Re: kern/60220 (kernel panic after modifying floating-point status register)

[Bruno Haible <bruno%clisp.org@localhost>]

Taylor R Campbell wrote:
> Still haven't reproduced it, but I don't have vbox handy to test.
> Were you able to reproduce it in any environment other than vbox?

Yes, I have so far reproduced it with
  - VirtualBox 7.0.x in an Ubuntu 24.04 host,
  - VirtualBox 7.2.x in an Ubuntu 26.04 host,
  - qemu 10.1.1 in an Ubuntu 26.04 host, with qemu option -enable-kvm.

I do *not* reproduce it with
  - qemu 10.1.1 in an Ubuntu 26.04 host, without qemu option -enable-kvm.
In this setting "make check" fails (unlike in NetBSD 10.0, where it succeeded),
but it does not crash the kernel.

> But the first line of ps-part1.png shows that the process running your
> test program is still on the CPU:

"make check" runs the same program few dozens of times.

> Can you try running the NetBSD ATF tests in this environment to see if
> they also reproduce the crash?
> 
> cd /usr/tests/lib/libc && atf-run | atf-report
> cd /usr/tests/lib/libm && atf-run | atf-report
> cd /usr/tests/kernel && atf-run | atf-report

* In the VirtualBox 7.2.x environment:
  sigfpe_flt: Passed.
  No crash.

* In the QEMU environment (both with and without -enable-kvm):
  sigfpe_flt: Skipped: Test does not run correctly under QEMU
  No crash.