[PATCH] KVM: PPC: Book3S HV: Handle host system reset in guest mode

2017-11-05 Thread Nicholas Piggin 
If the host takes a system reset interrupt while a guest is running,
the CPU must exit the guest before processing the host exception
handler.

After this patch, taking a sysrq+x with a CPU running in a guest
gives a trace like this:

   cpu 0x27: Vector: 100 (System Reset) at [c00fdf5776f0]
   pc: c00810158b80: kvmppc_run_core+0x16b8/0x1ad0 [kvm_hv]
   lr: c00810158b80: kvmppc_run_core+0x16b8/0x1ad0 [kvm_hv]
   sp: c00fdf577850
  msr: 92803033
 current = 0xc00fdf4b1e00
 paca= 0xcfd4d680softe: 3irq_happened: 0x01
   pid   = 6608, comm = qemu-system-ppc
   Linux version 4.14.0-rc7-01489-g47e1893a404a-dirty #26 SMP
   [c00fdf577a00] c00810159dd4 kvmppc_vcpu_run_hv+0x3dc/0x12d0 [kvm_hv]
   [c00fdf577b30] c008100a537c kvmppc_vcpu_run+0x44/0x60 [kvm]
   [c00fdf577b60] c008100a1ae0 kvm_arch_vcpu_ioctl_run+0x118/0x310 [kvm]
   [c00fdf577c00] c00810093e98 kvm_vcpu_ioctl+0x530/0x7c0 [kvm]
   [c00fdf577d50] c0357bf8 do_vfs_ioctl+0xd8/0x8c0
   [c00fdf577df0] c0358448 SyS_ioctl+0x68/0x100
   [c00fdf577e30] c000b220 system_call+0x58/0x6c
   --- Exception: c01 (System Call) at 7fff76868df0
   SP (7fff7069baf0) is in userspace

Fixes: e36d0a2ed5 ("powerpc/powernv: Implement NMI IPI with 
OPAL_SIGNAL_SYSTEM_RESET")
Signed-off-by: Nicholas Piggin 
--

It has always been possible to sreset the host with direct scom
access, but the patch e36d0a2ed5 has significantly expanded this
functionality so in practice this is a required as a fix for it.

Since RFC:
- Removed the last hunk as sugggested by Paul.
- Re-tested.

Thanks,
Nick
---
 arch/powerpc/include/asm/hw_irq.h| 1 +
 arch/powerpc/kernel/exceptions-64s.S | 2 ++
 arch/powerpc/kernel/irq.c| 3 ++-
 arch/powerpc/kvm/book3s_hv.c | 7 ++-
 4 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/include/asm/hw_irq.h 
b/arch/powerpc/include/asm/hw_irq.h
index 92a3e9a79cb4..a8bbac425ae6 100644
--- a/arch/powerpc/include/asm/hw_irq.h
+++ b/arch/powerpc/include/asm/hw_irq.h
@@ -40,6 +40,7 @@
 
 #ifndef __ASSEMBLY__
 
+extern void replay_system_reset(void);
 extern void __replay_interrupt(unsigned int vector);
 
 extern void timer_interrupt(struct pt_regs *);
diff --git a/arch/powerpc/kernel/exceptions-64s.S 
b/arch/powerpc/kernel/exceptions-64s.S
index 651e1a0114ed..bff2ed6e3c3c 100644
--- a/arch/powerpc/kernel/exceptions-64s.S
+++ b/arch/powerpc/kernel/exceptions-64s.S
@@ -113,6 +113,7 @@ EXC_VIRT_NONE(0x4000, 0x100)
cmpwi   cr3,r10,2 ; \
BRANCH_TO_C000(r10, system_reset_idle_common) ; \
 1: \
+   KVMTEST_PR(n) ; \
END_FTR_SECTION_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
 #else
 #define IDLETEST NOTEST
@@ -129,6 +130,7 @@ EXC_REAL_BEGIN(system_reset, 0x100, 0x100)
 
