Re: panic shortly after boot when amdgpu.ko is loaded (fpu?)

On 11/27/20 6:09 AM, Rebecca Cran wrote:
> I have a Threadripper 2990WX system that I recently installed an AMD 
> Radeon Pro W5700 into. It runs fine unless I load the amdgpu driver, at 
> which point it panics several seconds after boot: I have enough time to 
> login and run a few commands, but even if I just leave it it'll panic. 
> I'm running:
> 
> 
> FreeBSD photon.int.bluestop.org 13.0-CURRENT FreeBSD 13.0-CURRENT #0 
> 6db1a3e8098-c273171(master): Thu Nov 26 01:26:17 MST 2020 
> bcran_at_photon.int.bluestop.org:/usr/obj/usr/src/amd64.amd64/sys/GENERIC-NODEBUG 
> amd64
> 
> 
> I rebuilt the drm-current-kmod-5.4.62.g20201109_1 port today.
> 
> 
> The panic is:
> 
> Fatal trap 9: general protection fault while in kernel mode
> cpuid = 24; apic id = 18
> instruction pointer    = 0x20:0xffffffff81002dcf
> stack pointer            = 0x0:0xfffffe016e6ffaa0
> frame pointer            = 0x0:0xfffffe016e6ffaa0
> code segment        = base 0x0, limit 0xfffff, type 0x1b
>              = DPL 0, pres 1, long 1, def32 0, gran 1
> processor eflags    = interrupt enabled, resume, IOPL = 0
> current process        = 4372 (hald)
> trap number        = 9
> panic: general protection fault
> cpuid = 24
> time = 1606450595
> KDB: stack backtrace:
> db_trace_self_wrapper() at db_trace_self_wrapper+0x2b/frame 
> 0xfffffe016e6ff7b0
> vpanic() at vpanic+0x181/frame 0xfffffe016e6ff800
> panic() at panic+0x43/frame 0xfffffe016e6ff860
> trap_fatal() at trap_fatal+0x387/frame 0xfffffe016e6ff8c0
> trap() at trap+0x8e/frame 0xfffffe016e6ff9d0
> calltrap() at calltrap+0x8/frame 0xfffffe016e6ff9d0
> --- trap 0x9, rip = 0xffffffff81002dcf, rsp = 0xfffffe016e6ffaa0, rbp = 
> 0xfffffe016e6ffaa0 ---
> fpurestore_xrstor3264() at fpurestore_xrstor3264+0x2f/frame 
> 0xfffffe016e6ffaa0
> restore_fpu_curthread() at restore_fpu_curthread+0x85/frame 
> 0xfffffe016e6ffac0
> fpudna() at fpudna+0x3a/frame 0xfffffe016e6ffae0
> trap() at trap+0x246/frame 0xfffffe016e6ffbf0
> calltrap() at calltrap+0x8/frame 0xfffffe016e6ffbf0
> --- trap 0x16, rip = 0x80067137f, rsp = 0x7fffffffd8b0, rbp = 
> 0x7fffffffd8f0 ---
> Uptime: 1m4s
> Dumping 4193 out of 130894 
> MB:..1%..11%..21%..31%..41%..51%..61%..71%..81%..91%
> 
> 
> I've uploaded details (core.txt, dmesg.txt etc.) to 
> https://bsdio.com/freebsd/crashes/2020-11-26-amdgpu/ and the vmcore file 
> is available on request.
> 

Hi,

Is the problem always triggered by hald? If you disable hald in rc.conf, 
does the system run for a longer period of time?

--HPS