Re: hppa lazy FPU switching and QEMU bugs

[Martin Husemann <martin%duskware.de@localhost>]

On Wed, Aug 05, 2026 at 08:18:47AM -0400, Thor Lancelot Simon wrote:
> Unfortunately QEMU lies - it claims to support coprocessor disable but it
> never checks and never generates the trap.  So processes get their FP regs
> switched with each other, and since *integer* multiply uses the FP regs
> on PA, that's even worse than you'd expect.

I am confused, as I assume we initially start with coprocessor disabled
and get a trap to make the lwp owner of the FPU state.

If we never get that initial trap, we should not have FPU state to save
and all qemu runs should see massive garbling of any values in FPU registers
through context switches as soon as any other userland lwps run in betweeen.

Martin