Re: current now panics when starting VBox VM

On 11/3/21 11:13, Konstantin Belousov wrote:
> On Wed, Nov 03, 2021 at 11:05:11AM -0400, Michael Butler via freebsd-emulation wrote:
>> On 11/3/21 10:36, Ed Maste wrote:
>> The kgdb back-trace isn't any more enlightening to me :-(
>>
>> __curthread () at /usr/src/sys/amd64/include/pcpu_aux.h:55
>> 55              __asm("movq %%gs:%P1,%0" : "=r" (td) : "n" (offsetof(struct
>> pcpu,
>> (kgdb) bt
>> #0  __curthread () at /usr/src/sys/amd64/include/pcpu_aux.h:55
>> #1  doadump (textdump=<optimized out>) at
>> /usr/src/sys/kern/kern_shutdown.c:399
>> #2  0xffffffff808cbac5 in kern_reboot (howto=260) at
>> /usr/src/sys/kern/kern_shutdown.c:487
>> #3  0xffffffff808cbedb in vpanic (fmt=<optimized out>,
>> ap=0xfffffe0129a6a8d0) at /usr/src/sys/kern/kern_shutdown.c:920
>> #4  0xffffffff808cbd33 in panic (fmt=<unavailable>) at
>> /usr/src/sys/kern/kern_shutdown.c:844
>> #5  0xffffffff80ca920c in trap_fatal (frame=frame@entry=0xfffffe0129a6aac0,
>> eva=0) at /usr/src/sys/amd64/amd64/trap.c:946
>> #6  0xffffffff80ca95af in trap_pfault (frame=frame@entry=0xfffffe0129a6aac0,
>> usermode=false, signo=<optimized out>, signo@entry=0x0, ucode=<optimized
>> out>, ucode@entry=0x0)
>>      at /usr/src/sys/amd64/include/cpufunc.h:417
>> #7  0xffffffff80ca89bc in trap (frame=0xfffffe0129a6aac0) at
>> /usr/src/sys/amd64/amd64/trap.c:443
>> #8  <signal handler called>
>> #9  strlen () at /usr/src/sys/amd64/amd64/support.S:751
>> #10 0xffffffff808b4d79 in sysctl_kern_proc_pathname (oidp=<optimized out>,
>> arg1=0xfffffe0129a6ad8c, arg2=<optimized out>, req=0xfffffe0129a6acc0) at
>> /usr/src/sys/kern/kern_proc.c:2330
>> #11 0xffffffff808dc331 in sysctl_root_handler_locked
>> (oid=oid@entry=0xffffffff810cf0e0 <sysctl___kern_proc_pathname>,
>> arg1=arg1@entry=0xfffffe0129a6ad8c, arg2=arg2@entry=1,
>>      req=0xfffffe0129a6acc0, tracker=tracker@entry=0xfffffe0129a6ac38) at
>> /usr/src/sys/kern/kern_sysctl.c:185
>> #12 0xffffffff808db88b in sysctl_root (oidp=<optimized out>,
>> arg1=0xfffffe0129a6ad8c, arg1@entry=0xfffffe0129a6ad80, arg2=1,
>> arg2@entry=4, req=req@entry=0xfffffe0129a6acc0)
>>      at /usr/src/sys/kern/kern_sysctl.c:2305
>> #13 0xffffffff808dbdf3 in userland_sysctl (td=td@entry=0xfffffe012991a000,
>> name=name@entry=0xfffffe0129a6ad80, namelen=4, old=<optimized out>,
>> oldlenp=<optimized out>,
>>      inkernel=<optimized out>, inkernel@entry=0, new=0x0, newlen=0,
>> retval=0xfffffe0129a6ade8, flags=0) at /usr/src/sys/kern/kern_sysctl.c:2462
>> #14 0xffffffff808dbc3c in sys___sysctl (td=0xfffffe012991a000,
>> uap=0xfffffe012991a3f0) at /usr/src/sys/kern/kern_sysctl.c:2335
>> #15 0xffffffff80ca9b5c in syscallenter (td=0xfffffe012991a000) at
>> /usr/src/sys/amd64/amd64/../../kern/subr_syscall.c:189
>> #16 amd64_syscall (td=0xfffffe012991a000, traced=0) at
>> /usr/src/sys/amd64/amd64/trap.c:1191
>> #17 <signal handler called>
>> #18 0x000000080315a71a in ?? ()
>> Backtrace stopped: Cannot access memory at address 0x7fffffffc778
>> (kgdb)
>>
> 
> Try this
> 
> commit 2d3f95bd1fd4f71769f60b8037c1ff27c75d8258
> Author: Konstantin Belousov <kib@FreeBSD.org>
> Date:   Wed Nov 3 17:11:33 2021 +0200
> 
>      proc_get_binpath(): return empty string instead of NULL
>      
>      for strange case where process does not have text.
>      
>      Sponsored by:   The FreeBSD Foundation
>      MFC after:      3 days
> 
> diff --git a/sys/kern/kern_proc.c b/sys/kern/kern_proc.c
> index 2156c5c465ba..d11f651960c0 100644
> --- a/sys/kern/kern_proc.c
> +++ b/sys/kern/kern_proc.c
> @@ -2252,7 +2252,7 @@ proc_get_binpath(struct proc *p, char *binname, char **retbuf,
>   	vp = p->p_textvp;
>   	if (vp == NULL) {
>   		PROC_UNLOCK(p);
> -		*retbuf = NULL;
> +		*retbuf = "";
>   		*freebuf = NULL;
>   		return (0);
>   	}
> 

