All of lore.kernel.org
 help / color / mirror / Atom feed
* [PATCH 0/4] KVM: PPC: Convert P9 HV path to C
@ 2021-05-13 12:22 ` Nicholas Piggin
  0 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc; +Cc: linuxppc-dev, Nicholas Piggin

This applies on top of these series:

https://patchwork.ozlabs.org/project/linuxppc-dev/list/?series=238649
https://patchwork.ozlabs.org/project/linuxppc-dev/list/?series=238941
https://patchwork.ozlabs.org/project/linuxppc-dev/list/?series=238946

This was broken out from the large Cify series. Since that was last
posted, changes are:

- Rebase, reordering of patches, tweaking changelog and comments.
- Changed P9 radix exist SLB sanitising to use 4x slbmte to clear
  rather than 8x slbmfee/slbmfev, which turns out to be faster (and
  is what today's asm code does) [from review from Paul].
- Renamed book3s_hv_interrupt.c to book3s_hv_p9_entry.c, which reduces
  confusion with book3s_hv_interrupts.S.
- Fixed !HV compile [Alexey].

Nicholas Piggin (4):
  KVM: PPC: Book3S HV P9: Move xive vcpu context management into
    kvmhv_p9_guest_entry
  KVM: PPC: Book3S HV P9: Move radix MMU switching instructions together
  KVM: PPC: Book3S HV P9: Stop handling hcalls in real-mode in the P9
    path
  KVM: PPC: Book3S HV P9: Implement the rest of the P9 path in C

 arch/powerpc/include/asm/asm-prototypes.h |   3 +-
 arch/powerpc/include/asm/kvm_asm.h        |   1 +
 arch/powerpc/include/asm/kvm_book3s_64.h  |   8 +
 arch/powerpc/include/asm/kvm_host.h       |   7 +-
 arch/powerpc/include/asm/kvm_ppc.h        |   6 +
 arch/powerpc/kernel/security.c            |   5 +-
 arch/powerpc/kvm/Makefile                 |   1 +
 arch/powerpc/kvm/book3s.c                 |   6 +
 arch/powerpc/kvm/book3s_64_entry.S        | 254 ++++++++++++++++++++++
 arch/powerpc/kvm/book3s_hv.c              | 155 +++++++++----
 arch/powerpc/kvm/book3s_hv_p9_entry.c     | 207 ++++++++++++++++++
 arch/powerpc/kvm/book3s_hv_rmhandlers.S   | 120 +---------
 arch/powerpc/kvm/book3s_xive.c            |  64 ++++++
 13 files changed, 683 insertions(+), 154 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_hv_p9_entry.c

-- 
2.23.0


^ permalink raw reply	[flat|nested] 10+ messages in thread
* [PATCH 0/4] KVM: PPC: Convert P9 HV path to C
@ 2021-05-13 12:22 ` Nicholas Piggin
  0 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc; +Cc: linuxppc-dev, Nicholas Piggin

This applies on top of these series:

https://patchwork.ozlabs.org/project/linuxppc-dev/list/?series#8649
https://patchwork.ozlabs.org/project/linuxppc-dev/list/?series#8941
https://patchwork.ozlabs.org/project/linuxppc-dev/list/?series#8946

This was broken out from the large Cify series. Since that was last
posted, changes are:

- Rebase, reordering of patches, tweaking changelog and comments.
- Changed P9 radix exist SLB sanitising to use 4x slbmte to clear
  rather than 8x slbmfee/slbmfev, which turns out to be faster (and
  is what today's asm code does) [from review from Paul].
- Renamed book3s_hv_interrupt.c to book3s_hv_p9_entry.c, which reduces
  confusion with book3s_hv_interrupts.S.
- Fixed !HV compile [Alexey].

Nicholas Piggin (4):
  KVM: PPC: Book3S HV P9: Move xive vcpu context management into
    kvmhv_p9_guest_entry
  KVM: PPC: Book3S HV P9: Move radix MMU switching instructions together
  KVM: PPC: Book3S HV P9: Stop handling hcalls in real-mode in the P9
    path
  KVM: PPC: Book3S HV P9: Implement the rest of the P9 path in C

 arch/powerpc/include/asm/asm-prototypes.h |   3 +-
 arch/powerpc/include/asm/kvm_asm.h        |   1 +
 arch/powerpc/include/asm/kvm_book3s_64.h  |   8 +
 arch/powerpc/include/asm/kvm_host.h       |   7 +-
 arch/powerpc/include/asm/kvm_ppc.h        |   6 +
 arch/powerpc/kernel/security.c            |   5 +-
 arch/powerpc/kvm/Makefile                 |   1 +
 arch/powerpc/kvm/book3s.c                 |   6 +
 arch/powerpc/kvm/book3s_64_entry.S        | 254 ++++++++++++++++++++++
 arch/powerpc/kvm/book3s_hv.c              | 155 +++++++++----
 arch/powerpc/kvm/book3s_hv_p9_entry.c     | 207 ++++++++++++++++++
 arch/powerpc/kvm/book3s_hv_rmhandlers.S   | 120 +---------
 arch/powerpc/kvm/book3s_xive.c            |  64 ++++++
 13 files changed, 683 insertions(+), 154 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_hv_p9_entry.c

-- 
2.23.0

^ permalink raw reply	[flat|nested] 10+ messages in thread
* [PATCH 1/4] KVM: PPC: Book3S HV P9: Move xive vcpu context management into kvmhv_p9_guest_entry
  2021-05-13 12:22 ` Nicholas Piggin
@ 2021-05-13 12:22   ` Nicholas Piggin
  -1 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc; +Cc: Alexey Kardashevskiy, linuxppc-dev, Nicholas Piggin

Move the xive management up so the low level register switching can be
pushed further down in a later patch. XIVE MMIO CI operations can run in
higher level code with machine checks, tracing, etc., available.

Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/kvm/book3s_hv.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 8a31df067c65..68914b26017b 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -3558,15 +3558,11 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	 */
 	mtspr(SPRN_HDEC, hdec);
 
-	kvmppc_xive_push_vcpu(vcpu);
-
 	mtspr(SPRN_SRR0, vcpu->arch.shregs.srr0);
 	mtspr(SPRN_SRR1, vcpu->arch.shregs.srr1);
 
 	trap = __kvmhv_vcpu_entry_p9(vcpu);
 
-	kvmppc_xive_pull_vcpu(vcpu);
-
 	/* Advance host PURR/SPURR by the amount used by guest */
 	purr = mfspr(SPRN_PURR);
 	spurr = mfspr(SPRN_SPURR);
@@ -3764,7 +3760,10 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 			trap = 0;
 		}
 	} else {
+		kvmppc_xive_push_vcpu(vcpu);
 		trap = kvmhv_load_hv_regs_and_go(vcpu, time_limit, lpcr);
+		kvmppc_xive_pull_vcpu(vcpu);
+
 	}
 
 	vcpu->arch.slb_max = 0;
-- 
2.23.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread
* [PATCH 1/4] KVM: PPC: Book3S HV P9: Move xive vcpu context management into kvmhv_p9_guest_entry
@ 2021-05-13 12:22   ` Nicholas Piggin
  0 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc; +Cc: Alexey Kardashevskiy, linuxppc-dev, Nicholas Piggin

Move the xive management up so the low level register switching can be
pushed further down in a later patch. XIVE MMIO CI operations can run in
higher level code with machine checks, tracing, etc., available.

Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/kvm/book3s_hv.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 8a31df067c65..68914b26017b 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -3558,15 +3558,11 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	 */
 	mtspr(SPRN_HDEC, hdec);
 
-	kvmppc_xive_push_vcpu(vcpu);
-
 	mtspr(SPRN_SRR0, vcpu->arch.shregs.srr0);
 	mtspr(SPRN_SRR1, vcpu->arch.shregs.srr1);
 
 	trap = __kvmhv_vcpu_entry_p9(vcpu);
 
-	kvmppc_xive_pull_vcpu(vcpu);
-
 	/* Advance host PURR/SPURR by the amount used by guest */
 	purr = mfspr(SPRN_PURR);
 	spurr = mfspr(SPRN_SPURR);
@@ -3764,7 +3760,10 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 			trap = 0;
 		}
 	} else {
+		kvmppc_xive_push_vcpu(vcpu);
 		trap = kvmhv_load_hv_regs_and_go(vcpu, time_limit, lpcr);
+		kvmppc_xive_pull_vcpu(vcpu);
+
 	}
 
 	vcpu->arch.slb_max = 0;
-- 
2.23.0

^ permalink raw reply related	[flat|nested] 10+ messages in thread
* [PATCH 2/4] KVM: PPC: Book3S HV P9: Move radix MMU switching instructions together
  2021-05-13 12:22 ` Nicholas Piggin
@ 2021-05-13 12:22   ` Nicholas Piggin
  -1 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc; +Cc: Alexey Kardashevskiy, linuxppc-dev, Nicholas Piggin

Switching the MMU from radix<->radix mode is tricky particularly as the
MMU can remain enabled and requires a certain sequence of SPR updates.
Move these together into their own functions.

This also includes the radix TLB check / flush because it's tied in to
MMU switching due to tlbiel getting LPID from LPIDR.

Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/kvm/book3s_hv.c | 62 ++++++++++++++++++++++++------------
 1 file changed, 41 insertions(+), 21 deletions(-)

diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 68914b26017b..287cd3e1b918 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -3478,12 +3478,49 @@ static noinline void kvmppc_run_core(struct kvmppc_vcore *vc)
 	trace_kvmppc_run_core(vc, 1);
 }
 
+static void switch_mmu_to_guest_radix(struct kvm *kvm, struct kvm_vcpu *vcpu, u64 lpcr)
+{
+	struct kvmppc_vcore *vc = vcpu->arch.vcore;
+	struct kvm_nested_guest *nested = vcpu->arch.nested;
+	u32 lpid;
+
+	lpid = nested ? nested->shadow_lpid : kvm->arch.lpid;
+
+	/*
+	 * All the isync()s are overkill but trivially follow the ISA
+	 * requirements. Some can likely be replaced with justification
+	 * comment for why they are not needed.
+	 */
+	isync();
+	mtspr(SPRN_LPID, lpid);
+	isync();
+	mtspr(SPRN_LPCR, lpcr);
+	isync();
+	mtspr(SPRN_PID, vcpu->arch.pid);
+	isync();
+
+	/* TLBIEL must have LPIDR set, so set guest LPID before flushing. */
+	kvmppc_check_need_tlb_flush(kvm, vc->pcpu, nested);
+}
+
+static void switch_mmu_to_host_radix(struct kvm *kvm, u32 pid)
+{
+	isync();
+	mtspr(SPRN_PID, pid);
+	isync();
+	mtspr(SPRN_LPID, kvm->arch.host_lpid);
+	isync();
+	mtspr(SPRN_LPCR, kvm->arch.host_lpcr);
+	isync();
+}
+
 /*
  * Load up hypervisor-mode registers on P9.
  */
 static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 				     unsigned long lpcr)
 {
+	struct kvm *kvm = vcpu->kvm;
 	struct kvmppc_vcore *vc = vcpu->arch.vcore;
 	s64 hdec;
 	u64 tb, purr, spurr;
@@ -3535,7 +3572,6 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	}
 	mtspr(SPRN_CIABR, vcpu->arch.ciabr);
 	mtspr(SPRN_IC, vcpu->arch.ic);
-	mtspr(SPRN_PID, vcpu->arch.pid);
 
 	mtspr(SPRN_PSSCR, vcpu->arch.psscr | PSSCR_EC |
 	      (local_paca->kvm_hstate.fake_suspend << PSSCR_FAKE_SUSPEND_LG));
@@ -3549,8 +3585,7 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 
 	mtspr(SPRN_AMOR, ~0UL);
 
-	mtspr(SPRN_LPCR, lpcr);
-	isync();
+	switch_mmu_to_guest_radix(kvm, vcpu, lpcr);
 
 	/*
 	 * P9 suppresses the HDEC exception when LPCR[HDICE] = 0,
@@ -3593,7 +3628,6 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 		mtspr(SPRN_DAWR1, host_dawr1);
 		mtspr(SPRN_DAWRX1, host_dawrx1);
 	}
-	mtspr(SPRN_PID, host_pidr);
 
 	/*
 	 * Since this is radix, do a eieio; tlbsync; ptesync sequence in
@@ -3608,9 +3642,6 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	if (cpu_has_feature(CPU_FTR_ARCH_31))
 		asm volatile(PPC_CP_ABORT);
 
-	mtspr(SPRN_LPID, vcpu->kvm->arch.host_lpid);	/* restore host LPID */
-	isync();
-
 	vc->dpdes = mfspr(SPRN_DPDES);
 	vc->vtb = mfspr(SPRN_VTB);
 	mtspr(SPRN_DPDES, 0);
@@ -3627,7 +3658,8 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	}
 
 	mtspr(SPRN_HDEC, 0x7fffffff);
-	mtspr(SPRN_LPCR, vcpu->kvm->arch.host_lpcr);
+
+	switch_mmu_to_host_radix(kvm, host_pidr);
 
 	return trap;
 }
@@ -4181,7 +4213,7 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 {
 	struct kvm_run *run = vcpu->run;
 	int trap, r, pcpu;
-	int srcu_idx, lpid;
+	int srcu_idx;
 	struct kvmppc_vcore *vc;
 	struct kvm *kvm = vcpu->kvm;
 	struct kvm_nested_guest *nested = vcpu->arch.nested;
@@ -4255,13 +4287,6 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 	vc->vcore_state = VCORE_RUNNING;
 	trace_kvmppc_run_core(vc, 0);
 
-	if (cpu_has_feature(CPU_FTR_HVMODE)) {
-		lpid = nested ? nested->shadow_lpid : kvm->arch.lpid;
-		mtspr(SPRN_LPID, lpid);
-		isync();
-		kvmppc_check_need_tlb_flush(kvm, pcpu, nested);
-	}
-
 	guest_enter_irqoff();
 
 	srcu_idx = srcu_read_lock(&kvm->srcu);
@@ -4280,11 +4305,6 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 
 	srcu_read_unlock(&kvm->srcu, srcu_idx);
 
-	if (cpu_has_feature(CPU_FTR_HVMODE)) {
-		mtspr(SPRN_LPID, kvm->arch.host_lpid);
-		isync();
-	}
-
 	set_irq_happened(trap);
 
 	kvmppc_set_host_core(pcpu);
-- 
2.23.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread
* [PATCH 2/4] KVM: PPC: Book3S HV P9: Move radix MMU switching instructions together
@ 2021-05-13 12:22   ` Nicholas Piggin
  0 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc; +Cc: Alexey Kardashevskiy, linuxppc-dev, Nicholas Piggin

Switching the MMU from radix<->radix mode is tricky particularly as the
MMU can remain enabled and requires a certain sequence of SPR updates.
Move these together into their own functions.

This also includes the radix TLB check / flush because it's tied in to
MMU switching due to tlbiel getting LPID from LPIDR.

Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/kvm/book3s_hv.c | 62 ++++++++++++++++++++++++------------
 1 file changed, 41 insertions(+), 21 deletions(-)

diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 68914b26017b..287cd3e1b918 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -3478,12 +3478,49 @@ static noinline void kvmppc_run_core(struct kvmppc_vcore *vc)
 	trace_kvmppc_run_core(vc, 1);
 }
 
+static void switch_mmu_to_guest_radix(struct kvm *kvm, struct kvm_vcpu *vcpu, u64 lpcr)
+{
+	struct kvmppc_vcore *vc = vcpu->arch.vcore;
+	struct kvm_nested_guest *nested = vcpu->arch.nested;
+	u32 lpid;
+
+	lpid = nested ? nested->shadow_lpid : kvm->arch.lpid;
+
+	/*
+	 * All the isync()s are overkill but trivially follow the ISA
+	 * requirements. Some can likely be replaced with justification
+	 * comment for why they are not needed.
+	 */
+	isync();
+	mtspr(SPRN_LPID, lpid);
+	isync();
+	mtspr(SPRN_LPCR, lpcr);
+	isync();
+	mtspr(SPRN_PID, vcpu->arch.pid);
+	isync();
+
+	/* TLBIEL must have LPIDR set, so set guest LPID before flushing. */
+	kvmppc_check_need_tlb_flush(kvm, vc->pcpu, nested);
+}
+
+static void switch_mmu_to_host_radix(struct kvm *kvm, u32 pid)
+{
+	isync();
+	mtspr(SPRN_PID, pid);
+	isync();
+	mtspr(SPRN_LPID, kvm->arch.host_lpid);
+	isync();
+	mtspr(SPRN_LPCR, kvm->arch.host_lpcr);
+	isync();
+}
+
 /*
  * Load up hypervisor-mode registers on P9.
  */
 static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 				     unsigned long lpcr)
 {
+	struct kvm *kvm = vcpu->kvm;
 	struct kvmppc_vcore *vc = vcpu->arch.vcore;
 	s64 hdec;
 	u64 tb, purr, spurr;
@@ -3535,7 +3572,6 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	}
 	mtspr(SPRN_CIABR, vcpu->arch.ciabr);
 	mtspr(SPRN_IC, vcpu->arch.ic);
-	mtspr(SPRN_PID, vcpu->arch.pid);
 
 	mtspr(SPRN_PSSCR, vcpu->arch.psscr | PSSCR_EC |
 	      (local_paca->kvm_hstate.fake_suspend << PSSCR_FAKE_SUSPEND_LG));
@@ -3549,8 +3585,7 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 
 	mtspr(SPRN_AMOR, ~0UL);
 
-	mtspr(SPRN_LPCR, lpcr);
-	isync();
+	switch_mmu_to_guest_radix(kvm, vcpu, lpcr);
 
 	/*
 	 * P9 suppresses the HDEC exception when LPCR[HDICE] = 0,
@@ -3593,7 +3628,6 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 		mtspr(SPRN_DAWR1, host_dawr1);
 		mtspr(SPRN_DAWRX1, host_dawrx1);
 	}
-	mtspr(SPRN_PID, host_pidr);
 
 	/*
 	 * Since this is radix, do a eieio; tlbsync; ptesync sequence in
@@ -3608,9 +3642,6 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	if (cpu_has_feature(CPU_FTR_ARCH_31))
 		asm volatile(PPC_CP_ABORT);
 
-	mtspr(SPRN_LPID, vcpu->kvm->arch.host_lpid);	/* restore host LPID */
-	isync();
-
 	vc->dpdes = mfspr(SPRN_DPDES);
 	vc->vtb = mfspr(SPRN_VTB);
 	mtspr(SPRN_DPDES, 0);
@@ -3627,7 +3658,8 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	}
 
 	mtspr(SPRN_HDEC, 0x7fffffff);
-	mtspr(SPRN_LPCR, vcpu->kvm->arch.host_lpcr);
+
+	switch_mmu_to_host_radix(kvm, host_pidr);
 
 	return trap;
 }
@@ -4181,7 +4213,7 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 {
 	struct kvm_run *run = vcpu->run;
 	int trap, r, pcpu;
-	int srcu_idx, lpid;
+	int srcu_idx;
 	struct kvmppc_vcore *vc;
 	struct kvm *kvm = vcpu->kvm;
 	struct kvm_nested_guest *nested = vcpu->arch.nested;
@@ -4255,13 +4287,6 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 	vc->vcore_state = VCORE_RUNNING;
 	trace_kvmppc_run_core(vc, 0);
 
-	if (cpu_has_feature(CPU_FTR_HVMODE)) {
-		lpid = nested ? nested->shadow_lpid : kvm->arch.lpid;
-		mtspr(SPRN_LPID, lpid);
-		isync();
-		kvmppc_check_need_tlb_flush(kvm, pcpu, nested);
-	}
-
 	guest_enter_irqoff();
 
 	srcu_idx = srcu_read_lock(&kvm->srcu);
@@ -4280,11 +4305,6 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 
 	srcu_read_unlock(&kvm->srcu, srcu_idx);
 
-	if (cpu_has_feature(CPU_FTR_HVMODE)) {
-		mtspr(SPRN_LPID, kvm->arch.host_lpid);
-		isync();
-	}
-
 	set_irq_happened(trap);
 
 	kvmppc_set_host_core(pcpu);
-- 
2.23.0

^ permalink raw reply related	[flat|nested] 10+ messages in thread
* [PATCH 3/4] KVM: PPC: Book3S HV P9: Stop handling hcalls in real-mode in the P9 path
  2021-05-13 12:22 ` Nicholas Piggin
@ 2021-05-13 12:22   ` Nicholas Piggin
  -1 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc
  Cc: Alexey Kardashevskiy, linuxppc-dev, Cédric Le Goater,
	Nicholas Piggin

In the interest of minimising the amount of code that is run in
"real-mode", don't handle hcalls in real mode in the P9 path. This
requires some new handlers for H_CEDE and xics-on-xive to be added
before xive is pulled or cede logic is checked.

This introduces a change in radix guest behaviour where radix guests
that execute 'sc 1' in userspace now get a privilege fault whereas
previously the 'sc 1' would be reflected as a syscall interrupt to the
guest kernel. That reflection is only required for hash guests that run
PR KVM.

Background:

In POWER8 and earlier processors, it is very expensive to exit from the
HV real mode context of a guest hypervisor interrupt, and switch to host
virtual mode. On those processors, guest->HV interrupts reach the
hypervisor with the MMU off because the MMU is loaded with guest context
(LPCR, SDR1, SLB), and the other threads in the sub-core need to be
pulled out of the guest too. Then the primary must save off guest state,
invalidate SLB and ERAT, and load up host state before the MMU can be
enabled to run in host virtual mode (~= regular Linux mode).

Hash guests also require a lot of hcalls to run due to the nature of the
MMU architecture and paravirtualisation design. The XICS interrupt
controller requires hcalls to run.

So KVM traditionally tries hard to avoid the full exit, by handling
hcalls and other interrupts in real mode as much as possible.

By contrast, POWER9 has independent MMU context per-thread, and in radix
mode the hypervisor is in host virtual memory mode when the HV interrupt
is taken. Radix guests do not require significant hcalls to manage their
translations, and xive guests don't need hcalls to handle interrupts.

So it's much less important from a performance standpoint to handle
hcalls in real mode in P9.

The TCE hcalls are performance critical, introduced for P8 in order to
achieve 10GbE performance. They was not required on P9 (which was able
to drive 40GbE networking with just the virt mode hcalls), but
performance is quite important. After later changes that add hash guest
support to the P9 path, hash hcalls are also performance critical.  The
full entry/exit performance can still be improved significanty, but if
this is found to be inadequate then real-mode hcall handlers could be
reintroduced for the P9 path (in C), but this would be a last resort.

Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/include/asm/kvm_ppc.h      |  6 ++
 arch/powerpc/kvm/book3s.c               |  6 ++
 arch/powerpc/kvm/book3s_hv.c            | 79 +++++++++++++++++++++----
 arch/powerpc/kvm/book3s_hv_rmhandlers.S |  5 ++
 arch/powerpc/kvm/book3s_xive.c          | 64 ++++++++++++++++++++
 5 files changed, 149 insertions(+), 11 deletions(-)

diff --git a/arch/powerpc/include/asm/kvm_ppc.h b/arch/powerpc/include/asm/kvm_ppc.h
index 8c10c3427166..cb9e3c85c605 100644
--- a/arch/powerpc/include/asm/kvm_ppc.h
+++ b/arch/powerpc/include/asm/kvm_ppc.h
@@ -129,6 +129,7 @@ extern void kvmppc_core_vcpu_put(struct kvm_vcpu *vcpu);
 extern int kvmppc_core_prepare_to_enter(struct kvm_vcpu *vcpu);
 extern int kvmppc_core_pending_dec(struct kvm_vcpu *vcpu);
 extern void kvmppc_core_queue_machine_check(struct kvm_vcpu *vcpu, ulong flags);
+extern void kvmppc_core_queue_syscall(struct kvm_vcpu *vcpu);
 extern void kvmppc_core_queue_program(struct kvm_vcpu *vcpu, ulong flags);
 extern void kvmppc_core_queue_fpunavail(struct kvm_vcpu *vcpu);
 extern void kvmppc_core_queue_vec_unavail(struct kvm_vcpu *vcpu);
@@ -606,6 +607,7 @@ extern void kvmppc_free_pimap(struct kvm *kvm);
 extern int kvmppc_xics_rm_complete(struct kvm_vcpu *vcpu, u32 hcall);
 extern void kvmppc_xics_free_icp(struct kvm_vcpu *vcpu);
 extern int kvmppc_xics_hcall(struct kvm_vcpu *vcpu, u32 cmd);
+extern int kvmppc_xive_xics_hcall(struct kvm_vcpu *vcpu, u32 req);
 extern u64 kvmppc_xics_get_icp(struct kvm_vcpu *vcpu);
 extern int kvmppc_xics_set_icp(struct kvm_vcpu *vcpu, u64 icpval);
 extern int kvmppc_xics_connect_vcpu(struct kvm_device *dev,
@@ -638,6 +640,8 @@ static inline int kvmppc_xics_enabled(struct kvm_vcpu *vcpu)
 static inline void kvmppc_xics_free_icp(struct kvm_vcpu *vcpu) { }
 static inline int kvmppc_xics_hcall(struct kvm_vcpu *vcpu, u32 cmd)
 	{ return 0; }
+static inline int kvmppc_xive_xics_hcall(struct kvm_vcpu *vcpu, u32 req)
+	{ return 0; }
 #endif
 
 #ifdef CONFIG_KVM_XIVE
@@ -672,6 +676,7 @@ extern int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 			       int level, bool line_status);
 extern void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_pull_vcpu(struct kvm_vcpu *vcpu);
+extern void kvmppc_xive_rearm_escalation(struct kvm_vcpu *vcpu);
 
 static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
 {
@@ -713,6 +718,7 @@ static inline int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 ir
 				      int level, bool line_status) { return -ENODEV; }
 static inline void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_pull_vcpu(struct kvm_vcpu *vcpu) { }
+static inline void kvmppc_xive_rearm_escalation(struct kvm_vcpu *vcpu) { }
 
 static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
 	{ return 0; }
diff --git a/arch/powerpc/kvm/book3s.c b/arch/powerpc/kvm/book3s.c
index 2b691f4d1f26..d69560d5bf16 100644
--- a/arch/powerpc/kvm/book3s.c
+++ b/arch/powerpc/kvm/book3s.c
@@ -171,6 +171,12 @@ void kvmppc_core_queue_machine_check(struct kvm_vcpu *vcpu, ulong flags)
 }
 EXPORT_SYMBOL_GPL(kvmppc_core_queue_machine_check);
 
+void kvmppc_core_queue_syscall(struct kvm_vcpu *vcpu)
+{
+	kvmppc_inject_interrupt(vcpu, BOOK3S_INTERRUPT_SYSCALL, 0);
+}
+EXPORT_SYMBOL(kvmppc_core_queue_syscall);
+
 void kvmppc_core_queue_program(struct kvm_vcpu *vcpu, ulong flags)
 {
 	/* might as well deliver this straight away */
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 287cd3e1b918..5815c04a48a3 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -899,6 +899,10 @@ static int kvm_arch_vcpu_yield_to(struct kvm_vcpu *target)
 	 * H_SUCCESS if the source vcore wasn't idle (e.g. if it may
 	 * have useful work to do and should not confer) so we don't
 	 * recheck that here.
+	 *
+	 * In the case of the P9 single vcpu per vcore case, the real
+	 * mode handler is not called but no other threads are in the
+	 * source vcore.
 	 */
 
 	spin_lock(&vcore->lock);
@@ -1142,12 +1146,13 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu)
 }
 
 /*
- * Handle H_CEDE in the nested virtualization case where we haven't
- * called the real-mode hcall handlers in book3s_hv_rmhandlers.S.
+ * Handle H_CEDE in the P9 path where we don't call the real-mode hcall
+ * handlers in book3s_hv_rmhandlers.S.
+ *
  * This has to be done early, not in kvmppc_pseries_do_hcall(), so
  * that the cede logic in kvmppc_run_single_vcpu() works properly.
  */
-static void kvmppc_nested_cede(struct kvm_vcpu *vcpu)
+static void kvmppc_cede(struct kvm_vcpu *vcpu)
 {
 	vcpu->arch.shregs.msr |= MSR_EE;
 	vcpu->arch.ceded = 1;
@@ -1400,13 +1405,29 @@ static int kvmppc_handle_exit_hv(struct kvm_vcpu *vcpu,
 	}
 	case BOOK3S_INTERRUPT_SYSCALL:
 	{
-		/* hcall - punt to userspace */
 		int i;
 
-		/* hypercall with MSR_PR has already been handled in rmode,
-		 * and never reaches here.
-		 */
+		if (unlikely(vcpu->arch.shregs.msr & MSR_PR)) {
+			/*
+			 * Guest userspace executed sc 1. This can only be
+			 * reached by the P9 path because the old path
+			 * handles this case in realmode hcall handlers.
+			 *
+			 * Radix guests can not run PR KVM or nested HV hash
+			 * guests which might run PR KVM, so this is always
+			 * a privilege fault. Send a program check to guest
+			 * kernel.
+			 */
+			kvmppc_core_queue_program(vcpu, SRR1_PROGPRIV);
+			r = RESUME_GUEST;
+			break;
+		}
 
+		/*
+		 * hcall - gather args and set exit_reason. This will next be
+		 * handled by kvmppc_pseries_do_hcall which may be able to deal
+		 * with it and resume guest, or may punt to userspace.
+		 */
 		run->papr_hcall.nr = kvmppc_get_gpr(vcpu, 3);
 		for (i = 0; i < 9; ++i)
 			run->papr_hcall.args[i] = kvmppc_get_gpr(vcpu, 4 + i);
@@ -3664,6 +3685,12 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	return trap;
 }
 
+static inline bool hcall_is_xics(unsigned long req)
+{
+	return req == H_EOI || req == H_CPPR || req == H_IPI ||
+		req == H_IPOLL || req == H_XIRR || req == H_XIRR_X;
+}
+
 /*
  * Virtual-mode guest entry for POWER9 and later when the host and
  * guest are both using the radix MMU.  The LPIDR has already been set.
@@ -3787,15 +3814,36 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 		/* H_CEDE has to be handled now, not later */
 		if (trap == BOOK3S_INTERRUPT_SYSCALL && !vcpu->arch.nested &&
 		    kvmppc_get_gpr(vcpu, 3) == H_CEDE) {
-			kvmppc_nested_cede(vcpu);
+			kvmppc_cede(vcpu);
 			kvmppc_set_gpr(vcpu, 3, 0);
 			trap = 0;
 		}
 	} else {
 		kvmppc_xive_push_vcpu(vcpu);
 		trap = kvmhv_load_hv_regs_and_go(vcpu, time_limit, lpcr);
+		if (trap == BOOK3S_INTERRUPT_SYSCALL && !vcpu->arch.nested &&
+		    !(vcpu->arch.shregs.msr & MSR_PR)) {
+			unsigned long req = kvmppc_get_gpr(vcpu, 3);
+
+			/* H_CEDE has to be handled now, not later */
+			if (req == H_CEDE) {
+				kvmppc_cede(vcpu);
+				kvmppc_xive_rearm_escalation(vcpu); /* may un-cede */
+				kvmppc_set_gpr(vcpu, 3, 0);
+				trap = 0;
+
+			/* XICS hcalls must be handled before xive is pulled */
+			} else if (hcall_is_xics(req)) {
+				int ret;
+
+				ret = kvmppc_xive_xics_hcall(vcpu, req);
+				if (ret != H_TOO_HARD) {
+					kvmppc_set_gpr(vcpu, 3, ret);
+					trap = 0;
+				}
+			}
+		}
 		kvmppc_xive_pull_vcpu(vcpu);
-
 	}
 
 	vcpu->arch.slb_max = 0;
@@ -4461,8 +4509,17 @@ static int kvmppc_vcpu_run_hv(struct kvm_vcpu *vcpu)
 		else
 			r = kvmppc_run_vcpu(vcpu);
 
-		if (run->exit_reason == KVM_EXIT_PAPR_HCALL &&
-		    !(vcpu->arch.shregs.msr & MSR_PR)) {
+		if (run->exit_reason == KVM_EXIT_PAPR_HCALL) {
+			if (WARN_ON_ONCE(vcpu->arch.shregs.msr & MSR_PR)) {
+				/*
+				 * These should have been caught reflected
+				 * into the guest by now. Final sanity check:
+				 * don't allow userspace to execute hcalls in
+				 * the hypervisor.
+				 */
+				r = RESUME_GUEST;
+				continue;
+			}
 			trace_kvm_hcall_enter(vcpu);
 			r = kvmppc_pseries_do_hcall(vcpu);
 			trace_kvm_hcall_exit(vcpu, r);
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
index c11597f815e4..2d0d14ed1d92 100644
--- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
@@ -1397,9 +1397,14 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
 	mr	r4,r9
 	bge	fast_guest_return
 2:
+	/* If we came in through the P9 short path, no real mode hcalls */
+	lwz	r0, STACK_SLOT_SHORT_PATH(r1)
+	cmpwi	r0, 0
+	bne	no_try_real
 	/* See if this is an hcall we can handle in real mode */
 	cmpwi	r12,BOOK3S_INTERRUPT_SYSCALL
 	beq	hcall_try_real_mode
+no_try_real:
 
