Intermittent MemMgtFault on stm32l4x6-nucleo

[agkaminski]

For now seen once on CI (https://github.com/phoenix-rtos/phoenix-rtos-project/actions/runs/9502217223/job/26189746156):

 phoenix-rtos-tests/psh/history: FAIL
SHELL ERROR: Couldn't find a prompt!
EXPECTED: (psh)% 
OUTPUT:
ciE3fGUFgW2uIV~_

Exception: 4 #MemMgtFault
 r0=00000001  r1=00000071  r2=00000071  r3=40004400
 r4=2000d434  r5=2000d150  r6=2000d370  r7=2000d430
 r8=00000002  r9=20009d30 r10=2000d32c r11=20013e88
r12=00000001 psr=21000000  lr=0802195d  pc=08021966
psp=2000d318 msp=20017200 exr=fffffffd bfa=40004400
cfs=00000082 mma=40004400

Seems like psh might be trying to access USART2 CR1 register for some reason. Did not look further into this issue.

[damianloew]

• Discovered for a second time on 5eabbc3 with the following output:

Exception: 4 #MemMgtFault  
r0=00000001  r1=0000003f  r2=0000003f  r3=40004400
r4=2000d434  r5=2000d150  r6=2000d370  r7=2000d430
r8=00000002  r9=20009d30 r10=2000d32c r11=20013e88
r12=00000001 psr=21000000  lr=0802195d  pc=08021966
psp=2000d318 msp=20017200 exr=fffffffd bfa=40004400
cfs=00000082 mma=40004400

https://github.com/phoenix-rtos/plo/actions/runs/9546062432/job/26308199690

• Seems to happen from time to time

[agkaminski]

There's a slight chance that phoenix-rtos/phoenix-rtos-kernel#561 might fix it

[anglov]

PC:

269     static int tty_uartenabled(tty_ctx_t *ctx)
270     {
271             return (*(ctx->base + cr1) & 1) ? 1 : 0;
   0x08021962 <+530>:   ldr.w   r3, [r5, #512]  @ 0x200
   0x08021966 <+534>:   ldr     r3, [r3, #0]

285                             while (((tty_dmarxready(ctx) == 0) && (tty_dmatxready(ctx) == 0)) || (tty_uartenabled(ctx) == 0)) {
   0x08021968 <+536>:   lsls    r3, r3, #31
   0x0802196a <+538>:   bpl.w   0x80217c2 <tty_dmathread+114>
   0x0802196e <+542>:   b.n     0x8021804 <tty_dmathread+180>

235                     libtty_wake_reader(&ctx->ttyCommon);
   0x08021970 <+544>:   mov     r0, r6
   0x08021972 <+546>:   bl      0x8024048 <libtty_wake_reader>
   0x08021976 <+550>:   b.n     0x80218f0 <tty_dmathread+416>
   0x08021978 <+552>:   lsls    r0, r2, #3
   0x0802197a <+554>:   movs    r0, r0

LR:

257     static int tty_dmatxready(tty_ctx_t *ctx)
258     {
259             return ((ctx->data.dma.txDoneFlag != 0) && (libtty_txready(&ctx->ttyCommon) != 0)) ? 1 : 0;
   0x08021956 <+518>:   mov     r0, r6
   0x08021958 <+520>:   bl      0x8023d5c <libtty_txready>
   0x0802195c <+524>:   cmp     r0, #0
   0x0802195e <+526>:   beq.w   0x80217c2 <tty_dmathread+114>

260     }

(gdb) info registers 
r0             0x1                 1
r1             0x39                57
r2             0x39                57
r3             0x40004400          1073759232
r4             0x2000d434          536925236
r5             0x2000d150          536924496
r6             0x2000d370          536925040
r7             0x2000d430          536925232
r8             0x1                 1
r9             0x20009d30          536911152
r10            0x2000d32c          536924972
r11            0x20013e88          536952456
r12            0x1                 1
sp             0x2000d318          0x2000d318
lr             0x802195d           134355293
pc             0x8021966           0x8021966 <tty_dmathread+534>
xPSR           0x21000000          553648128
fpscr          0x0                 0
msp            0x20017200          0x20017200
psp            0x2000d318          0x2000d318
primask        0x0                 0
basepri        0x0                 0
faultmask      0x0                 0
control        0x3                 3

Looks like code path that are reached in normal work and works fine.

[adamdebek]

I checked on 5beec45 and 69bcce0. Errors still occur, in my opinion it looks like error pertaining to process creation.

[Darchiv]

I checked on 5beec45 and errors still occur. In my opinion it looks like error pertaining to process creation.

Maybe you can try phoenix-rtos/phoenix-rtos-kernel#553?

[adamdebek]

I checked on 5beec45 and errors still occur. In my opinion it looks like error pertaining to process creation.

Maybe you can try phoenix-rtos/phoenix-rtos-kernel#553?

I see it failed during merge to master, so I assume this pr needs further changes.

[anglov]

At the first glance it looks fixed by phoenix-rtos/phoenix-rtos-kernel#562

@adamdebek could you also check?

[adamdebek]

At the first glance it looks fixed by phoenix-rtos/phoenix-rtos-kernel#562

@adamdebek could you also check?

Yea, tests are running already.

[adamdebek]

I ran over 200 test companies during night and not single one failed, so this confirm the issue is resolved.