Interestingly, when I went to log out of KDE afer applying this patch 
and rebuilding the kernel, it also paniced without VBox ruuning. The log 
looks similar but the back-trace points to this possibility .. see 
frames 8 through 10 .

__curthread () at /usr/src/sys/amd64/include/pcpu_aux.h:55
55              __asm("movq %%gs:%P1,%0" : "=r" (td) : "n" 
(offsetof(struct pcpu,
(kgdb) #0  __curthread () at /usr/src/sys/amd64/include/pcpu_aux.h:55
#1  doadump (textdump=<optimized out>)
     at /usr/src/sys/kern/kern_shutdown.c:399
#2  0xffffffff808cbac5 in kern_reboot (howto=260)
     at /usr/src/sys/kern/kern_shutdown.c:487
#3  0xffffffff808cbedb in vpanic (fmt=<optimized out>, 
ap=0xfffffe011c1858d0)
     at /usr/src/sys/kern/kern_shutdown.c:920
#4  0xffffffff808cbd33 in panic (fmt=<unavailable>)
     at /usr/src/sys/kern/kern_shutdown.c:844
#5  0xffffffff80ca920c in trap_fatal (frame=frame@entry=0xfffffe011c185ac0,
     eva=0) at /usr/src/sys/amd64/amd64/trap.c:946
#6  0xffffffff80ca95af in trap_pfault 
(frame=frame@entry=0xfffffe011c185ac0,
     usermode=false, signo=<optimized out>, signo@entry=0x0,
     ucode=<optimized out>, ucode@entry=0x0)
     at /usr/src/sys/amd64/include/cpufunc.h:417
#7  0xffffffff80ca89bc in trap (frame=0xfffffe011c185ac0)
     at /usr/src/sys/amd64/amd64/trap.c:443
#8  <signal handler called>
#9  strlen () at /usr/src/sys/amd64/amd64/support.S:751
#10 0xffffffff808b4d79 in sysctl_kern_proc_pathname (oidp=<optimized out>,
     arg1=0xfffffe011c185d8c, arg2=<optimized out>, req=0xfffffe011c185cc0)
     at /usr/src/sys/kern/kern_proc.c:2330
#11 0xffffffff808dc331 in sysctl_root_handler_locked (
     oid=oid@entry=0xffffffff810cf0e0 <sysctl___kern_proc_pathname>,
     arg1=arg1@entry=0xfffffe011c185d8c, arg2=arg2@entry=1,
     req=0xfffffe011c185cc0, tracker=tracker@entry=0xfffffe011c185c38)
     at /usr/src/sys/kern/kern_sysctl.c:185

After rebooting with this patch, I can now run a VBox VM without 
provoking a panic. Thanks! :-)

	Michael