check-trace

name	check-trace
description	Analyze the JIT dispatch trace to diagnose boot stalls or crashes
allowed-tools	Bash, Read, Grep
argument-hint	["pattern-or-address"]

Check JIT Dispatch Trace

Analyze jitc_trace.log for the pattern or address $ARGUMENTS.

If no argument, check the tail of the trace to diagnose what the CPU is doing.

Steps

1. Check trace tail (last 30 lines):

   tail -30 jitc_trace.log
   

2. Identify the pattern:

   • Same 2-3 PCs repeating with MSR_POW (0x4xxxx) → idle loop
   • Same 2-3 PCs, DEC frozen → DEC timer not firing
   • DEC changes but idle continues → scheduler not switching
   • Dispatches to PA 0x00000300 → DSI exceptions
   • Dispatches to PA 0x00000900 → DEC interrupts
   • Dispatches to PA 0x00000C00 → syscalls (sc instruction)

3. Count exception dispatches:

   grep -c 'pa=00000300' jitc_trace.log   # DSI count
   grep -c 'pa=00000900' jitc_trace.log   # DEC count
   grep -c 'pa=00000c00' jitc_trace.log   # SC count
   

4. If argument is a hex address, search for dispatches to/from that address:

   grep 'pa=ADDRESS\|ea=ADDRESS' jitc_trace.log | head -20
   

5. Check DEC timer status:

   grep '\[DEC\]' BOOT_LOG | tail -5
   

6. Check WATCH events:

   grep '\[WATCH\]' BOOT_LOG | tail -10
   

Key fields in trace lines

DISPATCH ea=EFFECTIVE_ADDR pa=PHYSICAL_ADDR msr=MSR_VALUE
SEQ_NUM pc=PA msr=MSR cr=CR lr=LR ctr=CTR r0=R0 r1=SP r2=R2 r3=R3 r4=R4 r5=R5 dec=DEC

• ea = effective (virtual) address the kernel sees
• pa = physical address after translation
• dec = decrementer register value (should change over time)
• r1 = stack pointer (0xC01Cxxxx for idle thread, 0xC04Dxxxx for init thread)