 EXC_REAL_END(system_reset, 0x100, 0x100)
 EXC_VIRT_NONE(0x4100, 0x100)
+TRAMP_KVM(PACA_EXNMI, 0x100)
 
 #ifdef CONFIG_PPC_P7_NAP
 EXC_COMMON_BEGIN(system_reset_idle_common)
diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
index efbadcbbf694..7e8259106944 100644
--- a/arch/powerpc/kernel/irq.c
+++ b/arch/powerpc/kernel/irq.c
@@ -437,7 +437,7 @@ static const u8 srr1_to_lazyirq[0x10] = {
PACA_IRQ_HMI,
0, 0, 0, 0, 0 };
 
-static noinline void replay_system_reset(void)
+void replay_system_reset(void)
 {
struct pt_regs regs;
 
@@ -447,6 +447,7 @@ static noinline void replay_system_reset(void)
system_reset_exception();
get_paca()->in_nmi = 0;
 }
+EXPORT_SYMBOL_GPL(replay_system_reset);
 
 void irq_set_pending_from_srr1(unsigned long srr1)
 {
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 8f34715cfbff..31a362669fea 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -47,6 +47,7 @@
 
 #include 
 #include 
+#include 
 #include 
 #include 
 #include 
@@ -1089,9 +1090,10 @@ static int kvmppc_handle_exit_hv(struct kvm_run *run, 
struct kvm_vcpu *vcpu,
vcpu->stat.ext_intr_exits++;
r = RESUME_GUEST;
break;
-   /* HMI is hypervisor interrupt and host has handled it. Resume guest.*/
+   /* SR/HMI/PMI are HV interrupts that host has handled. Resume guest.*/
case BOOK3S_INTERRUPT_HMI:
case BOOK3S_INTERRUPT_PERFMON:
+   case BOOK3S_INTERRUPT_SYSTEM_RESET:
r = RESUME_GUEST;
break;
case BOOK3S_INTERRUPT_MACHINE_CHECK:
@@ -2604,6 +2606,9 @@ static void set_irq_happened(int trap)
case BOOK3S_INTERRUPT_HMI:
local_paca->irq_happened |= PACA_IRQ_HMI;
break;
+   case BOOK3S_INTERRUPT_SYSTEM_RESET:
+   replay_system_reset();
+

Re: [RFC PATCH] KVM: PPC: Book3S HV: Handle host system reset in guest mode

2017-11-02 Thread Paul Mackerras 
On Fri, Nov 03, 2017 at 03:38:03PM +1100, Nicholas Piggin wrote:
> If the host takes a system reset interrupt while a guest is running,
> the CPU must exit the guest before processing the host exception
> handler.
> 
> After this patch, taking a sysrq+x with a CPU running in a guest
> gives a trace like this:
> 
>cpu 0x27: Vector: 100 (System Reset) at [c00fdf5776f0]
>pc: c00810158b80: kvmppc_run_core+0x16b8/0x1ad0 [kvm_hv]
>lr: c00810158b80: kvmppc_run_core+0x16b8/0x1ad0 [kvm_hv]
>sp: c00fdf577850
>   msr: 92803033
>  current = 0xc00fdf4b1e00
>  paca= 0xcfd4d680  softe: 3irq_happened: 0x01
>pid   = 6608, comm = qemu-system-ppc
>Linux version 4.14.0-rc7-01489-g47e1893a404a-dirty #26 SMP
>[c00fdf577a00] c00810159dd4 kvmppc_vcpu_run_hv+0x3dc/0x12d0 
> [kvm_hv]
>[c00fdf577b30] c008100a537c kvmppc_vcpu_run+0x44/0x60 [kvm]
>[c00fdf577b60] c008100a1ae0 kvm_arch_vcpu_ioctl_run+0x118/0x310 
> [kvm]
>[c00fdf577c00] c00810093e98 kvm_vcpu_ioctl+0x530/0x7c0 [kvm]
>[c00fdf577d50] c0357bf8 do_vfs_ioctl+0xd8/0x8c0
>[c00fdf577df0] c0358448 SyS_ioctl+0x68/0x100
>[c00fdf577e30] c000b220 system_call+0x58/0x6c
>--- Exception: c01 (System Call) at 7fff76868df0
>SP (7fff7069baf0) is in userspace
> 
> Fixes: e36d0a2ed5 ("powerpc/powernv: Implement NMI IPI with 
> OPAL_SIGNAL_SYSTEM_RESET")
> Signed-off-by: Nicholas Piggin 