 	/* Hypervisor doorbell - exit only if host IPI flag set */
 	cmpwi	r12, BOOK3S_INTERRUPT_H_DOORBELL
diff --git a/arch/powerpc/kvm/book3s_xive.c b/arch/powerpc/kvm/book3s_xive.c
index 741bf1f4387a..24c07094651a 100644
--- a/arch/powerpc/kvm/book3s_xive.c
+++ b/arch/powerpc/kvm/book3s_xive.c
@@ -158,6 +158,40 @@ void kvmppc_xive_pull_vcpu(struct kvm_vcpu *vcpu)
 }
 EXPORT_SYMBOL_GPL(kvmppc_xive_pull_vcpu);
 
+void kvmppc_xive_rearm_escalation(struct kvm_vcpu *vcpu)
+{
+	void __iomem *esc_vaddr = (void __iomem *)vcpu->arch.xive_esc_vaddr;
+
+	if (!esc_vaddr)
+		return;
+
+	/* we are using XIVE with single escalation */
+
+	if (vcpu->arch.xive_esc_on) {
+		/*
+		 * If we still have a pending escalation, abort the cede,
+		 * and we must set PQ to 10 rather than 00 so that we don't
+		 * potentially end up with two entries for the escalation
+		 * interrupt in the XIVE interrupt queue.  In that case
+		 * we also don't want to set xive_esc_on to 1 here in
+		 * case we race with xive_esc_irq().
+		 */
+		vcpu->arch.ceded = 0;
+		/*
+		 * The escalation interrupts are special as we don't EOI them.
+		 * There is no need to use the load-after-store ordering offset
+		 * to set PQ to 10 as we won't use StoreEOI.
+		 */
+		__raw_readq(esc_vaddr + XIVE_ESB_SET_PQ_10);
+	} else {
+		vcpu->arch.xive_esc_on = true;
+		mb();
+		__raw_readq(esc_vaddr + XIVE_ESB_SET_PQ_00);
+	}
+	mb();
+}
+EXPORT_SYMBOL_GPL(kvmppc_xive_rearm_escalation);
+
 /*
  * This is a simple trigger for a generic XIVE IRQ. This must
  * only be called for interrupts that support a trigger page
@@ -2106,6 +2140,36 @@ static int kvmppc_xive_create(struct kvm_device *dev, u32 type)
 	return 0;
 }
 
+int kvmppc_xive_xics_hcall(struct kvm_vcpu *vcpu, u32 req)
+{
+	struct kvmppc_vcore *vc = vcpu->arch.vcore;
+
+	/* The VM should have configured XICS mode before doing XICS hcalls. */
+	if (!kvmppc_xics_enabled(vcpu))
+		return H_TOO_HARD;
+
+	switch (req) {
+	case H_XIRR:
+		return xive_vm_h_xirr(vcpu);
+	case H_CPPR:
+		return xive_vm_h_cppr(vcpu, kvmppc_get_gpr(vcpu, 4));
+	case H_EOI:
+		return xive_vm_h_eoi(vcpu, kvmppc_get_gpr(vcpu, 4));
+	case H_IPI:
+		return xive_vm_h_ipi(vcpu, kvmppc_get_gpr(vcpu, 4),
+					  kvmppc_get_gpr(vcpu, 5));
+	case H_IPOLL:
+		return xive_vm_h_ipoll(vcpu, kvmppc_get_gpr(vcpu, 4));
+	case H_XIRR_X:
+		xive_vm_h_xirr(vcpu);
+		kvmppc_set_gpr(vcpu, 5, get_tb() + vc->tb_offset);
+		return H_SUCCESS;
+	}
+
+	return H_UNSUPPORTED;
+}
+EXPORT_SYMBOL_GPL(kvmppc_xive_xics_hcall);
+
 int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
-- 
2.23.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread
* [PATCH 3/4] KVM: PPC: Book3S HV P9: Stop handling hcalls in real-mode in the P9 path
@ 2021-05-13 12:22   ` Nicholas Piggin
  0 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc
  Cc: Alexey Kardashevskiy, linuxppc-dev, Cédric Le Goater,
	Nicholas Piggin

In the interest of minimising the amount of code that is run in
"real-mode", don't handle hcalls in real mode in the P9 path. This
requires some new handlers for H_CEDE and xics-on-xive to be added
before xive is pulled or cede logic is checked.

This introduces a change in radix guest behaviour where radix guests
that execute 'sc 1' in userspace now get a privilege fault whereas
previously the 'sc 1' would be reflected as a syscall interrupt to the
guest kernel. That reflection is only required for hash guests that run
PR KVM.

Background:

In POWER8 and earlier processors, it is very expensive to exit from the
HV real mode context of a guest hypervisor interrupt, and switch to host
virtual mode. On those processors, guest->HV interrupts reach the
hypervisor with the MMU off because the MMU is loaded with guest context
(LPCR, SDR1, SLB), and the other threads in the sub-core need to be
pulled out of the guest too. Then the primary must save off guest state,
invalidate SLB and ERAT, and load up host state before the MMU can be
enabled to run in host virtual mode (~= regular Linux mode).

Hash guests also require a lot of hcalls to run due to the nature of the
MMU architecture and paravirtualisation design. The XICS interrupt
controller requires hcalls to run.

So KVM traditionally tries hard to avoid the full exit, by handling
hcalls and other interrupts in real mode as much as possible.

By contrast, POWER9 has independent MMU context per-thread, and in radix
mode the hypervisor is in host virtual memory mode when the HV interrupt
is taken. Radix guests do not require significant hcalls to manage their
translations, and xive guests don't need hcalls to handle interrupts.

So it's much less important from a performance standpoint to handle
hcalls in real mode in P9.

The TCE hcalls are performance critical, introduced for P8 in order to
achieve 10GbE performance. They was not required on P9 (which was able
to drive 40GbE networking with just the virt mode hcalls), but
performance is quite important. After later changes that add hash guest
support to the P9 path, hash hcalls are also performance critical.  The
full entry/exit performance can still be improved significanty, but if
this is found to be inadequate then real-mode hcall handlers could be
reintroduced for the P9 path (in C), but this would be a last resort.

Reviewed-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/include/asm/kvm_ppc.h      |  6 ++
 arch/powerpc/kvm/book3s.c               |  6 ++
 arch/powerpc/kvm/book3s_hv.c            | 79 +++++++++++++++++++++----
 arch/powerpc/kvm/book3s_hv_rmhandlers.S |  5 ++
 arch/powerpc/kvm/book3s_xive.c          | 64 ++++++++++++++++++++
 5 files changed, 149 insertions(+), 11 deletions(-)

diff --git a/arch/powerpc/include/asm/kvm_ppc.h b/arch/powerpc/include/asm/kvm_ppc.h
index 8c10c3427166..cb9e3c85c605 100644
--- a/arch/powerpc/include/asm/kvm_ppc.h
+++ b/arch/powerpc/include/asm/kvm_ppc.h
@@ -129,6 +129,7 @@ extern void kvmppc_core_vcpu_put(struct kvm_vcpu *vcpu);
 extern int kvmppc_core_prepare_to_enter(struct kvm_vcpu *vcpu);
 extern int kvmppc_core_pending_dec(struct kvm_vcpu *vcpu);
 extern void kvmppc_core_queue_machine_check(struct kvm_vcpu *vcpu, ulong flags);
+extern void kvmppc_core_queue_syscall(struct kvm_vcpu *vcpu);
 extern void kvmppc_core_queue_program(struct kvm_vcpu *vcpu, ulong flags);
 extern void kvmppc_core_queue_fpunavail(struct kvm_vcpu *vcpu);
 extern void kvmppc_core_queue_vec_unavail(struct kvm_vcpu *vcpu);
@@ -606,6 +607,7 @@ extern void kvmppc_free_pimap(struct kvm *kvm);
 extern int kvmppc_xics_rm_complete(struct kvm_vcpu *vcpu, u32 hcall);
 extern void kvmppc_xics_free_icp(struct kvm_vcpu *vcpu);
 extern int kvmppc_xics_hcall(struct kvm_vcpu *vcpu, u32 cmd);
+extern int kvmppc_xive_xics_hcall(struct kvm_vcpu *vcpu, u32 req);
 extern u64 kvmppc_xics_get_icp(struct kvm_vcpu *vcpu);
 extern int kvmppc_xics_set_icp(struct kvm_vcpu *vcpu, u64 icpval);
 extern int kvmppc_xics_connect_vcpu(struct kvm_device *dev,
@@ -638,6 +640,8 @@ static inline int kvmppc_xics_enabled(struct kvm_vcpu *vcpu)
 static inline void kvmppc_xics_free_icp(struct kvm_vcpu *vcpu) { }
 static inline int kvmppc_xics_hcall(struct kvm_vcpu *vcpu, u32 cmd)
 	{ return 0; }
+static inline int kvmppc_xive_xics_hcall(struct kvm_vcpu *vcpu, u32 req)
+	{ return 0; }
 #endif
 
 #ifdef CONFIG_KVM_XIVE
@@ -672,6 +676,7 @@ extern int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 irq,
 			       int level, bool line_status);
 extern void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu);
 extern void kvmppc_xive_pull_vcpu(struct kvm_vcpu *vcpu);
+extern void kvmppc_xive_rearm_escalation(struct kvm_vcpu *vcpu);
 
 static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
 {
@@ -713,6 +718,7 @@ static inline int kvmppc_xive_set_irq(struct kvm *kvm, int irq_source_id, u32 ir
 				      int level, bool line_status) { return -ENODEV; }
 static inline void kvmppc_xive_push_vcpu(struct kvm_vcpu *vcpu) { }
 static inline void kvmppc_xive_pull_vcpu(struct kvm_vcpu *vcpu) { }
+static inline void kvmppc_xive_rearm_escalation(struct kvm_vcpu *vcpu) { }
 
 static inline int kvmppc_xive_enabled(struct kvm_vcpu *vcpu)
 	{ return 0; }
diff --git a/arch/powerpc/kvm/book3s.c b/arch/powerpc/kvm/book3s.c
index 2b691f4d1f26..d69560d5bf16 100644
--- a/arch/powerpc/kvm/book3s.c
+++ b/arch/powerpc/kvm/book3s.c
@@ -171,6 +171,12 @@ void kvmppc_core_queue_machine_check(struct kvm_vcpu *vcpu, ulong flags)
 }
 EXPORT_SYMBOL_GPL(kvmppc_core_queue_machine_check);
 
+void kvmppc_core_queue_syscall(struct kvm_vcpu *vcpu)
+{
+	kvmppc_inject_interrupt(vcpu, BOOK3S_INTERRUPT_SYSCALL, 0);
+}
+EXPORT_SYMBOL(kvmppc_core_queue_syscall);
+
 void kvmppc_core_queue_program(struct kvm_vcpu *vcpu, ulong flags)
 {
 	/* might as well deliver this straight away */
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 287cd3e1b918..5815c04a48a3 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -899,6 +899,10 @@ static int kvm_arch_vcpu_yield_to(struct kvm_vcpu *target)
 	 * H_SUCCESS if the source vcore wasn't idle (e.g. if it may
 	 * have useful work to do and should not confer) so we don't
 	 * recheck that here.
+	 *
+	 * In the case of the P9 single vcpu per vcore case, the real
+	 * mode handler is not called but no other threads are in the
+	 * source vcore.
 	 */
 
 	spin_lock(&vcore->lock);
@@ -1142,12 +1146,13 @@ int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu)
 }
 
 /*
- * Handle H_CEDE in the nested virtualization case where we haven't
- * called the real-mode hcall handlers in book3s_hv_rmhandlers.S.
+ * Handle H_CEDE in the P9 path where we don't call the real-mode hcall
+ * handlers in book3s_hv_rmhandlers.S.
+ *
  * This has to be done early, not in kvmppc_pseries_do_hcall(), so
  * that the cede logic in kvmppc_run_single_vcpu() works properly.
  */
-static void kvmppc_nested_cede(struct kvm_vcpu *vcpu)
+static void kvmppc_cede(struct kvm_vcpu *vcpu)
 {
 	vcpu->arch.shregs.msr |= MSR_EE;
 	vcpu->arch.ceded = 1;
@@ -1400,13 +1405,29 @@ static int kvmppc_handle_exit_hv(struct kvm_vcpu *vcpu,
 	}
 	case BOOK3S_INTERRUPT_SYSCALL:
 	{
-		/* hcall - punt to userspace */
 		int i;
 
-		/* hypercall with MSR_PR has already been handled in rmode,
-		 * and never reaches here.
-		 */
+		if (unlikely(vcpu->arch.shregs.msr & MSR_PR)) {
+			/*
+			 * Guest userspace executed sc 1. This can only be
+			 * reached by the P9 path because the old path
+			 * handles this case in realmode hcall handlers.
+			 *
+			 * Radix guests can not run PR KVM or nested HV hash
+			 * guests which might run PR KVM, so this is always
+			 * a privilege fault. Send a program check to guest
+			 * kernel.
+			 */
+			kvmppc_core_queue_program(vcpu, SRR1_PROGPRIV);
+			r = RESUME_GUEST;
+			break;
+		}
 
+		/*
+		 * hcall - gather args and set exit_reason. This will next be
+		 * handled by kvmppc_pseries_do_hcall which may be able to deal
+		 * with it and resume guest, or may punt to userspace.
+		 */
 		run->papr_hcall.nr = kvmppc_get_gpr(vcpu, 3);
 		for (i = 0; i < 9; ++i)
 			run->papr_hcall.args[i] = kvmppc_get_gpr(vcpu, 4 + i);
@@ -3664,6 +3685,12 @@ static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
 	return trap;
 }
 
+static inline bool hcall_is_xics(unsigned long req)
+{
+	return req = H_EOI || req = H_CPPR || req = H_IPI ||
+		req = H_IPOLL || req = H_XIRR || req = H_XIRR_X;
+}
+
 /*
  * Virtual-mode guest entry for POWER9 and later when the host and
  * guest are both using the radix MMU.  The LPIDR has already been set.
@@ -3787,15 +3814,36 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 		/* H_CEDE has to be handled now, not later */
 		if (trap = BOOK3S_INTERRUPT_SYSCALL && !vcpu->arch.nested &&
 		    kvmppc_get_gpr(vcpu, 3) = H_CEDE) {
-			kvmppc_nested_cede(vcpu);
+			kvmppc_cede(vcpu);
 			kvmppc_set_gpr(vcpu, 3, 0);
 			trap = 0;
 		}
 	} else {
 		kvmppc_xive_push_vcpu(vcpu);
 		trap = kvmhv_load_hv_regs_and_go(vcpu, time_limit, lpcr);
+		if (trap = BOOK3S_INTERRUPT_SYSCALL && !vcpu->arch.nested &&
+		    !(vcpu->arch.shregs.msr & MSR_PR)) {
+			unsigned long req = kvmppc_get_gpr(vcpu, 3);
+
+			/* H_CEDE has to be handled now, not later */
+			if (req = H_CEDE) {
+				kvmppc_cede(vcpu);
+				kvmppc_xive_rearm_escalation(vcpu); /* may un-cede */
+				kvmppc_set_gpr(vcpu, 3, 0);
+				trap = 0;
+
+			/* XICS hcalls must be handled before xive is pulled */
+			} else if (hcall_is_xics(req)) {
+				int ret;
+
+				ret = kvmppc_xive_xics_hcall(vcpu, req);
+				if (ret != H_TOO_HARD) {
+					kvmppc_set_gpr(vcpu, 3, ret);
+					trap = 0;
+				}
+			}
+		}
 		kvmppc_xive_pull_vcpu(vcpu);
-
 	}
 
 	vcpu->arch.slb_max = 0;
@@ -4461,8 +4509,17 @@ static int kvmppc_vcpu_run_hv(struct kvm_vcpu *vcpu)
 		else
 			r = kvmppc_run_vcpu(vcpu);
 
-		if (run->exit_reason = KVM_EXIT_PAPR_HCALL &&
-		    !(vcpu->arch.shregs.msr & MSR_PR)) {
+		if (run->exit_reason = KVM_EXIT_PAPR_HCALL) {
+			if (WARN_ON_ONCE(vcpu->arch.shregs.msr & MSR_PR)) {
+				/*
+				 * These should have been caught reflected
+				 * into the guest by now. Final sanity check:
+				 * don't allow userspace to execute hcalls in
+				 * the hypervisor.
+				 */
+				r = RESUME_GUEST;
+				continue;
+			}
 			trace_kvm_hcall_enter(vcpu);
 			r = kvmppc_pseries_do_hcall(vcpu);
 			trace_kvm_hcall_exit(vcpu, r);
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
index c11597f815e4..2d0d14ed1d92 100644
--- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
@@ -1397,9 +1397,14 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
 	mr	r4,r9
 	bge	fast_guest_return
 2:
+	/* If we came in through the P9 short path, no real mode hcalls */
+	lwz	r0, STACK_SLOT_SHORT_PATH(r1)
+	cmpwi	r0, 0
+	bne	no_try_real
 	/* See if this is an hcall we can handle in real mode */
 	cmpwi	r12,BOOK3S_INTERRUPT_SYSCALL
 	beq	hcall_try_real_mode
+no_try_real:
 
