[PATCH v3 00/13] arm64/firmware: Software Delegated Exception Interface

[PATCH v3 00/13] arm64/firmware: Software Delegated Exception Interface

[James Morse]

Hello!

The Software Delegated Exception Interface (SDEI) is an ARM specification
for registering callbacks from the platform firmware into the OS.
This is intended to be used to implement firmware-first RAS notifications,
but also supports vendor-defined events and binding IRQs as events.

The document is here:
http://infocenter.arm.com/help/topic/com.arm.doc.den0054a/ARM_DEN0054A_Software_Delegated_Exception_Interface.pdf

I anticpate once reviewed this series will go via the arm-soc tree, as the
bulk of the code is under /drivers/firmware/.

The major change in v3 is due to VMAP stacks. We can't trust sp when we take
an event, so need our own SDEI stack. Events can nest, so we need two. Private
events can be delivered per-cpu, so we need two per-cpu SDEI stacks.

The APCICA support for the SDEI table is now in mainline, so this series has
grown the detection code for non-DT systems. The APEI/GHES wire-up is still
being held back as APEI's ghes_proc() isn't (yet) safe for multiple sources
of NMI.

Exposing the HVC range through KVM to user-space is gone, this will be
re-incarnated as something that better fits KVMs architecture emulation,
as opposed to assuming all-the-world's-SMC-CC.


This series (juggles some registers with KVM+VHE, then) adds a DT binding to
trigger probing of the interface and support for the SDEI API.

SDEI runs between adjacent exception levels, so events will always be delivered
to EL2 if firmware is at EL3. For VHE hosts we run the SDEI event handler
behind KVM's back with all exceptions masked. Once the handler has done its
work we return to the appropriate vbar+irq entry. This allows KVM to
world-switch and deliver any signals sent by the handler to Qemu/kvmtool. We
do the same thing if we interrupt host EL0. If we interrupted code with
interrupts masked, we use a different API call to return to the interrupted
context.

What about non-VHE KVM? If you don't have VHE support and boot at EL2, the
kernel drops to EL1. This driver will print an error message then give up. This
is because events would still be delivered to EL2 hitting either KVM, or the
hyp-stub. Supporting this is complicated, but because the main use-case is
RAS, and ARM v8.2's RAS extensions imply v8.1's Virtual Host Extensions, we
can assume all platforms with SDEI will support VHE too. (I have some ideas
on how to support non-VHE).

Running the event handler behind VHE-KVM's back has some side effects: The
event handler will blindly use any registers that are shared between the host
and guest. The two that I think matter are TPIDR_EL1, and the debug state. The
guest may have set MDSCR_EL1 so debug exceptions must remain masked. The
guest's TPIDR_EL1 will be used by the event handler if it accesses per-cpu
variables. This needs fixing. The first part of this series juggles KVMs use
of TPIDR_EL2 so that we share it with the host on VHE systems. An equivalent
change for 32bit is (still) on my todo list. (the alternative to this is to
have a parody world switch in the SDEI event handler, but this would mean
special casing interrupted guests, and be an ABI link to KVM.)

Is this another begins-with-S RAS mechanism for arm64? Yes.
Why? Any notification delivered as an exception will overwrite the exception
registers. This is fatal for the running thread if it happens during entry.S's
kernel_enter or kernel_exit. Instead of adding masking and routing controls,
events are delivered to a registered address at a fixed exception level and
don't change the exception registers when delivered.

This series can be retrieved from:
git://linux-arm.org/linux-jm.git -b sdei/v3/base

Questions and contradictions welcome!

Thanks,

James

[Changes since previous versions are noted on each patch]


James Morse (13):
  KVM: arm64: Store vcpu on the stack during __guest_enter()
  KVM: arm/arm64: Convert kvm_host_cpu_state to a static per-cpu
    allocation
  KVM: arm64: Change hyp_panic()s dependency on tpidr_el2
  arm64: alternatives: use tpidr_el2 on VHE hosts
  KVM: arm64: Stop save/restoring host tpidr_el1 on VHE
  Docs: dt: add devicetree binding for describing arm64 SDEI firmware
  firmware: arm_sdei: Add driver for Software Delegated Exceptions
  arm64: Add vmap_stack header file
  arm64: kernel: Add arch-specific SDEI entry code and CPU masking
  firmware: arm_sdei: Add support for CPU and system power states
  firmware: arm_sdei: add support for CPU private events
  arm64: acpi: Remove __init from acpi_psci_use_hvc() for use by SDEI
  firmware: arm_sdei: Discover SDEI support via ACPI

 .../devicetree/bindings/arm/firmware/sdei.txt      |   42 +
 arch/arm64/Kconfig                                 |    2 +-
 arch/arm64/include/asm/assembler.h                 |    8 +
 arch/arm64/include/asm/kvm_host.h                  |    2 +
 arch/arm64/include/asm/percpu.h                    |   11 +-
 arch/arm64/include/asm/processor.h                 |    1 +
 arch/arm64/include/asm/sdei.h                      |   63 ++
 arch/arm64/include/asm/stacktrace.h                |    3 +
 arch/arm64/include/asm/vmap_stack.h                |   41 +
 arch/arm64/kernel/Makefile                         |    1 +
 arch/arm64/kernel/acpi.c                           |    2 +-
 arch/arm64/kernel/asm-offsets.c                    |    5 +
 arch/arm64/kernel/cpufeature.c                     |   23 +
 arch/arm64/kernel/irq.c                            |   13 +-
 arch/arm64/kernel/sdei-entry.S                     |  122 +++
 arch/arm64/kernel/sdei.c                           |  235 +++++
 arch/arm64/kernel/smp.c                            |   11 +-
 arch/arm64/kvm/hyp-init.S                          |    4 +
 arch/arm64/kvm/hyp/entry.S                         |   10 +-
 arch/arm64/kvm/hyp/hyp-entry.S                     |   18 +-
 arch/arm64/kvm/hyp/switch.c                        |   25 +-
 arch/arm64/kvm/hyp/sysreg-sr.c                     |   16 +-
 arch/arm64/mm/proc.S                               |    8 +
 drivers/firmware/Kconfig                           |    8 +
 drivers/firmware/Makefile                          |    1 +
 drivers/firmware/arm_sdei.c                        | 1096 ++++++++++++++++++++
 include/linux/cpuhotplug.h                         |    1 +
 include/linux/psci.h                               |    3 +-
 include/linux/sdei.h                               |  105 ++
 include/uapi/linux/sdei.h                          |   91 ++
 virt/kvm/arm/arm.c                                 |   18 +-
 31 files changed, 1931 insertions(+), 58 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/arm/firmware/sdei.txt
 create mode 100644 arch/arm64/include/asm/sdei.h
 create mode 100644 arch/arm64/include/asm/vmap_stack.h
 create mode 100644 arch/arm64/kernel/sdei-entry.S
 create mode 100644 arch/arm64/kernel/sdei.c
 create mode 100644 drivers/firmware/arm_sdei.c
 create mode 100644 include/linux/sdei.h
 create mode 100644 include/uapi/linux/sdei.h

-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 00/13] arm64/firmware: Software Delegated Exception Interface

[James Morse]

Hello!

The Software Delegated Exception Interface (SDEI) is an ARM specification
for registering callbacks from the platform firmware into the OS.
This is intended to be used to implement firmware-first RAS notifications,
but also supports vendor-defined events and binding IRQs as events.

The document is here:
http://infocenter.arm.com/help/topic/com.arm.doc.den0054a/ARM_DEN0054A_Software_Delegated_Exception_Interface.pdf

I anticpate once reviewed this series will go via the arm-soc tree, as the
bulk of the code is under /drivers/firmware/.

The major change in v3 is due to VMAP stacks. We can't trust sp when we take
an event, so need our own SDEI stack. Events can nest, so we need two. Private
events can be delivered per-cpu, so we need two per-cpu SDEI stacks.

The APCICA support for the SDEI table is now in mainline, so this series has
grown the detection code for non-DT systems. The APEI/GHES wire-up is still
being held back as APEI's ghes_proc() isn't (yet) safe for multiple sources
of NMI.

Exposing the HVC range through KVM to user-space is gone, this will be
re-incarnated as something that better fits KVMs architecture emulation,
as opposed to assuming all-the-world's-SMC-CC.


This series (juggles some registers with KVM+VHE, then) adds a DT binding to
trigger probing of the interface and support for the SDEI API.

SDEI runs between adjacent exception levels, so events will always be delivered
to EL2 if firmware is at EL3. For VHE hosts we run the SDEI event handler
behind KVM's back with all exceptions masked. Once the handler has done its
work we return to the appropriate vbar+irq entry. This allows KVM to
world-switch and deliver any signals sent by the handler to Qemu/kvmtool. We
do the same thing if we interrupt host EL0. If we interrupted code with
interrupts masked, we use a different API call to return to the interrupted
context.

What about non-VHE KVM? If you don't have VHE support and boot at EL2, the
kernel drops to EL1. This driver will print an error message then give up. This
is because events would still be delivered to EL2 hitting either KVM, or the
hyp-stub. Supporting this is complicated, but because the main use-case is
RAS, and ARM v8.2's RAS extensions imply v8.1's Virtual Host Extensions, we
can assume all platforms with SDEI will support VHE too. (I have some ideas
on how to support non-VHE).

Running the event handler behind VHE-KVM's back has some side effects: The
event handler will blindly use any registers that are shared between the host
and guest. The two that I think matter are TPIDR_EL1, and the debug state. The
guest may have set MDSCR_EL1 so debug exceptions must remain masked. The
guest's TPIDR_EL1 will be used by the event handler if it accesses per-cpu
variables. This needs fixing. The first part of this series juggles KVMs use
of TPIDR_EL2 so that we share it with the host on VHE systems. An equivalent
change for 32bit is (still) on my todo list. (the alternative to this is to
have a parody world switch in the SDEI event handler, but this would mean
special casing interrupted guests, and be an ABI link to KVM.)

Is this another begins-with-S RAS mechanism for arm64? Yes.
Why? Any notification delivered as an exception will overwrite the exception
registers. This is fatal for the running thread if it happens during entry.S's
kernel_enter or kernel_exit. Instead of adding masking and routing controls,
events are delivered to a registered address at a fixed exception level and
don't change the exception registers when delivered.

This series can be retrieved from:
git://linux-arm.org/linux-jm.git -b sdei/v3/base

Questions and contradictions welcome!

Thanks,

James

[Changes since previous versions are noted on each patch]


James Morse (13):
  KVM: arm64: Store vcpu on the stack during __guest_enter()
  KVM: arm/arm64: Convert kvm_host_cpu_state to a static per-cpu
    allocation
  KVM: arm64: Change hyp_panic()s dependency on tpidr_el2
  arm64: alternatives: use tpidr_el2 on VHE hosts
  KVM: arm64: Stop save/restoring host tpidr_el1 on VHE
  Docs: dt: add devicetree binding for describing arm64 SDEI firmware
  firmware: arm_sdei: Add driver for Software Delegated Exceptions
  arm64: Add vmap_stack header file
  arm64: kernel: Add arch-specific SDEI entry code and CPU masking
  firmware: arm_sdei: Add support for CPU and system power states
  firmware: arm_sdei: add support for CPU private events
  arm64: acpi: Remove __init from acpi_psci_use_hvc() for use by SDEI
  firmware: arm_sdei: Discover SDEI support via ACPI

 .../devicetree/bindings/arm/firmware/sdei.txt      |   42 +
 arch/arm64/Kconfig                                 |    2 +-
 arch/arm64/include/asm/assembler.h                 |    8 +
 arch/arm64/include/asm/kvm_host.h                  |    2 +
 arch/arm64/include/asm/percpu.h                    |   11 +-
 arch/arm64/include/asm/processor.h                 |    1 +
 arch/arm64/include/asm/sdei.h                      |   63 ++
 arch/arm64/include/asm/stacktrace.h                |    3 +
 arch/arm64/include/asm/vmap_stack.h                |   41 +
 arch/arm64/kernel/Makefile                         |    1 +
 arch/arm64/kernel/acpi.c                           |    2 +-
 arch/arm64/kernel/asm-offsets.c                    |    5 +
 arch/arm64/kernel/cpufeature.c                     |   23 +
 arch/arm64/kernel/irq.c                            |   13 +-
 arch/arm64/kernel/sdei-entry.S                     |  122 +++
 arch/arm64/kernel/sdei.c                           |  235 +++++
 arch/arm64/kernel/smp.c                            |   11 +-
 arch/arm64/kvm/hyp-init.S                          |    4 +
 arch/arm64/kvm/hyp/entry.S                         |   10 +-
 arch/arm64/kvm/hyp/hyp-entry.S                     |   18 +-
 arch/arm64/kvm/hyp/switch.c                        |   25 +-
 arch/arm64/kvm/hyp/sysreg-sr.c                     |   16 +-
 arch/arm64/mm/proc.S                               |    8 +
 drivers/firmware/Kconfig                           |    8 +
 drivers/firmware/Makefile                          |    1 +
 drivers/firmware/arm_sdei.c                        | 1096 ++++++++++++++++++++
 include/linux/cpuhotplug.h                         |    1 +
 include/linux/psci.h                               |    3 +-
 include/linux/sdei.h                               |  105 ++
 include/uapi/linux/sdei.h                          |   91 ++
 virt/kvm/arm/arm.c                                 |   18 +-
 31 files changed, 1931 insertions(+), 58 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/arm/firmware/sdei.txt
 create mode 100644 arch/arm64/include/asm/sdei.h
 create mode 100644 arch/arm64/include/asm/vmap_stack.h
 create mode 100644 arch/arm64/kernel/sdei-entry.S
 create mode 100644 arch/arm64/kernel/sdei.c
 create mode 100644 drivers/firmware/arm_sdei.c
 create mode 100644 include/linux/sdei.h
 create mode 100644 include/uapi/linux/sdei.h

-- 
2.13.3

[PATCH v3 01/13] KVM: arm64: Store vcpu on the stack during __guest_enter()

[James Morse]

KVM uses tpidr_el2 as its private vcpu register, which makes sense for
non-vhe world switch as only KVM can access this register. This means
vhe Linux has to use tpidr_el1, which KVM has to save/restore as part
of the host context.

If the SDEI handler code runs behind KVMs back, it mustn't access any
per-cpu variables. To allow this on systems with vhe we need to make
the host use tpidr_el2, saving KVM from save/restoring it.

__guest_enter() stores the host_ctxt on the stack, do the same with
the vcpu.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
Reviewed-by: Christoffer Dall <cdall-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
Changes since v2:
 * Added middle paragraph of commit message.

 arch/arm64/kvm/hyp/entry.S     | 10 +++++++---
 arch/arm64/kvm/hyp/hyp-entry.S |  6 +++---
 2 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/arch/arm64/kvm/hyp/entry.S b/arch/arm64/kvm/hyp/entry.S
index 12ee62d6d410..9a8ab5dddd9e 100644
--- a/arch/arm64/kvm/hyp/entry.S
+++ b/arch/arm64/kvm/hyp/entry.S
@@ -62,8 +62,8 @@ ENTRY(__guest_enter)
 	// Store the host regs
 	save_callee_saved_regs x1
 
-	// Store the host_ctxt for use at exit time
-	str	x1, [sp, #-16]!
+	// Store host_ctxt and vcpu for use at exit time
+	stp	x1, x0, [sp, #-16]!
 
 	add	x18, x0, #VCPU_CONTEXT
 
@@ -159,6 +159,10 @@ abort_guest_exit_end:
 ENDPROC(__guest_exit)
 
 ENTRY(__fpsimd_guest_restore)
+	// x0: esr
+	// x1: vcpu
+	// x2-x29,lr: vcpu regs
+	// vcpu x0-x1 on the stack
 	stp	x2, x3, [sp, #-16]!
 	stp	x4, lr, [sp, #-16]!
 
@@ -173,7 +177,7 @@ alternative_else
 alternative_endif
 	isb
 
-	mrs	x3, tpidr_el2
+	mov	x3, x1
 
 	ldr	x0, [x3, #VCPU_HOST_CONTEXT]
 	kern_hyp_va x0
diff --git a/arch/arm64/kvm/hyp/hyp-entry.S b/arch/arm64/kvm/hyp/hyp-entry.S
index 5170ce1021da..fce7cc507e0a 100644
--- a/arch/arm64/kvm/hyp/hyp-entry.S
+++ b/arch/arm64/kvm/hyp/hyp-entry.S
@@ -104,6 +104,7 @@ el1_trap:
 	/*
 	 * x0: ESR_EC
 	 */
+	ldr	x1, [sp, #16 + 8]	// vcpu stored by __guest_enter
 
 	/*
 	 * We trap the first access to the FP/SIMD to save the host context
@@ -116,19 +117,18 @@ alternative_if_not ARM64_HAS_NO_FPSIMD
 	b.eq	__fpsimd_guest_restore
 alternative_else_nop_endif
 
-	mrs	x1, tpidr_el2
 	mov	x0, #ARM_EXCEPTION_TRAP
 	b	__guest_exit
 
 el1_irq:
 	stp     x0, x1, [sp, #-16]!
-	mrs	x1, tpidr_el2
+	ldr	x1, [sp, #16 + 8]
 	mov	x0, #ARM_EXCEPTION_IRQ
 	b	__guest_exit
 
 el1_error:
 	stp     x0, x1, [sp, #-16]!
-	mrs	x1, tpidr_el2
+	ldr	x1, [sp, #16 + 8]
 	mov	x0, #ARM_EXCEPTION_EL1_SERROR
 	b	__guest_exit
 
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 01/13] KVM: arm64: Store vcpu on the stack during __guest_enter()

[James Morse]

KVM uses tpidr_el2 as its private vcpu register, which makes sense for
non-vhe world switch as only KVM can access this register. This means
vhe Linux has to use tpidr_el1, which KVM has to save/restore as part
of the host context.

If the SDEI handler code runs behind KVMs back, it mustn't access any
per-cpu variables. To allow this on systems with vhe we need to make
the host use tpidr_el2, saving KVM from save/restoring it.

__guest_enter() stores the host_ctxt on the stack, do the same with
the vcpu.

Signed-off-by: James Morse <james.morse@arm.com>
Reviewed-by: Christoffer Dall <cdall@linaro.org>
---
Changes since v2:
 * Added middle paragraph of commit message.

 arch/arm64/kvm/hyp/entry.S     | 10 +++++++---
 arch/arm64/kvm/hyp/hyp-entry.S |  6 +++---
 2 files changed, 10 insertions(+), 6 deletions(-)

diff --git a/arch/arm64/kvm/hyp/entry.S b/arch/arm64/kvm/hyp/entry.S
index 12ee62d6d410..9a8ab5dddd9e 100644
--- a/arch/arm64/kvm/hyp/entry.S
+++ b/arch/arm64/kvm/hyp/entry.S
@@ -62,8 +62,8 @@ ENTRY(__guest_enter)
 	// Store the host regs
 	save_callee_saved_regs x1
 
-	// Store the host_ctxt for use at exit time
-	str	x1, [sp, #-16]!
+	// Store host_ctxt and vcpu for use at exit time
+	stp	x1, x0, [sp, #-16]!
 
 	add	x18, x0, #VCPU_CONTEXT
 
@@ -159,6 +159,10 @@ abort_guest_exit_end:
 ENDPROC(__guest_exit)
 
 ENTRY(__fpsimd_guest_restore)
+	// x0: esr
+	// x1: vcpu
+	// x2-x29,lr: vcpu regs
+	// vcpu x0-x1 on the stack
 	stp	x2, x3, [sp, #-16]!
 	stp	x4, lr, [sp, #-16]!
 
@@ -173,7 +177,7 @@ alternative_else
 alternative_endif
 	isb
 
-	mrs	x3, tpidr_el2
+	mov	x3, x1
 
 	ldr	x0, [x3, #VCPU_HOST_CONTEXT]
 	kern_hyp_va x0
diff --git a/arch/arm64/kvm/hyp/hyp-entry.S b/arch/arm64/kvm/hyp/hyp-entry.S
index 5170ce1021da..fce7cc507e0a 100644
--- a/arch/arm64/kvm/hyp/hyp-entry.S
+++ b/arch/arm64/kvm/hyp/hyp-entry.S
@@ -104,6 +104,7 @@ el1_trap:
 	/*
 	 * x0: ESR_EC
 	 */
+	ldr	x1, [sp, #16 + 8]	// vcpu stored by __guest_enter
 
 	/*
 	 * We trap the first access to the FP/SIMD to save the host context
@@ -116,19 +117,18 @@ alternative_if_not ARM64_HAS_NO_FPSIMD
 	b.eq	__fpsimd_guest_restore
 alternative_else_nop_endif
 
-	mrs	x1, tpidr_el2
 	mov	x0, #ARM_EXCEPTION_TRAP
 	b	__guest_exit
 
 el1_irq:
 	stp     x0, x1, [sp, #-16]!
-	mrs	x1, tpidr_el2
+	ldr	x1, [sp, #16 + 8]
 	mov	x0, #ARM_EXCEPTION_IRQ
 	b	__guest_exit
 
 el1_error:
 	stp     x0, x1, [sp, #-16]!
-	mrs	x1, tpidr_el2
+	ldr	x1, [sp, #16 + 8]
 	mov	x0, #ARM_EXCEPTION_EL1_SERROR
 	b	__guest_exit
 
-- 
2.13.3

[PATCH v3 02/13] KVM: arm/arm64: Convert kvm_host_cpu_state to a static per-cpu allocation

[James Morse]

kvm_host_cpu_state is a per-cpu allocation made from kvm_arch_init()
used to store the host EL1 registers when KVM switches to a guest.

Make it easier for ASM to generate pointers into this per-cpu memory
by making it a static allocation.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
Acked-by: Christoffer Dall <cdall-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
 virt/kvm/arm/arm.c | 18 +++---------------
 1 file changed, 3 insertions(+), 15 deletions(-)

diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
index b9f68e4add71..3786dd7e277d 100644
--- a/virt/kvm/arm/arm.c
+++ b/virt/kvm/arm/arm.c
@@ -51,8 +51,8 @@
 __asm__(".arch_extension	virt");
 #endif
 
+DEFINE_PER_CPU(kvm_cpu_context_t, kvm_host_cpu_state);
 static DEFINE_PER_CPU(unsigned long, kvm_arm_hyp_stack_page);
-static kvm_cpu_context_t __percpu *kvm_host_cpu_state;
 
 /* Per-CPU variable containing the currently running vcpu. */
 static DEFINE_PER_CPU(struct kvm_vcpu *, kvm_arm_running_vcpu);
@@ -351,7 +351,7 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	}
 
 	vcpu->cpu = cpu;
-	vcpu->arch.host_cpu_context = this_cpu_ptr(kvm_host_cpu_state);
+	vcpu->arch.host_cpu_context = this_cpu_ptr(&kvm_host_cpu_state);
 
 	kvm_arm_set_running_vcpu(vcpu);
 
@@ -1255,19 +1255,8 @@ static inline void hyp_cpu_pm_exit(void)
 }
 #endif
 
-static void teardown_common_resources(void)
-{
-	free_percpu(kvm_host_cpu_state);
-}
-
 static int init_common_resources(void)
 {
-	kvm_host_cpu_state = alloc_percpu(kvm_cpu_context_t);
-	if (!kvm_host_cpu_state) {
-		kvm_err("Cannot allocate host CPU state\n");
-		return -ENOMEM;
-	}
-
 	/* set size of VMID supported by CPU */
 	kvm_vmid_bits = kvm_get_vmid_bits();
 	kvm_info("%d-bit VMID\n", kvm_vmid_bits);
@@ -1412,7 +1401,7 @@ static int init_hyp_mode(void)
 	for_each_possible_cpu(cpu) {
 		kvm_cpu_context_t *cpu_ctxt;
 
-		cpu_ctxt = per_cpu_ptr(kvm_host_cpu_state, cpu);
+		cpu_ctxt = per_cpu_ptr(&kvm_host_cpu_state, cpu);
 		err = create_hyp_mappings(cpu_ctxt, cpu_ctxt + 1, PAGE_HYP);
 
 		if (err) {
@@ -1490,7 +1479,6 @@ int kvm_arch_init(void *opaque)
 out_hyp:
 	teardown_hyp_mode();
 out_err:
-	teardown_common_resources();
 	return err;
 }
 
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 02/13] KVM: arm/arm64: Convert kvm_host_cpu_state to a static per-cpu allocation

[James Morse]

kvm_host_cpu_state is a per-cpu allocation made from kvm_arch_init()
used to store the host EL1 registers when KVM switches to a guest.

Make it easier for ASM to generate pointers into this per-cpu memory
by making it a static allocation.

Signed-off-by: James Morse <james.morse@arm.com>
Acked-by: Christoffer Dall <cdall@linaro.org>
---
 virt/kvm/arm/arm.c | 18 +++---------------
 1 file changed, 3 insertions(+), 15 deletions(-)

diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c
index b9f68e4add71..3786dd7e277d 100644
--- a/virt/kvm/arm/arm.c
+++ b/virt/kvm/arm/arm.c
@@ -51,8 +51,8 @@
 __asm__(".arch_extension	virt");
 #endif
 
+DEFINE_PER_CPU(kvm_cpu_context_t, kvm_host_cpu_state);
 static DEFINE_PER_CPU(unsigned long, kvm_arm_hyp_stack_page);
-static kvm_cpu_context_t __percpu *kvm_host_cpu_state;
 
 /* Per-CPU variable containing the currently running vcpu. */
 static DEFINE_PER_CPU(struct kvm_vcpu *, kvm_arm_running_vcpu);
@@ -351,7 +351,7 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
 	}
 
 	vcpu->cpu = cpu;
-	vcpu->arch.host_cpu_context = this_cpu_ptr(kvm_host_cpu_state);
+	vcpu->arch.host_cpu_context = this_cpu_ptr(&kvm_host_cpu_state);
 
 	kvm_arm_set_running_vcpu(vcpu);
 
@@ -1255,19 +1255,8 @@ static inline void hyp_cpu_pm_exit(void)
 }
 #endif
 
-static void teardown_common_resources(void)
-{
-	free_percpu(kvm_host_cpu_state);
-}
-
 static int init_common_resources(void)
 {
-	kvm_host_cpu_state = alloc_percpu(kvm_cpu_context_t);
-	if (!kvm_host_cpu_state) {
-		kvm_err("Cannot allocate host CPU state\n");
-		return -ENOMEM;
-	}
-
 	/* set size of VMID supported by CPU */
 	kvm_vmid_bits = kvm_get_vmid_bits();
 	kvm_info("%d-bit VMID\n", kvm_vmid_bits);
@@ -1412,7 +1401,7 @@ static int init_hyp_mode(void)
 	for_each_possible_cpu(cpu) {
 		kvm_cpu_context_t *cpu_ctxt;
 
-		cpu_ctxt = per_cpu_ptr(kvm_host_cpu_state, cpu);
+		cpu_ctxt = per_cpu_ptr(&kvm_host_cpu_state, cpu);
 		err = create_hyp_mappings(cpu_ctxt, cpu_ctxt + 1, PAGE_HYP);
 
 		if (err) {
@@ -1490,7 +1479,6 @@ int kvm_arch_init(void *opaque)
 out_hyp:
 	teardown_hyp_mode();
 out_err:
-	teardown_common_resources();
 	return err;
 }
 
-- 
2.13.3

[PATCH v3 03/13] KVM: arm64: Change hyp_panic()s dependency on tpidr_el2

[James Morse]

Make tpidr_el2 a cpu-offset for per-cpu variables in the same way the
host uses tpidr_el1. This lets tpidr_el{1,2} have the same value, and
on VHE they can be the same register.

