[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-ppc] [PATCH 09/77] ppc: Fix do_rfi() for rfi emulation
From: |
Benjamin Herrenschmidt |
Subject: |
Re: [Qemu-ppc] [PATCH 09/77] ppc: Fix do_rfi() for rfi emulation |
Date: |
Thu, 19 Nov 2015 21:23:58 +1100 |
On Thu, 2015-11-19 at 17:19 +1100, David Gibson wrote:
> On Wed, Nov 11, 2015 at 11:27:22AM +1100, Benjamin Herrenschmidt
> wrote:
> > XXX This patch needs double checking... It fixed 32-bit userspace
> > but I'm not sure it's right. I wonder whether msr_is_64bit() should
> > be applied to env->msr, not msr, but I need to double check the
> > architecture.
>
> Hrm, I'm not really sure where I'd look in the arch, but
> msr_is_64bit(env->msr) seems like it would make more sense to me.
> The current logic means that rfi, ostensibly a 32-bit instruction
> will
> have different behaviour depending on the upper bits of SRR1, which
> seems a unexpected.
I only just discovered that rfi is actually gone from arch 2.07 :-)
I'll dig a bit more tomorrow.
Cheers,
Ben.
>
> >
> > Signed-off-by: Benjamin Herrenschmidt <address@hidden>
> > ---
> > target-ppc/excp_helper.c | 4 ++--
> > 1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/target-ppc/excp_helper.c b/target-ppc/excp_helper.c
> > index c1d6605..00fae60 100644
> > --- a/target-ppc/excp_helper.c
> > +++ b/target-ppc/excp_helper.c
> > @@ -878,13 +878,13 @@ static inline void do_rfi(CPUPPCState *env,
> > target_ulong nip, target_ulong msr,
> > CPUState *cs = CPU(ppc_env_get_cpu(env));
> >
> > #if defined(TARGET_PPC64)
> > + msr = msr & msrm;
> > if (msr_is_64bit(env, msr)) {
> > nip = (uint64_t)nip;
> > - msr &= (uint64_t)msrm;
> > } else {
> > nip = (uint32_t)nip;
> > - msr = (uint32_t)(msr & msrm);
> > if (keep_msrh) {
> > + msr &= 0xffffffff;
> > msr |= env->msr & ~((uint64_t)0xFFFFFFFF);
> > }
> > }
>
- Re: [Qemu-ppc] [PATCH 02/77] ppc: Use split I/D mmu modes to avoid flushes on interrupts, (continued)
[Qemu-ppc] [PATCH 05/77] ppc: Update SPR definitions, Benjamin Herrenschmidt, 2015/11/10
[Qemu-ppc] [PATCH 11/77] ppc: Create cpu_ppc_set_papr() helper, Benjamin Herrenschmidt, 2015/11/10
[Qemu-ppc] [PATCH 12/77] ppc: Better figure out if processor has HV mode, Benjamin Herrenschmidt, 2015/11/10
[Qemu-ppc] [PATCH 09/77] ppc: Fix do_rfi() for rfi emulation, Benjamin Herrenschmidt, 2015/11/10
[Qemu-ppc] [PATCH 08/77] ppc: Add number of threads per core to the processor definition, Benjamin Herrenschmidt, 2015/11/10
[Qemu-ppc] [PATCH 03/77] ppc: Do some batching of TCG tlb flushes, Benjamin Herrenschmidt, 2015/11/10
[Qemu-ppc] [PATCH 06/77] ppc: Add macros to register hypervisor mode SPRs, Benjamin Herrenschmidt, 2015/11/10