 	/* Hypervisor doorbell - exit only if host IPI flag set */
 	cmpwi	r12, BOOK3S_INTERRUPT_H_DOORBELL
diff --git a/arch/powerpc/kvm/book3s_xive.c b/arch/powerpc/kvm/book3s_xive.c
index 741bf1f4387a..24c07094651a 100644
--- a/arch/powerpc/kvm/book3s_xive.c
+++ b/arch/powerpc/kvm/book3s_xive.c
@@ -158,6 +158,40 @@ void kvmppc_xive_pull_vcpu(struct kvm_vcpu *vcpu)
 }
 EXPORT_SYMBOL_GPL(kvmppc_xive_pull_vcpu);
 
+void kvmppc_xive_rearm_escalation(struct kvm_vcpu *vcpu)
+{
+	void __iomem *esc_vaddr = (void __iomem *)vcpu->arch.xive_esc_vaddr;
+
+	if (!esc_vaddr)
+		return;
+
+	/* we are using XIVE with single escalation */
+
+	if (vcpu->arch.xive_esc_on) {
+		/*
+		 * If we still have a pending escalation, abort the cede,
+		 * and we must set PQ to 10 rather than 00 so that we don't
+		 * potentially end up with two entries for the escalation
+		 * interrupt in the XIVE interrupt queue.  In that case
+		 * we also don't want to set xive_esc_on to 1 here in
+		 * case we race with xive_esc_irq().
+		 */
+		vcpu->arch.ceded = 0;
+		/*
+		 * The escalation interrupts are special as we don't EOI them.
+		 * There is no need to use the load-after-store ordering offset
+		 * to set PQ to 10 as we won't use StoreEOI.
+		 */
+		__raw_readq(esc_vaddr + XIVE_ESB_SET_PQ_10);
+	} else {
+		vcpu->arch.xive_esc_on = true;
+		mb();
+		__raw_readq(esc_vaddr + XIVE_ESB_SET_PQ_00);
+	}
+	mb();
+}
+EXPORT_SYMBOL_GPL(kvmppc_xive_rearm_escalation);
+
 /*
  * This is a simple trigger for a generic XIVE IRQ. This must
  * only be called for interrupts that support a trigger page
@@ -2106,6 +2140,36 @@ static int kvmppc_xive_create(struct kvm_device *dev, u32 type)
 	return 0;
 }
 
+int kvmppc_xive_xics_hcall(struct kvm_vcpu *vcpu, u32 req)
+{
+	struct kvmppc_vcore *vc = vcpu->arch.vcore;
+
+	/* The VM should have configured XICS mode before doing XICS hcalls. */
+	if (!kvmppc_xics_enabled(vcpu))
+		return H_TOO_HARD;
+
+	switch (req) {
+	case H_XIRR:
+		return xive_vm_h_xirr(vcpu);
+	case H_CPPR:
+		return xive_vm_h_cppr(vcpu, kvmppc_get_gpr(vcpu, 4));
+	case H_EOI:
+		return xive_vm_h_eoi(vcpu, kvmppc_get_gpr(vcpu, 4));
+	case H_IPI:
+		return xive_vm_h_ipi(vcpu, kvmppc_get_gpr(vcpu, 4),
+					  kvmppc_get_gpr(vcpu, 5));
+	case H_IPOLL:
+		return xive_vm_h_ipoll(vcpu, kvmppc_get_gpr(vcpu, 4));
+	case H_XIRR_X:
+		xive_vm_h_xirr(vcpu);
+		kvmppc_set_gpr(vcpu, 5, get_tb() + vc->tb_offset);
+		return H_SUCCESS;
+	}
+
+	return H_UNSUPPORTED;
+}
+EXPORT_SYMBOL_GPL(kvmppc_xive_xics_hcall);
+
 int kvmppc_xive_debug_show_queues(struct seq_file *m, struct kvm_vcpu *vcpu)
 {
 	struct kvmppc_xive_vcpu *xc = vcpu->arch.xive_vcpu;
-- 
2.23.0

^ permalink raw reply related	[flat|nested] 10+ messages in thread
* [PATCH 4/4] KVM: PPC: Book3S HV P9: Implement the rest of the P9 path in C
  2021-05-13 12:22 ` Nicholas Piggin
@ 2021-05-13 12:22   ` Nicholas Piggin
  -1 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc; +Cc: linuxppc-dev, Nicholas Piggin

Almost all logic is moved to C, by introducing a new in_guest mode for
the P9 path that branches very early in the KVM interrupt handler to P9
exit code.

The main P9 entry and exit assembly is now only about 160 lines of low
level stack setup and register save/restore, plus a bad-interrupt
handler.

There are two motivations for this, the first is just make the code more
maintainable being in C. The second is to reduce the amount of code
running in a special KVM mode, "realmode". In quotes because with radix
it is no longer necessarily real-mode in the MMU, but it still has to be
treated specially because it may be in real-mode, and has various
important registers like PID, DEC, TB, etc set to guest. This is hostile
to the rest of Linux and can't use arbitrary kernel functionality or be
instrumented well.

This initial patch is a reasonably faithful conversion of the asm code,
but it does lack any loop to return quickly back into the guest without
switching out of realmode in the case of unimportant or easily handled
interrupts. As explained in previous changes, handling HV interrupts
very quickly in this low level realmode is not so important for P9
performance, and are important to avoid for security, observability,
debugability reasons.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/include/asm/asm-prototypes.h |   3 +-
 arch/powerpc/include/asm/kvm_asm.h        |   1 +
 arch/powerpc/include/asm/kvm_book3s_64.h  |   8 +
 arch/powerpc/include/asm/kvm_host.h       |   7 +-
 arch/powerpc/kernel/security.c            |   5 +-
 arch/powerpc/kvm/Makefile                 |   1 +
 arch/powerpc/kvm/book3s_64_entry.S        | 254 ++++++++++++++++++++++
 arch/powerpc/kvm/book3s_hv.c              |   9 +-
 arch/powerpc/kvm/book3s_hv_p9_entry.c     | 207 ++++++++++++++++++
 arch/powerpc/kvm/book3s_hv_rmhandlers.S   | 125 +----------
 10 files changed, 496 insertions(+), 124 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_hv_p9_entry.c

diff --git a/arch/powerpc/include/asm/asm-prototypes.h b/arch/powerpc/include/asm/asm-prototypes.h
index 1c7b75834e04..02ee6f5ac9fe 100644
--- a/arch/powerpc/include/asm/asm-prototypes.h
+++ b/arch/powerpc/include/asm/asm-prototypes.h
@@ -120,6 +120,7 @@ extern s32 patch__call_flush_branch_caches3;
 extern s32 patch__flush_count_cache_return;
 extern s32 patch__flush_link_stack_return;
 extern s32 patch__call_kvm_flush_link_stack;
+extern s32 patch__call_kvm_flush_link_stack_p9;
 extern s32 patch__memset_nocache, patch__memcpy_nocache;
 
 extern long flush_branch_caches;
@@ -140,7 +141,7 @@ void kvmhv_load_host_pmu(void);
 void kvmhv_save_guest_pmu(struct kvm_vcpu *vcpu, bool pmu_in_use);
 void kvmhv_load_guest_pmu(struct kvm_vcpu *vcpu);
 
-int __kvmhv_vcpu_entry_p9(struct kvm_vcpu *vcpu);
+void kvmppc_p9_enter_guest(struct kvm_vcpu *vcpu);
 
 long kvmppc_h_set_dabr(struct kvm_vcpu *vcpu, unsigned long dabr);
 long kvmppc_h_set_xdabr(struct kvm_vcpu *vcpu, unsigned long dabr,
diff --git a/arch/powerpc/include/asm/kvm_asm.h b/arch/powerpc/include/asm/kvm_asm.h
index a3633560493b..43b1788e1f93 100644
--- a/arch/powerpc/include/asm/kvm_asm.h
+++ b/arch/powerpc/include/asm/kvm_asm.h
@@ -147,6 +147,7 @@
 #define KVM_GUEST_MODE_SKIP	2
 #define KVM_GUEST_MODE_GUEST_HV	3
 #define KVM_GUEST_MODE_HOST_HV	4
+#define KVM_GUEST_MODE_HV_FAST	5 /* ISA >= v3.0 host+guest radix, indep thr */
 
 #define KVM_INST_FETCH_FAILED	-1
 
diff --git a/arch/powerpc/include/asm/kvm_book3s_64.h b/arch/powerpc/include/asm/kvm_book3s_64.h
index 9bb9bb370b53..c214bcffb441 100644
--- a/arch/powerpc/include/asm/kvm_book3s_64.h
+++ b/arch/powerpc/include/asm/kvm_book3s_64.h
@@ -153,9 +153,17 @@ static inline bool kvmhv_vcpu_is_radix(struct kvm_vcpu *vcpu)
 	return radix;
 }
 
+int __kvmhv_vcpu_entry_p9(struct kvm_vcpu *vcpu);
+
 #define KVM_DEFAULT_HPT_ORDER	24	/* 16MB HPT by default */
 #endif
 
+/*
+ * Invalid HDSISR value which is used to indicate when HW has not set the reg.
+ * Used to work around an errata.
+ */
+#define HDSISR_CANARY	0x7fff
+
 /*
  * We use a lock bit in HPTE dword 0 to synchronize updates and
  * accesses to each HPTE, and another bit to indicate non-present
diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
index 1e83359f286b..69add9d662df 100644
--- a/arch/powerpc/include/asm/kvm_host.h
+++ b/arch/powerpc/include/asm/kvm_host.h
@@ -683,7 +683,12 @@ struct kvm_vcpu_arch {
 	ulong fault_dar;
 	u32 fault_dsisr;
 	unsigned long intr_msr;
-	ulong fault_gpa;	/* guest real address of page fault (POWER9) */
+	/*
+	 * POWER9 and later: fault_gpa contains the guest real address of page
+	 * fault for a radix guest, or segment descriptor (equivalent to result
+	 * from slbmfev of SLB entry that translated the EA) for hash guests.
+	 */
+	ulong fault_gpa;
 #endif
 
 #ifdef CONFIG_BOOKE
diff --git a/arch/powerpc/kernel/security.c b/arch/powerpc/kernel/security.c
index 0fdfcdd9d880..c17d1c9362b5 100644
--- a/arch/powerpc/kernel/security.c
+++ b/arch/powerpc/kernel/security.c
@@ -432,16 +432,19 @@ device_initcall(stf_barrier_debugfs_init);
 
 static void update_branch_cache_flush(void)
 {
-	u32 *site;
+	u32 *site, __maybe_unused *site2;
 
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 	site = &patch__call_kvm_flush_link_stack;
+	site2 = &patch__call_kvm_flush_link_stack_p9;
 	// This controls the branch from guest_exit_cont to kvm_flush_link_stack
 	if (link_stack_flush_type == BRANCH_CACHE_FLUSH_NONE) {
 		patch_instruction_site(site, ppc_inst(PPC_INST_NOP));
+		patch_instruction_site(site2, ppc_inst(PPC_INST_NOP));
 	} else {
 		// Could use HW flush, but that could also flush count cache
 		patch_branch_site(site, (u64)&kvm_flush_link_stack, BRANCH_SET_LINK);
+		patch_branch_site(site2, (u64)&kvm_flush_link_stack, BRANCH_SET_LINK);
 	}
 #endif
 
diff --git a/arch/powerpc/kvm/Makefile b/arch/powerpc/kvm/Makefile
index cdd119028f64..91eb67bb91e1 100644
--- a/arch/powerpc/kvm/Makefile
+++ b/arch/powerpc/kvm/Makefile
@@ -89,6 +89,7 @@ kvm-book3s_64-builtin-tm-objs-$(CONFIG_PPC_TRANSACTIONAL_MEM) += \
 ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 kvm-book3s_64-builtin-objs-$(CONFIG_KVM_BOOK3S_64_HANDLER) += \
 	book3s_hv_hmi.o \
+	book3s_hv_p9_entry.o \
 	book3s_hv_rmhandlers.o \
 	book3s_hv_rm_mmu.o \
 	book3s_hv_ras.o \
diff --git a/arch/powerpc/kvm/book3s_64_entry.S b/arch/powerpc/kvm/book3s_64_entry.S
index a01046202eef..177e8fad5c8d 100644
--- a/arch/powerpc/kvm/book3s_64_entry.S
+++ b/arch/powerpc/kvm/book3s_64_entry.S
@@ -1,11 +1,16 @@
 /* SPDX-License-Identifier: GPL-2.0-only */
 #include <asm/asm-offsets.h>
 #include <asm/cache.h>
+#include <asm/code-patching-asm.h>
 #include <asm/exception-64s.h>
+#include <asm/export.h>
 #include <asm/kvm_asm.h>
 #include <asm/kvm_book3s_asm.h>
+#include <asm/mmu.h>
 #include <asm/ppc_asm.h>
+#include <asm/ptrace.h>
 #include <asm/reg.h>
+#include <asm/ultravisor-api.h>
 