KVM calls hyp_panic() when anything unexpected happens. This may occur
while a guest owns the EL1 registers. KVM stashes the vcpu pointer in
tpidr_el2, which it uses to find the host context in order to restore
the host EL1 registers before parachuting into the host's panic().

The host context is a struct kvm_cpu_context allocated in the per-cpu
area, and mapped to hyp. Given the per-cpu offset for this CPU, this is
easy to find. Change hyp_panic() to take a pointer to the
struct kvm_cpu_context. Wrap these calls with an asm function that
retrieves the struct kvm_cpu_context from the host's per-cpu area.

Copy the per-cpu offset from the hosts tpidr_el1 into tpidr_el2 during
kvm init. (Later patches will make this unnecessary for VHE hosts)

We print out the vcpu pointer as part of the panic message. Add a back
reference to the 'running vcpu' in the host cpu context to preserve this.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
Reviewed-by: Christoffer Dall <cdall-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
Changes since v1:
 * Added a comment explaining how =kvm_host_cpu_state gets from a host-va
   to a hyp va.
 * Added the first paragraph to the commit message.

 arch/arm64/include/asm/kvm_host.h |  2 ++
 arch/arm64/kvm/hyp/hyp-entry.S    | 12 ++++++++++++
 arch/arm64/kvm/hyp/s2-setup.c     |  3 +++
 arch/arm64/kvm/hyp/switch.c       | 25 +++++++++++++------------
 4 files changed, 30 insertions(+), 12 deletions(-)

diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index e923b58606e2..806ccef7470a 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -191,6 +191,8 @@ struct kvm_cpu_context {
 		u64 sys_regs[NR_SYS_REGS];
 		u32 copro[NR_COPRO_REGS];
 	};
+
+	struct kvm_vcpu *__hyp_running_vcpu;
 };
 
 typedef struct kvm_cpu_context kvm_cpu_context_t;
diff --git a/arch/arm64/kvm/hyp/hyp-entry.S b/arch/arm64/kvm/hyp/hyp-entry.S
index fce7cc507e0a..e4f37b9dd47c 100644
--- a/arch/arm64/kvm/hyp/hyp-entry.S
+++ b/arch/arm64/kvm/hyp/hyp-entry.S
@@ -163,6 +163,18 @@ ENTRY(__hyp_do_panic)
 	eret
 ENDPROC(__hyp_do_panic)
 
+ENTRY(__hyp_panic)
+	/*
+	 * '=kvm_host_cpu_state' is a host VA from the constant pool, it may
+	 * not be accessible by this address from EL2, hyp_panic() converts
+	 * it with kern_hyp_va() before use.
+	 */
+	ldr	x0, =kvm_host_cpu_state
+	mrs	x1, tpidr_el2
+	add	x0, x0, x1
+	b	hyp_panic
+ENDPROC(__hyp_panic)
+
 .macro invalid_vector	label, target = __hyp_panic
 	.align	2
 \label:
diff --git a/arch/arm64/kvm/hyp/s2-setup.c b/arch/arm64/kvm/hyp/s2-setup.c
index a81f5e10fc8c..7fb88274eba1 100644
--- a/arch/arm64/kvm/hyp/s2-setup.c
+++ b/arch/arm64/kvm/hyp/s2-setup.c
@@ -84,5 +84,8 @@ u32 __hyp_text __init_stage2_translation(void)
 
 	write_sysreg(val, vtcr_el2);
 
+	/* copy tpidr_el1 into tpidr_el2 for use by HYP */
+	write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
+
 	return parange;
 }
diff --git a/arch/arm64/kvm/hyp/switch.c b/arch/arm64/kvm/hyp/switch.c
index 945e79c641c4..235d615cee30 100644
--- a/arch/arm64/kvm/hyp/switch.c
+++ b/arch/arm64/kvm/hyp/switch.c
@@ -286,9 +286,9 @@ int __hyp_text __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	u64 exit_code;
 
 	vcpu = kern_hyp_va(vcpu);
-	write_sysreg(vcpu, tpidr_el2);
 
 	host_ctxt = kern_hyp_va(vcpu->arch.host_cpu_context);
+	host_ctxt->__hyp_running_vcpu = vcpu;
 	guest_ctxt = &vcpu->arch.ctxt;
 
 	__sysreg_save_host_state(host_ctxt);
@@ -393,7 +393,8 @@ int __hyp_text __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 
 static const char __hyp_panic_string[] = "HYP panic:\nPS:%08llx PC:%016llx ESR:%08llx\nFAR:%016llx HPFAR:%016llx PAR:%016llx\nVCPU:%p\n";
 
-static void __hyp_text __hyp_call_panic_nvhe(u64 spsr, u64 elr, u64 par)
+static void __hyp_text __hyp_call_panic_nvhe(u64 spsr, u64 elr, u64 par,
+					     struct kvm_vcpu *vcpu)
 {
 	unsigned long str_va;
 
@@ -407,35 +408,35 @@ static void __hyp_text __hyp_call_panic_nvhe(u64 spsr, u64 elr, u64 par)
 	__hyp_do_panic(str_va,
 		       spsr,  elr,
 		       read_sysreg(esr_el2),   read_sysreg_el2(far),
-		       read_sysreg(hpfar_el2), par,
-		       (void *)read_sysreg(tpidr_el2));
+		       read_sysreg(hpfar_el2), par, vcpu);
 }
 
-static void __hyp_text __hyp_call_panic_vhe(u64 spsr, u64 elr, u64 par)
+static void __hyp_text __hyp_call_panic_vhe(u64 spsr, u64 elr, u64 par,
+					    struct kvm_vcpu *vcpu)
 {
 	panic(__hyp_panic_string,
 	      spsr,  elr,
 	      read_sysreg_el2(esr),   read_sysreg_el2(far),
-	      read_sysreg(hpfar_el2), par,
-	      (void *)read_sysreg(tpidr_el2));
+	      read_sysreg(hpfar_el2), par, vcpu);
 }
 
 static hyp_alternate_select(__hyp_call_panic,
 			    __hyp_call_panic_nvhe, __hyp_call_panic_vhe,
 			    ARM64_HAS_VIRT_HOST_EXTN);
 
-void __hyp_text __noreturn __hyp_panic(void)
+void __hyp_text __noreturn hyp_panic(struct kvm_cpu_context *__host_ctxt)
 {
+	struct kvm_vcpu *vcpu = NULL;
+
 	u64 spsr = read_sysreg_el2(spsr);
 	u64 elr = read_sysreg_el2(elr);
 	u64 par = read_sysreg(par_el1);
 
 	if (read_sysreg(vttbr_el2)) {
-		struct kvm_vcpu *vcpu;
 		struct kvm_cpu_context *host_ctxt;
 
-		vcpu = (struct kvm_vcpu *)read_sysreg(tpidr_el2);
-		host_ctxt = kern_hyp_va(vcpu->arch.host_cpu_context);
+		host_ctxt = kern_hyp_va(__host_ctxt);
+		vcpu = host_ctxt->__hyp_running_vcpu;
 		__timer_save_state(vcpu);
 		__deactivate_traps(vcpu);
 		__deactivate_vm(vcpu);
@@ -443,7 +444,7 @@ void __hyp_text __noreturn __hyp_panic(void)
 	}
 
 	/* Call panic for real */
-	__hyp_call_panic()(spsr, elr, par);
+	__hyp_call_panic()(spsr, elr, par, vcpu);
 
 	unreachable();
 }
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 03/13] KVM: arm64: Change hyp_panic()s dependency on tpidr_el2

[James Morse]

Make tpidr_el2 a cpu-offset for per-cpu variables in the same way the
host uses tpidr_el1. This lets tpidr_el{1,2} have the same value, and
on VHE they can be the same register.

KVM calls hyp_panic() when anything unexpected happens. This may occur
while a guest owns the EL1 registers. KVM stashes the vcpu pointer in
tpidr_el2, which it uses to find the host context in order to restore
the host EL1 registers before parachuting into the host's panic().

The host context is a struct kvm_cpu_context allocated in the per-cpu
area, and mapped to hyp. Given the per-cpu offset for this CPU, this is
easy to find. Change hyp_panic() to take a pointer to the
struct kvm_cpu_context. Wrap these calls with an asm function that
retrieves the struct kvm_cpu_context from the host's per-cpu area.

Copy the per-cpu offset from the hosts tpidr_el1 into tpidr_el2 during
kvm init. (Later patches will make this unnecessary for VHE hosts)

We print out the vcpu pointer as part of the panic message. Add a back
reference to the 'running vcpu' in the host cpu context to preserve this.

Signed-off-by: James Morse <james.morse@arm.com>
Reviewed-by: Christoffer Dall <cdall@linaro.org>
---
Changes since v1:
 * Added a comment explaining how =kvm_host_cpu_state gets from a host-va
   to a hyp va.
 * Added the first paragraph to the commit message.

 arch/arm64/include/asm/kvm_host.h |  2 ++
 arch/arm64/kvm/hyp/hyp-entry.S    | 12 ++++++++++++
 arch/arm64/kvm/hyp/s2-setup.c     |  3 +++
 arch/arm64/kvm/hyp/switch.c       | 25 +++++++++++++------------
 4 files changed, 30 insertions(+), 12 deletions(-)

diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index e923b58606e2..806ccef7470a 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -191,6 +191,8 @@ struct kvm_cpu_context {
 		u64 sys_regs[NR_SYS_REGS];
 		u32 copro[NR_COPRO_REGS];
 	};
+
+	struct kvm_vcpu *__hyp_running_vcpu;
 };
 
 typedef struct kvm_cpu_context kvm_cpu_context_t;
diff --git a/arch/arm64/kvm/hyp/hyp-entry.S b/arch/arm64/kvm/hyp/hyp-entry.S
index fce7cc507e0a..e4f37b9dd47c 100644
--- a/arch/arm64/kvm/hyp/hyp-entry.S
+++ b/arch/arm64/kvm/hyp/hyp-entry.S
@@ -163,6 +163,18 @@ ENTRY(__hyp_do_panic)
 	eret
 ENDPROC(__hyp_do_panic)
 
+ENTRY(__hyp_panic)
+	/*
+	 * '=kvm_host_cpu_state' is a host VA from the constant pool, it may
+	 * not be accessible by this address from EL2, hyp_panic() converts
+	 * it with kern_hyp_va() before use.
+	 */
+	ldr	x0, =kvm_host_cpu_state
+	mrs	x1, tpidr_el2
+	add	x0, x0, x1
+	b	hyp_panic
+ENDPROC(__hyp_panic)
+
 .macro invalid_vector	label, target = __hyp_panic
 	.align	2
 \label:
diff --git a/arch/arm64/kvm/hyp/s2-setup.c b/arch/arm64/kvm/hyp/s2-setup.c
index a81f5e10fc8c..7fb88274eba1 100644
--- a/arch/arm64/kvm/hyp/s2-setup.c
+++ b/arch/arm64/kvm/hyp/s2-setup.c
@@ -84,5 +84,8 @@ u32 __hyp_text __init_stage2_translation(void)
 
 	write_sysreg(val, vtcr_el2);
 
+	/* copy tpidr_el1 into tpidr_el2 for use by HYP */
+	write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
+
 	return parange;
 }
diff --git a/arch/arm64/kvm/hyp/switch.c b/arch/arm64/kvm/hyp/switch.c
index 945e79c641c4..235d615cee30 100644
--- a/arch/arm64/kvm/hyp/switch.c
+++ b/arch/arm64/kvm/hyp/switch.c
@@ -286,9 +286,9 @@ int __hyp_text __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 	u64 exit_code;
 
 	vcpu = kern_hyp_va(vcpu);
-	write_sysreg(vcpu, tpidr_el2);
 
 	host_ctxt = kern_hyp_va(vcpu->arch.host_cpu_context);
+	host_ctxt->__hyp_running_vcpu = vcpu;
 	guest_ctxt = &vcpu->arch.ctxt;
 
 	__sysreg_save_host_state(host_ctxt);
@@ -393,7 +393,8 @@ int __hyp_text __kvm_vcpu_run(struct kvm_vcpu *vcpu)
 
 static const char __hyp_panic_string[] = "HYP panic:\nPS:%08llx PC:%016llx ESR:%08llx\nFAR:%016llx HPFAR:%016llx PAR:%016llx\nVCPU:%p\n";
 
-static void __hyp_text __hyp_call_panic_nvhe(u64 spsr, u64 elr, u64 par)
+static void __hyp_text __hyp_call_panic_nvhe(u64 spsr, u64 elr, u64 par,
+					     struct kvm_vcpu *vcpu)
 {
 	unsigned long str_va;
 
@@ -407,35 +408,35 @@ static void __hyp_text __hyp_call_panic_nvhe(u64 spsr, u64 elr, u64 par)
 	__hyp_do_panic(str_va,
 		       spsr,  elr,
 		       read_sysreg(esr_el2),   read_sysreg_el2(far),
-		       read_sysreg(hpfar_el2), par,
-		       (void *)read_sysreg(tpidr_el2));
+		       read_sysreg(hpfar_el2), par, vcpu);
 }
 
-static void __hyp_text __hyp_call_panic_vhe(u64 spsr, u64 elr, u64 par)
+static void __hyp_text __hyp_call_panic_vhe(u64 spsr, u64 elr, u64 par,
+					    struct kvm_vcpu *vcpu)
 {
 	panic(__hyp_panic_string,
 	      spsr,  elr,
 	      read_sysreg_el2(esr),   read_sysreg_el2(far),
-	      read_sysreg(hpfar_el2), par,
-	      (void *)read_sysreg(tpidr_el2));
+	      read_sysreg(hpfar_el2), par, vcpu);
 }
 
 static hyp_alternate_select(__hyp_call_panic,
 			    __hyp_call_panic_nvhe, __hyp_call_panic_vhe,
 			    ARM64_HAS_VIRT_HOST_EXTN);
 
-void __hyp_text __noreturn __hyp_panic(void)
+void __hyp_text __noreturn hyp_panic(struct kvm_cpu_context *__host_ctxt)
 {
+	struct kvm_vcpu *vcpu = NULL;
+
 	u64 spsr = read_sysreg_el2(spsr);
 	u64 elr = read_sysreg_el2(elr);
 	u64 par = read_sysreg(par_el1);
 
 	if (read_sysreg(vttbr_el2)) {
-		struct kvm_vcpu *vcpu;
 		struct kvm_cpu_context *host_ctxt;
 
-		vcpu = (struct kvm_vcpu *)read_sysreg(tpidr_el2);
-		host_ctxt = kern_hyp_va(vcpu->arch.host_cpu_context);
+		host_ctxt = kern_hyp_va(__host_ctxt);
+		vcpu = host_ctxt->__hyp_running_vcpu;
 		__timer_save_state(vcpu);
 		__deactivate_traps(vcpu);
 		__deactivate_vm(vcpu);
@@ -443,7 +444,7 @@ void __hyp_text __noreturn __hyp_panic(void)
 	}
 
 	/* Call panic for real */
-	__hyp_call_panic()(spsr, elr, par);
+	__hyp_call_panic()(spsr, elr, par, vcpu);
 
 	unreachable();
 }
-- 
2.13.3

[PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[James Morse]

Now that KVM uses tpidr_el2 in the same way as Linux's cpu_offset in
tpidr_el1, merge the two. This saves KVM from save/restoring tpidr_el1
on VHE hosts, and allows future code to blindly access per-cpu variables
without triggering world-switch.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
Reviewed-by: Christoffer Dall <cdall-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
Changes since v1:
 * cpu_copy_el2regs()'s 'have I been patched' test now always sets a register,
   just in case the compiler optimises out part of the logic.

 arch/arm64/include/asm/assembler.h |  8 ++++++++
 arch/arm64/include/asm/percpu.h    | 11 +++++++++--
 arch/arm64/include/asm/processor.h |  1 +
 arch/arm64/kernel/cpufeature.c     | 23 +++++++++++++++++++++++
 arch/arm64/mm/proc.S               |  8 ++++++++
 5 files changed, 49 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/include/asm/assembler.h b/arch/arm64/include/asm/assembler.h
index d58a6253c6ab..1ba12f59dec0 100644
--- a/arch/arm64/include/asm/assembler.h
+++ b/arch/arm64/include/asm/assembler.h
@@ -242,7 +242,11 @@ lr	.req	x30		// link register
 #else
 	adr_l	\dst, \sym
 #endif
+alternative_if_not ARM64_HAS_VIRT_HOST_EXTN
 	mrs	\tmp, tpidr_el1
+alternative_else
+	mrs	\tmp, tpidr_el2
+alternative_endif
 	add	\dst, \dst, \tmp
 	.endm
 
@@ -253,7 +257,11 @@ lr	.req	x30		// link register
 	 */
 	.macro ldr_this_cpu dst, sym, tmp
 	adr_l	\dst, \sym
+alternative_if_not ARM64_HAS_VIRT_HOST_EXTN
 	mrs	\tmp, tpidr_el1
+alternative_else
+	mrs	\tmp, tpidr_el2
+alternative_endif
 	ldr	\dst, [\dst, \tmp]
 	.endm
 
diff --git a/arch/arm64/include/asm/percpu.h b/arch/arm64/include/asm/percpu.h
index 3bd498e4de4c..43393208229e 100644
--- a/arch/arm64/include/asm/percpu.h
+++ b/arch/arm64/include/asm/percpu.h
@@ -16,11 +16,15 @@
 #ifndef __ASM_PERCPU_H
 #define __ASM_PERCPU_H
 
+#include <asm/alternative.h>
 #include <asm/stack_pointer.h>
 
 static inline void set_my_cpu_offset(unsigned long off)
 {
-	asm volatile("msr tpidr_el1, %0" :: "r" (off) : "memory");
+	asm volatile(ALTERNATIVE("msr tpidr_el1, %0",
+				 "msr tpidr_el2, %0",
+				 ARM64_HAS_VIRT_HOST_EXTN)
+			:: "r" (off) : "memory");
 }
 
 static inline unsigned long __my_cpu_offset(void)
@@ -31,7 +35,10 @@ static inline unsigned long __my_cpu_offset(void)
 	 * We want to allow caching the value, so avoid using volatile and
 	 * instead use a fake stack read to hazard against barrier().
 	 */
-	asm("mrs %0, tpidr_el1" : "=r" (off) :
+	asm(ALTERNATIVE("mrs %0, tpidr_el1",
+			"mrs %0, tpidr_el2",
+			ARM64_HAS_VIRT_HOST_EXTN)
+		: "=r" (off) :
 		"Q" (*(const unsigned long *)current_stack_pointer));
 
 	return off;
diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h
index 29adab8138c3..8f2d0f7d193b 100644
--- a/arch/arm64/include/asm/processor.h
+++ b/arch/arm64/include/asm/processor.h
@@ -193,5 +193,6 @@ static inline void spin_lock_prefetch(const void *ptr)
 
 int cpu_enable_pan(void *__unused);
 int cpu_enable_cache_maint_trap(void *__unused);
+int cpu_copy_el2regs(void *__unused);
 
 #endif /* __ASM_PROCESSOR_H */
diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index cd52d365d1f0..8e4c7da2b126 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -865,6 +865,7 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
 		.capability = ARM64_HAS_VIRT_HOST_EXTN,
 		.def_scope = SCOPE_SYSTEM,
 		.matches = runs_at_el2,
+		.enable = cpu_copy_el2regs,
 	},
 	{
 		.desc = "32-bit EL0 Support",
@@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
 }
 
 late_initcall(enable_mrs_emulation);
+
+int cpu_copy_el2regs(void *__unused)
+{
+	int do_copyregs = 0;
+
+	/*
+	 * Copy register values that aren't redirected by hardware.
+	 *
+	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
+	 * this value to tpidr_el2 before we patch the code. Once we've done
+	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
+	 * do anything here.
+	 */
+	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
+				 ARM64_HAS_VIRT_HOST_EXTN)
+		    : "=r" (do_copyregs) : : );
+
+	if (do_copyregs)
+		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
+
+	return 0;
+}
diff --git a/arch/arm64/mm/proc.S b/arch/arm64/mm/proc.S
index 877d42fb0df6..109d43a15aaf 100644
--- a/arch/arm64/mm/proc.S
+++ b/arch/arm64/mm/proc.S
@@ -70,7 +70,11 @@ ENTRY(cpu_do_suspend)
 	mrs	x8, mdscr_el1
 	mrs	x9, oslsr_el1
 	mrs	x10, sctlr_el1
+alternative_if_not ARM64_HAS_VIRT_HOST_EXTN
 	mrs	x11, tpidr_el1
+alternative_else
+	mrs	x11, tpidr_el2
+alternative_endif
 	mrs	x12, sp_el0
 	stp	x2, x3, [x0]
 	stp	x4, xzr, [x0, #16]
@@ -116,7 +120,11 @@ ENTRY(cpu_do_resume)
 	msr	mdscr_el1, x10
 
 	msr	sctlr_el1, x12
+alternative_if_not ARM64_HAS_VIRT_HOST_EXTN
 	msr	tpidr_el1, x13
+alternative_else
+	msr	tpidr_el2, x13
+alternative_endif
 	msr	sp_el0, x14
 	/*
 	 * Restore oslsr_el1 by writing oslar_el1
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[James Morse]

Now that KVM uses tpidr_el2 in the same way as Linux's cpu_offset in
tpidr_el1, merge the two. This saves KVM from save/restoring tpidr_el1
on VHE hosts, and allows future code to blindly access per-cpu variables
without triggering world-switch.

Signed-off-by: James Morse <james.morse@arm.com>
Reviewed-by: Christoffer Dall <cdall@linaro.org>
---
Changes since v1:
 * cpu_copy_el2regs()'s 'have I been patched' test now always sets a register,
   just in case the compiler optimises out part of the logic.

 arch/arm64/include/asm/assembler.h |  8 ++++++++
 arch/arm64/include/asm/percpu.h    | 11 +++++++++--
 arch/arm64/include/asm/processor.h |  1 +
 arch/arm64/kernel/cpufeature.c     | 23 +++++++++++++++++++++++
 arch/arm64/mm/proc.S               |  8 ++++++++
 5 files changed, 49 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/include/asm/assembler.h b/arch/arm64/include/asm/assembler.h
index d58a6253c6ab..1ba12f59dec0 100644
--- a/arch/arm64/include/asm/assembler.h
+++ b/arch/arm64/include/asm/assembler.h
@@ -242,7 +242,11 @@ lr	.req	x30		// link register
 #else
 	adr_l	\dst, \sym
 #endif
+alternative_if_not ARM64_HAS_VIRT_HOST_EXTN
 	mrs	\tmp, tpidr_el1
+alternative_else
+	mrs	\tmp, tpidr_el2
+alternative_endif
 	add	\dst, \dst, \tmp
 	.endm
 
@@ -253,7 +257,11 @@ lr	.req	x30		// link register
 	 */
 	.macro ldr_this_cpu dst, sym, tmp
 	adr_l	\dst, \sym
+alternative_if_not ARM64_HAS_VIRT_HOST_EXTN
 	mrs	\tmp, tpidr_el1
+alternative_else
+	mrs	\tmp, tpidr_el2
+alternative_endif
 	ldr	\dst, [\dst, \tmp]
 	.endm
 
diff --git a/arch/arm64/include/asm/percpu.h b/arch/arm64/include/asm/percpu.h
index 3bd498e4de4c..43393208229e 100644
--- a/arch/arm64/include/asm/percpu.h
+++ b/arch/arm64/include/asm/percpu.h
@@ -16,11 +16,15 @@
 #ifndef __ASM_PERCPU_H
 #define __ASM_PERCPU_H
 
+#include <asm/alternative.h>
 #include <asm/stack_pointer.h>
 
 static inline void set_my_cpu_offset(unsigned long off)
 {
-	asm volatile("msr tpidr_el1, %0" :: "r" (off) : "memory");
+	asm volatile(ALTERNATIVE("msr tpidr_el1, %0",
+				 "msr tpidr_el2, %0",
+				 ARM64_HAS_VIRT_HOST_EXTN)
+			:: "r" (off) : "memory");
 }
 
 static inline unsigned long __my_cpu_offset(void)
@@ -31,7 +35,10 @@ static inline unsigned long __my_cpu_offset(void)
 	 * We want to allow caching the value, so avoid using volatile and
 	 * instead use a fake stack read to hazard against barrier().
 	 */
-	asm("mrs %0, tpidr_el1" : "=r" (off) :
+	asm(ALTERNATIVE("mrs %0, tpidr_el1",
+			"mrs %0, tpidr_el2",
+			ARM64_HAS_VIRT_HOST_EXTN)
+		: "=r" (off) :
 		"Q" (*(const unsigned long *)current_stack_pointer));
 
 	return off;
diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h
index 29adab8138c3..8f2d0f7d193b 100644
--- a/arch/arm64/include/asm/processor.h
+++ b/arch/arm64/include/asm/processor.h
@@ -193,5 +193,6 @@ static inline void spin_lock_prefetch(const void *ptr)
 
 int cpu_enable_pan(void *__unused);
 int cpu_enable_cache_maint_trap(void *__unused);
+int cpu_copy_el2regs(void *__unused);
 
 #endif /* __ASM_PROCESSOR_H */
diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index cd52d365d1f0..8e4c7da2b126 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -865,6 +865,7 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
 		.capability = ARM64_HAS_VIRT_HOST_EXTN,
 		.def_scope = SCOPE_SYSTEM,
 		.matches = runs_at_el2,
+		.enable = cpu_copy_el2regs,
 	},
 	{
 		.desc = "32-bit EL0 Support",
@@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
 }
 
 late_initcall(enable_mrs_emulation);
+
+int cpu_copy_el2regs(void *__unused)
+{
+	int do_copyregs = 0;
+
+	/*
+	 * Copy register values that aren't redirected by hardware.
+	 *
+	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
+	 * this value to tpidr_el2 before we patch the code. Once we've done
+	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
+	 * do anything here.
+	 */
+	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
+				 ARM64_HAS_VIRT_HOST_EXTN)
+		    : "=r" (do_copyregs) : : );
+
+	if (do_copyregs)
+		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
+
+	return 0;
+}
diff --git a/arch/arm64/mm/proc.S b/arch/arm64/mm/proc.S
index 877d42fb0df6..109d43a15aaf 100644
--- a/arch/arm64/mm/proc.S
+++ b/arch/arm64/mm/proc.S
@@ -70,7 +70,11 @@ ENTRY(cpu_do_suspend)
 	mrs	x8, mdscr_el1
 	mrs	x9, oslsr_el1
 	mrs	x10, sctlr_el1
+alternative_if_not ARM64_HAS_VIRT_HOST_EXTN
 	mrs	x11, tpidr_el1
+alternative_else
+	mrs	x11, tpidr_el2
+alternative_endif
 	mrs	x12, sp_el0
 	stp	x2, x3, [x0]
 	stp	x4, xzr, [x0, #16]
@@ -116,7 +120,11 @@ ENTRY(cpu_do_resume)
 	msr	mdscr_el1, x10
 
 	msr	sctlr_el1, x12
+alternative_if_not ARM64_HAS_VIRT_HOST_EXTN
 	msr	tpidr_el1, x13
+alternative_else
+	msr	tpidr_el2, x13
+alternative_endif
 	msr	sp_el0, x14
 	/*
 	 * Restore oslsr_el1 by writing oslar_el1
-- 
2.13.3

[PATCH v3 05/13] KVM: arm64: Stop save/restoring host tpidr_el1 on VHE

[James Morse]

Now that a VHE host uses tpidr_el2 for the cpu offset we no longer
need KVM to save/restore tpidr_el1. Move this from the 'common' code
into the non-vhe code. While we're at it, on VHE we don't need to
save the ELR or SPSR as kernel_entry in entry.S will have pushed these
onto the kernel stack, and will restore them from there. Move these
to the non-vhe code as we need them to get back to the host.

