Kaz Kylheku wrote:
Chris Friesem wrote:
In the trace below, is "epc" the program counter at the time of the
timer interrupt? How does "ra" fit into this, given that the function
whose address it contains isn't seen in the stack trace until quite a
Mind you, these Linux MIPS stack traces are not completely
trustworthy; the routine just walks the stack one word at a time
and prints out anything that looks like the address of code.
So the stack trace may include stale stack data from previous
call chains that have already returned. It also might not
include the full call chain, because the return address is not
always stored on the stack. For instance, here it looks like
RA is pointing into a kernel module. But you don't actually
see this in the fake stack trace. Nowhere in the stack trace
do you see _read_lock being called by c000000001b4ab9c.
Right...but it does look like it calls _read_unlock().
I would proceed by inspecting this vnb module and its use of locks.
Yeah...I'll do that.
scheduler_tick() calls BUG: soft lockup detected on CPU#0!
One thing I've noticed, all the softlockups are on cpu0 even though
other cpus are also complaining about scheduling delays. Is there
something special about cpu0?
This read lock is almost certainly a red herring; it's
stuff left over on the stack that hasn't been overwritten
by new activation chains.
Wouldn't it be part of the do_IRQ call below?
This is the timer interrupt going off, which ends up
detecting the lockup. So the problem is above here.
I assume you mean "above" in terms of chronological order, since in
terms of the call trace the problem would be below.
This may also be garbage. Basically, the stack
trace is not that useful, except that it provides
some additional circumstantial evidence implicating
the vnb module.
It's unfortunate that the call trace isn't accurate.
Thanks for the help,