 /*
  * These are branched to from interrupt handlers in exception-64s.S which set
@@ -29,6 +34,11 @@
 .global	kvmppc_hcall
 .balign IFETCH_ALIGN_BYTES
 kvmppc_hcall:
+#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+	lbz	r10,HSTATE_IN_GUEST(r13)
+	cmpwi	r10,KVM_GUEST_MODE_HV_FAST
+	beq	kvmppc_p9_exit_hcall
+#endif
 	ld	r10,PACA_EXGEN+EX_R13(r13)
 	SET_SCRATCH0(r10)
 	li	r10,0xc00
@@ -55,6 +65,13 @@ kvmppc_hcall:
 .global	kvmppc_interrupt
 .balign IFETCH_ALIGN_BYTES
 kvmppc_interrupt:
+#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+	std	r10,HSTATE_SCRATCH0(r13)
+	lbz	r10,HSTATE_IN_GUEST(r13)
+	cmpwi	r10,KVM_GUEST_MODE_HV_FAST
+	beq	kvmppc_p9_exit_interrupt
+	ld	r10,HSTATE_SCRATCH0(r13)
+#endif
 	li	r11,PACA_EXGEN
 	cmpdi	r10,0x200
 	bgt+	.Lgot_save_area
@@ -156,3 +173,240 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
 	GET_SCRATCH0(r13)
 	HRFI_TO_KERNEL
 #endif
+
+#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+
+/* Stack frame offsets for kvmppc_p9_enter_guest */
+#define SFS			(144 + STACK_FRAME_MIN_SIZE)
+#define STACK_SLOT_NVGPRS	(SFS - 144)	/* 18 gprs */
+
+/*
+ * void kvmppc_p9_enter_guest(struct vcpu *vcpu);
+ *
+ * Enter the guest on a ISAv3.0 or later system where we have exactly
+ * one vcpu per vcore, and both the host and guest are radix, and threads
+ * are set to "indepdent mode".
+ */
+.balign	IFETCH_ALIGN_BYTES
+_GLOBAL(kvmppc_p9_enter_guest)
+EXPORT_SYMBOL_GPL(kvmppc_p9_enter_guest)
+	mflr	r0
+	std	r0,PPC_LR_STKOFF(r1)
+	stdu	r1,-SFS(r1)
+
+	std	r1,HSTATE_HOST_R1(r13)
+
+	mfcr	r4
+	stw	r4,SFS+8(r1)
+
+	reg = 14
+	.rept	18
+	std	reg,STACK_SLOT_NVGPRS + ((reg - 14) * 8)(r1)
+	reg = reg + 1
+	.endr
+
+	ld	r4,VCPU_LR(r3)
+	mtlr	r4
+	ld	r4,VCPU_CTR(r3)
+	mtctr	r4
+	ld	r4,VCPU_XER(r3)
+	mtspr	SPRN_XER,r4
+
+	ld	r1,VCPU_CR(r3)
+
+BEGIN_FTR_SECTION
+	ld	r4,VCPU_CFAR(r3)
+	mtspr	SPRN_CFAR,r4
+END_FTR_SECTION_IFSET(CPU_FTR_CFAR)
+BEGIN_FTR_SECTION
+	ld	r4,VCPU_PPR(r3)
+	mtspr	SPRN_PPR,r4
+END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
+
+	reg = 4
+	.rept	28
+	ld	reg,__VCPU_GPR(reg)(r3)
+	reg = reg + 1
+	.endr
+
+	ld	r4,VCPU_KVM(r3)
+	lbz	r4,KVM_SECURE_GUEST(r4)
+	cmpdi	r4,0
+	ld	r4,VCPU_GPR(R4)(r3)
+	bne	.Lret_to_ultra
+
+	mtcr	r1
+
+	ld	r0,VCPU_GPR(R0)(r3)
+	ld	r1,VCPU_GPR(R1)(r3)
+	ld	r2,VCPU_GPR(R2)(r3)
+	ld	r3,VCPU_GPR(R3)(r3)
+
+	HRFI_TO_GUEST
+	b	.
+
+	/*
+	 * Use UV_RETURN ultracall to return control back to the Ultravisor
+	 * after processing an hypercall or interrupt that was forwarded
+	 * (a.k.a. reflected) to the Hypervisor.
+	 *
+	 * All registers have already been reloaded except the ucall requires:
+	 *   R0 = hcall result
+	 *   R2 = SRR1, so UV can detect a synthesized interrupt (if any)
+	 *   R3 = UV_RETURN
+	 */
+.Lret_to_ultra:
+	mtcr	r1
+	ld	r1,VCPU_GPR(R1)(r3)
+
+	ld	r0,VCPU_GPR(R3)(r3)
+	mfspr	r2,SPRN_SRR1
+	LOAD_REG_IMMEDIATE(r3, UV_RETURN)
+	sc	2
+
+/*
+ * kvmppc_p9_exit_hcall and kvmppc_p9_exit_interrupt are branched to from
+ * above if the interrupt was taken for a guest that was entered via
+ * kvmppc_p9_enter_guest().
+ *
+ * The exit code recovers the host stack and vcpu pointer, saves all guest GPRs
+ * and CR, LR, XER as well as guest MSR and NIA into the VCPU, then re-
+ * establishes the host stack and registers to return from the
+ * kvmppc_p9_enter_guest() function, which saves CTR and other guest registers
+ * (SPRs and FP, VEC, etc).
+ */
+.balign	IFETCH_ALIGN_BYTES
+kvmppc_p9_exit_hcall:
+	mfspr	r11,SPRN_SRR0
+	mfspr	r12,SPRN_SRR1
+	li	r10,0xc00
+	std	r10,HSTATE_SCRATCH0(r13)
+
+.balign	IFETCH_ALIGN_BYTES
+kvmppc_p9_exit_interrupt:
+	/*
+	 * If set to KVM_GUEST_MODE_HV_FAST but we're still in the
+	 * hypervisor, that means we can't return from the entry stack.
+	 */
+	rldicl. r10,r12,64-MSR_HV_LG,63
+	bne-	kvmppc_p9_bad_interrupt
+
+	std     r1,HSTATE_SCRATCH1(r13)
+	std     r3,HSTATE_SCRATCH2(r13)
+	ld	r1,HSTATE_HOST_R1(r13)
+	ld	r3,HSTATE_KVM_VCPU(r13)
+
+	std	r9,VCPU_CR(r3)
+
+1:
+	std	r11,VCPU_PC(r3)
+	std	r12,VCPU_MSR(r3)
+
+	reg = 14
+	.rept	18
+	std	reg,__VCPU_GPR(reg)(r3)
+	reg = reg + 1
+	.endr
+
+	/* r1, r3, r9-r13 are saved to vcpu by C code */
+	std	r0,VCPU_GPR(R0)(r3)
+	std	r2,VCPU_GPR(R2)(r3)
+	reg = 4
+	.rept	5
+	std	reg,__VCPU_GPR(reg)(r3)
+	reg = reg + 1
+	.endr
+
+	ld	r2,PACATOC(r13)
+
+	mflr	r4
+	std	r4,VCPU_LR(r3)
+	mfspr	r4,SPRN_XER
+	std	r4,VCPU_XER(r3)
+
+	reg = 14
+	.rept	18
+	ld	reg,STACK_SLOT_NVGPRS + ((reg - 14) * 8)(r1)
+	reg = reg + 1
+	.endr
+
+	lwz	r4,SFS+8(r1)
+	mtcr	r4
+
+	/*
+	 * Flush the link stack here, before executing the first blr on the
+	 * way out of the guest.
+	 *
+	 * The link stack won't match coming out of the guest anyway so the
+	 * only cost is the flush itself. The call clobbers r0.
+	 */
+1:	nop
+	patch_site 1b patch__call_kvm_flush_link_stack_p9
+
+	addi	r1,r1,SFS
+	ld	r0,PPC_LR_STKOFF(r1)
+	mtlr	r0
+	blr
+
+/*
+ * Took an interrupt somewhere right before HRFID to guest, so registers are
+ * in a bad way. Return things hopefully enough to run host virtual code and
+ * run the Linux interrupt handler (SRESET or MCE) to print something useful.
+ *
+ * We could be really clever and save all host registers in known locations
+ * before setting HSTATE_IN_GUEST, then restoring them all here, and setting
+ * return address to a fixup that sets them up again. But that's a lot of
+ * effort for a small bit of code. Lots of other things to do first.
+ */
+kvmppc_p9_bad_interrupt:
+	/*
+	 * Clean up guest registers to give host a chance to run.
+	 */
+	li	r10,0
+	mtspr	SPRN_AMR,r10
+	mtspr	SPRN_IAMR,r10
+	mtspr	SPRN_CIABR,r10
+	mtspr	SPRN_DAWRX0,r10
+BEGIN_FTR_SECTION
+	mtspr	SPRN_DAWRX1,r10
+END_FTR_SECTION_IFSET(CPU_FTR_DAWR1)
+	mtspr	SPRN_PID,r10
+
+	/*
+	 * Switch to host MMU mode
+	 */
+	ld	r10, HSTATE_KVM_VCPU(r13)
+	ld	r10, VCPU_KVM(r10)
+	lwz	r10, KVM_HOST_LPID(r10)
+	mtspr	SPRN_LPID,r10
+
+	ld	r10, HSTATE_KVM_VCPU(r13)
+	ld	r10, VCPU_KVM(r10)
+	ld	r10, KVM_HOST_LPCR(r10)
+	mtspr	SPRN_LPCR,r10
+
+	/*
+	 * Set GUEST_MODE_NONE so the handler won't branch to KVM, and clear
+	 * MSR_RI in r12 ([H]SRR1) so the handler won't try to return.
+	 */
+	li	r10,KVM_GUEST_MODE_NONE
+	stb	r10,HSTATE_IN_GUEST(r13)
+	li	r10,MSR_RI
+	andc	r12,r12,r10
+
+	/*
+	 * Go back to interrupt handler. MCE and SRESET have their specific
+	 * PACA save area so they should be used directly. They set up their
+	 * own stack. The other handlers all use EXGEN. They will use the
+	 * guest r1 if it looks like a kernel stack, so just load the
+	 * emergency stack and go to program check for all other interrupts.
+	 */
+	ld	r10,HSTATE_SCRATCH0(r13)
+	cmpwi	r10,BOOK3S_INTERRUPT_MACHINE_CHECK
+	beq	machine_check_common
+
+	cmpwi	r10,BOOK3S_INTERRUPT_SYSTEM_RESET
+	beq	system_reset_common
+
+	b	.
+#endif
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 5815c04a48a3..e338da9d9312 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -1445,6 +1445,8 @@ static int kvmppc_handle_exit_hv(struct kvm_vcpu *vcpu,
 	 */
 	case BOOK3S_INTERRUPT_H_DATA_STORAGE:
 		r = RESUME_PAGE_FAULT;
+		if (vcpu->arch.fault_dsisr == HDSISR_CANARY)
+			r = RESUME_GUEST; /* Just retry if it's the canary */
 		break;
 	case BOOK3S_INTERRUPT_H_INST_STORAGE:
 		vcpu->arch.fault_dar = kvmppc_get_pc(vcpu);
@@ -3708,6 +3710,8 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 	u64 tb;
 	int trap, save_pmu;
 
+	WARN_ON_ONCE(vcpu->arch.ceded);
+
 	dec = mfspr(SPRN_DEC);
 	tb = mftb();
 	if (dec < 0)
@@ -3716,8 +3720,6 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 	if (local_paca->kvm_hstate.dec_expires < time_limit)
 		time_limit = local_paca->kvm_hstate.dec_expires;
 
-	vcpu->arch.ceded = 0;
-
 	kvmhv_save_host_pmu();		/* saves it to PACA kvm_hstate */
 
 	kvmppc_subcore_enter_guest();
@@ -3844,9 +3846,10 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 			}
 		}
 		kvmppc_xive_pull_vcpu(vcpu);
+
+		vcpu->arch.slb_max = 0;
 	}
 
-	vcpu->arch.slb_max = 0;
 	dec = mfspr(SPRN_DEC);
 	if (!(lpcr & LPCR_LD)) /* Sign extend if not using large decrementer */
 		dec = (s32) dec;
diff --git a/arch/powerpc/kvm/book3s_hv_p9_entry.c b/arch/powerpc/kvm/book3s_hv_p9_entry.c
new file mode 100644
index 000000000000..9db0e031a443
--- /dev/null
+++ b/arch/powerpc/kvm/book3s_hv_p9_entry.c
@@ -0,0 +1,207 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include <linux/kernel.h>
+#include <linux/kvm_host.h>
+#include <asm/asm-prototypes.h>
+#include <asm/dbell.h>
+#include <asm/kvm_ppc.h>
+
+#ifdef CONFIG_KVM_BOOK3S_HV_EXIT_TIMING
+static void __start_timing(struct kvm_vcpu *vcpu, struct kvmhv_tb_accumulator *next)
+{
+	struct kvmppc_vcore *vc = vcpu->arch.vcore;
+	u64 tb = mftb() - vc->tb_offset_applied;
+
+	vcpu->arch.cur_activity = next;
+	vcpu->arch.cur_tb_start = tb;
+}
+
+static void __accumulate_time(struct kvm_vcpu *vcpu, struct kvmhv_tb_accumulator *next)
+{
+	struct kvmppc_vcore *vc = vcpu->arch.vcore;
+	struct kvmhv_tb_accumulator *curr;
+	u64 tb = mftb() - vc->tb_offset_applied;
+	u64 prev_tb;
+	u64 delta;
+	u64 seq;
+
+	curr = vcpu->arch.cur_activity;
+	vcpu->arch.cur_activity = next;
+	prev_tb = vcpu->arch.cur_tb_start;
+	vcpu->arch.cur_tb_start = tb;
+
+	if (!curr)
+		return;
+
+	delta = tb - prev_tb;
+
+	seq = curr->seqcount;
+	curr->seqcount = seq + 1;
+	smp_wmb();
+	curr->tb_total += delta;
+	if (seq == 0 || delta < curr->tb_min)
+		curr->tb_min = delta;
+	if (delta > curr->tb_max)
+		curr->tb_max = delta;
+	smp_wmb();
+	curr->seqcount = seq + 2;
+}
+
+#define start_timing(vcpu, next) __start_timing(vcpu, next)
+#define end_timing(vcpu) __start_timing(vcpu, NULL)
+#define accumulate_time(vcpu, next) __accumulate_time(vcpu, next)
+#else
+#define start_timing(vcpu, next) do {} while (0)
+#define end_timing(vcpu) do {} while (0)
+#define accumulate_time(vcpu, next) do {} while (0)
+#endif
+
+static inline void mtslb(u64 slbee, u64 slbev)
+{
+	asm volatile("slbmte %0,%1" :: "r" (slbev), "r" (slbee));
+}
+
+static inline void clear_slb_entry(unsigned int idx)
+{
+	mtslb(idx, 0);
+}
+
+/*
+ * Malicious or buggy radix guests may have inserted SLB entries
+ * (only 0..3 because radix always runs with UPRT=1), so these must
+ * be cleared here to avoid side-channels. slbmte is used rather
+ * than slbia, as it won't clear cached translations.
+ */
+static void radix_clear_slb(void)
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		clear_slb_entry(i);
+}
+
+int __kvmhv_vcpu_entry_p9(struct kvm_vcpu *vcpu)
+{
+	u64 *exsave;
+	unsigned long msr = mfmsr();
+	int trap;
+
+	start_timing(vcpu, &vcpu->arch.rm_entry);
+
+	vcpu->arch.ceded = 0;
+
+	WARN_ON_ONCE(vcpu->arch.shregs.msr & MSR_HV);
+	WARN_ON_ONCE(!(vcpu->arch.shregs.msr & MSR_ME));
+
+	mtspr(SPRN_HSRR0, vcpu->arch.regs.nip);
+	mtspr(SPRN_HSRR1, (vcpu->arch.shregs.msr & ~MSR_HV) | MSR_ME);
+
+	/*
+	 * On POWER9 DD2.1 and below, sometimes on a Hypervisor Data Storage
+	 * Interrupt (HDSI) the HDSISR is not be updated at all.
+	 *
+	 * To work around this we put a canary value into the HDSISR before
+	 * returning to a guest and then check for this canary when we take a
+	 * HDSI. If we find the canary on a HDSI, we know the hardware didn't
+	 * update the HDSISR. In this case we return to the guest to retake the
+	 * HDSI which should correctly update the HDSISR the second time HDSI
+	 * entry.
+	 *
+	 * Just do this on all p9 processors for now.
+	 */
+	mtspr(SPRN_HDSISR, HDSISR_CANARY);
+
+	accumulate_time(vcpu, &vcpu->arch.guest_time);
+
+	local_paca->kvm_hstate.in_guest = KVM_GUEST_MODE_HV_FAST;
+	kvmppc_p9_enter_guest(vcpu);
+	// Radix host and guest means host never runs with guest MMU state
+	local_paca->kvm_hstate.in_guest = KVM_GUEST_MODE_NONE;
+
+	accumulate_time(vcpu, &vcpu->arch.rm_intr);
+
+	/* XXX: Could get these from r11/12 and paca exsave instead */
+	vcpu->arch.shregs.srr0 = mfspr(SPRN_SRR0);
+	vcpu->arch.shregs.srr1 = mfspr(SPRN_SRR1);
+	vcpu->arch.shregs.dar = mfspr(SPRN_DAR);
+	vcpu->arch.shregs.dsisr = mfspr(SPRN_DSISR);
+
+	/* 0x2 bit for HSRR is only used by PR and P7/8 HV paths, clear it */
+	trap = local_paca->kvm_hstate.scratch0 & ~0x2;
+	if (likely(trap > BOOK3S_INTERRUPT_MACHINE_CHECK)) {
+		exsave = local_paca->exgen;
+	} else if (trap == BOOK3S_INTERRUPT_SYSTEM_RESET) {
+		exsave = local_paca->exnmi;
+	} else { /* trap == 0x200 */
+		exsave = local_paca->exmc;
+	}
+
+	vcpu->arch.regs.gpr[1] = local_paca->kvm_hstate.scratch1;
+	vcpu->arch.regs.gpr[3] = local_paca->kvm_hstate.scratch2;
+	vcpu->arch.regs.gpr[9] = exsave[EX_R9/sizeof(u64)];
+	vcpu->arch.regs.gpr[10] = exsave[EX_R10/sizeof(u64)];
+	vcpu->arch.regs.gpr[11] = exsave[EX_R11/sizeof(u64)];
+	vcpu->arch.regs.gpr[12] = exsave[EX_R12/sizeof(u64)];
+	vcpu->arch.regs.gpr[13] = exsave[EX_R13/sizeof(u64)];
+	vcpu->arch.ppr = exsave[EX_PPR/sizeof(u64)];
+	vcpu->arch.cfar = exsave[EX_CFAR/sizeof(u64)];
+	vcpu->arch.regs.ctr = exsave[EX_CTR/sizeof(u64)];
+
+	vcpu->arch.last_inst = KVM_INST_FETCH_FAILED;
+
+	if (unlikely(trap == BOOK3S_INTERRUPT_MACHINE_CHECK)) {
+		vcpu->arch.fault_dar = exsave[EX_DAR/sizeof(u64)];
+		vcpu->arch.fault_dsisr = exsave[EX_DSISR/sizeof(u64)];
+		kvmppc_realmode_machine_check(vcpu);
+
+	} else if (unlikely(trap == BOOK3S_INTERRUPT_HMI)) {
+		kvmppc_realmode_hmi_handler();
+
+	} else if (trap == BOOK3S_INTERRUPT_H_EMUL_ASSIST) {
+		vcpu->arch.emul_inst = mfspr(SPRN_HEIR);
+
+	} else if (trap == BOOK3S_INTERRUPT_H_DATA_STORAGE) {
+		vcpu->arch.fault_dar = exsave[EX_DAR/sizeof(u64)];
+		vcpu->arch.fault_dsisr = exsave[EX_DSISR/sizeof(u64)];
+		vcpu->arch.fault_gpa = mfspr(SPRN_ASDR);
+
+	} else if (trap == BOOK3S_INTERRUPT_H_INST_STORAGE) {
+		vcpu->arch.fault_gpa = mfspr(SPRN_ASDR);
+
+	} else if (trap == BOOK3S_INTERRUPT_H_FAC_UNAVAIL) {
+		vcpu->arch.hfscr = mfspr(SPRN_HFSCR);
+
+#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
+	/*
+	 * Softpatch interrupt for transactional memory emulation cases
+	 * on POWER9 DD2.2.  This is early in the guest exit path - we
+	 * haven't saved registers or done a treclaim yet.
+	 */
+	} else if (trap == BOOK3S_INTERRUPT_HV_SOFTPATCH) {
+		vcpu->arch.emul_inst = mfspr(SPRN_HEIR);
+
+		/*
+		 * The cases we want to handle here are those where the guest
+		 * is in real suspend mode and is trying to transition to
+		 * transactional mode.
+		 */
+		if (local_paca->kvm_hstate.fake_suspend &&
+				(vcpu->arch.shregs.msr & MSR_TS_S)) {
+			if (kvmhv_p9_tm_emulation_early(vcpu)) {
+				/* Prevent it being handled again. */
+				trap = 0;
+			}
+		}
+#endif
+	}
+
+	radix_clear_slb();
+
+	__mtmsrd(msr, 0);
+
+	accumulate_time(vcpu, &vcpu->arch.rm_exit);
+
+	end_timing(vcpu);
+
+	return trap;
+}
+EXPORT_SYMBOL_GPL(__kvmhv_vcpu_entry_p9);
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
index 2d0d14ed1d92..85c2595ead8d 100644
--- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
@@ -44,9 +44,8 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_300)
 #define NAPPING_UNSPLIT	3
 