Finally remove the always-copy-tpidr we hid in the stage2 setup
code, cpufeature's enable callback will do this for VHE, we only
need KVM to do it for non-vhe. Add the copy into kvm-init instead.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
Reviewed-by: Christoffer Dall <cdall-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org>
---
Changes since v1:
 * Switched KVM<->arm64 in the subject.

 arch/arm64/kvm/hyp-init.S      |  4 ++++
 arch/arm64/kvm/hyp/s2-setup.c  |  3 ---
 arch/arm64/kvm/hyp/sysreg-sr.c | 16 ++++++++--------
 3 files changed, 12 insertions(+), 11 deletions(-)

diff --git a/arch/arm64/kvm/hyp-init.S b/arch/arm64/kvm/hyp-init.S
index 3f9615582377..fbf259893f6a 100644
--- a/arch/arm64/kvm/hyp-init.S
+++ b/arch/arm64/kvm/hyp-init.S
@@ -122,6 +122,10 @@ CPU_BE(	orr	x4, x4, #SCTLR_ELx_EE)
 	kern_hyp_va	x2
 	msr	vbar_el2, x2
 
+	/* copy tpidr_el1 into tpidr_el2 for use by HYP */
+	mrs	x1, tpidr_el1
+	msr	tpidr_el2, x1
+
 	/* Hello, World! */
 	eret
 ENDPROC(__kvm_hyp_init)
diff --git a/arch/arm64/kvm/hyp/s2-setup.c b/arch/arm64/kvm/hyp/s2-setup.c
index 7fb88274eba1..a81f5e10fc8c 100644
--- a/arch/arm64/kvm/hyp/s2-setup.c
+++ b/arch/arm64/kvm/hyp/s2-setup.c
@@ -84,8 +84,5 @@ u32 __hyp_text __init_stage2_translation(void)
 
 	write_sysreg(val, vtcr_el2);
 
-	/* copy tpidr_el1 into tpidr_el2 for use by HYP */
-	write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
-
 	return parange;
 }
diff --git a/arch/arm64/kvm/hyp/sysreg-sr.c b/arch/arm64/kvm/hyp/sysreg-sr.c
index 934137647837..c54cc2afb92b 100644
--- a/arch/arm64/kvm/hyp/sysreg-sr.c
+++ b/arch/arm64/kvm/hyp/sysreg-sr.c
@@ -27,8 +27,8 @@ static void __hyp_text __sysreg_do_nothing(struct kvm_cpu_context *ctxt) { }
 /*
  * Non-VHE: Both host and guest must save everything.
  *
- * VHE: Host must save tpidr*_el[01], actlr_el1, mdscr_el1, sp0, pc,
- * pstate, and guest must save everything.
+ * VHE: Host must save tpidr*_el0, actlr_el1, mdscr_el1, sp_el0,
+ * and guest must save everything.
  */
 
 static void __hyp_text __sysreg_save_common_state(struct kvm_cpu_context *ctxt)
@@ -36,11 +36,8 @@ static void __hyp_text __sysreg_save_common_state(struct kvm_cpu_context *ctxt)
 	ctxt->sys_regs[ACTLR_EL1]	= read_sysreg(actlr_el1);
 	ctxt->sys_regs[TPIDR_EL0]	= read_sysreg(tpidr_el0);
 	ctxt->sys_regs[TPIDRRO_EL0]	= read_sysreg(tpidrro_el0);
-	ctxt->sys_regs[TPIDR_EL1]	= read_sysreg(tpidr_el1);
 	ctxt->sys_regs[MDSCR_EL1]	= read_sysreg(mdscr_el1);
 	ctxt->gp_regs.regs.sp		= read_sysreg(sp_el0);
-	ctxt->gp_regs.regs.pc		= read_sysreg_el2(elr);
-	ctxt->gp_regs.regs.pstate	= read_sysreg_el2(spsr);
 }
 
 static void __hyp_text __sysreg_save_state(struct kvm_cpu_context *ctxt)
@@ -62,10 +59,13 @@ static void __hyp_text __sysreg_save_state(struct kvm_cpu_context *ctxt)
 	ctxt->sys_regs[AMAIR_EL1]	= read_sysreg_el1(amair);
 	ctxt->sys_regs[CNTKCTL_EL1]	= read_sysreg_el1(cntkctl);
 	ctxt->sys_regs[PAR_EL1]		= read_sysreg(par_el1);
+	ctxt->sys_regs[TPIDR_EL1]	= read_sysreg(tpidr_el1);
 
 	ctxt->gp_regs.sp_el1		= read_sysreg(sp_el1);
 	ctxt->gp_regs.elr_el1		= read_sysreg_el1(elr);
 	ctxt->gp_regs.spsr[KVM_SPSR_EL1]= read_sysreg_el1(spsr);
+	ctxt->gp_regs.regs.pc		= read_sysreg_el2(elr);
+	ctxt->gp_regs.regs.pstate	= read_sysreg_el2(spsr);
 }
 
 static hyp_alternate_select(__sysreg_call_save_host_state,
@@ -89,11 +89,8 @@ static void __hyp_text __sysreg_restore_common_state(struct kvm_cpu_context *ctx
 	write_sysreg(ctxt->sys_regs[ACTLR_EL1],	  actlr_el1);
 	write_sysreg(ctxt->sys_regs[TPIDR_EL0],	  tpidr_el0);
 	write_sysreg(ctxt->sys_regs[TPIDRRO_EL0], tpidrro_el0);
-	write_sysreg(ctxt->sys_regs[TPIDR_EL1],	  tpidr_el1);
 	write_sysreg(ctxt->sys_regs[MDSCR_EL1],	  mdscr_el1);
 	write_sysreg(ctxt->gp_regs.regs.sp,	  sp_el0);
-	write_sysreg_el2(ctxt->gp_regs.regs.pc,	  elr);
-	write_sysreg_el2(ctxt->gp_regs.regs.pstate, spsr);
 }
 
 static void __hyp_text __sysreg_restore_state(struct kvm_cpu_context *ctxt)
@@ -115,10 +112,13 @@ static void __hyp_text __sysreg_restore_state(struct kvm_cpu_context *ctxt)
 	write_sysreg_el1(ctxt->sys_regs[AMAIR_EL1],	amair);
 	write_sysreg_el1(ctxt->sys_regs[CNTKCTL_EL1], 	cntkctl);
 	write_sysreg(ctxt->sys_regs[PAR_EL1],		par_el1);
+	write_sysreg(ctxt->sys_regs[TPIDR_EL1],		tpidr_el1);
 
 	write_sysreg(ctxt->gp_regs.sp_el1,		sp_el1);
 	write_sysreg_el1(ctxt->gp_regs.elr_el1,		elr);
 	write_sysreg_el1(ctxt->gp_regs.spsr[KVM_SPSR_EL1],spsr);
+	write_sysreg_el2(ctxt->gp_regs.regs.pc,		elr);
+	write_sysreg_el2(ctxt->gp_regs.regs.pstate,	spsr);
 }
 
 static hyp_alternate_select(__sysreg_call_restore_host_state,
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 05/13] KVM: arm64: Stop save/restoring host tpidr_el1 on VHE

[James Morse]

Now that a VHE host uses tpidr_el2 for the cpu offset we no longer
need KVM to save/restore tpidr_el1. Move this from the 'common' code
into the non-vhe code. While we're at it, on VHE we don't need to
save the ELR or SPSR as kernel_entry in entry.S will have pushed these
onto the kernel stack, and will restore them from there. Move these
to the non-vhe code as we need them to get back to the host.

Finally remove the always-copy-tpidr we hid in the stage2 setup
code, cpufeature's enable callback will do this for VHE, we only
need KVM to do it for non-vhe. Add the copy into kvm-init instead.

Signed-off-by: James Morse <james.morse@arm.com>
Reviewed-by: Christoffer Dall <cdall@linaro.org>
---
Changes since v1:
 * Switched KVM<->arm64 in the subject.

 arch/arm64/kvm/hyp-init.S      |  4 ++++
 arch/arm64/kvm/hyp/s2-setup.c  |  3 ---
 arch/arm64/kvm/hyp/sysreg-sr.c | 16 ++++++++--------
 3 files changed, 12 insertions(+), 11 deletions(-)

diff --git a/arch/arm64/kvm/hyp-init.S b/arch/arm64/kvm/hyp-init.S
index 3f9615582377..fbf259893f6a 100644
--- a/arch/arm64/kvm/hyp-init.S
+++ b/arch/arm64/kvm/hyp-init.S
@@ -122,6 +122,10 @@ CPU_BE(	orr	x4, x4, #SCTLR_ELx_EE)
 	kern_hyp_va	x2
 	msr	vbar_el2, x2
 
+	/* copy tpidr_el1 into tpidr_el2 for use by HYP */
+	mrs	x1, tpidr_el1
+	msr	tpidr_el2, x1
+
 	/* Hello, World! */
 	eret
 ENDPROC(__kvm_hyp_init)
diff --git a/arch/arm64/kvm/hyp/s2-setup.c b/arch/arm64/kvm/hyp/s2-setup.c
index 7fb88274eba1..a81f5e10fc8c 100644
--- a/arch/arm64/kvm/hyp/s2-setup.c
+++ b/arch/arm64/kvm/hyp/s2-setup.c
@@ -84,8 +84,5 @@ u32 __hyp_text __init_stage2_translation(void)
 
 	write_sysreg(val, vtcr_el2);
 
-	/* copy tpidr_el1 into tpidr_el2 for use by HYP */
-	write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
-
 	return parange;
 }
diff --git a/arch/arm64/kvm/hyp/sysreg-sr.c b/arch/arm64/kvm/hyp/sysreg-sr.c
index 934137647837..c54cc2afb92b 100644
--- a/arch/arm64/kvm/hyp/sysreg-sr.c
+++ b/arch/arm64/kvm/hyp/sysreg-sr.c
@@ -27,8 +27,8 @@ static void __hyp_text __sysreg_do_nothing(struct kvm_cpu_context *ctxt) { }
 /*
  * Non-VHE: Both host and guest must save everything.
  *
- * VHE: Host must save tpidr*_el[01], actlr_el1, mdscr_el1, sp0, pc,
- * pstate, and guest must save everything.
+ * VHE: Host must save tpidr*_el0, actlr_el1, mdscr_el1, sp_el0,
+ * and guest must save everything.
  */
 
 static void __hyp_text __sysreg_save_common_state(struct kvm_cpu_context *ctxt)
@@ -36,11 +36,8 @@ static void __hyp_text __sysreg_save_common_state(struct kvm_cpu_context *ctxt)
 	ctxt->sys_regs[ACTLR_EL1]	= read_sysreg(actlr_el1);
 	ctxt->sys_regs[TPIDR_EL0]	= read_sysreg(tpidr_el0);
 	ctxt->sys_regs[TPIDRRO_EL0]	= read_sysreg(tpidrro_el0);
-	ctxt->sys_regs[TPIDR_EL1]	= read_sysreg(tpidr_el1);
 	ctxt->sys_regs[MDSCR_EL1]	= read_sysreg(mdscr_el1);
 	ctxt->gp_regs.regs.sp		= read_sysreg(sp_el0);
-	ctxt->gp_regs.regs.pc		= read_sysreg_el2(elr);
-	ctxt->gp_regs.regs.pstate	= read_sysreg_el2(spsr);
 }
 
 static void __hyp_text __sysreg_save_state(struct kvm_cpu_context *ctxt)
@@ -62,10 +59,13 @@ static void __hyp_text __sysreg_save_state(struct kvm_cpu_context *ctxt)
 	ctxt->sys_regs[AMAIR_EL1]	= read_sysreg_el1(amair);
 	ctxt->sys_regs[CNTKCTL_EL1]	= read_sysreg_el1(cntkctl);
 	ctxt->sys_regs[PAR_EL1]		= read_sysreg(par_el1);
+	ctxt->sys_regs[TPIDR_EL1]	= read_sysreg(tpidr_el1);
 
 	ctxt->gp_regs.sp_el1		= read_sysreg(sp_el1);
 	ctxt->gp_regs.elr_el1		= read_sysreg_el1(elr);
 	ctxt->gp_regs.spsr[KVM_SPSR_EL1]= read_sysreg_el1(spsr);
+	ctxt->gp_regs.regs.pc		= read_sysreg_el2(elr);
+	ctxt->gp_regs.regs.pstate	= read_sysreg_el2(spsr);
 }
 
 static hyp_alternate_select(__sysreg_call_save_host_state,
@@ -89,11 +89,8 @@ static void __hyp_text __sysreg_restore_common_state(struct kvm_cpu_context *ctx
 	write_sysreg(ctxt->sys_regs[ACTLR_EL1],	  actlr_el1);
 	write_sysreg(ctxt->sys_regs[TPIDR_EL0],	  tpidr_el0);
 	write_sysreg(ctxt->sys_regs[TPIDRRO_EL0], tpidrro_el0);
-	write_sysreg(ctxt->sys_regs[TPIDR_EL1],	  tpidr_el1);
 	write_sysreg(ctxt->sys_regs[MDSCR_EL1],	  mdscr_el1);
 	write_sysreg(ctxt->gp_regs.regs.sp,	  sp_el0);
-	write_sysreg_el2(ctxt->gp_regs.regs.pc,	  elr);
-	write_sysreg_el2(ctxt->gp_regs.regs.pstate, spsr);
 }
 
 static void __hyp_text __sysreg_restore_state(struct kvm_cpu_context *ctxt)
@@ -115,10 +112,13 @@ static void __hyp_text __sysreg_restore_state(struct kvm_cpu_context *ctxt)
 	write_sysreg_el1(ctxt->sys_regs[AMAIR_EL1],	amair);
 	write_sysreg_el1(ctxt->sys_regs[CNTKCTL_EL1], 	cntkctl);
 	write_sysreg(ctxt->sys_regs[PAR_EL1],		par_el1);
+	write_sysreg(ctxt->sys_regs[TPIDR_EL1],		tpidr_el1);
 
 	write_sysreg(ctxt->gp_regs.sp_el1,		sp_el1);
 	write_sysreg_el1(ctxt->gp_regs.elr_el1,		elr);
 	write_sysreg_el1(ctxt->gp_regs.spsr[KVM_SPSR_EL1],spsr);
+	write_sysreg_el2(ctxt->gp_regs.regs.pc,		elr);
+	write_sysreg_el2(ctxt->gp_regs.regs.pstate,	spsr);
 }
 
 static hyp_alternate_select(__sysreg_call_restore_host_state,
-- 
2.13.3

[PATCH v3 06/13] Docs: dt: add devicetree binding for describing arm64 SDEI firmware

[James Morse]

The Software Delegated Exception Interface (SDEI) is an ARM standard
for registering callbacks from the platform firmware into the OS.
This is typically used to implement RAS notifications, or from an
IRQ that has been promoted to a firmware-assisted NMI.

Add a new devicetree binding to describe the SDE firmware interface.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
Acked-by: Rob Herring <robh-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
---
Changes since v2:
 * Added Rob's Ack
 * Fixed 'childe node' typo

Changes since v1:
* Added bound IRQ description for binding,
* Reference SMC-CC, not 'AAPCS like'
* Move sdei node under firmware node (and the file path)

 .../devicetree/bindings/arm/firmware/sdei.txt      | 42 ++++++++++++++++++++++
 1 file changed, 42 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/arm/firmware/sdei.txt

diff --git a/Documentation/devicetree/bindings/arm/firmware/sdei.txt b/Documentation/devicetree/bindings/arm/firmware/sdei.txt
new file mode 100644
index 000000000000..ee3f0ff49889
--- /dev/null
+++ b/Documentation/devicetree/bindings/arm/firmware/sdei.txt
@@ -0,0 +1,42 @@
+* Software Delegated Exception Interface (SDEI)
+
+Firmware implementing the SDEI functions described in ARM document number
+ARM DEN 0054A ("Software Delegated Exception Interface") can be used by
+Linux to receive notification of events such as those generated by
+firmware-first error handling, or from an IRQ that has been promoted to
+a firmware-assisted NMI.
+
+The interface provides a number of API functions for registering callbacks
+and enabling/disabling events. Functions are invoked by trapping to the
+privilege level of the SDEI firmware (specified as part of the binding
+below) and passing arguments in a manner specified by the "SMC Calling
+Convention (ARM DEN 0028B):
+
+	 r0		=> 32-bit Function ID / return value
+	{r1 - r3}	=> Parameters
+
+Note that the immediate field of the trapping instruction must be set
+to #0.
+
+The SDEI_EVENT_REGISTER function registers a callback in the kernel
+text to handle the specified event number.
+
+The sdei node should be a child node of '/firmware' and have required
+properties:
+
+ - compatible    : should contain:
+	* "arm,sdei-1.0" : For implementations complying to SDEI version 1.x.
+
+ - method        : The method of calling the SDEI firmware. Permitted
+                   values are:
+	* "smc" : SMC #0, with the register assignments specified in this
+	          binding.
+	* "hvc" : HVC #0, with the register assignments specified in this
+	          binding.
+Example:
+	firmware {
+		sdei {
+			compatible	= "arm,sdei-1.0";
+			method		= "smc";
+		};
+	};
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 06/13] Docs: dt: add devicetree binding for describing arm64 SDEI firmware

[James Morse]

The Software Delegated Exception Interface (SDEI) is an ARM standard
for registering callbacks from the platform firmware into the OS.
This is typically used to implement RAS notifications, or from an
IRQ that has been promoted to a firmware-assisted NMI.

Add a new devicetree binding to describe the SDE firmware interface.

Signed-off-by: James Morse <james.morse@arm.com>
Acked-by: Rob Herring <robh@kernel.org>
---
Changes since v2:
 * Added Rob's Ack
 * Fixed 'childe node' typo

Changes since v1:
* Added bound IRQ description for binding,
* Reference SMC-CC, not 'AAPCS like'
* Move sdei node under firmware node (and the file path)

 .../devicetree/bindings/arm/firmware/sdei.txt      | 42 ++++++++++++++++++++++
 1 file changed, 42 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/arm/firmware/sdei.txt

diff --git a/Documentation/devicetree/bindings/arm/firmware/sdei.txt b/Documentation/devicetree/bindings/arm/firmware/sdei.txt
new file mode 100644
index 000000000000..ee3f0ff49889
--- /dev/null
+++ b/Documentation/devicetree/bindings/arm/firmware/sdei.txt
@@ -0,0 +1,42 @@
+* Software Delegated Exception Interface (SDEI)
+
+Firmware implementing the SDEI functions described in ARM document number
+ARM DEN 0054A ("Software Delegated Exception Interface") can be used by
+Linux to receive notification of events such as those generated by
+firmware-first error handling, or from an IRQ that has been promoted to
+a firmware-assisted NMI.
+
+The interface provides a number of API functions for registering callbacks
+and enabling/disabling events. Functions are invoked by trapping to the
+privilege level of the SDEI firmware (specified as part of the binding
+below) and passing arguments in a manner specified by the "SMC Calling
+Convention (ARM DEN 0028B):
+
+	 r0		=> 32-bit Function ID / return value
+	{r1 - r3}	=> Parameters
+
+Note that the immediate field of the trapping instruction must be set
+to #0.
+
+The SDEI_EVENT_REGISTER function registers a callback in the kernel
+text to handle the specified event number.
+
+The sdei node should be a child node of '/firmware' and have required
+properties:
+
+ - compatible    : should contain:
+	* "arm,sdei-1.0" : For implementations complying to SDEI version 1.x.
+
+ - method        : The method of calling the SDEI firmware. Permitted
+                   values are:
+	* "smc" : SMC #0, with the register assignments specified in this
+	          binding.
+	* "hvc" : HVC #0, with the register assignments specified in this
+	          binding.
+Example:
+	firmware {
+		sdei {
+			compatible	= "arm,sdei-1.0";
+			method		= "smc";
+		};
+	};
-- 
2.13.3

[PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions

[James Morse]

The Software Delegated Exception Interface (SDEI) is an ARM standard
for registering callbacks from the platform firmware into the OS.
This is typically used to implement firmware notifications (such as
firmware-first RAS) or promote an IRQ that has been promoted to a
firmware-assisted NMI.

Add the code for detecting the SDEI version and the framework for
registering and unregistering events. Subsequent patches will add the
arch-specific backend code and the necessary power management hooks.

Only shared events are supported, power management, private events and
discovery for ACPI systems will be added by later patches.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
---
Changes since v2:
 * Copy the priority into the structure the arch asm handler gets. This
   is used for VMAP stacks where we can't know if a critical event interrupted
   a normal priority event, thus they need separate stacks.

