  a3, a4, 40
40103cb0:   30c442          addi    a4, a4, 48
**40103cb3: 9308        l32i.n  a0, a3, 36
40103cb5:   b02220          addx8   a2, a2, a2
40103cb8:   0008        l32i.n  a0, a0, 0
40103cba:   a02240          addx4   a2, a2, a4
40103cbd:   1060e7          bbci    a0, 14, 40103cd1 <lmacProcessAckTimeout+0x45>
40103cc0:   030c        movi.n  a3, 0
40103cc2:   070242          l8ui    a4, a2, 7
40103cc5:   064242          s8i a4, a2, 6
40103cc8:   094232          s8i a3, a2, 9
40103ccb:   ffe485          call0   40103b14 <lmacMSDUAged+0x238>
40103cce:   000106          j   40103cd6 <lmacProcessAckTimeout+0x4a>
40103cd1:   040c        movi.n  a4, 0
40103cd3:   ffc105          call0   401038e4 <lmacMSDUAged+0x8>
40103cd6:   0108        l32i.n  a0, a1, 0
40103cd8:   10c112          addi    a1, a1, 16
40103cdb:   f00d        ret.n
40103cdd:   000000          ill

[Duality4Y]

i'll try and do a pull and recompile

[DanielCasner]

There's an interesting note in the v1.3.0 SDK documentation now which says that interrupts may not be disabled for more than 10us. Since WiFi interrupts can't preempt other interrupts this also means that no interrupt handler should run longer than 10us or the wifi will screw up with a lmac error.

[igrr]

OTA related part of this issue is fixed in staging.