 /* Stack frame offsets for kvmppc_hv_entry */
-#define SFS			208
+#define SFS			160
 #define STACK_SLOT_TRAP		(SFS-4)
-#define STACK_SLOT_SHORT_PATH	(SFS-8)
 #define STACK_SLOT_TID		(SFS-16)
 #define STACK_SLOT_PSSCR	(SFS-24)
 #define STACK_SLOT_PID		(SFS-32)
@@ -59,8 +58,6 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_300)
 #define STACK_SLOT_UAMOR	(SFS-88)
 #define STACK_SLOT_DAWR1	(SFS-96)
 #define STACK_SLOT_DAWRX1	(SFS-104)
-/* the following is used by the P9 short path */
-#define STACK_SLOT_NVGPRS	(SFS-152)	/* 18 gprs */
 
 /*
  * Call kvmppc_hv_entry in real mode.
@@ -1008,9 +1005,6 @@ END_MMU_FTR_SECTION_IFCLR(MMU_FTR_TYPE_RADIX)
 no_xive:
 #endif /* CONFIG_KVM_XICS */
 
-	li	r0, 0
-	stw	r0, STACK_SLOT_SHORT_PATH(r1)
-
 deliver_guest_interrupt:	/* r4 = vcpu, r13 = paca */
 	/* Check if we can deliver an external or decrementer interrupt now */
 	ld	r0, VCPU_PENDING_EXC(r4)
@@ -1030,7 +1024,6 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300)
 	mtspr	SPRN_SRR0, r6
 	mtspr	SPRN_SRR1, r7
 
-fast_guest_entry_c:
 	ld	r10, VCPU_PC(r4)
 	ld	r11, VCPU_MSR(r4)
 	/* r11 = vcpu->arch.msr & ~MSR_HV */
@@ -1135,97 +1128,6 @@ ret_to_ultra:
 	ld	r4, VCPU_GPR(R4)(r4)
 	sc	2
 
-/*
- * Enter the guest on a P9 or later system where we have exactly
- * one vcpu per vcore and we don't need to go to real mode
- * (which implies that host and guest are both using radix MMU mode).
- * r3 = vcpu pointer
- * Most SPRs and all the VSRs have been loaded already.
- */
-_GLOBAL(__kvmhv_vcpu_entry_p9)
-EXPORT_SYMBOL_GPL(__kvmhv_vcpu_entry_p9)
-	mflr	r0
-	std	r0, PPC_LR_STKOFF(r1)
-	stdu	r1, -SFS(r1)
-
-	li	r0, 1
-	stw	r0, STACK_SLOT_SHORT_PATH(r1)
-
-	std	r3, HSTATE_KVM_VCPU(r13)
-	mfcr	r4
-	stw	r4, SFS+8(r1)
-
-	std	r1, HSTATE_HOST_R1(r13)
-
-	reg = 14
-	.rept	18
-	std	reg, STACK_SLOT_NVGPRS + ((reg - 14) * 8)(r1)
-	reg = reg + 1
-	.endr
-
-	reg = 14
-	.rept	18
-	ld	reg, __VCPU_GPR(reg)(r3)
-	reg = reg + 1
-	.endr
-
-	mfmsr	r10
-	std	r10, HSTATE_HOST_MSR(r13)
-
-	mr	r4, r3
-	b	fast_guest_entry_c
-guest_exit_short_path:
-	/*
-	 * Malicious or buggy radix guests may have inserted SLB entries
-	 * (only 0..3 because radix always runs with UPRT=1), so these must
-	 * be cleared here to avoid side-channels. slbmte is used rather
-	 * than slbia, as it won't clear cached translations.
-	 */
-	li	r0,0
-	slbmte	r0,r0
-	li	r4,1
-	slbmte	r0,r4
-	li	r4,2
-	slbmte	r0,r4
-	li	r4,3
-	slbmte	r0,r4
-
-	li	r0, KVM_GUEST_MODE_NONE
-	stb	r0, HSTATE_IN_GUEST(r13)
-
-	reg = 14
-	.rept	18
-	std	reg, __VCPU_GPR(reg)(r9)
-	reg = reg + 1
-	.endr
-
-	reg = 14
-	.rept	18
-	ld	reg, STACK_SLOT_NVGPRS + ((reg - 14) * 8)(r1)
-	reg = reg + 1
-	.endr
-
-	lwz	r4, SFS+8(r1)
-	mtcr	r4
-
-	mr	r3, r12		/* trap number */
-
-	addi	r1, r1, SFS
-	ld	r0, PPC_LR_STKOFF(r1)
-	mtlr	r0
-
-	/* If we are in real mode, do a rfid to get back to the caller */
-	mfmsr	r4
-	andi.	r5, r4, MSR_IR
-	bnelr
-	rldicl	r5, r4, 64 - MSR_TS_S_LG, 62	/* extract TS field */
-	mtspr	SPRN_SRR0, r0
-	ld	r10, HSTATE_HOST_MSR(r13)
-	rldimi	r10, r5, MSR_TS_S_LG, 63 - MSR_TS_T_LG
-	mtspr	SPRN_SRR1, r10
-	RFI_TO_KERNEL
-	b	.
-
 secondary_too_late:
 	li	r12, 0
 	stw	r12, STACK_SLOT_TRAP(r1)
@@ -1397,14 +1299,9 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
 	mr	r4,r9
 	bge	fast_guest_return
 2:
-	/* If we came in through the P9 short path, no real mode hcalls */
-	lwz	r0, STACK_SLOT_SHORT_PATH(r1)
-	cmpwi	r0, 0
-	bne	no_try_real
 	/* See if this is an hcall we can handle in real mode */
 	cmpwi	r12,BOOK3S_INTERRUPT_SYSCALL
 	beq	hcall_try_real_mode
-no_try_real:
 
 	/* Hypervisor doorbell - exit only if host IPI flag set */
 	cmpwi	r12, BOOK3S_INTERRUPT_H_DOORBELL
@@ -1447,11 +1344,6 @@ guest_exit_cont:		/* r9 = vcpu, r12 = trap, r13 = paca */
 	bl	kvmhv_accumulate_time
 #endif
 #ifdef CONFIG_KVM_XICS
-	/* If we came in through the P9 short path, xive pull is done in C */
-	lwz	r0, STACK_SLOT_SHORT_PATH(r1)
-	cmpwi	r0, 0
-	bne	1f
-
 	/* We are exiting, pull the VP from the XIVE */
 	lbz	r0, VCPU_XIVE_PUSHED(r9)
 	cmpwi	cr0, r0, 0
@@ -1491,16 +1383,11 @@ guest_exit_cont:		/* r9 = vcpu, r12 = trap, r13 = paca */
 
 	/*
 	 * Possibly flush the link stack here, before we do a blr in
-	 * guest_exit_short_path.
+	 * kvmhv_switch_to_host.
 	 */
 1:	nop
 	patch_site 1b patch__call_kvm_flush_link_stack
 
-	/* If we came in through the P9 short path, go back out to C now */
-	lwz	r0, STACK_SLOT_SHORT_PATH(r1)
-	cmpwi	r0, 0
-	bne	guest_exit_short_path
-
 	/* For hash guest, read the guest SLB and save it away */
 	ld	r5, VCPU_KVM(r9)
 	lbz	r0, KVM_RADIX(r5)
@@ -1548,8 +1435,10 @@ END_MMU_FTR_SECTION_IFSET(MMU_FTR_TYPE_RADIX)
 	b	guest_bypass
 
 0:	/*
-	 * Sanitise radix guest SLB, see guest_exit_short_path comment.
-	 * We clear vcpu->arch.slb_max to match earlier behaviour.
+	 * Malicious or buggy radix guests may have inserted SLB entries
+	 * (only 0..3 because radix always runs with UPRT=1), so these must
+	 * be cleared here to avoid side-channels. slbmte is used rather
+	 * than slbia, as it won't clear cached translations.
 	 */
 	li	r0,0
 	stw	r0,VCPU_SLB_MAX(r9)
@@ -3362,7 +3251,7 @@ BEGIN_FTR_SECTION
 	mtspr	SPRN_DAWRX1, r0
 END_FTR_SECTION_IFSET(CPU_FTR_DAWR1)
 
-	/* Clear hash and radix guest SLB, see guest_exit_short_path comment. */
+	/* Clear hash and radix guest SLB. */
 	slbmte	r0, r0
 	PPC_SLBIA(6)
 
-- 
2.23.0


^ permalink raw reply related	[flat|nested] 10+ messages in thread
* [PATCH 4/4] KVM: PPC: Book3S HV P9: Implement the rest of the P9 path in C
@ 2021-05-13 12:22   ` Nicholas Piggin
  0 siblings, 0 replies; 10+ messages in thread
From: Nicholas Piggin @ 2021-05-13 12:22 UTC (permalink / raw)
  To: kvm-ppc; +Cc: linuxppc-dev, Nicholas Piggin

Almost all logic is moved to C, by introducing a new in_guest mode for
the P9 path that branches very early in the KVM interrupt handler to P9
exit code.

The main P9 entry and exit assembly is now only about 160 lines of low
level stack setup and register save/restore, plus a bad-interrupt
handler.

There are two motivations for this, the first is just make the code more
maintainable being in C. The second is to reduce the amount of code
running in a special KVM mode, "realmode". In quotes because with radix
it is no longer necessarily real-mode in the MMU, but it still has to be
treated specially because it may be in real-mode, and has various
important registers like PID, DEC, TB, etc set to guest. This is hostile
to the rest of Linux and can't use arbitrary kernel functionality or be
instrumented well.

This initial patch is a reasonably faithful conversion of the asm code,
but it does lack any loop to return quickly back into the guest without
switching out of realmode in the case of unimportant or easily handled
interrupts. As explained in previous changes, handling HV interrupts
very quickly in this low level realmode is not so important for P9
performance, and are important to avoid for security, observability,
debugability reasons.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
---
 arch/powerpc/include/asm/asm-prototypes.h |   3 +-
 arch/powerpc/include/asm/kvm_asm.h        |   1 +
 arch/powerpc/include/asm/kvm_book3s_64.h  |   8 +
 arch/powerpc/include/asm/kvm_host.h       |   7 +-
 arch/powerpc/kernel/security.c            |   5 +-
 arch/powerpc/kvm/Makefile                 |   1 +
 arch/powerpc/kvm/book3s_64_entry.S        | 254 ++++++++++++++++++++++
 arch/powerpc/kvm/book3s_hv.c              |   9 +-
 arch/powerpc/kvm/book3s_hv_p9_entry.c     | 207 ++++++++++++++++++
 arch/powerpc/kvm/book3s_hv_rmhandlers.S   | 125 +----------
 10 files changed, 496 insertions(+), 124 deletions(-)
 create mode 100644 arch/powerpc/kvm/book3s_hv_p9_entry.c

diff --git a/arch/powerpc/include/asm/asm-prototypes.h b/arch/powerpc/include/asm/asm-prototypes.h
index 1c7b75834e04..02ee6f5ac9fe 100644
--- a/arch/powerpc/include/asm/asm-prototypes.h
+++ b/arch/powerpc/include/asm/asm-prototypes.h
@@ -120,6 +120,7 @@ extern s32 patch__call_flush_branch_caches3;
 extern s32 patch__flush_count_cache_return;
 extern s32 patch__flush_link_stack_return;
 extern s32 patch__call_kvm_flush_link_stack;
+extern s32 patch__call_kvm_flush_link_stack_p9;
 extern s32 patch__memset_nocache, patch__memcpy_nocache;
 
 extern long flush_branch_caches;
@@ -140,7 +141,7 @@ void kvmhv_load_host_pmu(void);
 void kvmhv_save_guest_pmu(struct kvm_vcpu *vcpu, bool pmu_in_use);
 void kvmhv_load_guest_pmu(struct kvm_vcpu *vcpu);
 
-int __kvmhv_vcpu_entry_p9(struct kvm_vcpu *vcpu);
+void kvmppc_p9_enter_guest(struct kvm_vcpu *vcpu);
 
 long kvmppc_h_set_dabr(struct kvm_vcpu *vcpu, unsigned long dabr);
 long kvmppc_h_set_xdabr(struct kvm_vcpu *vcpu, unsigned long dabr,
diff --git a/arch/powerpc/include/asm/kvm_asm.h b/arch/powerpc/include/asm/kvm_asm.h
index a3633560493b..43b1788e1f93 100644
--- a/arch/powerpc/include/asm/kvm_asm.h
+++ b/arch/powerpc/include/asm/kvm_asm.h
@@ -147,6 +147,7 @@
 #define KVM_GUEST_MODE_SKIP	2
 #define KVM_GUEST_MODE_GUEST_HV	3
 #define KVM_GUEST_MODE_HOST_HV	4
+#define KVM_GUEST_MODE_HV_FAST	5 /* ISA >= v3.0 host+guest radix, indep thr */
 
 #define KVM_INST_FETCH_FAILED	-1
 
diff --git a/arch/powerpc/include/asm/kvm_book3s_64.h b/arch/powerpc/include/asm/kvm_book3s_64.h
index 9bb9bb370b53..c214bcffb441 100644
--- a/arch/powerpc/include/asm/kvm_book3s_64.h
+++ b/arch/powerpc/include/asm/kvm_book3s_64.h
@@ -153,9 +153,17 @@ static inline bool kvmhv_vcpu_is_radix(struct kvm_vcpu *vcpu)
 	return radix;
 }
 
+int __kvmhv_vcpu_entry_p9(struct kvm_vcpu *vcpu);
+
 #define KVM_DEFAULT_HPT_ORDER	24	/* 16MB HPT by default */
 #endif
 
+/*
+ * Invalid HDSISR value which is used to indicate when HW has not set the reg.
+ * Used to work around an errata.
+ */
+#define HDSISR_CANARY	0x7fff
+
 /*
  * We use a lock bit in HPTE dword 0 to synchronize updates and
  * accesses to each HPTE, and another bit to indicate non-present
diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
index 1e83359f286b..69add9d662df 100644
--- a/arch/powerpc/include/asm/kvm_host.h
+++ b/arch/powerpc/include/asm/kvm_host.h
@@ -683,7 +683,12 @@ struct kvm_vcpu_arch {
 	ulong fault_dar;
 	u32 fault_dsisr;
 	unsigned long intr_msr;
-	ulong fault_gpa;	/* guest real address of page fault (POWER9) */
+	/*
+	 * POWER9 and later: fault_gpa contains the guest real address of page
+	 * fault for a radix guest, or segment descriptor (equivalent to result
+	 * from slbmfev of SLB entry that translated the EA) for hash guests.
+	 */
+	ulong fault_gpa;
 #endif
 
 #ifdef CONFIG_BOOKE
diff --git a/arch/powerpc/kernel/security.c b/arch/powerpc/kernel/security.c
index 0fdfcdd9d880..c17d1c9362b5 100644
--- a/arch/powerpc/kernel/security.c
+++ b/arch/powerpc/kernel/security.c
@@ -432,16 +432,19 @@ device_initcall(stf_barrier_debugfs_init);
 
 static void update_branch_cache_flush(void)
 {
-	u32 *site;
+	u32 *site, __maybe_unused *site2;
 