Changes since v1:
 * Changed entry point to unsigned long, if we support non-vhe systems this
   won't be a valid pointer
 * Made invoke_sdei_fn() pass the only register we are interested in, instead
   of the whole arm_smccc_res
 * Made all the locking WARN_ON()s lockdep_assert_held()s.
 * Moved more messages from 'err' to 'warn'.
 * Made IS_SDEI_CALL() not depend on whether the config option is selected.
 * Made 'event failed' messages rate limited.

 drivers/firmware/Kconfig    |   7 +
 drivers/firmware/Makefile   |   1 +
 drivers/firmware/arm_sdei.c | 616 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/sdei.h        | 100 +++++++
 include/uapi/linux/sdei.h   |  91 +++++++
 5 files changed, 815 insertions(+)
 create mode 100644 drivers/firmware/arm_sdei.c
 create mode 100644 include/linux/sdei.h
 create mode 100644 include/uapi/linux/sdei.h

diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
index 6e4ed5a9c6fd..d8a9859f6102 100644
--- a/drivers/firmware/Kconfig
+++ b/drivers/firmware/Kconfig
@@ -48,6 +48,13 @@ config ARM_SCPI_POWER_DOMAIN
 	  This enables support for the SCPI power domains which can be
 	  enabled or disabled via the SCP firmware
 
+config ARM_SDE_INTERFACE
+	bool "ARM Software Delegated Exception Interface (SDEI)"
+	help
+	  The Software Delegated Exception Interface (SDEI) is an ARM
+	  standard for registering callbacks from the platform firmware
+	  into the OS. This is typically used to implement RAS notifications.
+
 config EDD
 	tristate "BIOS Enhanced Disk Drive calls determine boot disk"
 	depends on X86
diff --git a/drivers/firmware/Makefile b/drivers/firmware/Makefile
index a37f12e8d137..229984feca4a 100644
--- a/drivers/firmware/Makefile
+++ b/drivers/firmware/Makefile
@@ -5,6 +5,7 @@ obj-$(CONFIG_ARM_PSCI_FW)	+= psci.o
 obj-$(CONFIG_ARM_PSCI_CHECKER)	+= psci_checker.o
 obj-$(CONFIG_ARM_SCPI_PROTOCOL)	+= arm_scpi.o
 obj-$(CONFIG_ARM_SCPI_POWER_DOMAIN) += scpi_pm_domain.o
+obj-$(CONFIG_ARM_SDE_INTERFACE)	+= arm_sdei.o
 obj-$(CONFIG_DMI)		+= dmi_scan.o
 obj-$(CONFIG_DMI_SYSFS)		+= dmi-sysfs.o
 obj-$(CONFIG_EDD)		+= edd.o
diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
new file mode 100644
index 000000000000..4b3c7fd99c34
--- /dev/null
+++ b/drivers/firmware/arm_sdei.c
@@ -0,0 +1,616 @@
+/*
+ * Software Delegated Exception Interface (SDEI) driver code.
+ *
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define pr_fmt(fmt) "sdei: " fmt
+
+#include <linux/acpi.h>
+#include <linux/arm-smccc.h>
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/errno.h>
+#include <linux/hardirq.h>
+#include <linux/kernel.h>
+#include <linux/kprobes.h>
+#include <linux/kvm_host.h>
+#include <linux/list.h>
+#include <linux/lockdep.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/percpu.h>
+#include <linux/platform_device.h>
+#include <linux/ptrace.h>
+#include <linux/preempt.h>
+#include <linux/sdei.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/spinlock.h>
+#include <linux/uaccess.h>
+#include <uapi/linux/sdei.h>
+
+/*
+ * The call to use to reach the firmware.
+ */
+static asmlinkage void (*sdei_firmware_call)(unsigned long function_id,
+		      unsigned long arg0, unsigned long arg1,
+		      unsigned long arg2, unsigned long arg3,
+		      unsigned long arg4, struct arm_smccc_res *res);
+
+/* entry point from firmware to arch asm code */
+static unsigned long sdei_entry_point;
+
+struct sdei_event {
+	struct list_head	list;
+	u32			event_num;
+	u8			type;
+	u8			priority;
+
+	/* This pointer is handed to firmware as the event argument. */
+	struct sdei_registered_event *registered;
+};
+
+static LIST_HEAD(sdei_events);
+static DEFINE_SPINLOCK(sdei_events_lock);
+
+static int sdei_to_linux_errno(unsigned long sdei_err)
+{
+	switch (sdei_err) {
+	case SDEI_NOT_SUPPORTED:
+		return -EOPNOTSUPP;
+	case SDEI_INVALID_PARAMETERS:
+		return -EINVAL;
+	case SDEI_DENIED:
+		return -EPERM;
+	case SDEI_PENDING:
+		return -EINPROGRESS;
+	case SDEI_OUT_OF_RESOURCE:
+		return -ENOMEM;
+	}
+
+	/* Not an error value ... */
+	return sdei_err;
+}
+
+/*
+ * If x0 is any of these values, then the call failed, use sdei_to_linux_errno()
+ * to translate.
+ */
+static int sdei_is_err(struct arm_smccc_res *res)
+{
+	switch (res->a0) {
+	case SDEI_NOT_SUPPORTED:
+	case SDEI_INVALID_PARAMETERS:
+	case SDEI_DENIED:
+	case SDEI_PENDING:
+	case SDEI_OUT_OF_RESOURCE:
+		return true;
+	}
+
+	return false;
+}
+
+static int invoke_sdei_fn(unsigned long function_id, unsigned long arg0,
+			  unsigned long arg1, unsigned long arg2,
+			  unsigned long arg3, unsigned long arg4,
+			  u64 *result)
+{
+	int err = 0;
+	struct arm_smccc_res res;
+
+	if (sdei_firmware_call) {
+		sdei_firmware_call(function_id, arg0, arg1, arg2, arg3, arg4,
+				   &res);
+		if (sdei_is_err(&res))
+			err = sdei_to_linux_errno(res.a0);
+	} else {
+		/*
+		 * !sdei_firmware_call means we failed to probe or called
+		 * sdei_mark_interface_broken(). -EIO is not an error returned
+		 * by sdei_to_linux_errno() and is used to suppress messages
+		 * from this driver.
+		 */
+		err = -EIO;
+		res.a0 = SDEI_NOT_SUPPORTED;
+	}
+
+	if (result)
+		*result = res.a0;
+
+	return err;
+}
+
+static struct sdei_event *sdei_event_find(u32 event_num)
+{
+	struct sdei_event *e, *found = NULL;
+
+	lockdep_assert_held(&sdei_events_lock);
+
+	list_for_each_entry(e, &sdei_events, list) {
+		if (e->event_num == event_num) {
+			found = e;
+			break;
+		}
+	}
+
+	return found;
+}
+
+int sdei_api_event_context(u32 query, u64 *result)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_CONTEXT, query, 0, 0, 0, 0,
+			      result);
+}
+
+static int sdei_api_event_get_info(u32 event, u32 info, u64 *result)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_GET_INFO, event, info, 0,
+			      0, 0, result);
+}
+
+static struct sdei_event *sdei_event_create(u32 event_num,
+					    sdei_event_callback *cb,
+					    void *cb_arg, gfp_t gfp_flags)
+{
+	int err;
+	u64 result;
+	struct sdei_event *event;
+	struct sdei_registered_event *reg;
+
+	lockdep_assert_held(&sdei_events_lock);
+
+	event = kzalloc(sizeof(*event), gfp_flags);
+	if (!event)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&event->list);
+	event->event_num = event_num;
+
+	err = sdei_api_event_get_info(event_num, SDEI_EVENT_INFO_EV_PRIORITY,
+				      &result);
+	if (err) {
+		kfree(event);
+		return ERR_PTR(err);
+	}
+	event->priority = result;
+
+	err = sdei_api_event_get_info(event_num, SDEI_EVENT_INFO_EV_TYPE,
+				      &result);
+	if (err) {
+		kfree(event);
+		return ERR_PTR(err);
+	}
+	event->type = result;
+
+	if (event->type == SDEI_EVENT_TYPE_SHARED) {
+		reg = kzalloc(sizeof(*reg), GFP_KERNEL);
+		if (!reg) {
+			kfree(event);
+			return ERR_PTR(-ENOMEM);
+		}
+
+		reg->event_num = event_num;
+		reg->priority = event->priority;
+
+		reg->callback = cb;
+		reg->callback_arg = cb_arg;
+		event->registered = reg;
+	}
+
+	if (sdei_event_find(event_num)) {
+		kfree(event->registered);
+		kfree(event);
+		event = ERR_PTR(-EBUSY);
+	} else {
+		list_add(&event->list, &sdei_events);
+	}
+
+	return event;
+}
+
+static void sdei_event_destroy(struct sdei_event *event)
+{
+	lockdep_assert_held(&sdei_events_lock);
+
+	list_del(&event->list);
+
+	if (event->type == SDEI_EVENT_TYPE_SHARED)
+		kfree(event->registered);
+
+	kfree(event);
+}
+
+static int sdei_api_get_version(u64 *version)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_VERSION, 0, 0, 0, 0, 0, version);
+}
+
+int sdei_mask_local_cpu(void)
+{
+	int err;
+
+	WARN_ON_ONCE(preemptible());
+
+	err = invoke_sdei_fn(SDEI_1_0_FN_SDEI_PE_MASK, 0, 0, 0, 0, 0, NULL);
+	if (err && err != -EIO) {
+		pr_warn_once("failed to mask CPU[%u]: %d\n",
+			      smp_processor_id(), err);
+		return err;
+	}
+
+	return 0;
+}
+
+static void _ipi_mask_cpu(void *ignored)
+{
+	sdei_mask_local_cpu();
+}
+
+int sdei_unmask_local_cpu(void)
+{
+	int err;
+
+	WARN_ON_ONCE(preemptible());
+
+	err = invoke_sdei_fn(SDEI_1_0_FN_SDEI_PE_UNMASK, 0, 0, 0, 0, 0, NULL);
+	if (err && err != -EIO) {
+		pr_warn_once("failed to unmask CPU[%u]: %d\n",
+			     smp_processor_id(), err);
+		return err;
+	}
+
+	return 0;
+}
+
+static void _ipi_unmask_cpu(void *ignored)
+{
+	sdei_unmask_local_cpu();
+}
+
+static void _ipi_private_reset(void *ignored)
+{
+	int err;
+
+	err = invoke_sdei_fn(SDEI_1_0_FN_SDEI_PRIVATE_RESET, 0, 0, 0, 0, 0,
+			     NULL);
+	if (err && err != -EIO)
+		pr_warn_once("failed to reset CPU[%u]: %d\n",
+			     smp_processor_id(), err);
+}
+
+static int sdei_api_shared_reset(void)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_SHARED_RESET, 0, 0, 0, 0, 0,
+			      NULL);
+}
+
+static void sdei_mark_interface_broken(void)
+{
+	pr_err("disabling SDEI firmware interface\n");
+	on_each_cpu(&_ipi_mask_cpu, NULL, true);
+	sdei_firmware_call = NULL;
+}
+
+static int sdei_platform_reset(void)
+{
+	int err;
+
+	on_each_cpu(&_ipi_private_reset, NULL, true);
+	err = sdei_api_shared_reset();
+	if (err) {
+		pr_err("Failed to reset platform: %d\n", err);
+		sdei_mark_interface_broken();
+	}
+
+	return err;
+}
+
+static int sdei_api_event_enable(u32 event_num)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_ENABLE, event_num, 0, 0, 0,
+			      0, NULL);
+}
+
+int sdei_event_enable(u32 event_num)
+{
+	int err = -EINVAL;
+	struct sdei_event *event;
+
+	spin_lock(&sdei_events_lock);
+	event = sdei_event_find(event_num);
+	if (!event)
+		err = -ENOENT;
+	else if (event->type == SDEI_EVENT_TYPE_SHARED)
+		err = sdei_api_event_enable(event->event_num);
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(sdei_event_enable);
+
+static int sdei_api_event_disable(u32 event_num)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_DISABLE, event_num, 0, 0,
+			      0, 0, NULL);
+}
+
+int sdei_event_disable(u32 event_num)
+{
+	int err = -EINVAL;
+	struct sdei_event *event;
+
+	spin_lock(&sdei_events_lock);
+	event = sdei_event_find(event_num);
+	if (!event)
+		err = -ENOENT;
+	else if (event->type == SDEI_EVENT_TYPE_SHARED)
+		err = sdei_api_event_disable(event->event_num);
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(sdei_event_disable);
+
+static int sdei_api_event_unregister(u32 event_num)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_UNREGISTER, event_num, 0,
+			      0, 0, 0, NULL);
+}
+
+static int _sdei_event_unregister(struct sdei_event *event)
+{
+	if (event->type == SDEI_EVENT_TYPE_SHARED)
+		return sdei_api_event_unregister(event->event_num);
+
+	return -EINVAL;
+}
+
+int sdei_event_unregister(u32 event_num)
+{
+	int err;
+	struct sdei_event *event;
+
+	WARN_ON(in_nmi());
+
+	spin_lock(&sdei_events_lock);
+	event = sdei_event_find(event_num);
+	do {
+		if (!event) {
+			pr_warn("Event %u not registered\n", event_num);
+			err = -ENOENT;
+			break;
+		}
+
+		err = _sdei_event_unregister(event);
+		if (err)
+			break;
+
+		sdei_event_destroy(event);
+	} while (0);
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(sdei_event_unregister);
+
+static int sdei_api_event_register(u32 event_num, unsigned long entry_point,
+				   void *arg, u64 flags, u64 affinity)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_REGISTER, event_num,
+			      (unsigned long)entry_point, (unsigned long)arg,
+			      flags, affinity, NULL);
+}
+
+static int _sdei_event_register(struct sdei_event *event)
+{
+	if (event->type == SDEI_EVENT_TYPE_SHARED)
+		return sdei_api_event_register(event->event_num,
+					       sdei_entry_point,
+					       event->registered,
+					       SDEI_EVENT_REGISTER_RM_ANY, 0);
+
+	return -EINVAL;
+}
+
+int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
+			gfp_t gfp_flags)
+{
+	int err;
+	struct sdei_event *event;
+
+	WARN_ON(in_nmi());
+
+	spin_lock(&sdei_events_lock);
+	do {
+		if (sdei_event_find(event_num)) {
+			pr_warn("Event %u already registered\n", event_num);
+			err = -EBUSY;
+			break;
+		}
+
+		event = sdei_event_create(event_num, cb, arg, gfp_flags);
+		if (IS_ERR(event)) {
+			err = PTR_ERR(event);
+			pr_warn("Failed to create event %u: %d\n", event_num,
+				err);
+			break;
+		}
+
+		err = _sdei_event_register(event);
+		if (err) {
+			sdei_event_destroy(event);
+			pr_warn("Failed to register event %u: %d\n", event_num,
+				err);
+		}
+	} while (0);
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(sdei_event_register);
+
+static void sdei_smccc_smc(unsigned long function_id,
+			   unsigned long arg0, unsigned long arg1,
+			   unsigned long arg2, unsigned long arg3,
+			   unsigned long arg4, struct arm_smccc_res *res)
+{
+	arm_smccc_smc(function_id, arg0, arg1, arg2, arg3, arg4, 0, 0, res);
+}
+
+static void sdei_smccc_hvc(unsigned long function_id,
+			   unsigned long arg0, unsigned long arg1,
+			   unsigned long arg2, unsigned long arg3,
+			   unsigned long arg4, struct arm_smccc_res *res)
+{
+	arm_smccc_hvc(function_id, arg0, arg1, arg2, arg3, arg4, 0, 0, res);
+}
+
+static int sdei_get_conduit(struct platform_device *pdev)
+{
+	const char *method;
+	struct device_node *np = pdev->dev.of_node;
+
+	sdei_firmware_call = NULL;
+	if (np) {
+		if (of_property_read_string(np, "method", &method)) {
+			pr_warn("missing \"method\" property\n");
+			return CONDUIT_INVALID;
+		}
+
+		if (!strcmp("hvc", method)) {
+			sdei_firmware_call = &sdei_smccc_hvc;
+			return CONDUIT_HVC;
+		} else if (!strcmp("smc", method)) {
+			sdei_firmware_call = &sdei_smccc_smc;
+			return CONDUIT_SMC;
+		}
+
+		pr_warn("invalid \"method\" property: %s\n", method);
+	}
+
+	return CONDUIT_INVALID;
+}
+
+static int sdei_probe(struct platform_device *pdev)
+{
+	int err;
+	u64 ver = 0;
+	int conduit;
+
+	conduit = sdei_get_conduit(pdev);
+	if (!sdei_firmware_call)
+		return 0;
+
+	err = sdei_api_get_version(&ver);
+	if (err == -EOPNOTSUPP)
+		pr_err("advertised but not implemented in platform firmware\n");
+	if (err) {
+		pr_err("Failed to get SDEI version: %d\n", err);
+		sdei_mark_interface_broken();
+		return err;
+	}
+
+	pr_info("SDEIv%d.%d (0x%x) detected in firmware.\n",
+		(int)SDEI_VERSION_MAJOR(ver), (int)SDEI_VERSION_MINOR(ver),
+		(int)SDEI_VERSION_VENDOR(ver));
+
+	if (SDEI_VERSION_MAJOR(ver) != 1) {
+		pr_warn("Conflicting SDEI version detected.\n");
+		sdei_mark_interface_broken();
+		return -EINVAL;
+	}
+
+	err = sdei_platform_reset();
+	if (err)
+		return err;
+
+	sdei_entry_point = sdei_arch_get_entry_point(conduit);
+	if (!sdei_entry_point) {
+		/* Not supported due to hardware or boot configuration */
+		sdei_mark_interface_broken();
+		return 0;
+	}
+
+	on_each_cpu(&_ipi_unmask_cpu, NULL, false);
+
+	return 0;
+}
+
+static const struct of_device_id sdei_of_match[] = {
+	{ .compatible = "arm,sdei-1.0" },
+	{}
+};
+
+static struct platform_driver sdei_driver = {
+	.driver		= {
+		.name			= "sdei",
+		.of_match_table		= sdei_of_match,
+	},
+	.probe		= sdei_probe,
+};
+
+static bool __init sdei_present_dt(void)
+{
+	struct platform_device *pdev;
+	struct device_node *np, *fw_np;
+
+	fw_np = of_find_node_by_name(NULL, "firmware");
+	if (!fw_np)
+		return false;
+
+	np = of_find_matching_node(fw_np, sdei_of_match);
+	of_node_put(fw_np);
+	if (!np)
+		return false;
+
+	pdev = of_platform_device_create(np, sdei_driver.driver.name, NULL);
+	of_node_put(np);
+	if (IS_ERR(pdev))
+		return false;
+
+	return true;
+}
+
+static int __init sdei_init(void)
+{
+	if (sdei_present_dt())
+		platform_driver_register(&sdei_driver);
+
+	return 0;
+}
+
+subsys_initcall_sync(sdei_init);
+
+int sdei_event_handler(struct pt_regs *regs,
+		       struct sdei_registered_event *arg)
+{
+	int err;
+	mm_segment_t orig_addr_limit;
+	u32 event_num = arg->event_num;
+
+	orig_addr_limit = get_fs();
+	set_fs(USER_DS);
+
+	err = arg->callback(event_num, regs, arg->callback_arg);
+	if (err)
+		pr_err_ratelimited("event %u on CPU %u failed with error: %d\n",
+				   event_num, smp_processor_id(), err);
+
+	set_fs(orig_addr_limit);
+
+	return err;
+}
+NOKPROBE_SYMBOL(sdei_event_handler);
diff --git a/include/linux/sdei.h b/include/linux/sdei.h
new file mode 100644
index 000000000000..bb3dd000771e
--- /dev/null
+++ b/include/linux/sdei.h
@@ -0,0 +1,100 @@
+/*
+ * Software Delegated Exception Interface (SDEI) API
+ *
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __LINUX_SDEI_H
+#define __LINUX_SDEI_H
+
+#include <linux/gfp.h>
+
+#include <uapi/linux/sdei.h>
+
+/* kvm needs to know if hvcs are for SDEI or PSCI */
+#define IS_SDEI_CALL(x)		((x & SDEI_1_0_MASK) == SDEI_1_0_FN_BASE)
+
+enum sdei_conduit_types {
+	CONDUIT_INVALID = 0,
+	CONDUIT_SMC,
+	CONDUIT_HVC,
+};
+
+/* Arch code should override this to set the entry point from firmware... */
+#ifndef sdei_arch_get_entry_point
+#define sdei_arch_get_entry_point(conduit)	(0)
+#endif
+
+/*
+ * When an event occurs sdei_event_handler() will call a user-provided callback
+ * like this in NMI context on the CPU that received the event.
+ */
+typedef int (sdei_event_callback)(u32 event, struct pt_regs *regs, void *arg);
+
+/*
+ * Register your callback to claim an event. The event must be described
+ * by firmware.
+ * This function allocates memory, pass an appropriate gfp_flags.
+ */
+int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
+			gfp_t gfp_flags);
+
+/*
+ * Calls to sdei_event_unregister() may return EINPROGRESS. Keep calling
+ * it until it succeeds.
+ */
+int sdei_event_unregister(u32 event_num);
+
+int sdei_event_enable(u32 event_num);
+int sdei_event_disable(u32 event_num);
+
+#ifdef CONFIG_ARM_SDE_INTERFACE
+/* For use by arch code when CPU hotplug notifiers are not appropriate. */
+int sdei_mask_local_cpu(void);
+int sdei_unmask_local_cpu(void);
+#else
+static inline int sdei_mask_local_cpu(void) { return 0; }
+static inline int sdei_unmask_local_cpu(void) { return 0; }
+#endif /* CONFIG_ARM_SDE_INTERFACE */
+
+
+/*
+ * This struct represents an event that has been registered. The driver
+ * maintains a list of all events, and which ones are registered. (Private
+ * events have one entry in the list, but are registered on each CPU).
+ * A pointer to this struct is passed to firmware, and back to the event
+ * handler. The event handler can then use this to invoke the registered
+ * callback, without having to walk the list.
+ *
+ * For CPU private events, this structure is per-cpu.
+ */
+struct sdei_registered_event {
+	/* For use by arch code: */
+	struct pt_regs          interrupted_regs;
+
+	sdei_event_callback	*callback;
+	void			*callback_arg;
+	u32			 event_num;
+	u8			 priority;
+};
+
+/* The arch code entry point should then call this when an event arrives. */
+int notrace sdei_event_handler(struct pt_regs *regs,
+			       struct sdei_registered_event *arg);
+
+/* arch code may use this to retrieve the extra registers. */
+int sdei_api_event_context(u32 query, u64 *result);
+
+#endif /* __LINUX_SDEI_H */
diff --git a/include/uapi/linux/sdei.h b/include/uapi/linux/sdei.h
new file mode 100644
index 000000000000..535c9d8d14cc
--- /dev/null
+++ b/include/uapi/linux/sdei.h
@@ -0,0 +1,91 @@
+/*
+ *Software Delegated Exception Interface (SDEI) Firmware API
+ *
+ * This header holds common SDEI defines and macros shared
+ * by: ARM64 kernel, KVM and user space.
+ *
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef _UAPI_LINUX_SDEI_H
+#define _UAPI_LINUX_SDEI_H
+
+#define SDEI_1_0_FN_BASE			0xC4000020
+#define SDEI_1_0_MASK				0xFFFFFFE0
+#define SDEI_1_0_FN(n)				(SDEI_1_0_FN_BASE + (n))
+
+#define SDEI_1_0_FN_SDEI_VERSION			SDEI_1_0_FN(0x00)
+#define SDEI_1_0_FN_SDEI_EVENT_REGISTER			SDEI_1_0_FN(0x01)
+#define SDEI_1_0_FN_SDEI_EVENT_ENABLE			SDEI_1_0_FN(0x02)
+#define SDEI_1_0_FN_SDEI_EVENT_DISABLE			SDEI_1_0_FN(0x03)
+#define SDEI_1_0_FN_SDEI_EVENT_CONTEXT			SDEI_1_0_FN(0x04)
+#define SDEI_1_0_FN_SDEI_EVENT_COMPLETE			SDEI_1_0_FN(0x05)
+#define SDEI_1_0_FN_SDEI_EVENT_COMPLETE_AND_RESUME	SDEI_1_0_FN(0x06)
+#define SDEI_1_0_FN_SDEI_EVENT_UNREGISTER		SDEI_1_0_FN(0x07)
+#define SDEI_1_0_FN_SDEI_EVENT_STATUS			SDEI_1_0_FN(0x08)
+#define SDEI_1_0_FN_SDEI_EVENT_GET_INFO			SDEI_1_0_FN(0x09)
+#define SDEI_1_0_FN_SDEI_EVENT_ROUTING_SET		SDEI_1_0_FN(0x0A)
+#define SDEI_1_0_FN_SDEI_PE_MASK			SDEI_1_0_FN(0x0B)
+#define SDEI_1_0_FN_SDEI_PE_UNMASK			SDEI_1_0_FN(0x0C)
+#define SDEI_1_0_FN_SDEI_INTERRUPT_BIND			SDEI_1_0_FN(0x0D)
+#define SDEI_1_0_FN_SDEI_INTERRUPT_RELEASE		SDEI_1_0_FN(0x0E)
+#define SDEI_1_0_FN_SDEI_PRIVATE_RESET			SDEI_1_0_FN(0x11)
+#define SDEI_1_0_FN_SDEI_SHARED_RESET			SDEI_1_0_FN(0x12)
+
+#define SDEI_VERSION_MAJOR_SHIFT			48
+#define SDEI_VERSION_MAJOR_MASK				0x7fff
+#define SDEI_VERSION_MINOR_SHIFT			32
+#define SDEI_VERSION_MINOR_MASK				0xffff
+#define SDEI_VERSION_VENDOR_SHIFT			0
+#define SDEI_VERSION_VENDOR_MASK			0xffffffff
+
+#define SDEI_VERSION_MAJOR(x)	(x>>SDEI_VERSION_MAJOR_SHIFT & SDEI_VERSION_MAJOR_MASK)
+#define SDEI_VERSION_MINOR(x)	(x>>SDEI_VERSION_MINOR_SHIFT & SDEI_VERSION_MINOR_MASK)
+#define SDEI_VERSION_VENDOR(x)	(x>>SDEI_VERSION_VENDOR_SHIFT & SDEI_VERSION_VENDOR_MASK)
+
+/* SDEI return values */
+#define SDEI_SUCCESS		0
+#define SDEI_NOT_SUPPORTED	-1
+#define SDEI_INVALID_PARAMETERS	-2
+#define SDEI_DENIED		-3
+#define SDEI_PENDING		-5
+#define SDEI_OUT_OF_RESOURCE	-10
+
+/* EVENT_REGISTER flags */
+#define SDEI_EVENT_REGISTER_RM_ANY	0
+#define SDEI_EVENT_REGISTER_RM_PE	1
+
+/* EVENT_STATUS return value bits */
+#define SDEI_EVENT_STATUS_RUNNING	2
+#define SDEI_EVENT_STATUS_ENABLED	1
+#define SDEI_EVENT_STATUS_REGISTERED	0
+
+/* EVENT_COMPLETE status values */
+#define SDEI_EV_HANDLED	0
+#define SDEI_EV_FAILED	1
+
+/* GET_INFO values */
+#define SDEI_EVENT_INFO_EV_TYPE			0
+#define SDEI_EVENT_INFO_EV_SIGNALED		1
+#define SDEI_EVENT_INFO_EV_PRIORITY		2
+#define SDEI_EVENT_INFO_EV_ROUTING_MODE		3
+#define SDEI_EVENT_INFO_EV_ROUTING_AFF		4
+
+/* and their results */
+#define SDEI_EVENT_TYPE_PRIVATE			0
+#define SDEI_EVENT_TYPE_SHARED			1
+#define SDEI_EVENT_PRIORITY_NORMAL		0
+#define SDEI_EVENT_PRIORITY_CRITICAL		1
+
+#endif /* _UAPI_LINUX_SDEI_H */
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions

[James Morse]

The Software Delegated Exception Interface (SDEI) is an ARM standard
for registering callbacks from the platform firmware into the OS.
This is typically used to implement firmware notifications (such as
firmware-first RAS) or promote an IRQ that has been promoted to a
firmware-assisted NMI.

Add the code for detecting the SDEI version and the framework for
registering and unregistering events. Subsequent patches will add the
arch-specific backend code and the necessary power management hooks.

Only shared events are supported, power management, private events and
discovery for ACPI systems will be added by later patches.

Signed-off-by: James Morse <james.morse@arm.com>
---
Changes since v2:
 * Copy the priority into the structure the arch asm handler gets. This
   is used for VMAP stacks where we can't know if a critical event interrupted
   a normal priority event, thus they need separate stacks.

