On Thu, Feb 20, 2025 at 06:49:43AM +0100, Martin Husemann wrote:
> On Wed, Feb 19, 2025 at 07:24:41PM +0100, Thomas Klausner wrote:
> > #6 0xffffffff805af065 in pmap_unmap_ptes (pmap=<optimized out>, pmap2=0x0)
> > at /usr/src/sys/arch/x86/x86/pmap.c:873
>
> That is very strange, do you have crash dumps?
Yes, two of them, let me know if you want to look at them in more detail.
> /* If not our own pmap, mark whatever's on the CPU now as lazy. */
> KASSERT(ci->ci_tlbstate == TLBSTATE_VALID);
> >> mypmap = vm_map_pmap(&l->l_proc->p_vmspace->vm_map);
> if (ci->ci_pmap == vm_map_pmap(&l->l_proc->p_vmspace->vm_map)) {
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
> shouldn't the mypmap be used the line below instead of the second vm_map_pmap
> call?
>
> > Any ideas? Time for a bug report?
>
> Deinitively worth a bug report.
I've filed "kern/59089: panic when running qemu + nvmm".
Cheers,
Thomas