 #ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 	site = &patch__call_kvm_flush_link_stack;
+	site2 = &patch__call_kvm_flush_link_stack_p9;
 	// This controls the branch from guest_exit_cont to kvm_flush_link_stack
 	if (link_stack_flush_type = BRANCH_CACHE_FLUSH_NONE) {
 		patch_instruction_site(site, ppc_inst(PPC_INST_NOP));
+		patch_instruction_site(site2, ppc_inst(PPC_INST_NOP));
 	} else {
 		// Could use HW flush, but that could also flush count cache
 		patch_branch_site(site, (u64)&kvm_flush_link_stack, BRANCH_SET_LINK);
+		patch_branch_site(site2, (u64)&kvm_flush_link_stack, BRANCH_SET_LINK);
 	}
 #endif
 
diff --git a/arch/powerpc/kvm/Makefile b/arch/powerpc/kvm/Makefile
index cdd119028f64..91eb67bb91e1 100644
--- a/arch/powerpc/kvm/Makefile
+++ b/arch/powerpc/kvm/Makefile
@@ -89,6 +89,7 @@ kvm-book3s_64-builtin-tm-objs-$(CONFIG_PPC_TRANSACTIONAL_MEM) += \
 ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
 kvm-book3s_64-builtin-objs-$(CONFIG_KVM_BOOK3S_64_HANDLER) += \
 	book3s_hv_hmi.o \
+	book3s_hv_p9_entry.o \
 	book3s_hv_rmhandlers.o \
 	book3s_hv_rm_mmu.o \
 	book3s_hv_ras.o \
diff --git a/arch/powerpc/kvm/book3s_64_entry.S b/arch/powerpc/kvm/book3s_64_entry.S
index a01046202eef..177e8fad5c8d 100644
--- a/arch/powerpc/kvm/book3s_64_entry.S
+++ b/arch/powerpc/kvm/book3s_64_entry.S
@@ -1,11 +1,16 @@
 /* SPDX-License-Identifier: GPL-2.0-only */
 #include <asm/asm-offsets.h>
 #include <asm/cache.h>
+#include <asm/code-patching-asm.h>
 #include <asm/exception-64s.h>
+#include <asm/export.h>
 #include <asm/kvm_asm.h>
 #include <asm/kvm_book3s_asm.h>
+#include <asm/mmu.h>
 #include <asm/ppc_asm.h>
+#include <asm/ptrace.h>
 #include <asm/reg.h>
+#include <asm/ultravisor-api.h>
 
 /*
  * These are branched to from interrupt handlers in exception-64s.S which set
@@ -29,6 +34,11 @@
 .global	kvmppc_hcall
 .balign IFETCH_ALIGN_BYTES
 kvmppc_hcall:
+#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+	lbz	r10,HSTATE_IN_GUEST(r13)
+	cmpwi	r10,KVM_GUEST_MODE_HV_FAST
+	beq	kvmppc_p9_exit_hcall
+#endif
 	ld	r10,PACA_EXGEN+EX_R13(r13)
 	SET_SCRATCH0(r10)
 	li	r10,0xc00
@@ -55,6 +65,13 @@ kvmppc_hcall:
 .global	kvmppc_interrupt
 .balign IFETCH_ALIGN_BYTES
 kvmppc_interrupt:
+#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+	std	r10,HSTATE_SCRATCH0(r13)
+	lbz	r10,HSTATE_IN_GUEST(r13)
+	cmpwi	r10,KVM_GUEST_MODE_HV_FAST
+	beq	kvmppc_p9_exit_interrupt
+	ld	r10,HSTATE_SCRATCH0(r13)
+#endif
 	li	r11,PACA_EXGEN
 	cmpdi	r10,0x200
 	bgt+	.Lgot_save_area
@@ -156,3 +173,240 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
 	GET_SCRATCH0(r13)
 	HRFI_TO_KERNEL
 #endif
+
+#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+
+/* Stack frame offsets for kvmppc_p9_enter_guest */
+#define SFS			(144 + STACK_FRAME_MIN_SIZE)
+#define STACK_SLOT_NVGPRS	(SFS - 144)	/* 18 gprs */
+
+/*
+ * void kvmppc_p9_enter_guest(struct vcpu *vcpu);
+ *
+ * Enter the guest on a ISAv3.0 or later system where we have exactly
+ * one vcpu per vcore, and both the host and guest are radix, and threads
+ * are set to "indepdent mode".
+ */
+.balign	IFETCH_ALIGN_BYTES
+_GLOBAL(kvmppc_p9_enter_guest)
+EXPORT_SYMBOL_GPL(kvmppc_p9_enter_guest)
+	mflr	r0
+	std	r0,PPC_LR_STKOFF(r1)
+	stdu	r1,-SFS(r1)
+
+	std	r1,HSTATE_HOST_R1(r13)
+
+	mfcr	r4
+	stw	r4,SFS+8(r1)
+
+	reg = 14
+	.rept	18
+	std	reg,STACK_SLOT_NVGPRS + ((reg - 14) * 8)(r1)
+	reg = reg + 1
+	.endr
+
+	ld	r4,VCPU_LR(r3)
+	mtlr	r4
+	ld	r4,VCPU_CTR(r3)
+	mtctr	r4
+	ld	r4,VCPU_XER(r3)
+	mtspr	SPRN_XER,r4
+
+	ld	r1,VCPU_CR(r3)
+
+BEGIN_FTR_SECTION
+	ld	r4,VCPU_CFAR(r3)
+	mtspr	SPRN_CFAR,r4
+END_FTR_SECTION_IFSET(CPU_FTR_CFAR)
+BEGIN_FTR_SECTION
+	ld	r4,VCPU_PPR(r3)
+	mtspr	SPRN_PPR,r4
+END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
+
+	reg = 4
+	.rept	28
+	ld	reg,__VCPU_GPR(reg)(r3)
+	reg = reg + 1
+	.endr
+
+	ld	r4,VCPU_KVM(r3)
+	lbz	r4,KVM_SECURE_GUEST(r4)
+	cmpdi	r4,0
+	ld	r4,VCPU_GPR(R4)(r3)
+	bne	.Lret_to_ultra
+
+	mtcr	r1
+
+	ld	r0,VCPU_GPR(R0)(r3)
+	ld	r1,VCPU_GPR(R1)(r3)
+	ld	r2,VCPU_GPR(R2)(r3)
+	ld	r3,VCPU_GPR(R3)(r3)
+
+	HRFI_TO_GUEST
+	b	.
+
+	/*
+	 * Use UV_RETURN ultracall to return control back to the Ultravisor
+	 * after processing an hypercall or interrupt that was forwarded
+	 * (a.k.a. reflected) to the Hypervisor.
+	 *
+	 * All registers have already been reloaded except the ucall requires:
+	 *   R0 = hcall result
+	 *   R2 = SRR1, so UV can detect a synthesized interrupt (if any)
+	 *   R3 = UV_RETURN
+	 */
+.Lret_to_ultra:
+	mtcr	r1
+	ld	r1,VCPU_GPR(R1)(r3)
+
+	ld	r0,VCPU_GPR(R3)(r3)
+	mfspr	r2,SPRN_SRR1
+	LOAD_REG_IMMEDIATE(r3, UV_RETURN)
+	sc	2
+
+/*
+ * kvmppc_p9_exit_hcall and kvmppc_p9_exit_interrupt are branched to from
+ * above if the interrupt was taken for a guest that was entered via
+ * kvmppc_p9_enter_guest().
+ *
+ * The exit code recovers the host stack and vcpu pointer, saves all guest GPRs
+ * and CR, LR, XER as well as guest MSR and NIA into the VCPU, then re-
+ * establishes the host stack and registers to return from the
+ * kvmppc_p9_enter_guest() function, which saves CTR and other guest registers
+ * (SPRs and FP, VEC, etc).
+ */
+.balign	IFETCH_ALIGN_BYTES
+kvmppc_p9_exit_hcall:
+	mfspr	r11,SPRN_SRR0
+	mfspr	r12,SPRN_SRR1
+	li	r10,0xc00
+	std	r10,HSTATE_SCRATCH0(r13)
+
+.balign	IFETCH_ALIGN_BYTES
+kvmppc_p9_exit_interrupt:
+	/*
+	 * If set to KVM_GUEST_MODE_HV_FAST but we're still in the
+	 * hypervisor, that means we can't return from the entry stack.
+	 */
+	rldicl. r10,r12,64-MSR_HV_LG,63
+	bne-	kvmppc_p9_bad_interrupt
+
+	std     r1,HSTATE_SCRATCH1(r13)
+	std     r3,HSTATE_SCRATCH2(r13)
+	ld	r1,HSTATE_HOST_R1(r13)
+	ld	r3,HSTATE_KVM_VCPU(r13)
+
+	std	r9,VCPU_CR(r3)
+
+1:
+	std	r11,VCPU_PC(r3)
+	std	r12,VCPU_MSR(r3)
+
+	reg = 14
+	.rept	18
+	std	reg,__VCPU_GPR(reg)(r3)
+	reg = reg + 1
+	.endr
+
+	/* r1, r3, r9-r13 are saved to vcpu by C code */
+	std	r0,VCPU_GPR(R0)(r3)
+	std	r2,VCPU_GPR(R2)(r3)
+	reg = 4
+	.rept	5
+	std	reg,__VCPU_GPR(reg)(r3)
+	reg = reg + 1
+	.endr
+
+	ld	r2,PACATOC(r13)
+
+	mflr	r4
+	std	r4,VCPU_LR(r3)
+	mfspr	r4,SPRN_XER
+	std	r4,VCPU_XER(r3)
+
+	reg = 14
+	.rept	18
+	ld	reg,STACK_SLOT_NVGPRS + ((reg - 14) * 8)(r1)
+	reg = reg + 1
+	.endr
+
+	lwz	r4,SFS+8(r1)
+	mtcr	r4
+
+	/*
+	 * Flush the link stack here, before executing the first blr on the
+	 * way out of the guest.
+	 *
+	 * The link stack won't match coming out of the guest anyway so the
+	 * only cost is the flush itself. The call clobbers r0.
+	 */
+1:	nop
+	patch_site 1b patch__call_kvm_flush_link_stack_p9
+
+	addi	r1,r1,SFS
+	ld	r0,PPC_LR_STKOFF(r1)
+	mtlr	r0
+	blr
+
+/*
+ * Took an interrupt somewhere right before HRFID to guest, so registers are
+ * in a bad way. Return things hopefully enough to run host virtual code and
+ * run the Linux interrupt handler (SRESET or MCE) to print something useful.
+ *
+ * We could be really clever and save all host registers in known locations
+ * before setting HSTATE_IN_GUEST, then restoring them all here, and setting
+ * return address to a fixup that sets them up again. But that's a lot of
+ * effort for a small bit of code. Lots of other things to do first.
+ */
+kvmppc_p9_bad_interrupt:
+	/*
+	 * Clean up guest registers to give host a chance to run.
+	 */
+	li	r10,0
+	mtspr	SPRN_AMR,r10
+	mtspr	SPRN_IAMR,r10
+	mtspr	SPRN_CIABR,r10
+	mtspr	SPRN_DAWRX0,r10
+BEGIN_FTR_SECTION
+	mtspr	SPRN_DAWRX1,r10
+END_FTR_SECTION_IFSET(CPU_FTR_DAWR1)
+	mtspr	SPRN_PID,r10
+
+	/*
+	 * Switch to host MMU mode
+	 */
+	ld	r10, HSTATE_KVM_VCPU(r13)
+	ld	r10, VCPU_KVM(r10)
+	lwz	r10, KVM_HOST_LPID(r10)
+	mtspr	SPRN_LPID,r10
+
+	ld	r10, HSTATE_KVM_VCPU(r13)
+	ld	r10, VCPU_KVM(r10)
+	ld	r10, KVM_HOST_LPCR(r10)
+	mtspr	SPRN_LPCR,r10
+
+	/*
+	 * Set GUEST_MODE_NONE so the handler won't branch to KVM, and clear
+	 * MSR_RI in r12 ([H]SRR1) so the handler won't try to return.
+	 */
+	li	r10,KVM_GUEST_MODE_NONE
+	stb	r10,HSTATE_IN_GUEST(r13)
+	li	r10,MSR_RI
+	andc	r12,r12,r10
+
+	/*
+	 * Go back to interrupt handler. MCE and SRESET have their specific
+	 * PACA save area so they should be used directly. They set up their
+	 * own stack. The other handlers all use EXGEN. They will use the
+	 * guest r1 if it looks like a kernel stack, so just load the
+	 * emergency stack and go to program check for all other interrupts.
+	 */
+	ld	r10,HSTATE_SCRATCH0(r13)
+	cmpwi	r10,BOOK3S_INTERRUPT_MACHINE_CHECK
+	beq	machine_check_common
+
+	cmpwi	r10,BOOK3S_INTERRUPT_SYSTEM_RESET
+	beq	system_reset_common
+
+	b	.
+#endif
diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 5815c04a48a3..e338da9d9312 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c
@@ -1445,6 +1445,8 @@ static int kvmppc_handle_exit_hv(struct kvm_vcpu *vcpu,
 	 */
 	case BOOK3S_INTERRUPT_H_DATA_STORAGE:
 		r = RESUME_PAGE_FAULT;
+		if (vcpu->arch.fault_dsisr = HDSISR_CANARY)
+			r = RESUME_GUEST; /* Just retry if it's the canary */
 		break;
 	case BOOK3S_INTERRUPT_H_INST_STORAGE:
 		vcpu->arch.fault_dar = kvmppc_get_pc(vcpu);
@@ -3708,6 +3710,8 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 	u64 tb;
 	int trap, save_pmu;
 
+	WARN_ON_ONCE(vcpu->arch.ceded);
+
 	dec = mfspr(SPRN_DEC);
 	tb = mftb();
 	if (dec < 0)
@@ -3716,8 +3720,6 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 	if (local_paca->kvm_hstate.dec_expires < time_limit)
 		time_limit = local_paca->kvm_hstate.dec_expires;
 
-	vcpu->arch.ceded = 0;
-
 	kvmhv_save_host_pmu();		/* saves it to PACA kvm_hstate */
 
 	kvmppc_subcore_enter_guest();
@@ -3844,9 +3846,10 @@ static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
 			}
 		}
 		kvmppc_xive_pull_vcpu(vcpu);
+
+		vcpu->arch.slb_max = 0;
 	}
 