Changes since v1:
 * Changed entry point to unsigned long, if we support non-vhe systems this
   won't be a valid pointer
 * Made invoke_sdei_fn() pass the only register we are interested in, instead
   of the whole arm_smccc_res
 * Made all the locking WARN_ON()s lockdep_assert_held()s.
 * Moved more messages from 'err' to 'warn'.
 * Made IS_SDEI_CALL() not depend on whether the config option is selected.
 * Made 'event failed' messages rate limited.

 drivers/firmware/Kconfig    |   7 +
 drivers/firmware/Makefile   |   1 +
 drivers/firmware/arm_sdei.c | 616 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/sdei.h        | 100 +++++++
 include/uapi/linux/sdei.h   |  91 +++++++
 5 files changed, 815 insertions(+)
 create mode 100644 drivers/firmware/arm_sdei.c
 create mode 100644 include/linux/sdei.h
 create mode 100644 include/uapi/linux/sdei.h

diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
index 6e4ed5a9c6fd..d8a9859f6102 100644
--- a/drivers/firmware/Kconfig
+++ b/drivers/firmware/Kconfig
@@ -48,6 +48,13 @@ config ARM_SCPI_POWER_DOMAIN
 	  This enables support for the SCPI power domains which can be
 	  enabled or disabled via the SCP firmware
 
+config ARM_SDE_INTERFACE
+	bool "ARM Software Delegated Exception Interface (SDEI)"
+	help
+	  The Software Delegated Exception Interface (SDEI) is an ARM
+	  standard for registering callbacks from the platform firmware
+	  into the OS. This is typically used to implement RAS notifications.
+
 config EDD
 	tristate "BIOS Enhanced Disk Drive calls determine boot disk"
 	depends on X86
diff --git a/drivers/firmware/Makefile b/drivers/firmware/Makefile
index a37f12e8d137..229984feca4a 100644
--- a/drivers/firmware/Makefile
+++ b/drivers/firmware/Makefile
@@ -5,6 +5,7 @@ obj-$(CONFIG_ARM_PSCI_FW)	+= psci.o
 obj-$(CONFIG_ARM_PSCI_CHECKER)	+= psci_checker.o
 obj-$(CONFIG_ARM_SCPI_PROTOCOL)	+= arm_scpi.o
 obj-$(CONFIG_ARM_SCPI_POWER_DOMAIN) += scpi_pm_domain.o
+obj-$(CONFIG_ARM_SDE_INTERFACE)	+= arm_sdei.o
 obj-$(CONFIG_DMI)		+= dmi_scan.o
 obj-$(CONFIG_DMI_SYSFS)		+= dmi-sysfs.o
 obj-$(CONFIG_EDD)		+= edd.o
diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
new file mode 100644
index 000000000000..4b3c7fd99c34
--- /dev/null
+++ b/drivers/firmware/arm_sdei.c
@@ -0,0 +1,616 @@
+/*
+ * Software Delegated Exception Interface (SDEI) driver code.
+ *
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define pr_fmt(fmt) "sdei: " fmt
+
+#include <linux/acpi.h>
+#include <linux/arm-smccc.h>
+#include <linux/bitops.h>
+#include <linux/compiler.h>
+#include <linux/errno.h>
+#include <linux/hardirq.h>
+#include <linux/kernel.h>
+#include <linux/kprobes.h>
+#include <linux/kvm_host.h>
+#include <linux/list.h>
+#include <linux/lockdep.h>
+#include <linux/of.h>
+#include <linux/of_platform.h>
+#include <linux/percpu.h>
+#include <linux/platform_device.h>
+#include <linux/ptrace.h>
+#include <linux/preempt.h>
+#include <linux/sdei.h>
+#include <linux/slab.h>
+#include <linux/smp.h>
+#include <linux/spinlock.h>
+#include <linux/uaccess.h>
+#include <uapi/linux/sdei.h>
+
+/*
+ * The call to use to reach the firmware.
+ */
+static asmlinkage void (*sdei_firmware_call)(unsigned long function_id,
+		      unsigned long arg0, unsigned long arg1,
+		      unsigned long arg2, unsigned long arg3,
+		      unsigned long arg4, struct arm_smccc_res *res);
+
+/* entry point from firmware to arch asm code */
+static unsigned long sdei_entry_point;
+
+struct sdei_event {
+	struct list_head	list;
+	u32			event_num;
+	u8			type;
+	u8			priority;
+
+	/* This pointer is handed to firmware as the event argument. */
+	struct sdei_registered_event *registered;
+};
+
+static LIST_HEAD(sdei_events);
+static DEFINE_SPINLOCK(sdei_events_lock);
+
+static int sdei_to_linux_errno(unsigned long sdei_err)
+{
+	switch (sdei_err) {
+	case SDEI_NOT_SUPPORTED:
+		return -EOPNOTSUPP;
+	case SDEI_INVALID_PARAMETERS:
+		return -EINVAL;
+	case SDEI_DENIED:
+		return -EPERM;
+	case SDEI_PENDING:
+		return -EINPROGRESS;
+	case SDEI_OUT_OF_RESOURCE:
+		return -ENOMEM;
+	}
+
+	/* Not an error value ... */
+	return sdei_err;
+}
+
+/*
+ * If x0 is any of these values, then the call failed, use sdei_to_linux_errno()
+ * to translate.
+ */
+static int sdei_is_err(struct arm_smccc_res *res)
+{
+	switch (res->a0) {
+	case SDEI_NOT_SUPPORTED:
+	case SDEI_INVALID_PARAMETERS:
+	case SDEI_DENIED:
+	case SDEI_PENDING:
+	case SDEI_OUT_OF_RESOURCE:
+		return true;
+	}
+
+	return false;
+}
+
+static int invoke_sdei_fn(unsigned long function_id, unsigned long arg0,
+			  unsigned long arg1, unsigned long arg2,
+			  unsigned long arg3, unsigned long arg4,
+			  u64 *result)
+{
+	int err = 0;
+	struct arm_smccc_res res;
+
+	if (sdei_firmware_call) {
+		sdei_firmware_call(function_id, arg0, arg1, arg2, arg3, arg4,
+				   &res);
+		if (sdei_is_err(&res))
+			err = sdei_to_linux_errno(res.a0);
+	} else {
+		/*
+		 * !sdei_firmware_call means we failed to probe or called
+		 * sdei_mark_interface_broken(). -EIO is not an error returned
+		 * by sdei_to_linux_errno() and is used to suppress messages
+		 * from this driver.
+		 */
+		err = -EIO;
+		res.a0 = SDEI_NOT_SUPPORTED;
+	}
+
+	if (result)
+		*result = res.a0;
+
+	return err;
+}
+
+static struct sdei_event *sdei_event_find(u32 event_num)
+{
+	struct sdei_event *e, *found = NULL;
+
+	lockdep_assert_held(&sdei_events_lock);
+
+	list_for_each_entry(e, &sdei_events, list) {
+		if (e->event_num == event_num) {
+			found = e;
+			break;
+		}
+	}
+
+	return found;
+}
+
+int sdei_api_event_context(u32 query, u64 *result)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_CONTEXT, query, 0, 0, 0, 0,
+			      result);
+}
+
+static int sdei_api_event_get_info(u32 event, u32 info, u64 *result)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_GET_INFO, event, info, 0,
+			      0, 0, result);
+}
+
+static struct sdei_event *sdei_event_create(u32 event_num,
+					    sdei_event_callback *cb,
+					    void *cb_arg, gfp_t gfp_flags)
+{
+	int err;
+	u64 result;
+	struct sdei_event *event;
+	struct sdei_registered_event *reg;
+
+	lockdep_assert_held(&sdei_events_lock);
+
+	event = kzalloc(sizeof(*event), gfp_flags);
+	if (!event)
+		return ERR_PTR(-ENOMEM);
+
+	INIT_LIST_HEAD(&event->list);
+	event->event_num = event_num;
+
+	err = sdei_api_event_get_info(event_num, SDEI_EVENT_INFO_EV_PRIORITY,
+				      &result);
+	if (err) {
+		kfree(event);
+		return ERR_PTR(err);
+	}
+	event->priority = result;
+
+	err = sdei_api_event_get_info(event_num, SDEI_EVENT_INFO_EV_TYPE,
+				      &result);
+	if (err) {
+		kfree(event);
+		return ERR_PTR(err);
+	}
+	event->type = result;
+
+	if (event->type == SDEI_EVENT_TYPE_SHARED) {
+		reg = kzalloc(sizeof(*reg), GFP_KERNEL);
+		if (!reg) {
+			kfree(event);
+			return ERR_PTR(-ENOMEM);
+		}
+
+		reg->event_num = event_num;
+		reg->priority = event->priority;
+
+		reg->callback = cb;
+		reg->callback_arg = cb_arg;
+		event->registered = reg;
+	}
+
+	if (sdei_event_find(event_num)) {
+		kfree(event->registered);
+		kfree(event);
+		event = ERR_PTR(-EBUSY);
+	} else {
+		list_add(&event->list, &sdei_events);
+	}
+
+	return event;
+}
+
+static void sdei_event_destroy(struct sdei_event *event)
+{
+	lockdep_assert_held(&sdei_events_lock);
+
+	list_del(&event->list);
+
+	if (event->type == SDEI_EVENT_TYPE_SHARED)
+		kfree(event->registered);
+
+	kfree(event);
+}
+
+static int sdei_api_get_version(u64 *version)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_VERSION, 0, 0, 0, 0, 0, version);
+}
+
+int sdei_mask_local_cpu(void)
+{
+	int err;
+
+	WARN_ON_ONCE(preemptible());
+
+	err = invoke_sdei_fn(SDEI_1_0_FN_SDEI_PE_MASK, 0, 0, 0, 0, 0, NULL);
+	if (err && err != -EIO) {
+		pr_warn_once("failed to mask CPU[%u]: %d\n",
+			      smp_processor_id(), err);
+		return err;
+	}
+
+	return 0;
+}
+
+static void _ipi_mask_cpu(void *ignored)
+{
+	sdei_mask_local_cpu();
+}
+
+int sdei_unmask_local_cpu(void)
+{
+	int err;
+
+	WARN_ON_ONCE(preemptible());
+
+	err = invoke_sdei_fn(SDEI_1_0_FN_SDEI_PE_UNMASK, 0, 0, 0, 0, 0, NULL);
+	if (err && err != -EIO) {
+		pr_warn_once("failed to unmask CPU[%u]: %d\n",
+			     smp_processor_id(), err);
+		return err;
+	}
+
+	return 0;
+}
+
+static void _ipi_unmask_cpu(void *ignored)
+{
+	sdei_unmask_local_cpu();
+}
+
+static void _ipi_private_reset(void *ignored)
+{
+	int err;
+
+	err = invoke_sdei_fn(SDEI_1_0_FN_SDEI_PRIVATE_RESET, 0, 0, 0, 0, 0,
+			     NULL);
+	if (err && err != -EIO)
+		pr_warn_once("failed to reset CPU[%u]: %d\n",
+			     smp_processor_id(), err);
+}
+
+static int sdei_api_shared_reset(void)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_SHARED_RESET, 0, 0, 0, 0, 0,
+			      NULL);
+}
+
+static void sdei_mark_interface_broken(void)
+{
+	pr_err("disabling SDEI firmware interface\n");
+	on_each_cpu(&_ipi_mask_cpu, NULL, true);
+	sdei_firmware_call = NULL;
+}
+
+static int sdei_platform_reset(void)
+{
+	int err;
+
+	on_each_cpu(&_ipi_private_reset, NULL, true);
+	err = sdei_api_shared_reset();
+	if (err) {
+		pr_err("Failed to reset platform: %d\n", err);
+		sdei_mark_interface_broken();
+	}
+
+	return err;
+}
+
+static int sdei_api_event_enable(u32 event_num)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_ENABLE, event_num, 0, 0, 0,
+			      0, NULL);
+}
+
+int sdei_event_enable(u32 event_num)
+{
+	int err = -EINVAL;
+	struct sdei_event *event;
+
+	spin_lock(&sdei_events_lock);
+	event = sdei_event_find(event_num);
+	if (!event)
+		err = -ENOENT;
+	else if (event->type == SDEI_EVENT_TYPE_SHARED)
+		err = sdei_api_event_enable(event->event_num);
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(sdei_event_enable);
+
+static int sdei_api_event_disable(u32 event_num)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_DISABLE, event_num, 0, 0,
+			      0, 0, NULL);
+}
+
+int sdei_event_disable(u32 event_num)
+{
+	int err = -EINVAL;
+	struct sdei_event *event;
+
+	spin_lock(&sdei_events_lock);
+	event = sdei_event_find(event_num);
+	if (!event)
+		err = -ENOENT;
+	else if (event->type == SDEI_EVENT_TYPE_SHARED)
+		err = sdei_api_event_disable(event->event_num);
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(sdei_event_disable);
+
+static int sdei_api_event_unregister(u32 event_num)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_UNREGISTER, event_num, 0,
+			      0, 0, 0, NULL);
+}
+
+static int _sdei_event_unregister(struct sdei_event *event)
+{
+	if (event->type == SDEI_EVENT_TYPE_SHARED)
+		return sdei_api_event_unregister(event->event_num);
+
+	return -EINVAL;
+}
+
+int sdei_event_unregister(u32 event_num)
+{
+	int err;
+	struct sdei_event *event;
+
+	WARN_ON(in_nmi());
+
+	spin_lock(&sdei_events_lock);
+	event = sdei_event_find(event_num);
+	do {
+		if (!event) {
+			pr_warn("Event %u not registered\n", event_num);
+			err = -ENOENT;
+			break;
+		}
+
+		err = _sdei_event_unregister(event);
+		if (err)
+			break;
+
+		sdei_event_destroy(event);
+	} while (0);
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(sdei_event_unregister);
+
+static int sdei_api_event_register(u32 event_num, unsigned long entry_point,
+				   void *arg, u64 flags, u64 affinity)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_REGISTER, event_num,
+			      (unsigned long)entry_point, (unsigned long)arg,
+			      flags, affinity, NULL);
+}
+
+static int _sdei_event_register(struct sdei_event *event)
+{
+	if (event->type == SDEI_EVENT_TYPE_SHARED)
+		return sdei_api_event_register(event->event_num,
+					       sdei_entry_point,
+					       event->registered,
+					       SDEI_EVENT_REGISTER_RM_ANY, 0);
+
+	return -EINVAL;
+}
+
+int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
+			gfp_t gfp_flags)
+{
+	int err;
+	struct sdei_event *event;
+
+	WARN_ON(in_nmi());
+
+	spin_lock(&sdei_events_lock);
+	do {
+		if (sdei_event_find(event_num)) {
+			pr_warn("Event %u already registered\n", event_num);
+			err = -EBUSY;
+			break;
+		}
+
+		event = sdei_event_create(event_num, cb, arg, gfp_flags);
+		if (IS_ERR(event)) {
+			err = PTR_ERR(event);
+			pr_warn("Failed to create event %u: %d\n", event_num,
+				err);
+			break;
+		}
+
+		err = _sdei_event_register(event);
+		if (err) {
+			sdei_event_destroy(event);
+			pr_warn("Failed to register event %u: %d\n", event_num,
+				err);
+		}
+	} while (0);
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+EXPORT_SYMBOL(sdei_event_register);
+
+static void sdei_smccc_smc(unsigned long function_id,
+			   unsigned long arg0, unsigned long arg1,
+			   unsigned long arg2, unsigned long arg3,
+			   unsigned long arg4, struct arm_smccc_res *res)
+{
+	arm_smccc_smc(function_id, arg0, arg1, arg2, arg3, arg4, 0, 0, res);
+}
+
+static void sdei_smccc_hvc(unsigned long function_id,
+			   unsigned long arg0, unsigned long arg1,
+			   unsigned long arg2, unsigned long arg3,
+			   unsigned long arg4, struct arm_smccc_res *res)
+{
+	arm_smccc_hvc(function_id, arg0, arg1, arg2, arg3, arg4, 0, 0, res);
+}
+
+static int sdei_get_conduit(struct platform_device *pdev)
+{
+	const char *method;
+	struct device_node *np = pdev->dev.of_node;
+
+	sdei_firmware_call = NULL;
+	if (np) {
+		if (of_property_read_string(np, "method", &method)) {
+			pr_warn("missing \"method\" property\n");
+			return CONDUIT_INVALID;
+		}
+
+		if (!strcmp("hvc", method)) {
+			sdei_firmware_call = &sdei_smccc_hvc;
+			return CONDUIT_HVC;
+		} else if (!strcmp("smc", method)) {
+			sdei_firmware_call = &sdei_smccc_smc;
+			return CONDUIT_SMC;
+		}
+
+		pr_warn("invalid \"method\" property: %s\n", method);
+	}
+
+	return CONDUIT_INVALID;
+}
+
+static int sdei_probe(struct platform_device *pdev)
+{
+	int err;
+	u64 ver = 0;
+	int conduit;
+
+	conduit = sdei_get_conduit(pdev);
+	if (!sdei_firmware_call)
+		return 0;
+
+	err = sdei_api_get_version(&ver);
+	if (err == -EOPNOTSUPP)
+		pr_err("advertised but not implemented in platform firmware\n");
+	if (err) {
+		pr_err("Failed to get SDEI version: %d\n", err);
+		sdei_mark_interface_broken();
+		return err;
+	}
+
+	pr_info("SDEIv%d.%d (0x%x) detected in firmware.\n",
+		(int)SDEI_VERSION_MAJOR(ver), (int)SDEI_VERSION_MINOR(ver),
+		(int)SDEI_VERSION_VENDOR(ver));
+
+	if (SDEI_VERSION_MAJOR(ver) != 1) {
+		pr_warn("Conflicting SDEI version detected.\n");
+		sdei_mark_interface_broken();
+		return -EINVAL;
+	}
+
+	err = sdei_platform_reset();
+	if (err)
+		return err;
+
+	sdei_entry_point = sdei_arch_get_entry_point(conduit);
+	if (!sdei_entry_point) {
+		/* Not supported due to hardware or boot configuration */
+		sdei_mark_interface_broken();
+		return 0;
+	}
+
+	on_each_cpu(&_ipi_unmask_cpu, NULL, false);
+
+	return 0;
+}
+
+static const struct of_device_id sdei_of_match[] = {
+	{ .compatible = "arm,sdei-1.0" },
+	{}
+};
+
+static struct platform_driver sdei_driver = {
+	.driver		= {
+		.name			= "sdei",
+		.of_match_table		= sdei_of_match,
+	},
+	.probe		= sdei_probe,
+};
+
+static bool __init sdei_present_dt(void)
+{
+	struct platform_device *pdev;
+	struct device_node *np, *fw_np;
+
+	fw_np = of_find_node_by_name(NULL, "firmware");
+	if (!fw_np)
+		return false;
+
+	np = of_find_matching_node(fw_np, sdei_of_match);
+	of_node_put(fw_np);
+	if (!np)
+		return false;
+
+	pdev = of_platform_device_create(np, sdei_driver.driver.name, NULL);
+	of_node_put(np);
+	if (IS_ERR(pdev))
+		return false;
+
+	return true;
+}
+
+static int __init sdei_init(void)
+{
+	if (sdei_present_dt())
+		platform_driver_register(&sdei_driver);
+
+	return 0;
+}
+
+subsys_initcall_sync(sdei_init);
+
+int sdei_event_handler(struct pt_regs *regs,
+		       struct sdei_registered_event *arg)
+{
+	int err;
+	mm_segment_t orig_addr_limit;
+	u32 event_num = arg->event_num;
+
+	orig_addr_limit = get_fs();
+	set_fs(USER_DS);
+
+	err = arg->callback(event_num, regs, arg->callback_arg);
+	if (err)
+		pr_err_ratelimited("event %u on CPU %u failed with error: %d\n",
+				   event_num, smp_processor_id(), err);
+
+	set_fs(orig_addr_limit);
+
+	return err;
+}
+NOKPROBE_SYMBOL(sdei_event_handler);
diff --git a/include/linux/sdei.h b/include/linux/sdei.h
new file mode 100644
index 000000000000..bb3dd000771e
--- /dev/null
+++ b/include/linux/sdei.h
@@ -0,0 +1,100 @@
+/*
+ * Software Delegated Exception Interface (SDEI) API
+ *
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __LINUX_SDEI_H
+#define __LINUX_SDEI_H
+
+#include <linux/gfp.h>
+
+#include <uapi/linux/sdei.h>
+
+/* kvm needs to know if hvcs are for SDEI or PSCI */
+#define IS_SDEI_CALL(x)		((x & SDEI_1_0_MASK) == SDEI_1_0_FN_BASE)
+
+enum sdei_conduit_types {
+	CONDUIT_INVALID = 0,
+	CONDUIT_SMC,
+	CONDUIT_HVC,
+};
+
+/* Arch code should override this to set the entry point from firmware... */
+#ifndef sdei_arch_get_entry_point
+#define sdei_arch_get_entry_point(conduit)	(0)
+#endif
+
+/*
+ * When an event occurs sdei_event_handler() will call a user-provided callback
+ * like this in NMI context on the CPU that received the event.
+ */
+typedef int (sdei_event_callback)(u32 event, struct pt_regs *regs, void *arg);
+
+/*
+ * Register your callback to claim an event. The event must be described
+ * by firmware.
+ * This function allocates memory, pass an appropriate gfp_flags.
+ */
+int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
+			gfp_t gfp_flags);
+
+/*
+ * Calls to sdei_event_unregister() may return EINPROGRESS. Keep calling
+ * it until it succeeds.
+ */
+int sdei_event_unregister(u32 event_num);
+
+int sdei_event_enable(u32 event_num);
+int sdei_event_disable(u32 event_num);
+
+#ifdef CONFIG_ARM_SDE_INTERFACE
+/* For use by arch code when CPU hotplug notifiers are not appropriate. */
+int sdei_mask_local_cpu(void);
+int sdei_unmask_local_cpu(void);
+#else
+static inline int sdei_mask_local_cpu(void) { return 0; }
+static inline int sdei_unmask_local_cpu(void) { return 0; }
+#endif /* CONFIG_ARM_SDE_INTERFACE */
+
+
+/*
+ * This struct represents an event that has been registered. The driver
+ * maintains a list of all events, and which ones are registered. (Private
+ * events have one entry in the list, but are registered on each CPU).
+ * A pointer to this struct is passed to firmware, and back to the event
+ * handler. The event handler can then use this to invoke the registered
+ * callback, without having to walk the list.
+ *
+ * For CPU private events, this structure is per-cpu.
+ */
+struct sdei_registered_event {
+	/* For use by arch code: */
+	struct pt_regs          interrupted_regs;
+
+	sdei_event_callback	*callback;
+	void			*callback_arg;
+	u32			 event_num;
+	u8			 priority;
+};
+
+/* The arch code entry point should then call this when an event arrives. */
+int notrace sdei_event_handler(struct pt_regs *regs,
+			       struct sdei_registered_event *arg);
+
+/* arch code may use this to retrieve the extra registers. */
+int sdei_api_event_context(u32 query, u64 *result);
+
+#endif /* __LINUX_SDEI_H */
diff --git a/include/uapi/linux/sdei.h b/include/uapi/linux/sdei.h
new file mode 100644
index 000000000000..535c9d8d14cc
--- /dev/null
+++ b/include/uapi/linux/sdei.h
@@ -0,0 +1,91 @@
+/*
+ *Software Delegated Exception Interface (SDEI) Firmware API
+ *
+ * This header holds common SDEI defines and macros shared
+ * by: ARM64 kernel, KVM and user space.
+ *
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef _UAPI_LINUX_SDEI_H
+#define _UAPI_LINUX_SDEI_H
+
+#define SDEI_1_0_FN_BASE			0xC4000020
+#define SDEI_1_0_MASK				0xFFFFFFE0
+#define SDEI_1_0_FN(n)				(SDEI_1_0_FN_BASE + (n))
+
+#define SDEI_1_0_FN_SDEI_VERSION			SDEI_1_0_FN(0x00)
+#define SDEI_1_0_FN_SDEI_EVENT_REGISTER			SDEI_1_0_FN(0x01)
+#define SDEI_1_0_FN_SDEI_EVENT_ENABLE			SDEI_1_0_FN(0x02)
+#define SDEI_1_0_FN_SDEI_EVENT_DISABLE			SDEI_1_0_FN(0x03)
+#define SDEI_1_0_FN_SDEI_EVENT_CONTEXT			SDEI_1_0_FN(0x04)
+#define SDEI_1_0_FN_SDEI_EVENT_COMPLETE			SDEI_1_0_FN(0x05)
+#define SDEI_1_0_FN_SDEI_EVENT_COMPLETE_AND_RESUME	SDEI_1_0_FN(0x06)
+#define SDEI_1_0_FN_SDEI_EVENT_UNREGISTER		SDEI_1_0_FN(0x07)
+#define SDEI_1_0_FN_SDEI_EVENT_STATUS			SDEI_1_0_FN(0x08)
+#define SDEI_1_0_FN_SDEI_EVENT_GET_INFO			SDEI_1_0_FN(0x09)
+#define SDEI_1_0_FN_SDEI_EVENT_ROUTING_SET		SDEI_1_0_FN(0x0A)
+#define SDEI_1_0_FN_SDEI_PE_MASK			SDEI_1_0_FN(0x0B)
+#define SDEI_1_0_FN_SDEI_PE_UNMASK			SDEI_1_0_FN(0x0C)
+#define SDEI_1_0_FN_SDEI_INTERRUPT_BIND			SDEI_1_0_FN(0x0D)
+#define SDEI_1_0_FN_SDEI_INTERRUPT_RELEASE		SDEI_1_0_FN(0x0E)
+#define SDEI_1_0_FN_SDEI_PRIVATE_RESET			SDEI_1_0_FN(0x11)
+#define SDEI_1_0_FN_SDEI_SHARED_RESET			SDEI_1_0_FN(0x12)
+
+#define SDEI_VERSION_MAJOR_SHIFT			48
+#define SDEI_VERSION_MAJOR_MASK				0x7fff
+#define SDEI_VERSION_MINOR_SHIFT			32
+#define SDEI_VERSION_MINOR_MASK				0xffff
+#define SDEI_VERSION_VENDOR_SHIFT			0
+#define SDEI_VERSION_VENDOR_MASK			0xffffffff
+
+#define SDEI_VERSION_MAJOR(x)	(x>>SDEI_VERSION_MAJOR_SHIFT & SDEI_VERSION_MAJOR_MASK)
+#define SDEI_VERSION_MINOR(x)	(x>>SDEI_VERSION_MINOR_SHIFT & SDEI_VERSION_MINOR_MASK)
+#define SDEI_VERSION_VENDOR(x)	(x>>SDEI_VERSION_VENDOR_SHIFT & SDEI_VERSION_VENDOR_MASK)
+
+/* SDEI return values */
+#define SDEI_SUCCESS		0
+#define SDEI_NOT_SUPPORTED	-1
+#define SDEI_INVALID_PARAMETERS	-2
+#define SDEI_DENIED		-3
+#define SDEI_PENDING		-5
+#define SDEI_OUT_OF_RESOURCE	-10
+
+/* EVENT_REGISTER flags */
+#define SDEI_EVENT_REGISTER_RM_ANY	0
+#define SDEI_EVENT_REGISTER_RM_PE	1
+
+/* EVENT_STATUS return value bits */
+#define SDEI_EVENT_STATUS_RUNNING	2
+#define SDEI_EVENT_STATUS_ENABLED	1
+#define SDEI_EVENT_STATUS_REGISTERED	0
+
+/* EVENT_COMPLETE status values */
+#define SDEI_EV_HANDLED	0
+#define SDEI_EV_FAILED	1
+
+/* GET_INFO values */
+#define SDEI_EVENT_INFO_EV_TYPE			0
+#define SDEI_EVENT_INFO_EV_SIGNALED		1
+#define SDEI_EVENT_INFO_EV_PRIORITY		2
+#define SDEI_EVENT_INFO_EV_ROUTING_MODE		3
+#define SDEI_EVENT_INFO_EV_ROUTING_AFF		4
+
+/* and their results */
+#define SDEI_EVENT_TYPE_PRIVATE			0
+#define SDEI_EVENT_TYPE_SHARED			1
+#define SDEI_EVENT_PRIORITY_NORMAL		0
+#define SDEI_EVENT_PRIORITY_CRITICAL		1
+
+#endif /* _UAPI_LINUX_SDEI_H */
-- 
2.13.3

[PATCH v3 08/13] arm64: Add vmap_stack header file

[James Morse]

Today the arm64 arch code allocates an extra IRQ stack per-cpu. If we
also have SDEI and VMAP stacks we need two extra per-cpu VMAP stacks.