Looks good, except that you don't need the last hunk (the change to
book3s_hv_rmhandlers.S) as far as I can see:

> diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S 
> b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
> index 68bf0f14a962..74958ad5efb9 100644
> --- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
> +++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
> @@ -1427,6 +1427,10 @@ guest_exit_cont:   /* r9 = vcpu, r12 = 
> trap, r13 = paca */
>   /* don't overwrite fault_dar/fault_dsisr if HDSI */
>   cmpwi   r12,BOOK3S_INTERRUPT_H_DATA_STORAGE
>   beq mc_cont
> +
> + cmpwi   r12, BOOK3S_INTERRUPT_SYSTEM_RESET
> + beq mc_cont
> +
>   std r6, VCPU_FAULT_DAR(r9)
>   stw r7, VCPU_FAULT_DSISR(r9)
>  
> -- 
> 2.15.0

Paul.

[RFC PATCH] KVM: PPC: Book3S HV: Handle host system reset in guest mode

2017-11-02 Thread Nicholas Piggin 
If the host takes a system reset interrupt while a guest is running,
the CPU must exit the guest before processing the host exception
handler.

After this patch, taking a sysrq+x with a CPU running in a guest
gives a trace like this:

   cpu 0x27: Vector: 100 (System Reset) at [c00fdf5776f0]
   pc: c00810158b80: kvmppc_run_core+0x16b8/0x1ad0 [kvm_hv]
   lr: c00810158b80: kvmppc_run_core+0x16b8/0x1ad0 [kvm_hv]
   sp: c00fdf577850
  msr: 92803033
 current = 0xc00fdf4b1e00
 paca= 0xcfd4d680softe: 3irq_happened: 0x01
   pid   = 6608, comm = qemu-system-ppc
   Linux version 4.14.0-rc7-01489-g47e1893a404a-dirty #26 SMP
   [c00fdf577a00] c00810159dd4 kvmppc_vcpu_run_hv+0x3dc/0x12d0 [kvm_hv]
   [c00fdf577b30] c008100a537c kvmppc_vcpu_run+0x44/0x60 [kvm]
   [c00fdf577b60] c008100a1ae0 kvm_arch_vcpu_ioctl_run+0x118/0x310 [kvm]
   [c00fdf577c00] c00810093e98 kvm_vcpu_ioctl+0x530/0x7c0 [kvm]
   [c00fdf577d50] c0357bf8 do_vfs_ioctl+0xd8/0x8c0
   [c00fdf577df0] c0358448 SyS_ioctl+0x68/0x100
   [c00fdf577e30] c000b220 system_call+0x58/0x6c
   --- Exception: c01 (System Call) at 7fff76868df0
   SP (7fff7069baf0) is in userspace

Fixes: e36d0a2ed5 ("powerpc/powernv: Implement NMI IPI with 
OPAL_SIGNAL_SYSTEM_RESET")
Signed-off-by: Nicholas Piggin 
--

It has always been possible to sreset the host with direct scom
access, but the patch e36d0a2ed5 has significantly expanded this
functionality so in practice this is a required as a fix for it.

For 4.14 I think we will either need to fix this, or disable
e36d0a2ed5, otherwise the host could be exposed to the guest MMU
(at least with hash).