-	vcpu->arch.slb_max = 0;
 	dec = mfspr(SPRN_DEC);
 	if (!(lpcr & LPCR_LD)) /* Sign extend if not using large decrementer */
 		dec = (s32) dec;
diff --git a/arch/powerpc/kvm/book3s_hv_p9_entry.c b/arch/powerpc/kvm/book3s_hv_p9_entry.c
new file mode 100644
index 000000000000..9db0e031a443
--- /dev/null
+++ b/arch/powerpc/kvm/book3s_hv_p9_entry.c
@@ -0,0 +1,207 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include <linux/kernel.h>
+#include <linux/kvm_host.h>
+#include <asm/asm-prototypes.h>
+#include <asm/dbell.h>
+#include <asm/kvm_ppc.h>
+
+#ifdef CONFIG_KVM_BOOK3S_HV_EXIT_TIMING
+static void __start_timing(struct kvm_vcpu *vcpu, struct kvmhv_tb_accumulator *next)
+{
+	struct kvmppc_vcore *vc = vcpu->arch.vcore;
+	u64 tb = mftb() - vc->tb_offset_applied;
+
+	vcpu->arch.cur_activity = next;
+	vcpu->arch.cur_tb_start = tb;
+}
+
+static void __accumulate_time(struct kvm_vcpu *vcpu, struct kvmhv_tb_accumulator *next)
+{
+	struct kvmppc_vcore *vc = vcpu->arch.vcore;
+	struct kvmhv_tb_accumulator *curr;
+	u64 tb = mftb() - vc->tb_offset_applied;
+	u64 prev_tb;
+	u64 delta;
+	u64 seq;
+
+	curr = vcpu->arch.cur_activity;
+	vcpu->arch.cur_activity = next;
+	prev_tb = vcpu->arch.cur_tb_start;
+	vcpu->arch.cur_tb_start = tb;
+
+	if (!curr)
+		return;
+
+	delta = tb - prev_tb;
+
+	seq = curr->seqcount;
+	curr->seqcount = seq + 1;
+	smp_wmb();
+	curr->tb_total += delta;
+	if (seq = 0 || delta < curr->tb_min)
+		curr->tb_min = delta;
+	if (delta > curr->tb_max)
+		curr->tb_max = delta;
+	smp_wmb();
+	curr->seqcount = seq + 2;
+}
+
+#define start_timing(vcpu, next) __start_timing(vcpu, next)
+#define end_timing(vcpu) __start_timing(vcpu, NULL)
+#define accumulate_time(vcpu, next) __accumulate_time(vcpu, next)
+#else
+#define start_timing(vcpu, next) do {} while (0)
+#define end_timing(vcpu) do {} while (0)
+#define accumulate_time(vcpu, next) do {} while (0)
+#endif
+
+static inline void mtslb(u64 slbee, u64 slbev)
+{
+	asm volatile("slbmte %0,%1" :: "r" (slbev), "r" (slbee));
+}
+
+static inline void clear_slb_entry(unsigned int idx)
+{
+	mtslb(idx, 0);
+}
+
+/*
+ * Malicious or buggy radix guests may have inserted SLB entries
+ * (only 0..3 because radix always runs with UPRT=1), so these must
+ * be cleared here to avoid side-channels. slbmte is used rather
+ * than slbia, as it won't clear cached translations.
+ */
+static void radix_clear_slb(void)
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		clear_slb_entry(i);
+}
+
+int __kvmhv_vcpu_entry_p9(struct kvm_vcpu *vcpu)
+{
+	u64 *exsave;
+	unsigned long msr = mfmsr();
+	int trap;
+
+	start_timing(vcpu, &vcpu->arch.rm_entry);
+
+	vcpu->arch.ceded = 0;
+
+	WARN_ON_ONCE(vcpu->arch.shregs.msr & MSR_HV);
+	WARN_ON_ONCE(!(vcpu->arch.shregs.msr & MSR_ME));
+
+	mtspr(SPRN_HSRR0, vcpu->arch.regs.nip);
+	mtspr(SPRN_HSRR1, (vcpu->arch.shregs.msr & ~MSR_HV) | MSR_ME);
+
+	/*
+	 * On POWER9 DD2.1 and below, sometimes on a Hypervisor Data Storage
+	 * Interrupt (HDSI) the HDSISR is not be updated at all.
+	 *
+	 * To work around this we put a canary value into the HDSISR before
+	 * returning to a guest and then check for this canary when we take a
+	 * HDSI. If we find the canary on a HDSI, we know the hardware didn't
+	 * update the HDSISR. In this case we return to the guest to retake the
+	 * HDSI which should correctly update the HDSISR the second time HDSI
+	 * entry.
+	 *
+	 * Just do this on all p9 processors for now.
+	 */
+	mtspr(SPRN_HDSISR, HDSISR_CANARY);
+
+	accumulate_time(vcpu, &vcpu->arch.guest_time);
+
+	local_paca->kvm_hstate.in_guest = KVM_GUEST_MODE_HV_FAST;
+	kvmppc_p9_enter_guest(vcpu);
+	// Radix host and guest means host never runs with guest MMU state
+	local_paca->kvm_hstate.in_guest = KVM_GUEST_MODE_NONE;
+
+	accumulate_time(vcpu, &vcpu->arch.rm_intr);
+
+	/* XXX: Could get these from r11/12 and paca exsave instead */
+	vcpu->arch.shregs.srr0 = mfspr(SPRN_SRR0);
+	vcpu->arch.shregs.srr1 = mfspr(SPRN_SRR1);
+	vcpu->arch.shregs.dar = mfspr(SPRN_DAR);
+	vcpu->arch.shregs.dsisr = mfspr(SPRN_DSISR);
+
+	/* 0x2 bit for HSRR is only used by PR and P7/8 HV paths, clear it */
+	trap = local_paca->kvm_hstate.scratch0 & ~0x2;
+	if (likely(trap > BOOK3S_INTERRUPT_MACHINE_CHECK)) {
+		exsave = local_paca->exgen;
+	} else if (trap = BOOK3S_INTERRUPT_SYSTEM_RESET) {
+		exsave = local_paca->exnmi;
+	} else { /* trap = 0x200 */
+		exsave = local_paca->exmc;
+	}
+
+	vcpu->arch.regs.gpr[1] = local_paca->kvm_hstate.scratch1;
+	vcpu->arch.regs.gpr[3] = local_paca->kvm_hstate.scratch2;
+	vcpu->arch.regs.gpr[9] = exsave[EX_R9/sizeof(u64)];
+	vcpu->arch.regs.gpr[10] = exsave[EX_R10/sizeof(u64)];
+	vcpu->arch.regs.gpr[11] = exsave[EX_R11/sizeof(u64)];
+	vcpu->arch.regs.gpr[12] = exsave[EX_R12/sizeof(u64)];
+	vcpu->arch.regs.gpr[13] = exsave[EX_R13/sizeof(u64)];
+	vcpu->arch.ppr = exsave[EX_PPR/sizeof(u64)];
+	vcpu->arch.cfar = exsave[EX_CFAR/sizeof(u64)];
+	vcpu->arch.regs.ctr = exsave[EX_CTR/sizeof(u64)];
+
+	vcpu->arch.last_inst = KVM_INST_FETCH_FAILED;
+
+	if (unlikely(trap = BOOK3S_INTERRUPT_MACHINE_CHECK)) {
+		vcpu->arch.fault_dar = exsave[EX_DAR/sizeof(u64)];
+		vcpu->arch.fault_dsisr = exsave[EX_DSISR/sizeof(u64)];
+		kvmppc_realmode_machine_check(vcpu);
+
+	} else if (unlikely(trap = BOOK3S_INTERRUPT_HMI)) {
+		kvmppc_realmode_hmi_handler();
+
+	} else if (trap = BOOK3S_INTERRUPT_H_EMUL_ASSIST) {
+		vcpu->arch.emul_inst = mfspr(SPRN_HEIR);
+
+	} else if (trap = BOOK3S_INTERRUPT_H_DATA_STORAGE) {
+		vcpu->arch.fault_dar = exsave[EX_DAR/sizeof(u64)];
+		vcpu->arch.fault_dsisr = exsave[EX_DSISR/sizeof(u64)];
+		vcpu->arch.fault_gpa = mfspr(SPRN_ASDR);
+
+	} else if (trap = BOOK3S_INTERRUPT_H_INST_STORAGE) {
+		vcpu->arch.fault_gpa = mfspr(SPRN_ASDR);
+
+	} else if (trap = BOOK3S_INTERRUPT_H_FAC_UNAVAIL) {
+		vcpu->arch.hfscr = mfspr(SPRN_HFSCR);
+
+#ifdef CONFIG_PPC_TRANSACTIONAL_MEM
+	/*
+	 * Softpatch interrupt for transactional memory emulation cases
+	 * on POWER9 DD2.2.  This is early in the guest exit path - we
+	 * haven't saved registers or done a treclaim yet.
+	 */
+	} else if (trap = BOOK3S_INTERRUPT_HV_SOFTPATCH) {
+		vcpu->arch.emul_inst = mfspr(SPRN_HEIR);
+
+		/*
+		 * The cases we want to handle here are those where the guest
+		 * is in real suspend mode and is trying to transition to
+		 * transactional mode.
+		 */
+		if (local_paca->kvm_hstate.fake_suspend &&
+				(vcpu->arch.shregs.msr & MSR_TS_S)) {
+			if (kvmhv_p9_tm_emulation_early(vcpu)) {
+				/* Prevent it being handled again. */
+				trap = 0;
+			}
+		}
+#endif
+	}
+
+	radix_clear_slb();
+
+	__mtmsrd(msr, 0);
+
+	accumulate_time(vcpu, &vcpu->arch.rm_exit);
+
+	end_timing(vcpu);
+
+	return trap;
+}
+EXPORT_SYMBOL_GPL(__kvmhv_vcpu_entry_p9);
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
index 2d0d14ed1d92..85c2595ead8d 100644
--- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
@@ -44,9 +44,8 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_300)
 #define NAPPING_UNSPLIT	3
 
 /* Stack frame offsets for kvmppc_hv_entry */
-#define SFS			208
+#define SFS			160
 #define STACK_SLOT_TRAP		(SFS-4)
-#define STACK_SLOT_SHORT_PATH	(SFS-8)
 #define STACK_SLOT_TID		(SFS-16)
 #define STACK_SLOT_PSSCR	(SFS-24)
 #define STACK_SLOT_PID		(SFS-32)
@@ -59,8 +58,6 @@ END_FTR_SECTION_IFCLR(CPU_FTR_ARCH_300)
 #define STACK_SLOT_UAMOR	(SFS-88)
 #define STACK_SLOT_DAWR1	(SFS-96)
 #define STACK_SLOT_DAWRX1	(SFS-104)
-/* the following is used by the P9 short path */
-#define STACK_SLOT_NVGPRS	(SFS-152)	/* 18 gprs */
 
 /*
  * Call kvmppc_hv_entry in real mode.
@@ -1008,9 +1005,6 @@ END_MMU_FTR_SECTION_IFCLR(MMU_FTR_TYPE_RADIX)
 no_xive:
 #endif /* CONFIG_KVM_XICS */
 
-	li	r0, 0
-	stw	r0, STACK_SLOT_SHORT_PATH(r1)
-
 deliver_guest_interrupt:	/* r4 = vcpu, r13 = paca */
 	/* Check if we can deliver an external or decrementer interrupt now */
 	ld	r0, VCPU_PENDING_EXC(r4)
@@ -1030,7 +1024,6 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_300)
 	mtspr	SPRN_SRR0, r6
 	mtspr	SPRN_SRR1, r7
 
-fast_guest_entry_c:
 	ld	r10, VCPU_PC(r4)
 	ld	r11, VCPU_MSR(r4)
 	/* r11 = vcpu->arch.msr & ~MSR_HV */
@@ -1135,97 +1128,6 @@ ret_to_ultra:
 	ld	r4, VCPU_GPR(R4)(r4)
 	sc	2
 
-/*
- * Enter the guest on a P9 or later system where we have exactly
- * one vcpu per vcore and we don't need to go to real mode
- * (which implies that host and guest are both using radix MMU mode).
- * r3 = vcpu pointer
- * Most SPRs and all the VSRs have been loaded already.
- */
-_GLOBAL(__kvmhv_vcpu_entry_p9)
-EXPORT_SYMBOL_GPL(__kvmhv_vcpu_entry_p9)
-	mflr	r0
-	std	r0, PPC_LR_STKOFF(r1)
-	stdu	r1, -SFS(r1)
-
-	li	r0, 1
-	stw	r0, STACK_SLOT_SHORT_PATH(r1)
-
-	std	r3, HSTATE_KVM_VCPU(r13)
-	mfcr	r4
-	stw	r4, SFS+8(r1)
-
-	std	r1, HSTATE_HOST_R1(r13)
-
-	reg = 14
-	.rept	18
-	std	reg, STACK_SLOT_NVGPRS + ((reg - 14) * 8)(r1)
-	reg = reg + 1
-	.endr
-
-	reg = 14
-	.rept	18
-	ld	reg, __VCPU_GPR(reg)(r3)
-	reg = reg + 1
-	.endr
-
-	mfmsr	r10
-	std	r10, HSTATE_HOST_MSR(r13)
-
-	mr	r4, r3
-	b	fast_guest_entry_c
-guest_exit_short_path:
-	/*
-	 * Malicious or buggy radix guests may have inserted SLB entries
-	 * (only 0..3 because radix always runs with UPRT=1), so these must
-	 * be cleared here to avoid side-channels. slbmte is used rather
-	 * than slbia, as it won't clear cached translations.
-	 */
-	li	r0,0
-	slbmte	r0,r0
-	li	r4,1
-	slbmte	r0,r4
-	li	r4,2
-	slbmte	r0,r4
-	li	r4,3
-	slbmte	r0,r4
-
-	li	r0, KVM_GUEST_MODE_NONE
-	stb	r0, HSTATE_IN_GUEST(r13)
-
-	reg = 14
-	.rept	18
-	std	reg, __VCPU_GPR(reg)(r9)
-	reg = reg + 1
-	.endr
-
-	reg = 14
-	.rept	18
-	ld	reg, STACK_SLOT_NVGPRS + ((reg - 14) * 8)(r1)
-	reg = reg + 1
-	.endr
-
-	lwz	r4, SFS+8(r1)
-	mtcr	r4
-
-	mr	r3, r12		/* trap number */
-
-	addi	r1, r1, SFS
-	ld	r0, PPC_LR_STKOFF(r1)
-	mtlr	r0
-
-	/* If we are in real mode, do a rfid to get back to the caller */
-	mfmsr	r4
-	andi.	r5, r4, MSR_IR
-	bnelr
-	rldicl	r5, r4, 64 - MSR_TS_S_LG, 62	/* extract TS field */
-	mtspr	SPRN_SRR0, r0
-	ld	r10, HSTATE_HOST_MSR(r13)
-	rldimi	r10, r5, MSR_TS_S_LG, 63 - MSR_TS_T_LG
-	mtspr	SPRN_SRR1, r10
-	RFI_TO_KERNEL
-	b	.
-
 secondary_too_late:
 	li	r12, 0
 	stw	r12, STACK_SLOT_TRAP(r1)
@@ -1397,14 +1299,9 @@ END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
 	mr	r4,r9
 	bge	fast_guest_return
 2:
-	/* If we came in through the P9 short path, no real mode hcalls */
-	lwz	r0, STACK_SLOT_SHORT_PATH(r1)
-	cmpwi	r0, 0
-	bne	no_try_real
 	/* See if this is an hcall we can handle in real mode */
 	cmpwi	r12,BOOK3S_INTERRUPT_SYSCALL
 	beq	hcall_try_real_mode
-no_try_real:
 
 	/* Hypervisor doorbell - exit only if host IPI flag set */
 	cmpwi	r12, BOOK3S_INTERRUPT_H_DOORBELL
@@ -1447,11 +1344,6 @@ guest_exit_cont:		/* r9 = vcpu, r12 = trap, r13 = paca */
 	bl	kvmhv_accumulate_time
 #endif
 #ifdef CONFIG_KVM_XICS
-	/* If we came in through the P9 short path, xive pull is done in C */
-	lwz	r0, STACK_SLOT_SHORT_PATH(r1)
-	cmpwi	r0, 0
-	bne	1f
-
 	/* We are exiting, pull the VP from the XIVE */
 	lbz	r0, VCPU_XIVE_PUSHED(r9)
 	cmpwi	cr0, r0, 0
@@ -1491,16 +1383,11 @@ guest_exit_cont:		/* r9 = vcpu, r12 = trap, r13 = paca */
 
 	/*
 	 * Possibly flush the link stack here, before we do a blr in
-	 * guest_exit_short_path.
+	 * kvmhv_switch_to_host.
 	 */
 1:	nop
 	patch_site 1b patch__call_kvm_flush_link_stack
 
-	/* If we came in through the P9 short path, go back out to C now */
-	lwz	r0, STACK_SLOT_SHORT_PATH(r1)
-	cmpwi	r0, 0
-	bne	guest_exit_short_path
-
 	/* For hash guest, read the guest SLB and save it away */
 	ld	r5, VCPU_KVM(r9)
 	lbz	r0, KVM_RADIX(r5)
@@ -1548,8 +1435,10 @@ END_MMU_FTR_SECTION_IFSET(MMU_FTR_TYPE_RADIX)
 	b	guest_bypass
 
 0:	/*
-	 * Sanitise radix guest SLB, see guest_exit_short_path comment.
-	 * We clear vcpu->arch.slb_max to match earlier behaviour.
+	 * Malicious or buggy radix guests may have inserted SLB entries
+	 * (only 0..3 because radix always runs with UPRT=1), so these must
+	 * be cleared here to avoid side-channels. slbmte is used rather
+	 * than slbia, as it won't clear cached translations.
 	 */
 	li	r0,0
 	stw	r0,VCPU_SLB_MAX(r9)
@@ -3362,7 +3251,7 @@ BEGIN_FTR_SECTION
 	mtspr	SPRN_DAWRX1, r0
 END_FTR_SECTION_IFSET(CPU_FTR_DAWR1)
 
-	/* Clear hash and radix guest SLB, see guest_exit_short_path comment. */
+	/* Clear hash and radix guest SLB. */
 	slbmte	r0, r0
 	PPC_SLBIA(6)
 
-- 
2.23.0

^ permalink raw reply related	[flat|nested] 10+ messages in thread
end of thread, other threads:[~2021-05-13 12:24 UTC | newest]

Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-05-13 12:22 [PATCH 0/4] KVM: PPC: Convert P9 HV path to C Nicholas Piggin
2021-05-13 12:22 ` Nicholas Piggin
2021-05-13 12:22 ` [PATCH 1/4] KVM: PPC: Book3S HV P9: Move xive vcpu context management into kvmhv_p9_guest_entry Nicholas Piggin
2021-05-13 12:22   ` Nicholas Piggin
2021-05-13 12:22 ` [PATCH 2/4] KVM: PPC: Book3S HV P9: Move radix MMU switching instructions together Nicholas Piggin
2021-05-13 12:22   ` Nicholas Piggin
2021-05-13 12:22 ` [PATCH 3/4] KVM: PPC: Book3S HV P9: Stop handling hcalls in real-mode in the P9 path Nicholas Piggin
2021-05-13 12:22   ` Nicholas Piggin
2021-05-13 12:22 ` [PATCH 4/4] KVM: PPC: Book3S HV P9: Implement the rest of the P9 path in C Nicholas Piggin
2021-05-13 12:22   ` Nicholas Piggin

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.