Move the VMAP stack allocation out to a helper in a new header file.
This avoids missing THREADINFO_GFP, or getting the all-important alignment
wrong.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>

---
This patch is new for v3. Having a bare prototype if VMAP stacks aren't
enabled allows us to avoid ifdeffery but still forbid building code using
the symbol.

 arch/arm64/include/asm/vmap_stack.h | 41 +++++++++++++++++++++++++++++++++++++
 arch/arm64/kernel/irq.c             | 13 ++----------
 2 files changed, 43 insertions(+), 11 deletions(-)
 create mode 100644 arch/arm64/include/asm/vmap_stack.h

diff --git a/arch/arm64/include/asm/vmap_stack.h b/arch/arm64/include/asm/vmap_stack.h
new file mode 100644
index 000000000000..f41d043cac31
--- /dev/null
+++ b/arch/arm64/include/asm/vmap_stack.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __ASM_VMAP_STACK_H
+#define __ASM_VMAP_STACK_H
+
+#include <linux/vmalloc.h>
+#include <asm/memory.h>
+#include <asm/pgtable.h>
+#include <asm/thread_info.h>
+
+#ifdef CONFIG_VMAP_STACK
+/*
+ * To ensure that VMAP'd stack overflow detection works correctly, all VMAP'd
+ * stacks need to have the same alignment.
+ */
+static inline unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node)
+{
+	return __vmalloc_node_range(stack_size, THREAD_ALIGN,
+				    VMALLOC_START, VMALLOC_END,
+				    THREADINFO_GFP, PAGE_KERNEL, 0, node,
+				    __builtin_return_address(0));
+}
+
+#else
+unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node); // link error
+
+#endif /* CONFIG_VMAP_STACK */
+#endif /* __ASM_VMAP_STACK_H */
diff --git a/arch/arm64/kernel/irq.c b/arch/arm64/kernel/irq.c
index 713561e5bcab..60e5fc661f74 100644
--- a/arch/arm64/kernel/irq.c
+++ b/arch/arm64/kernel/irq.c
@@ -29,6 +29,7 @@
 #include <linux/irqchip.h>
 #include <linux/seq_file.h>
 #include <linux/vmalloc.h>
+#include <asm/vmap_stack.h>
 
 unsigned long irq_err_count;
 
@@ -58,17 +59,7 @@ static void init_irq_stacks(void)
 	unsigned long *p;
 
 	for_each_possible_cpu(cpu) {
-		/*
-		* To ensure that VMAP'd stack overflow detection works
-		* correctly, the IRQ stacks need to have the same
-		* alignment as other stacks.
-		*/
-		p = __vmalloc_node_range(IRQ_STACK_SIZE, THREAD_ALIGN,
-					 VMALLOC_START, VMALLOC_END,
-					 THREADINFO_GFP, PAGE_KERNEL,
-					 0, cpu_to_node(cpu),
-					 __builtin_return_address(0));
-
+		p = arch_alloc_vmap_stack(IRQ_STACK_SIZE, cpu_to_node(cpu));
 		per_cpu(irq_stack_ptr, cpu) = p;
 	}
 }
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 08/13] arm64: Add vmap_stack header file

[James Morse]

Today the arm64 arch code allocates an extra IRQ stack per-cpu. If we
also have SDEI and VMAP stacks we need two extra per-cpu VMAP stacks.

Move the VMAP stack allocation out to a helper in a new header file.
This avoids missing THREADINFO_GFP, or getting the all-important alignment
wrong.

Signed-off-by: James Morse <james.morse@arm.com>

---
This patch is new for v3. Having a bare prototype if VMAP stacks aren't
enabled allows us to avoid ifdeffery but still forbid building code using
the symbol.

 arch/arm64/include/asm/vmap_stack.h | 41 +++++++++++++++++++++++++++++++++++++
 arch/arm64/kernel/irq.c             | 13 ++----------
 2 files changed, 43 insertions(+), 11 deletions(-)
 create mode 100644 arch/arm64/include/asm/vmap_stack.h

diff --git a/arch/arm64/include/asm/vmap_stack.h b/arch/arm64/include/asm/vmap_stack.h
new file mode 100644
index 000000000000..f41d043cac31
--- /dev/null
+++ b/arch/arm64/include/asm/vmap_stack.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __ASM_VMAP_STACK_H
+#define __ASM_VMAP_STACK_H
+
+#include <linux/vmalloc.h>
+#include <asm/memory.h>
+#include <asm/pgtable.h>
+#include <asm/thread_info.h>
+
+#ifdef CONFIG_VMAP_STACK
+/*
+ * To ensure that VMAP'd stack overflow detection works correctly, all VMAP'd
+ * stacks need to have the same alignment.
+ */
+static inline unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node)
+{
+	return __vmalloc_node_range(stack_size, THREAD_ALIGN,
+				    VMALLOC_START, VMALLOC_END,
+				    THREADINFO_GFP, PAGE_KERNEL, 0, node,
+				    __builtin_return_address(0));
+}
+
+#else
+unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node); // link error
+
+#endif /* CONFIG_VMAP_STACK */
+#endif /* __ASM_VMAP_STACK_H */
diff --git a/arch/arm64/kernel/irq.c b/arch/arm64/kernel/irq.c
index 713561e5bcab..60e5fc661f74 100644
--- a/arch/arm64/kernel/irq.c
+++ b/arch/arm64/kernel/irq.c
@@ -29,6 +29,7 @@
 #include <linux/irqchip.h>
 #include <linux/seq_file.h>
 #include <linux/vmalloc.h>
+#include <asm/vmap_stack.h>
 
 unsigned long irq_err_count;
 
@@ -58,17 +59,7 @@ static void init_irq_stacks(void)
 	unsigned long *p;
 
 	for_each_possible_cpu(cpu) {
-		/*
-		* To ensure that VMAP'd stack overflow detection works
-		* correctly, the IRQ stacks need to have the same
-		* alignment as other stacks.
-		*/
-		p = __vmalloc_node_range(IRQ_STACK_SIZE, THREAD_ALIGN,
-					 VMALLOC_START, VMALLOC_END,
-					 THREADINFO_GFP, PAGE_KERNEL,
-					 0, cpu_to_node(cpu),
-					 __builtin_return_address(0));
-
+		p = arch_alloc_vmap_stack(IRQ_STACK_SIZE, cpu_to_node(cpu));
 		per_cpu(irq_stack_ptr, cpu) = p;
 	}
 }
-- 
2.13.3

[PATCH v3 09/13] arm64: kernel: Add arch-specific SDEI entry code and CPU masking

[James Morse]

The Software Delegated Exception Interface (SDEI) is an ARM standard
for registering callbacks from the platform firmware into the OS.
This is typically used to implement RAS notifications.

Such notifications enter the kernel at the registered entry-point
with the register values of the interrupted CPU context. Because this
is not a CPU exception, it cannot reuse the existing entry code.
(crucially we don't implicitly know which exception level we interrupted),

Add an sdei-entry.S asm file to set us up for calling into C code. If
the event interrupted code that had interrupts masked, we always return
to that location. Otherwise we pretend this was an IRQ, and use SDEI's
complete_and_resume call to return to vbar_el1 + offset.

This allows the kernel to deliver signals to user space processes. For
KVM this triggers the world switch, a quick spin round vcpu_run, then
back into the guest, unless there are pending signals.

Add sdei_mask_local_cpu() calls to the smp_send_stop() code, this covers
the panic() code-path, which doesn't invoke cpuhotplug notifiers.

Because we can interrupt entry-from/exit-to another EL, we can't trust the
value in sp_el0 or x29, even if we interrupted the kernel, in this case
the code in entry.S will save/restore sp_el0 and use the value in
__entry_task.

When we have VMAP stacks we can interrupt the stack-overflow test, which
stirs x0 into sp, meaning we have to have our own VMAP stack.

Signed-off-by: James Morse <james.morse@arm.com>
---
Changes since v2:
 * Added PAN-setting as we didn't take an exception to get here.
 * Added VMAP stack allocation and switching.
 * overwrite x29 on entry from not-the-kernel.
 * Added a pe-mask to crash_smp_send_stop()s 'no secondaries' path (oops).
 * Added ELR-hasn't-changed test. Any exception in the handler will panic KVM
   as its switched VBAR_EL1, but go silently undetected otherwise. Generate a
   warning.

 arch/arm64/Kconfig                  |   2 +-
 arch/arm64/include/asm/sdei.h       |  63 ++++++++++
 arch/arm64/include/asm/stacktrace.h |   3 +
 arch/arm64/kernel/Makefile          |   1 +
 arch/arm64/kernel/asm-offsets.c     |   5 +
 arch/arm64/kernel/sdei-entry.S      | 122 +++++++++++++++++++
 arch/arm64/kernel/sdei.c            | 235 ++++++++++++++++++++++++++++++++++++
 arch/arm64/kernel/smp.c             |  11 +-
 drivers/firmware/Kconfig            |   1 +
 drivers/firmware/arm_sdei.c         |   1 +
 include/linux/sdei.h                |   2 +
 11 files changed, 444 insertions(+), 2 deletions(-)
 create mode 100644 arch/arm64/include/asm/sdei.h
 create mode 100644 arch/arm64/kernel/sdei-entry.S
 create mode 100644 arch/arm64/kernel/sdei.c

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index 0df64a6a56d4..70dfe4e9ccc5 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -98,7 +98,7 @@ config ARM64
 	select HAVE_IRQ_TIME_ACCOUNTING
 	select HAVE_MEMBLOCK
 	select HAVE_MEMBLOCK_NODE_MAP if NUMA
-	select HAVE_NMI if ACPI_APEI_SEA
+	select HAVE_NMI
 	select HAVE_PATA_PLATFORM
 	select HAVE_PERF_EVENTS
 	select HAVE_PERF_REGS
diff --git a/arch/arm64/include/asm/sdei.h b/arch/arm64/include/asm/sdei.h
new file mode 100644
index 000000000000..ed329e01a301
--- /dev/null
+++ b/arch/arm64/include/asm/sdei.h
@@ -0,0 +1,63 @@
+/*
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __ASM_SDEI_H
+#define __ASM_SDEI_H
+
+/* Values for sdei_exit_mode */
+#define SDEI_EXIT_HVC  0
+#define SDEI_EXIT_SMC  1
+
+#define SDEI_STACK_SIZE		IRQ_STACK_SIZE
+
+#ifndef __ASSEMBLY__
+
+#include <linux/linkage.h>
+#include <linux/types.h>
+
+#include <asm/virt.h>
+
+extern unsigned long sdei_exit_mode;
+
+/* Software Delegated Exception entry point from firmware*/
+asmlinkage void __sdei_asm_handler(unsigned long event_num, unsigned long arg,
+				   unsigned long pc, unsigned long pstate);
+
+/*
+ * The above entry point does the minimum to call C code. This function does
+ * anything else, before calling the driver.
+ */
+struct sdei_registered_event;
+asmlinkage unsigned long __sdei_handler(struct pt_regs *regs,
+					struct sdei_registered_event *arg);
+
+extern unsigned long sdei_arch_get_entry_point(int conduit);
+#define sdei_arch_get_entry_point(x)	sdei_arch_get_entry_point(x)
+
+bool _on_sdei_stack(unsigned long sp);
+static inline bool on_sdei_stack(unsigned long sp)
+{
+	if (!IS_ENABLED(CONFIG_VMAP_STACK))
+		return false;
+	if (!IS_ENABLED(CONFIG_ARM_SDE_INTERFACE))
+		return false;
+	if (in_nmi())
+		return _on_sdei_stack(sp);
+
+	return false;
+}
+
+#endif /* __ASSEMBLY__ */
+#endif	/* __ASM_SDEI_H */
diff --git a/arch/arm64/include/asm/stacktrace.h b/arch/arm64/include/asm/stacktrace.h
index 6ad30776e984..472ef944e932 100644
--- a/arch/arm64/include/asm/stacktrace.h
+++ b/arch/arm64/include/asm/stacktrace.h
@@ -22,6 +22,7 @@
 
 #include <asm/memory.h>
 #include <asm/ptrace.h>
+#include <asm/sdei.h>
 
 struct stackframe {
 	unsigned long fp;
@@ -85,6 +86,8 @@ static inline bool on_accessible_stack(struct task_struct *tsk, unsigned long sp
 		return true;
 	if (on_overflow_stack(sp))
 		return true;
+	if (on_sdei_stack(sp))
+		return true;
 
 	return false;
 }
diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index f2b4e816b6de..d22bae586ade 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -53,6 +53,7 @@ arm64-obj-$(CONFIG_KEXEC)		+= machine_kexec.o relocate_kernel.o	\
 arm64-obj-$(CONFIG_ARM64_RELOC_TEST)	+= arm64-reloc-test.o
 arm64-reloc-test-y := reloc_test_core.o reloc_test_syms.o
 arm64-obj-$(CONFIG_CRASH_DUMP)		+= crash_dump.o
+arm64-obj-$(CONFIG_ARM_SDE_INTERFACE)	+= sdei.o sdei-entry.o
 
 obj-y					+= $(arm64-obj-y) vdso/ probes/
 obj-m					+= $(arm64-obj-m)
diff --git a/arch/arm64/kernel/asm-offsets.c b/arch/arm64/kernel/asm-offsets.c
index 71bf088f1e4b..f6baa4ca79a9 100644
--- a/arch/arm64/kernel/asm-offsets.c
+++ b/arch/arm64/kernel/asm-offsets.c
@@ -22,6 +22,7 @@
 #include <linux/mm.h>
 #include <linux/dma-mapping.h>
 #include <linux/kvm_host.h>
+#include <linux/sdei.h>
 #include <linux/suspend.h>
 #include <asm/cpufeature.h>
 #include <asm/thread_info.h>
@@ -154,5 +155,9 @@ int main(void)
   DEFINE(HIBERN_PBE_ADDR,	offsetof(struct pbe, address));
   DEFINE(HIBERN_PBE_NEXT,	offsetof(struct pbe, next));
   DEFINE(ARM64_FTR_SYSVAL,	offsetof(struct arm64_ftr_reg, sys_val));
+#ifdef CONFIG_ARM_SDE_INTERFACE
+  DEFINE(SDEI_EVENT_INTREGS,	offsetof(struct sdei_registered_event, interrupted_regs));
+  DEFINE(SDEI_EVENT_PRIORITY,	offsetof(struct sdei_registered_event, priority));
+#endif
   return 0;
 }
diff --git a/arch/arm64/kernel/sdei-entry.S b/arch/arm64/kernel/sdei-entry.S
new file mode 100644
index 000000000000..cf12f8da0789
--- /dev/null
+++ b/arch/arm64/kernel/sdei-entry.S
@@ -0,0 +1,122 @@
+/*
+ * Software Delegated Exception entry point from firmware to the kernel.
+ *
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/linkage.h>
+
+#include <asm/alternative.h>
+#include <asm/asm-offsets.h>
+#include <asm/assembler.h>
+#include <asm/memory.h>
+#include <asm/ptrace.h>
+#include <asm/sdei.h>
+#include <uapi/linux/sdei.h>
+
+/*
+ * Software Delegated Exception entry point.
+ *
+ * x0: Event number
+ * x1: struct sdei_registered_event argument from registration time.
+ * x2: interrupted PC
+ * x3: interrupted PSTATE
+ *
+ * Firmware has preserved x0->x17 for us, we must save/restore the rest to
+ * follow SMC-CC. We save (or retrieve) all the registers as the handler may
+ * want them.
+ */
+ENTRY(__sdei_asm_handler)
+	stp     x2, x3, [x1, #SDEI_EVENT_INTREGS + S_PC]
+	stp     x4, x5, [x1, #SDEI_EVENT_INTREGS + 16 * 2]
+	stp     x6, x7, [x1, #SDEI_EVENT_INTREGS + 16 * 3]
+	stp     x8, x9, [x1, #SDEI_EVENT_INTREGS + 16 * 4]
+	stp     x10, x11, [x1, #SDEI_EVENT_INTREGS + 16 * 5]
+	stp     x12, x13, [x1, #SDEI_EVENT_INTREGS + 16 * 6]
+	stp     x14, x15, [x1, #SDEI_EVENT_INTREGS + 16 * 7]
+	stp     x16, x17, [x1, #SDEI_EVENT_INTREGS + 16 * 8]
+	stp     x18, x19, [x1, #SDEI_EVENT_INTREGS + 16 * 9]
+	stp     x20, x21, [x1, #SDEI_EVENT_INTREGS + 16 * 10]
+	stp     x22, x23, [x1, #SDEI_EVENT_INTREGS + 16 * 11]
+	stp     x24, x25, [x1, #SDEI_EVENT_INTREGS + 16 * 12]
+	stp     x26, x27, [x1, #SDEI_EVENT_INTREGS + 16 * 13]
+	stp     x28, x29, [x1, #SDEI_EVENT_INTREGS + 16 * 14]
+	mov	x4, sp
+	stp     lr, x4, [x1, #SDEI_EVENT_INTREGS + S_LR]
+
+	mov	x19, x1
+
+#ifdef CONFIG_VMAP_STACK
+	/*
+	 * entry.S may have been using sp as a scratch register, find whether
+	 * this is a normal or critical event and switch to the appropriate
+	 * stack for this CPU.
+	 */
+	ldrb	w4, [x19, #SDEI_EVENT_PRIORITY]
+	cbnz	w4, 1f
+	ldr_this_cpu dst=x5, sym=sdei_stack_normal_ptr, tmp=x6
+	b	2f
+1:	ldr_this_cpu dst=x5, sym=sdei_stack_critical_ptr, tmp=x6
+2:	mov	x6, #SDEI_STACK_SIZE
+	add	x5, x5, x6
+	mov	sp, x5
+#endif
+
+	/*
+	 * We may have interrupted userspace, or a guest, or exit-from or
+	 * return-to either of these. We can't trust sp_el0, restore it.
+	 */
+	mrs	x28, sp_el0
+	ldr_this_cpu	dst=x0, sym=__entry_task, tmp=x1
+	msr	sp_el0, x0
+
+	/* If we interrupted the kernel point to the previous stack/frame. */
+	and     x0, x3, #0xc
+	mrs     x1, CurrentEL
+	cmp     x0, x1
+	csel	x29, x29, xzr, eq	// fp, or zero
+	csel	x4, x2, xzr, eq		// elr, or zero
+
+	stp	x29, x4, [sp, #-16]!
+	mov	x29, sp
+
+	add	x0, x19, #SDEI_EVENT_INTREGS
+	mov	x1, x19
+	bl	__sdei_handler
+
+	msr	sp_el0, x28
+	/* restore regs >x17 that we clobbered */
+	ldp     x28, x29, [x19, #SDEI_EVENT_INTREGS + 16 * 14]
+	ldp     lr, x4, [x19, #SDEI_EVENT_INTREGS + S_LR]
+	mov	sp, x4
+	ldp     x18, x19, [x19, #SDEI_EVENT_INTREGS + 16 * 9]
+
+	mov	x1, x0			// address to complete_and_resume
+	/* x0 = (x0 <= 1) ? EVENT_COMPLETE:EVENT_COMPLETE_AND_RESUME */
+	cmp	x0, #1
+	mov_q	x2, SDEI_1_0_FN_SDEI_EVENT_COMPLETE
+	mov_q	x3, SDEI_1_0_FN_SDEI_EVENT_COMPLETE_AND_RESUME
+	csel	x0, x2, x3, ls
+
+	/* On success, this call never returns... */
+	ldr_l	x2, sdei_exit_mode
+	cmp	x2, #SDEI_EXIT_SMC
+	b.ne	1f
+	smc	#0
+	b	.
+1:	hvc	#0
+	b	.
+ENDPROC(__sdei_asm_handler)
+NOKPROBE(__sdei_asm_handler)
diff --git a/arch/arm64/kernel/sdei.c b/arch/arm64/kernel/sdei.c
new file mode 100644
index 000000000000..16fff51a699f
--- /dev/null
+++ b/arch/arm64/kernel/sdei.c
@@ -0,0 +1,235 @@
+/*
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define pr_fmt(fmt) "sdei: " fmt
+
+#include <linux/sched/task_stack.h>
+#include <linux/irqflags.h>
+#include <linux/hardirq.h>
+#include <linux/sdei.h>
+
+#include <asm/alternative.h>
+#include <asm/kprobes.h>
+#include <asm/ptrace.h>
+#include <asm/sysreg.h>
+#include <asm/vmap_stack.h>
+
+unsigned long sdei_exit_mode;
+
+/*
+ * VMAP'd stacks checking for stack overflow on exception using sp as a scratch
+ * register, meaning SDEI has to switch to its own stack. We need two stacks as
+ * a critical event may interrupt a normal event that has just taken a
+ * synchronous exception, and is using sp as scratch register. For a critical
+ * event interrupting a normal event, we can't reliably tell if we were on the
+ * sdei stack.
+ * For private events we need per-cpu stacks, these can't be allocated lazily
+ * as THREADINFO_GFP is a subset of GFP_KERNEL which can't necessarily be used
+ * when the first private event is registered.
+ */
+DECLARE_PER_CPU(unsigned long *, sdei_stack_normal_ptr);
+DECLARE_PER_CPU(unsigned long *, sdei_stack_critical_ptr);
+
+#ifdef CONFIG_VMAP_STACK
+DEFINE_PER_CPU(unsigned long *, sdei_stack_normal_ptr);
+DEFINE_PER_CPU(unsigned long *, sdei_stack_critical_ptr);
+#endif
+
+static void _free_sdei_stack(unsigned long * __percpu *ptr, int cpu)
+{
+	unsigned long *p;
+
+	p = per_cpu(*ptr, cpu);
+	if (p) {
+		per_cpu(*ptr, cpu) = NULL;
+		vfree(p);
+	}
+}
+
+static void free_sdei_stacks(void)
+{
+	int cpu;
+
+	for_each_possible_cpu(cpu) {
+		_free_sdei_stack(&sdei_stack_normal_ptr, cpu);
+		_free_sdei_stack(&sdei_stack_critical_ptr, cpu);
+	}
+}
+
+static int _init_sdei_stack(unsigned long * __percpu *ptr, int cpu)
+{
+	unsigned long *p;
+
+	p = arch_alloc_vmap_stack(SDEI_STACK_SIZE, cpu_to_node(cpu));
+	if (!p)
+		return -ENOMEM;
+	per_cpu(*ptr, cpu) = p;
+
+	return 0;
+}
+
+static int init_sdei_stacks(void)
+{
+	int cpu;
+	int err = 0;
+
+	for_each_possible_cpu(cpu) {
+		err = _init_sdei_stack(&sdei_stack_normal_ptr, cpu);
+		if (err)
+			break;
+		err = _init_sdei_stack(&sdei_stack_critical_ptr, cpu);
+		if (err)
+			break;
+	}
+
+	if (err)
+		free_sdei_stacks();
+
+	return err;
+}
+
+bool _on_sdei_stack(unsigned long sp)
+{
+	unsigned long low, high;
+
+	if (!IS_ENABLED(CONFIG_VMAP_STACK))
+		return false;
+
+	low = (unsigned long)raw_cpu_read(sdei_stack_critical_ptr);
+	high = low + SDEI_STACK_SIZE;
+
+	if (low <= sp && sp < high)
+		return true;
+
+	low = (unsigned long)raw_cpu_read(sdei_stack_normal_ptr);
+	high = low + SDEI_STACK_SIZE;
+
+	return (low <= sp && sp < high);
+}
+
+unsigned long sdei_arch_get_entry_point(int conduit)
+{
+	/*
+	 * SDEI works between adjacent exception levels. If we booted at EL1 we
+	 * assume a hypervisor is marshalling events. If we booted at EL2 and
+	 * dropped to EL1 because we don't support VHE, then we can't support
+	 * SDEI.
+	 */
+	if (is_hyp_mode_available() && !is_kernel_in_hyp_mode()) {
+		pr_err("Not supported on this hardware/boot configuration\n");
+		return 0;
+	}
+
+	if (IS_ENABLED(CONFIG_VMAP_STACK)) {
+		if (init_sdei_stacks())
+			return 0;
+	}
+
+	sdei_exit_mode = (conduit == CONDUIT_HVC) ? SDEI_EXIT_HVC : SDEI_EXIT_SMC;
+	return (unsigned long)__sdei_asm_handler;
+}
+
+/*
+ * __sdei_handler() returns one of:
+ *  SDEI_EV_HANDLED -  success, return to the interrupted context.
+ *  SDEI_EV_FAILED  -  failure, return this error code to firmare.
+ *  virtual-address -  success, return to this address.
+ */
+static __kprobes unsigned long _sdei_handler(struct pt_regs *regs,
+					     struct sdei_registered_event *arg)
+{
+	u32 mode;
+	int i, err = 0;
+	int clobbered_registers = 4;
+	u64 elr = read_sysreg(elr_el1);
+	u32 kernel_mode = read_sysreg(CurrentEL) | 1;	/* +SPSel */
+	unsigned long vbar = read_sysreg(vbar_el1);
+
+	/* Retrieve the missing registers values */
+	for (i = 0; i < clobbered_registers; i++) {
+		/* from within the handler, this call always succeeds */
+		sdei_api_event_context(i, &regs->regs[i]);
+	}
+
+	/*
+	 * We didn't take an exception to get here, set PAN. UAO will be cleared
+	 * by sdei_event_handler()s set_fs(USER_DS) call.
+	 */
+	asm(ALTERNATIVE("nop", SET_PSTATE_PAN(1), ARM64_HAS_PAN,
+			CONFIG_ARM64_PAN));
+
+	err = sdei_event_handler(regs, arg);
+	if (err)
+		return SDEI_EV_FAILED;
+
+	if (elr != read_sysreg(elr_el1)) {
+		/*
+		 * We took a synchronous exception from the SDEI handler.
+		 * This could deadlock, and if you interrupt KVM it will
+		 * hyp-panic instead.
+		 */
+		pr_warn("unsafe: exception during handler\n");
+	}
+
+	mode = regs->pstate & (PSR_MODE32_BIT | PSR_MODE_MASK);
+
+	/*
+	 * If we interrupted the kernel with interrupts masked, we always go
+	 * back to wherever we came from.
+	 */
+	if (mode == kernel_mode && !interrupts_enabled(regs))
+		return SDEI_EV_HANDLED;
+
+	/*
+	 * Otherwise, we pretend this was an IRQ. This lets user space tasks
+	 * receive signals before we return to them, and KVM to invoke it's
+	 * world switch to do the same.
+	 *
+	 * See DDI0487B.a Table D1-7 'Vector offsets from vector table base
+	 * address'.
+	 */
+	if (mode == kernel_mode)
+		return vbar + 0x280;
+	else if (mode & PSR_MODE32_BIT)
+		return vbar + 0x680;
+
+	return vbar + 0x480;
+}
+
+
+asmlinkage __kprobes notrace unsigned long
+__sdei_handler(struct pt_regs *regs, struct sdei_registered_event *arg)
+{
+	unsigned long ret;
+	bool do_nmi_exit = false;
+
+	/*
+	 * nmi_enter() deals with printk() re-entrance and use of RCU when
+	 * RCU believed this CPU was idle. Because critical events can
+	 * interrupt normal events, we may already be in_nmi().
+	 */
+	if (!in_nmi()) {
+		nmi_enter();
+		do_nmi_exit = true;
+	}
+
+	ret = _sdei_handler(regs, arg);
+
+	if (do_nmi_exit)
+		nmi_exit();
+
+	return ret;
+}
diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
index 9f7195a5773e..5976c44931a0 100644
--- a/arch/arm64/kernel/smp.c
+++ b/arch/arm64/kernel/smp.c
@@ -31,6 +31,7 @@
 #include <linux/mm.h>
 #include <linux/err.h>
 #include <linux/cpu.h>
+#include <linux/sdei.h>
 #include <linux/smp.h>
 #include <linux/seq_file.h>
 #include <linux/irq.h>
@@ -838,6 +839,7 @@ static void ipi_cpu_stop(unsigned int cpu)
 	set_cpu_online(cpu, false);
 
 	local_irq_disable();
+	sdei_mask_local_cpu();
 
 	while (1)
 		cpu_relax();
@@ -855,6 +857,7 @@ static void ipi_cpu_crash_stop(unsigned int cpu, struct pt_regs *regs)
 	atomic_dec(&waiting_for_crash_ipi);
 
 	local_irq_disable();
+	sdei_mask_local_cpu();
 
