r355009 dies with a kernel panic

[Mitchell Horne]

On Thu, Nov 28, 2019 at 3:15 PM Dennis Clarke <dclarke at blastwave.org> wrote:
>
>
> While is it fine that the CI does a bare bones compile I don't know
> if anyone actually tries to run it for a period of time.  If r355009
> is left running for a few hours ( on QEMU ) it merely dies with a
> panic :
>
>
> panic: deadlres_td_sleep_q: possible deadlock detected for
> 0xffffffd04630a560, blocked for 180225 ticks
>
> cpuid = 0
> time = 1574915402
> KDB: stack backtrace:
> db_trace_self() at db_read_token_flags+0x696
>           pc = 0xffffffc0004ca28a ra = 0xffffffc0000d289c
>           sp = 0xffffffc024bb4790 fp = 0xffffffc024bb49b0
>
> db_read_token_flags() at kdb_backtrace+0x2e
>           pc = 0xffffffc0000d289c ra = 0xffffffc00026b98c
>           sp = 0xffffffc024bb49b0 fp = 0xffffffc024bb4a60
>
> kdb_backtrace() at vpanic+0x14a
>           pc = 0xffffffc00026b98c ra = 0xffffffc00022dfac
>           sp = 0xffffffc024bb4a60 fp = 0xffffffc024bb4aa0
>
> vpanic() at panic+0x22
>           pc = 0xffffffc00022dfac ra = 0xffffffc00022dfe4
>           sp = 0xffffffc024bb4aa0 fp = 0xffffffc024bb4ac0
>
> panic() at acct_process+0x676
>           pc = 0xffffffc00022dfe4 ra = 0xffffffc0001de0aa
>           sp = 0xffffffc024bb4ac0 fp = 0xffffffc024bb4b80
>
> acct_process() at fork_exit+0x5c
>           pc = 0xffffffc0001de0aa ra = 0xffffffc0001fbd02
>           sp = 0xffffffc024bb4b80 fp = 0xffffffc024bb4bc0
>
> fork_exit() at fork_trampoline+0xa
>           pc = 0xffffffc0001fbd02 ra = 0xffffffc0004d768a
>           sp = 0xffffffc024bb4bc0 fp = 0xffffffc0001ddea4
>
> fork_trampoline() at 0xb7e1dbe985ce8b81
>           pc = 0xffffffc0004d768a ra = 0xb7e1dbe985ce8b81
>           sp = 0xffffffc0001ddea4 fp = 0x40a7d7b3000a3783
>
> KDB: enter: panic
> [ thread pid 0 tid 100030 ]
> Stopped at      kdb_enter+0x3a: c.ebreak
> db> Killed
>

Hi Dennis, thanks for the report. I have encountered this panic in
the past as well. Were you running something on your RISC-V system
when this happened or was it idle? How many cpus did you run qemu
with?

I tried to recreate it with an idle system as I have done in the past but
have seen nothing yet:
root at qemu:~ # uptime
10:15PM  up 4 days, 19:45, 1 user, load averages: 0.18, 0.28, 0.30

Could you try again with a more recent HEAD (note there is some breakage
at the exact time of writing) and let me know if the issue is still present?

Mitchell

>
> So is anyone anywhere trying to run FreeBSD on RISC-V at all or just me?
>
> --
> Dennis Clarke
> RISC-V/SPARC/PPC/ARM/CISC
> UNIX and Linux spoken
> GreyBeard and suspenders optional
> _______________________________________________
> freebsd-riscv at freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-riscv
> To unsubscribe, send any mail to "freebsd-riscv-unsubscribe at freebsd.org"