Thanks,
Nick
---
 arch/powerpc/include/asm/hw_irq.h   | 1 +
 arch/powerpc/kernel/exceptions-64s.S| 2 ++
 arch/powerpc/kernel/irq.c   | 3 ++-
 arch/powerpc/kvm/book3s_hv.c| 7 ++-
 arch/powerpc/kvm/book3s_hv_rmhandlers.S | 4 
 5 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/include/asm/hw_irq.h 
b/arch/powerpc/include/asm/hw_irq.h
index 92a3e9a79cb4..a8bbac425ae6 100644
--- a/arch/powerpc/include/asm/hw_irq.h
+++ b/arch/powerpc/include/asm/hw_irq.h
@@ -40,6 +40,7 @@
 
 #ifndef __ASSEMBLY__
 
+extern void replay_system_reset(void);
 extern void __replay_interrupt(unsigned int vector);
 
 extern void timer_interrupt(struct pt_regs *);
diff --git a/arch/powerpc/kernel/exceptions-64s.S 
b/arch/powerpc/kernel/exceptions-64s.S
index 651e1a0114ed..bff2ed6e3c3c 100644
--- a/arch/powerpc/kernel/exceptions-64s.S
+++ b/arch/powerpc/kernel/exceptions-64s.S
@@ -113,6 +113,7 @@ EXC_VIRT_NONE(0x4000, 0x100)
cmpwi   cr3,r10,2 ; \
BRANCH_TO_C000(r10, system_reset_idle_common) ; \
 1: \
+   KVMTEST_PR(n) ; \
END_FTR_SECTION_IFSET(CPU_FTR_HVMODE | CPU_FTR_ARCH_206)
 #else
 #define IDLETEST NOTEST
@@ -129,6 +130,7 @@ EXC_REAL_BEGIN(system_reset, 0x100, 0x100)
 
 EXC_REAL_END(system_reset, 0x100, 0x100)
 EXC_VIRT_NONE(0x4100, 0x100)
+TRAMP_KVM(PACA_EXNMI, 0x100)
 
 #ifdef CONFIG_PPC_P7_NAP
 EXC_COMMON_BEGIN(system_reset_idle_common)
diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c
index efbadcbbf694..7e8259106944 100644
--- a/arch/powerpc/kernel/irq.c
+++ b/arch/powerpc/kernel/irq.c
@@ -437,7 +437,7 @@ static const u8 srr1_to_lazyirq[0x10] = {
PACA_IRQ_HMI,
0, 0, 0, 0, 0 };
 
-static noinline void replay_system_reset(void)
+void replay_system_reset(void)
 {
struct pt_regs regs;
 
@@ -447,6 +447,7 @@ static noinline void replay_system_reset(void)
system_reset_exception();
get_paca()->in_nmi = 0;
 }
+EXPORT_SYMBOL_GPL(replay_system_reset);
 
 void irq_set_pending_from_srr1(unsigned long srr1)
 {
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 8f34715cfbff..31a362669fea 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -47,6 +47,7 @@
 
 #include 
 #include 
+#include 
 #include 
 #include 
 #include 
@@ -1089,9 +1090,10 @@ static int kvmppc_handle_exit_hv(struct kvm_run *run, 
struct kvm_vcpu *vcpu,
vcpu->stat.ext_intr_exits++;
r = RESUME_GUEST;
break;
-   /* HMI is hypervisor interrupt and host has handled it. Resume guest.*/
+   /* SR/HMI/PMI are HV interrupts that host has handled. Resume guest.*/
case BOOK3S_INTERRUPT_HMI:
case BOOK3S_INTERRUPT_PERFMON:
+   case BOOK3S_INTERRUPT_SYSTEM_RESET:
r = RESUME_GUEST;
break;
case BOOK3S_INTERRUPT_MACHINE_CHECK:
@@ -2604,6 +2606,9 @@ static void set_irq_happened(int trap)
case BOOK3S_INTERRUPT_HMI:
local_paca->irq_happened