 #ifdef CONFIG_HOTPLUG_CPU
 	if (cpu_ops[cpu]->cpu_die)
@@ -974,6 +977,8 @@ void smp_send_stop(void)
 	if (num_online_cpus() > 1)
 		pr_warning("SMP: failed to stop secondary CPUs %*pbl\n",
 			   cpumask_pr_args(cpu_online_mask));
+
+	sdei_mask_local_cpu();
 }
 
 #ifdef CONFIG_KEXEC_CORE
@@ -992,8 +997,10 @@ void crash_smp_send_stop(void)
 
 	cpus_stopped = 1;
 
-	if (num_online_cpus() == 1)
+	if (num_online_cpus() == 1) {
+		sdei_mask_local_cpu();
 		return;
+	}
 
 	cpumask_copy(&mask, cpu_online_mask);
 	cpumask_clear_cpu(smp_processor_id(), &mask);
@@ -1011,6 +1018,8 @@ void crash_smp_send_stop(void)
 	if (atomic_read(&waiting_for_crash_ipi) > 0)
 		pr_warning("SMP: failed to stop secondary CPUs %*pbl\n",
 			   cpumask_pr_args(&mask));
+
+	sdei_mask_local_cpu();
 }
 
 bool smp_crash_stop_failed(void)
diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
index d8a9859f6102..1f6633b337aa 100644
--- a/drivers/firmware/Kconfig
+++ b/drivers/firmware/Kconfig
@@ -50,6 +50,7 @@ config ARM_SCPI_POWER_DOMAIN
 
 config ARM_SDE_INTERFACE
 	bool "ARM Software Delegated Exception Interface (SDEI)"
+	depends on ARM64
 	help
 	  The Software Delegated Exception Interface (SDEI) is an ARM
 	  standard for registering callbacks from the platform firmware
diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
index 4b3c7fd99c34..3ea6a19ae439 100644
--- a/drivers/firmware/arm_sdei.c
+++ b/drivers/firmware/arm_sdei.c
@@ -154,6 +154,7 @@ int sdei_api_event_context(u32 query, u64 *result)
 	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_CONTEXT, query, 0, 0, 0, 0,
 			      result);
 }
+NOKPROBE_SYMBOL(sdei_api_event_context);
 
 static int sdei_api_event_get_info(u32 event, u32 info, u64 *result)
 {
diff --git a/include/linux/sdei.h b/include/linux/sdei.h
index bb3dd000771e..df3fe6373e32 100644
--- a/include/linux/sdei.h
+++ b/include/linux/sdei.h
@@ -32,6 +32,8 @@ enum sdei_conduit_types {
 	CONDUIT_HVC,
 };
 
+#include <asm/sdei.h>
+
 /* Arch code should override this to set the entry point from firmware... */
 #ifndef sdei_arch_get_entry_point
 #define sdei_arch_get_entry_point(conduit)	(0)
-- 
2.13.3

[PATCH v3 09/13] arm64: kernel: Add arch-specific SDEI entry code and CPU masking

[James Morse]

The Software Delegated Exception Interface (SDEI) is an ARM standard
for registering callbacks from the platform firmware into the OS.
This is typically used to implement RAS notifications.

Such notifications enter the kernel at the registered entry-point
with the register values of the interrupted CPU context. Because this
is not a CPU exception, it cannot reuse the existing entry code.
(crucially we don't implicitly know which exception level we interrupted),

Add an sdei-entry.S asm file to set us up for calling into C code. If
the event interrupted code that had interrupts masked, we always return
to that location. Otherwise we pretend this was an IRQ, and use SDEI's
complete_and_resume call to return to vbar_el1 + offset.

This allows the kernel to deliver signals to user space processes. For
KVM this triggers the world switch, a quick spin round vcpu_run, then
back into the guest, unless there are pending signals.

Add sdei_mask_local_cpu() calls to the smp_send_stop() code, this covers
the panic() code-path, which doesn't invoke cpuhotplug notifiers.

Because we can interrupt entry-from/exit-to another EL, we can't trust the
value in sp_el0 or x29, even if we interrupted the kernel, in this case
the code in entry.S will save/restore sp_el0 and use the value in
__entry_task.

When we have VMAP stacks we can interrupt the stack-overflow test, which
stirs x0 into sp, meaning we have to have our own VMAP stack.

Signed-off-by: James Morse <james.morse@arm.com>
---
Changes since v2:
 * Added PAN-setting as we didn't take an exception to get here.
 * Added VMAP stack allocation and switching.
 * overwrite x29 on entry from not-the-kernel.
 * Added a pe-mask to crash_smp_send_stop()s 'no secondaries' path (oops).
 * Added ELR-hasn't-changed test. Any exception in the handler will panic KVM
   as its switched VBAR_EL1, but go silently undetected otherwise. Generate a
   warning.

 arch/arm64/Kconfig                  |   2 +-
 arch/arm64/include/asm/sdei.h       |  63 ++++++++++
 arch/arm64/include/asm/stacktrace.h |   3 +
 arch/arm64/kernel/Makefile          |   1 +
 arch/arm64/kernel/asm-offsets.c     |   5 +
 arch/arm64/kernel/sdei-entry.S      | 122 +++++++++++++++++++
 arch/arm64/kernel/sdei.c            | 235 ++++++++++++++++++++++++++++++++++++
 arch/arm64/kernel/smp.c             |  11 +-
 drivers/firmware/Kconfig            |   1 +
 drivers/firmware/arm_sdei.c         |   1 +
 include/linux/sdei.h                |   2 +
 11 files changed, 444 insertions(+), 2 deletions(-)
 create mode 100644 arch/arm64/include/asm/sdei.h
 create mode 100644 arch/arm64/kernel/sdei-entry.S
 create mode 100644 arch/arm64/kernel/sdei.c

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index 0df64a6a56d4..70dfe4e9ccc5 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -98,7 +98,7 @@ config ARM64
 	select HAVE_IRQ_TIME_ACCOUNTING
 	select HAVE_MEMBLOCK
 	select HAVE_MEMBLOCK_NODE_MAP if NUMA
-	select HAVE_NMI if ACPI_APEI_SEA
+	select HAVE_NMI
 	select HAVE_PATA_PLATFORM
 	select HAVE_PERF_EVENTS
 	select HAVE_PERF_REGS
diff --git a/arch/arm64/include/asm/sdei.h b/arch/arm64/include/asm/sdei.h
new file mode 100644
index 000000000000..ed329e01a301
--- /dev/null
+++ b/arch/arm64/include/asm/sdei.h
@@ -0,0 +1,63 @@
+/*
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __ASM_SDEI_H
+#define __ASM_SDEI_H
+
+/* Values for sdei_exit_mode */
+#define SDEI_EXIT_HVC  0
+#define SDEI_EXIT_SMC  1
+
+#define SDEI_STACK_SIZE		IRQ_STACK_SIZE
+
+#ifndef __ASSEMBLY__
+
+#include <linux/linkage.h>
+#include <linux/types.h>
+
+#include <asm/virt.h>
+
+extern unsigned long sdei_exit_mode;
+
+/* Software Delegated Exception entry point from firmware*/
+asmlinkage void __sdei_asm_handler(unsigned long event_num, unsigned long arg,
+				   unsigned long pc, unsigned long pstate);
+
+/*
+ * The above entry point does the minimum to call C code. This function does
+ * anything else, before calling the driver.
+ */
+struct sdei_registered_event;
+asmlinkage unsigned long __sdei_handler(struct pt_regs *regs,
+					struct sdei_registered_event *arg);
+
+extern unsigned long sdei_arch_get_entry_point(int conduit);
+#define sdei_arch_get_entry_point(x)	sdei_arch_get_entry_point(x)
+
+bool _on_sdei_stack(unsigned long sp);
+static inline bool on_sdei_stack(unsigned long sp)
+{
+	if (!IS_ENABLED(CONFIG_VMAP_STACK))
+		return false;
+	if (!IS_ENABLED(CONFIG_ARM_SDE_INTERFACE))
+		return false;
+	if (in_nmi())
+		return _on_sdei_stack(sp);
+
+	return false;
+}
+
+#endif /* __ASSEMBLY__ */
+#endif	/* __ASM_SDEI_H */
diff --git a/arch/arm64/include/asm/stacktrace.h b/arch/arm64/include/asm/stacktrace.h
index 6ad30776e984..472ef944e932 100644
--- a/arch/arm64/include/asm/stacktrace.h
+++ b/arch/arm64/include/asm/stacktrace.h
@@ -22,6 +22,7 @@
 
 #include <asm/memory.h>
 #include <asm/ptrace.h>
+#include <asm/sdei.h>
 
 struct stackframe {
 	unsigned long fp;
@@ -85,6 +86,8 @@ static inline bool on_accessible_stack(struct task_struct *tsk, unsigned long sp
 		return true;
 	if (on_overflow_stack(sp))
 		return true;
+	if (on_sdei_stack(sp))
+		return true;
 
 	return false;
 }
diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index f2b4e816b6de..d22bae586ade 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -53,6 +53,7 @@ arm64-obj-$(CONFIG_KEXEC)		+= machine_kexec.o relocate_kernel.o	\
 arm64-obj-$(CONFIG_ARM64_RELOC_TEST)	+= arm64-reloc-test.o
 arm64-reloc-test-y := reloc_test_core.o reloc_test_syms.o
 arm64-obj-$(CONFIG_CRASH_DUMP)		+= crash_dump.o
+arm64-obj-$(CONFIG_ARM_SDE_INTERFACE)	+= sdei.o sdei-entry.o
 
 obj-y					+= $(arm64-obj-y) vdso/ probes/
 obj-m					+= $(arm64-obj-m)
diff --git a/arch/arm64/kernel/asm-offsets.c b/arch/arm64/kernel/asm-offsets.c
index 71bf088f1e4b..f6baa4ca79a9 100644
--- a/arch/arm64/kernel/asm-offsets.c
+++ b/arch/arm64/kernel/asm-offsets.c
@@ -22,6 +22,7 @@
 #include <linux/mm.h>
 #include <linux/dma-mapping.h>
 #include <linux/kvm_host.h>
+#include <linux/sdei.h>
 #include <linux/suspend.h>
 #include <asm/cpufeature.h>
 #include <asm/thread_info.h>
@@ -154,5 +155,9 @@ int main(void)
   DEFINE(HIBERN_PBE_ADDR,	offsetof(struct pbe, address));
   DEFINE(HIBERN_PBE_NEXT,	offsetof(struct pbe, next));
   DEFINE(ARM64_FTR_SYSVAL,	offsetof(struct arm64_ftr_reg, sys_val));
+#ifdef CONFIG_ARM_SDE_INTERFACE
+  DEFINE(SDEI_EVENT_INTREGS,	offsetof(struct sdei_registered_event, interrupted_regs));
+  DEFINE(SDEI_EVENT_PRIORITY,	offsetof(struct sdei_registered_event, priority));
+#endif
   return 0;
 }
diff --git a/arch/arm64/kernel/sdei-entry.S b/arch/arm64/kernel/sdei-entry.S
new file mode 100644
index 000000000000..cf12f8da0789
--- /dev/null
+++ b/arch/arm64/kernel/sdei-entry.S
@@ -0,0 +1,122 @@
+/*
+ * Software Delegated Exception entry point from firmware to the kernel.
+ *
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/linkage.h>
+
+#include <asm/alternative.h>
+#include <asm/asm-offsets.h>
+#include <asm/assembler.h>
+#include <asm/memory.h>
+#include <asm/ptrace.h>
+#include <asm/sdei.h>
+#include <uapi/linux/sdei.h>
+
+/*
+ * Software Delegated Exception entry point.
+ *
+ * x0: Event number
+ * x1: struct sdei_registered_event argument from registration time.
+ * x2: interrupted PC
+ * x3: interrupted PSTATE
+ *
+ * Firmware has preserved x0->x17 for us, we must save/restore the rest to
+ * follow SMC-CC. We save (or retrieve) all the registers as the handler may
+ * want them.
+ */
+ENTRY(__sdei_asm_handler)
+	stp     x2, x3, [x1, #SDEI_EVENT_INTREGS + S_PC]
+	stp     x4, x5, [x1, #SDEI_EVENT_INTREGS + 16 * 2]
+	stp     x6, x7, [x1, #SDEI_EVENT_INTREGS + 16 * 3]
+	stp     x8, x9, [x1, #SDEI_EVENT_INTREGS + 16 * 4]
+	stp     x10, x11, [x1, #SDEI_EVENT_INTREGS + 16 * 5]
+	stp     x12, x13, [x1, #SDEI_EVENT_INTREGS + 16 * 6]
+	stp     x14, x15, [x1, #SDEI_EVENT_INTREGS + 16 * 7]
+	stp     x16, x17, [x1, #SDEI_EVENT_INTREGS + 16 * 8]
+	stp     x18, x19, [x1, #SDEI_EVENT_INTREGS + 16 * 9]
+	stp     x20, x21, [x1, #SDEI_EVENT_INTREGS + 16 * 10]
+	stp     x22, x23, [x1, #SDEI_EVENT_INTREGS + 16 * 11]
+	stp     x24, x25, [x1, #SDEI_EVENT_INTREGS + 16 * 12]
+	stp     x26, x27, [x1, #SDEI_EVENT_INTREGS + 16 * 13]
+	stp     x28, x29, [x1, #SDEI_EVENT_INTREGS + 16 * 14]
+	mov	x4, sp
+	stp     lr, x4, [x1, #SDEI_EVENT_INTREGS + S_LR]
+
+	mov	x19, x1
+
+#ifdef CONFIG_VMAP_STACK
+	/*
+	 * entry.S may have been using sp as a scratch register, find whether
+	 * this is a normal or critical event and switch to the appropriate
+	 * stack for this CPU.
+	 */
+	ldrb	w4, [x19, #SDEI_EVENT_PRIORITY]
+	cbnz	w4, 1f
+	ldr_this_cpu dst=x5, sym=sdei_stack_normal_ptr, tmp=x6
+	b	2f
+1:	ldr_this_cpu dst=x5, sym=sdei_stack_critical_ptr, tmp=x6
+2:	mov	x6, #SDEI_STACK_SIZE
+	add	x5, x5, x6
+	mov	sp, x5
+#endif
+
+	/*
+	 * We may have interrupted userspace, or a guest, or exit-from or
+	 * return-to either of these. We can't trust sp_el0, restore it.
+	 */
+	mrs	x28, sp_el0
+	ldr_this_cpu	dst=x0, sym=__entry_task, tmp=x1
+	msr	sp_el0, x0
+
+	/* If we interrupted the kernel point to the previous stack/frame. */
+	and     x0, x3, #0xc
+	mrs     x1, CurrentEL
+	cmp     x0, x1
+	csel	x29, x29, xzr, eq	// fp, or zero
+	csel	x4, x2, xzr, eq		// elr, or zero
+
+	stp	x29, x4, [sp, #-16]!
+	mov	x29, sp
+
+	add	x0, x19, #SDEI_EVENT_INTREGS
+	mov	x1, x19
+	bl	__sdei_handler
+
+	msr	sp_el0, x28
+	/* restore regs >x17 that we clobbered */
+	ldp     x28, x29, [x19, #SDEI_EVENT_INTREGS + 16 * 14]
+	ldp     lr, x4, [x19, #SDEI_EVENT_INTREGS + S_LR]
+	mov	sp, x4
+	ldp     x18, x19, [x19, #SDEI_EVENT_INTREGS + 16 * 9]
+
+	mov	x1, x0			// address to complete_and_resume
+	/* x0 = (x0 <= 1) ? EVENT_COMPLETE:EVENT_COMPLETE_AND_RESUME */
+	cmp	x0, #1
+	mov_q	x2, SDEI_1_0_FN_SDEI_EVENT_COMPLETE
+	mov_q	x3, SDEI_1_0_FN_SDEI_EVENT_COMPLETE_AND_RESUME
+	csel	x0, x2, x3, ls
+
+	/* On success, this call never returns... */
+	ldr_l	x2, sdei_exit_mode
+	cmp	x2, #SDEI_EXIT_SMC
+	b.ne	1f
+	smc	#0
+	b	.
+1:	hvc	#0
+	b	.
+ENDPROC(__sdei_asm_handler)
+NOKPROBE(__sdei_asm_handler)
diff --git a/arch/arm64/kernel/sdei.c b/arch/arm64/kernel/sdei.c
new file mode 100644
index 000000000000..16fff51a699f
--- /dev/null
+++ b/arch/arm64/kernel/sdei.c
@@ -0,0 +1,235 @@
+/*
+ * Copyright (C) 2017 ARM Ltd.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define pr_fmt(fmt) "sdei: " fmt
+
+#include <linux/sched/task_stack.h>
+#include <linux/irqflags.h>
+#include <linux/hardirq.h>
+#include <linux/sdei.h>
+
+#include <asm/alternative.h>
+#include <asm/kprobes.h>
+#include <asm/ptrace.h>
+#include <asm/sysreg.h>
+#include <asm/vmap_stack.h>
+
+unsigned long sdei_exit_mode;
+
+/*
+ * VMAP'd stacks checking for stack overflow on exception using sp as a scratch
+ * register, meaning SDEI has to switch to its own stack. We need two stacks as
+ * a critical event may interrupt a normal event that has just taken a
+ * synchronous exception, and is using sp as scratch register. For a critical
+ * event interrupting a normal event, we can't reliably tell if we were on the
+ * sdei stack.
+ * For private events we need per-cpu stacks, these can't be allocated lazily
+ * as THREADINFO_GFP is a subset of GFP_KERNEL which can't necessarily be used
+ * when the first private event is registered.
+ */
+DECLARE_PER_CPU(unsigned long *, sdei_stack_normal_ptr);
+DECLARE_PER_CPU(unsigned long *, sdei_stack_critical_ptr);
+
+#ifdef CONFIG_VMAP_STACK
+DEFINE_PER_CPU(unsigned long *, sdei_stack_normal_ptr);
+DEFINE_PER_CPU(unsigned long *, sdei_stack_critical_ptr);
+#endif
+
+static void _free_sdei_stack(unsigned long * __percpu *ptr, int cpu)
+{
+	unsigned long *p;
+
+	p = per_cpu(*ptr, cpu);
+	if (p) {
+		per_cpu(*ptr, cpu) = NULL;
+		vfree(p);
+	}
+}
+
+static void free_sdei_stacks(void)
+{
+	int cpu;
+
+	for_each_possible_cpu(cpu) {
+		_free_sdei_stack(&sdei_stack_normal_ptr, cpu);
+		_free_sdei_stack(&sdei_stack_critical_ptr, cpu);
+	}
+}
+
+static int _init_sdei_stack(unsigned long * __percpu *ptr, int cpu)
+{
+	unsigned long *p;
+
+	p = arch_alloc_vmap_stack(SDEI_STACK_SIZE, cpu_to_node(cpu));
+	if (!p)
+		return -ENOMEM;
+	per_cpu(*ptr, cpu) = p;
+
+	return 0;
+}
+
+static int init_sdei_stacks(void)
+{
+	int cpu;
+	int err = 0;
+
+	for_each_possible_cpu(cpu) {
+		err = _init_sdei_stack(&sdei_stack_normal_ptr, cpu);
+		if (err)
+			break;
+		err = _init_sdei_stack(&sdei_stack_critical_ptr, cpu);
+		if (err)
+			break;
+	}
+
+	if (err)
+		free_sdei_stacks();
+
+	return err;
+}
+
+bool _on_sdei_stack(unsigned long sp)
+{
+	unsigned long low, high;
+
+	if (!IS_ENABLED(CONFIG_VMAP_STACK))
+		return false;
+
+	low = (unsigned long)raw_cpu_read(sdei_stack_critical_ptr);
+	high = low + SDEI_STACK_SIZE;
+
+	if (low <= sp && sp < high)
+		return true;
+
+	low = (unsigned long)raw_cpu_read(sdei_stack_normal_ptr);
+	high = low + SDEI_STACK_SIZE;
+
+	return (low <= sp && sp < high);
+}
+
+unsigned long sdei_arch_get_entry_point(int conduit)
+{
+	/*
+	 * SDEI works between adjacent exception levels. If we booted at EL1 we
+	 * assume a hypervisor is marshalling events. If we booted@EL2 and
+	 * dropped to EL1 because we don't support VHE, then we can't support
+	 * SDEI.
+	 */
+	if (is_hyp_mode_available() && !is_kernel_in_hyp_mode()) {
+		pr_err("Not supported on this hardware/boot configuration\n");
+		return 0;
+	}
+
+	if (IS_ENABLED(CONFIG_VMAP_STACK)) {
+		if (init_sdei_stacks())
+			return 0;
+	}
+
+	sdei_exit_mode = (conduit == CONDUIT_HVC) ? SDEI_EXIT_HVC : SDEI_EXIT_SMC;
+	return (unsigned long)__sdei_asm_handler;
+}
+
+/*
+ * __sdei_handler() returns one of:
+ *  SDEI_EV_HANDLED -  success, return to the interrupted context.
+ *  SDEI_EV_FAILED  -  failure, return this error code to firmare.
+ *  virtual-address -  success, return to this address.
+ */
+static __kprobes unsigned long _sdei_handler(struct pt_regs *regs,
+					     struct sdei_registered_event *arg)
+{
+	u32 mode;
+	int i, err = 0;
+	int clobbered_registers = 4;
+	u64 elr = read_sysreg(elr_el1);
+	u32 kernel_mode = read_sysreg(CurrentEL) | 1;	/* +SPSel */
+	unsigned long vbar = read_sysreg(vbar_el1);
+
+	/* Retrieve the missing registers values */
+	for (i = 0; i < clobbered_registers; i++) {
+		/* from within the handler, this call always succeeds */
+		sdei_api_event_context(i, &regs->regs[i]);
+	}
+
+	/*
+	 * We didn't take an exception to get here, set PAN. UAO will be cleared
+	 * by sdei_event_handler()s set_fs(USER_DS) call.
+	 */
+	asm(ALTERNATIVE("nop", SET_PSTATE_PAN(1), ARM64_HAS_PAN,
+			CONFIG_ARM64_PAN));
+
+	err = sdei_event_handler(regs, arg);
+	if (err)
+		return SDEI_EV_FAILED;
+
+	if (elr != read_sysreg(elr_el1)) {
+		/*
+		 * We took a synchronous exception from the SDEI handler.
+		 * This could deadlock, and if you interrupt KVM it will
+		 * hyp-panic instead.
+		 */
+		pr_warn("unsafe: exception during handler\n");
+	}
+
+	mode = regs->pstate & (PSR_MODE32_BIT | PSR_MODE_MASK);
+
+	/*
+	 * If we interrupted the kernel with interrupts masked, we always go
+	 * back to wherever we came from.
+	 */
+	if (mode == kernel_mode && !interrupts_enabled(regs))
+		return SDEI_EV_HANDLED;
+
+	/*
+	 * Otherwise, we pretend this was an IRQ. This lets user space tasks
+	 * receive signals before we return to them, and KVM to invoke it's
+	 * world switch to do the same.
+	 *
+	 * See DDI0487B.a Table D1-7 'Vector offsets from vector table base
+	 * address'.
+	 */
+	if (mode == kernel_mode)
+		return vbar + 0x280;
+	else if (mode & PSR_MODE32_BIT)
+		return vbar + 0x680;
+
+	return vbar + 0x480;
+}
+
+
+asmlinkage __kprobes notrace unsigned long
+__sdei_handler(struct pt_regs *regs, struct sdei_registered_event *arg)
+{
+	unsigned long ret;
+	bool do_nmi_exit = false;
+
+	/*
+	 * nmi_enter() deals with printk() re-entrance and use of RCU when
+	 * RCU believed this CPU was idle. Because critical events can
+	 * interrupt normal events, we may already be in_nmi().
+	 */
+	if (!in_nmi()) {
+		nmi_enter();
+		do_nmi_exit = true;
+	}
+
+	ret = _sdei_handler(regs, arg);
+
+	if (do_nmi_exit)
+		nmi_exit();
+
+	return ret;
+}
diff --git a/arch/arm64/kernel/smp.c b/arch/arm64/kernel/smp.c
index 9f7195a5773e..5976c44931a0 100644
--- a/arch/arm64/kernel/smp.c
+++ b/arch/arm64/kernel/smp.c
@@ -31,6 +31,7 @@
 #include <linux/mm.h>
 #include <linux/err.h>
 #include <linux/cpu.h>
+#include <linux/sdei.h>
 #include <linux/smp.h>
 #include <linux/seq_file.h>
 #include <linux/irq.h>
@@ -838,6 +839,7 @@ static void ipi_cpu_stop(unsigned int cpu)
 	set_cpu_online(cpu, false);
 
 	local_irq_disable();
+	sdei_mask_local_cpu();
 
 	while (1)
 		cpu_relax();
@@ -855,6 +857,7 @@ static void ipi_cpu_crash_stop(unsigned int cpu, struct pt_regs *regs)
 	atomic_dec(&waiting_for_crash_ipi);
 
 	local_irq_disable();
+	sdei_mask_local_cpu();
 
 #ifdef CONFIG_HOTPLUG_CPU
 	if (cpu_ops[cpu]->cpu_die)
@@ -974,6 +977,8 @@ void smp_send_stop(void)
 	if (num_online_cpus() > 1)
 		pr_warning("SMP: failed to stop secondary CPUs %*pbl\n",
 			   cpumask_pr_args(cpu_online_mask));
+
+	sdei_mask_local_cpu();
 }
 
 #ifdef CONFIG_KEXEC_CORE
@@ -992,8 +997,10 @@ void crash_smp_send_stop(void)
 
 	cpus_stopped = 1;
 
-	if (num_online_cpus() == 1)
+	if (num_online_cpus() == 1) {
+		sdei_mask_local_cpu();
 		return;
+	}
 
 	cpumask_copy(&mask, cpu_online_mask);
 	cpumask_clear_cpu(smp_processor_id(), &mask);
@@ -1011,6 +1018,8 @@ void crash_smp_send_stop(void)
 	if (atomic_read(&waiting_for_crash_ipi) > 0)
 		pr_warning("SMP: failed to stop secondary CPUs %*pbl\n",
 			   cpumask_pr_args(&mask));
+
+	sdei_mask_local_cpu();
 }
 
 bool smp_crash_stop_failed(void)
diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
index d8a9859f6102..1f6633b337aa 100644
--- a/drivers/firmware/Kconfig
+++ b/drivers/firmware/Kconfig
@@ -50,6 +50,7 @@ config ARM_SCPI_POWER_DOMAIN
 
 config ARM_SDE_INTERFACE
 	bool "ARM Software Delegated Exception Interface (SDEI)"
+	depends on ARM64
 	help
 	  The Software Delegated Exception Interface (SDEI) is an ARM
 	  standard for registering callbacks from the platform firmware
diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
index 4b3c7fd99c34..3ea6a19ae439 100644
--- a/drivers/firmware/arm_sdei.c
+++ b/drivers/firmware/arm_sdei.c
@@ -154,6 +154,7 @@ int sdei_api_event_context(u32 query, u64 *result)
 	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_CONTEXT, query, 0, 0, 0, 0,
 			      result);
 }
+NOKPROBE_SYMBOL(sdei_api_event_context);
 
 static int sdei_api_event_get_info(u32 event, u32 info, u64 *result)
 {
diff --git a/include/linux/sdei.h b/include/linux/sdei.h
index bb3dd000771e..df3fe6373e32 100644
--- a/include/linux/sdei.h
+++ b/include/linux/sdei.h
@@ -32,6 +32,8 @@ enum sdei_conduit_types {
 	CONDUIT_HVC,
 };
 
+#include <asm/sdei.h>
+
 /* Arch code should override this to set the entry point from firmware... */
 #ifndef sdei_arch_get_entry_point
 #define sdei_arch_get_entry_point(conduit)	(0)
-- 
2.13.3

[PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states

[James Morse]

When a CPU enters an idle lower-power state or is powering off, we
need to mask SDE events so that no events can be delivered while we
are messing with the MMU as the registered entry points won't be valid.

If the system reboots, we want to unregister all events and mask the CPUs.
For kexec this allows us to hand a clean slate to the next kernel
instead of relying on it to call sdei_{private,system}_data_reset().

For hibernate we unregister all events and re-register them on restore,
in case we restored with the SDE code loaded at a different address.
(e.g. KASLR).

Add all the notifiers necessary to do this. We only support shared events
so all events are left registered and enabled over CPU hotplug.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
---
 drivers/firmware/arm_sdei.c | 228 +++++++++++++++++++++++++++++++++++++++++++-
 include/linux/cpuhotplug.h  |   1 +
 include/linux/sdei.h        |   3 +
 3 files changed, 231 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
index 3ea6a19ae439..5656830efff4 100644
--- a/drivers/firmware/arm_sdei.c
+++ b/drivers/firmware/arm_sdei.c
@@ -22,6 +22,8 @@
 #include <linux/arm-smccc.h>
 #include <linux/bitops.h>
 #include <linux/compiler.h>
+#include <linux/cpuhotplug.h>
+#include <linux/cpu_pm.h>
 #include <linux/errno.h>
 #include <linux/hardirq.h>
 #include <linux/kernel.h>
@@ -29,12 +31,15 @@
 #include <linux/kvm_host.h>
 #include <linux/list.h>
 #include <linux/lockdep.h>
+#include <linux/notifier.h>
 #include <linux/of.h>
 #include <linux/of_platform.h>
 #include <linux/percpu.h>
 #include <linux/platform_device.h>
+#include <linux/pm.h>
 #include <linux/ptrace.h>
 #include <linux/preempt.h>
+#include <linux/reboot.h>
 #include <linux/sdei.h>
 #include <linux/slab.h>
 #include <linux/smp.h>
@@ -260,6 +265,11 @@ static void _ipi_mask_cpu(void *ignored)
 	sdei_mask_local_cpu();
 }
 
+static int sdei_cpuhp_down(unsigned int ignored)
+{
+	return sdei_mask_local_cpu();
+}
+
 int sdei_unmask_local_cpu(void)
 {
 	int err;
@@ -281,6 +291,11 @@ static void _ipi_unmask_cpu(void *ignored)
 	sdei_unmask_local_cpu();
 }
 
+static int sdei_cpuhp_up(unsigned int ignored)
+{
+	return sdei_unmask_local_cpu();
+}
+
 static void _ipi_private_reset(void *ignored)
 {
 	int err;
@@ -319,6 +334,12 @@ static int sdei_platform_reset(void)
 	return err;
 }
 
+static int sdei_api_event_status(u32 event_num, u64 *result)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_STATUS, event_num, 0, 0, 0,
+			      0, result);
+}
+
 static int sdei_api_event_enable(u32 event_num)
 {
 	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_ENABLE, event_num, 0, 0, 0,
@@ -373,8 +394,23 @@ static int sdei_api_event_unregister(u32 event_num)
 
 static int _sdei_event_unregister(struct sdei_event *event)
 {
-	if (event->type == SDEI_EVENT_TYPE_SHARED)
+	int err;
+	u64 result;
+	bool enabled;
+
+	if (event->type == SDEI_EVENT_TYPE_SHARED) {
+		/*
+		 * We re-register events after power-management, remember if
+		 * this event was previously enabled.
+		 */
+		err = sdei_api_event_status(event->event_num, &result);
+		if (err)
+			return err;
+		enabled = !!(result & BIT(SDEI_EVENT_STATUS_ENABLED));
+		event->registered->was_enabled = enabled;
+
 		return sdei_api_event_unregister(event->event_num);
+	}
 
 	return -EINVAL;
 }
@@ -407,6 +443,26 @@ int sdei_event_unregister(u32 event_num)
 }
 EXPORT_SYMBOL(sdei_event_unregister);
 
+/*
+ * unregister events, but don't destroy them as they are re-registered by
+ * sdei_reregister_events().
+ */
+static int sdei_event_unregister_all(void)
+{
+	int err = 0;
+	struct sdei_event *event;
+
+	spin_lock(&sdei_events_lock);
+	list_for_each_entry(event, &sdei_events, list) {
+		err = _sdei_event_unregister(event);
+		if (err)
+			break;
+	}
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+
 static int sdei_api_event_register(u32 event_num, unsigned long entry_point,
 				   void *arg, u64 flags, u64 affinity)
 {
@@ -463,6 +519,148 @@ int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
 }
 EXPORT_SYMBOL(sdei_event_register);
 
+static int sdei_reregister_event(struct sdei_event *event)
+{
+	int err;
+
+	lockdep_assert_held(&sdei_events_lock);
+
+	err = _sdei_event_register(event);
+	if (err) {
+		pr_err("Failed to re-register event %u\n", event->event_num);
+		sdei_event_destroy(event);
+		return err;
+	}
+
+	if (event->type == SDEI_EVENT_TYPE_SHARED) {
+		if (event->registered->was_enabled)
+			err = sdei_api_event_enable(event->event_num);
+	}
+
+	if (err)
+		pr_err("Failed to re-enable event %u\n", event->event_num);
+
+	return err;
+}
+
+static int sdei_reregister_events(void)
+{
+	int err = 0;
+	struct sdei_event *event;
+
+	spin_lock(&sdei_events_lock);
+	list_for_each_entry(event, &sdei_events, list) {
+		err = sdei_reregister_event(event);
+		if (err)
+			break;
+	}
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+
+/* When entering idle, mask/unmask events for this cpu */
+static int sdei_pm_notifier(struct notifier_block *nb, unsigned long action,
+			    void *data)
+{
+	int rv;
+
+	switch (action) {
+	case CPU_PM_ENTER:
+		rv = sdei_mask_local_cpu();
+		break;
+	case CPU_PM_EXIT:
+		rv = sdei_unmask_local_cpu();
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+
+	if (rv)
+		return notifier_from_errno(rv);
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block sdei_pm_nb = {
+	.notifier_call = sdei_pm_notifier,
+};
+
+static int sdei_device_suspend(struct device *dev)
+{
+	on_each_cpu(_ipi_mask_cpu, NULL, true);
+
+	return 0;
+}
+
+static int sdei_device_resume(struct device *dev)
+{
+	on_each_cpu(_ipi_unmask_cpu, NULL, true);
+
+	return 0;
+}
+
+/*
+ * We need all events to be reregistered when we resume from hibernate.
+ *
+ * The sequence is freeze->thaw. Reboot. freeze->restore. We unregister
+ * events during freeze and save the enabled state, then re-register and
+ * enable them during thaw and restore.
+ */
+static int sdei_device_freeze(struct device *dev)
+{
+	int err;
+
+	err = sdei_event_unregister_all();
+	if (err)
+		return err;
+
+	return sdei_device_suspend(dev);
+}
+
+static int sdei_device_thaw(struct device *dev)
+{
+	sdei_device_resume(dev);
+
+	return sdei_reregister_events();
+}
+
+static int sdei_device_restore(struct device *dev)
+{
+	int err;
+
+	err = sdei_platform_reset();
+	if (err)
+		return err;
+
+	return sdei_device_thaw(dev);
+}
+
+static const struct dev_pm_ops sdei_pm_ops = {
+	.suspend = sdei_device_suspend,
+	.resume = sdei_device_resume,
+	.freeze = sdei_device_freeze,
+	.thaw = sdei_device_thaw,
+	.restore = sdei_device_restore,
+};
+
+/*
+ * Mask all CPUs and unregister all events on panic, reboot or kexec.
+ */
+static int sdei_reboot_notifier(struct notifier_block *nb, unsigned long action,
+				void *data)
+{
+	on_each_cpu(&_ipi_mask_cpu, NULL, true);
+
+	sdei_platform_reset();
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block sdei_reboot_nb = {
+	.notifier_call = sdei_reboot_notifier,
+};
+
 static void sdei_smccc_smc(unsigned long function_id,
 			   unsigned long arg0, unsigned long arg1,
 			   unsigned long arg2, unsigned long arg3,
@@ -545,9 +743,36 @@ static int sdei_probe(struct platform_device *pdev)
 		return 0;
 	}
 
+	err = cpuhp_setup_state_nocalls(CPUHP_AP_SDEI_STARTING, "SDEI",
+				&sdei_cpuhp_up, &sdei_cpuhp_down);
+	if (err) {
+		pr_warn("Failed to register CPU hotplug notifier...\n");
+		return err;
+	}
+
+	err = cpu_pm_register_notifier(&sdei_pm_nb);
+	if (err) {
+		pr_warn("Failed to register CPU PM notifier...\n");
+		goto remove_cpuhp;
+	}
+
+	err = register_reboot_notifier(&sdei_reboot_nb);
+	if (err) {
+		pr_warn("Failed to register reboot notifier...\n");
+		goto remove_cpupm;
+	}
+
 	on_each_cpu(&_ipi_unmask_cpu, NULL, false);
 
 	return 0;
+
+remove_cpupm:
+	cpu_pm_unregister_notifier(&sdei_pm_nb);
+
+remove_cpuhp:
+	cpuhp_remove_state_nocalls(CPUHP_AP_SDEI_STARTING);
+
+	return err;
 }
 
 static const struct of_device_id sdei_of_match[] = {
@@ -558,6 +783,7 @@ static const struct of_device_id sdei_of_match[] = {
 static struct platform_driver sdei_driver = {
 	.driver		= {
 		.name			= "sdei",
+		.pm			= &sdei_pm_ops,
 		.of_match_table		= sdei_of_match,
 	},
 	.probe		= sdei_probe,
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index f24bfb2b9a2d..466b949474df 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -88,6 +88,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_PERF_METAG_STARTING,
 	CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
+	CPUHP_AP_SDEI_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_ARM64_DEBUG_MONITORS_STARTING,
 	CPUHP_AP_PERF_ARM_HW_BREAKPOINT_STARTING,
diff --git a/include/linux/sdei.h b/include/linux/sdei.h
index df3fe6373e32..f25b4b2cd69c 100644
--- a/include/linux/sdei.h
+++ b/include/linux/sdei.h
@@ -90,6 +90,9 @@ struct sdei_registered_event {
 	void			*callback_arg;
 	u32			 event_num;
 	u8			 priority;
+
+	/* Only valid over power management events. */
+	bool			 was_enabled;
 };
 
 /* The arch code entry point should then call this when an event arrives. */
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states

[James Morse]

When a CPU enters an idle lower-power state or is powering off, we
need to mask SDE events so that no events can be delivered while we
are messing with the MMU as the registered entry points won't be valid.

If the system reboots, we want to unregister all events and mask the CPUs.
For kexec this allows us to hand a clean slate to the next kernel
instead of relying on it to call sdei_{private,system}_data_reset().

For hibernate we unregister all events and re-register them on restore,
in case we restored with the SDE code loaded at a different address.
(e.g. KASLR).

Add all the notifiers necessary to do this. We only support shared events
so all events are left registered and enabled over CPU hotplug.

Signed-off-by: James Morse <james.morse@arm.com>
---
 drivers/firmware/arm_sdei.c | 228 +++++++++++++++++++++++++++++++++++++++++++-
 include/linux/cpuhotplug.h  |   1 +
 include/linux/sdei.h        |   3 +
 3 files changed, 231 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
index 3ea6a19ae439..5656830efff4 100644
--- a/drivers/firmware/arm_sdei.c
+++ b/drivers/firmware/arm_sdei.c
@@ -22,6 +22,8 @@
 #include <linux/arm-smccc.h>
 #include <linux/bitops.h>
 #include <linux/compiler.h>
+#include <linux/cpuhotplug.h>
+#include <linux/cpu_pm.h>
 #include <linux/errno.h>
 #include <linux/hardirq.h>
 #include <linux/kernel.h>
@@ -29,12 +31,15 @@
 #include <linux/kvm_host.h>
 #include <linux/list.h>
 #include <linux/lockdep.h>
+#include <linux/notifier.h>
 #include <linux/of.h>
 #include <linux/of_platform.h>
 #include <linux/percpu.h>
 #include <linux/platform_device.h>
+#include <linux/pm.h>
 #include <linux/ptrace.h>
 #include <linux/preempt.h>
+#include <linux/reboot.h>
 #include <linux/sdei.h>
 #include <linux/slab.h>
 #include <linux/smp.h>
@@ -260,6 +265,11 @@ static void _ipi_mask_cpu(void *ignored)
 	sdei_mask_local_cpu();
 }
 
+static int sdei_cpuhp_down(unsigned int ignored)
+{
+	return sdei_mask_local_cpu();
+}
+
 int sdei_unmask_local_cpu(void)
 {
 	int err;
@@ -281,6 +291,11 @@ static void _ipi_unmask_cpu(void *ignored)
 	sdei_unmask_local_cpu();
 }
 
+static int sdei_cpuhp_up(unsigned int ignored)
+{
+	return sdei_unmask_local_cpu();
+}
+
 static void _ipi_private_reset(void *ignored)
 {
 	int err;
@@ -319,6 +334,12 @@ static int sdei_platform_reset(void)
 	return err;
 }
 
+static int sdei_api_event_status(u32 event_num, u64 *result)
+{
+	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_STATUS, event_num, 0, 0, 0,
+			      0, result);
+}
+
 static int sdei_api_event_enable(u32 event_num)
 {
 	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_ENABLE, event_num, 0, 0, 0,
@@ -373,8 +394,23 @@ static int sdei_api_event_unregister(u32 event_num)
 
 static int _sdei_event_unregister(struct sdei_event *event)
 {
-	if (event->type == SDEI_EVENT_TYPE_SHARED)
+	int err;
+	u64 result;
+	bool enabled;
+
+	if (event->type == SDEI_EVENT_TYPE_SHARED) {
+		/*
+		 * We re-register events after power-management, remember if
+		 * this event was previously enabled.
+		 */
+		err = sdei_api_event_status(event->event_num, &result);
+		if (err)
+			return err;
+		enabled = !!(result & BIT(SDEI_EVENT_STATUS_ENABLED));
+		event->registered->was_enabled = enabled;
+
 		return sdei_api_event_unregister(event->event_num);
+	}
 
 	return -EINVAL;
 }
@@ -407,6 +443,26 @@ int sdei_event_unregister(u32 event_num)
 }
 EXPORT_SYMBOL(sdei_event_unregister);
 
+/*
+ * unregister events, but don't destroy them as they are re-registered by
+ * sdei_reregister_events().
+ */
+static int sdei_event_unregister_all(void)
+{
+	int err = 0;
+	struct sdei_event *event;
+
+	spin_lock(&sdei_events_lock);
+	list_for_each_entry(event, &sdei_events, list) {
+		err = _sdei_event_unregister(event);
+		if (err)
+			break;
+	}
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+
 static int sdei_api_event_register(u32 event_num, unsigned long entry_point,
 				   void *arg, u64 flags, u64 affinity)
 {
@@ -463,6 +519,148 @@ int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
 }
 EXPORT_SYMBOL(sdei_event_register);
 
+static int sdei_reregister_event(struct sdei_event *event)
+{
+	int err;
+
+	lockdep_assert_held(&sdei_events_lock);
+
+	err = _sdei_event_register(event);
+	if (err) {
+		pr_err("Failed to re-register event %u\n", event->event_num);
+		sdei_event_destroy(event);
+		return err;
+	}
+
+	if (event->type == SDEI_EVENT_TYPE_SHARED) {
+		if (event->registered->was_enabled)
+			err = sdei_api_event_enable(event->event_num);
+	}
+
+	if (err)
+		pr_err("Failed to re-enable event %u\n", event->event_num);
+
+	return err;
+}
+
+static int sdei_reregister_events(void)
+{
+	int err = 0;
+	struct sdei_event *event;
+
+	spin_lock(&sdei_events_lock);
+	list_for_each_entry(event, &sdei_events, list) {
+		err = sdei_reregister_event(event);
+		if (err)
+			break;
+	}
+	spin_unlock(&sdei_events_lock);
+
+	return err;
+}
+
+/* When entering idle, mask/unmask events for this cpu */
+static int sdei_pm_notifier(struct notifier_block *nb, unsigned long action,
+			    void *data)
+{
+	int rv;
+
+	switch (action) {
+	case CPU_PM_ENTER:
+		rv = sdei_mask_local_cpu();
+		break;
+	case CPU_PM_EXIT:
+		rv = sdei_unmask_local_cpu();
+		break;
+	default:
+		return NOTIFY_DONE;
+	}
+
+	if (rv)
+		return notifier_from_errno(rv);
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block sdei_pm_nb = {
+	.notifier_call = sdei_pm_notifier,
+};
+
+static int sdei_device_suspend(struct device *dev)
+{
+	on_each_cpu(_ipi_mask_cpu, NULL, true);
+
+	return 0;
+}
+
+static int sdei_device_resume(struct device *dev)
+{
+	on_each_cpu(_ipi_unmask_cpu, NULL, true);
+
+	return 0;
+}
+
+/*
+ * We need all events to be reregistered when we resume from hibernate.
+ *
+ * The sequence is freeze->thaw. Reboot. freeze->restore. We unregister
+ * events during freeze and save the enabled state, then re-register and
+ * enable them during thaw and restore.
+ */
+static int sdei_device_freeze(struct device *dev)
+{
+	int err;
+
+	err = sdei_event_unregister_all();
+	if (err)
+		return err;
+
+	return sdei_device_suspend(dev);
+}
+
+static int sdei_device_thaw(struct device *dev)
+{
+	sdei_device_resume(dev);
+
+	return sdei_reregister_events();
+}
+
+static int sdei_device_restore(struct device *dev)
+{
+	int err;
+
+	err = sdei_platform_reset();
+	if (err)
+		return err;
+
+	return sdei_device_thaw(dev);
+}
+
+static const struct dev_pm_ops sdei_pm_ops = {
+	.suspend = sdei_device_suspend,
+	.resume = sdei_device_resume,
+	.freeze = sdei_device_freeze,
+	.thaw = sdei_device_thaw,
+	.restore = sdei_device_restore,
+};
+
+/*
+ * Mask all CPUs and unregister all events on panic, reboot or kexec.
+ */
+static int sdei_reboot_notifier(struct notifier_block *nb, unsigned long action,
+				void *data)
+{
+	on_each_cpu(&_ipi_mask_cpu, NULL, true);
+
+	sdei_platform_reset();
+
+	return NOTIFY_OK;
+}
+
+static struct notifier_block sdei_reboot_nb = {
+	.notifier_call = sdei_reboot_notifier,
+};
+
 static void sdei_smccc_smc(unsigned long function_id,
 			   unsigned long arg0, unsigned long arg1,
 			   unsigned long arg2, unsigned long arg3,
@@ -545,9 +743,36 @@ static int sdei_probe(struct platform_device *pdev)
 		return 0;
 	}
 
+	err = cpuhp_setup_state_nocalls(CPUHP_AP_SDEI_STARTING, "SDEI",
+				&sdei_cpuhp_up, &sdei_cpuhp_down);
+	if (err) {
+		pr_warn("Failed to register CPU hotplug notifier...\n");
+		return err;
+	}
+
+	err = cpu_pm_register_notifier(&sdei_pm_nb);
+	if (err) {
+		pr_warn("Failed to register CPU PM notifier...\n");
+		goto remove_cpuhp;
+	}
+
+	err = register_reboot_notifier(&sdei_reboot_nb);
+	if (err) {
+		pr_warn("Failed to register reboot notifier...\n");
+		goto remove_cpupm;
+	}
+
 	on_each_cpu(&_ipi_unmask_cpu, NULL, false);
 
 	return 0;
+
+remove_cpupm:
+	cpu_pm_unregister_notifier(&sdei_pm_nb);
+
+remove_cpuhp:
+	cpuhp_remove_state_nocalls(CPUHP_AP_SDEI_STARTING);
+
+	return err;
 }
 
 static const struct of_device_id sdei_of_match[] = {
@@ -558,6 +783,7 @@ static const struct of_device_id sdei_of_match[] = {
 static struct platform_driver sdei_driver = {
 	.driver		= {
 		.name			= "sdei",
+		.pm			= &sdei_pm_ops,
 		.of_match_table		= sdei_of_match,
 	},
 	.probe		= sdei_probe,
diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
index f24bfb2b9a2d..466b949474df 100644
--- a/include/linux/cpuhotplug.h
+++ b/include/linux/cpuhotplug.h
@@ -88,6 +88,7 @@ enum cpuhp_state {
 	CPUHP_AP_PERF_XTENSA_STARTING,
 	CPUHP_AP_PERF_METAG_STARTING,
 	CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
+	CPUHP_AP_SDEI_STARTING,
 	CPUHP_AP_ARM_VFP_STARTING,
 	CPUHP_AP_ARM64_DEBUG_MONITORS_STARTING,
 	CPUHP_AP_PERF_ARM_HW_BREAKPOINT_STARTING,
diff --git a/include/linux/sdei.h b/include/linux/sdei.h
index df3fe6373e32..f25b4b2cd69c 100644
--- a/include/linux/sdei.h
+++ b/include/linux/sdei.h
@@ -90,6 +90,9 @@ struct sdei_registered_event {
 	void			*callback_arg;
 	u32			 event_num;
 	u8			 priority;
+
+	/* Only valid over power management events. */
+	bool			 was_enabled;
 };
 
 /* The arch code entry point should then call this when an event arrives. */
-- 
2.13.3

[PATCH v3 11/13] firmware: arm_sdei: add support for CPU private events

[James Morse]

Private SDE events are per-cpu, and need to be registered and enabled
on each CPU.

Hide this detail from the caller by adapting our {,un}register and
{en,dis}able calls to send an IPI to each CPU if the event is private.

CPU private events are unregistered when the CPU is powered-off, and
re-registered when the CPU is brought back online. This saves bringing
secondary cores back online to call private_reset() on shutdown, kexec
and resume from hibernate.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
---
 drivers/firmware/arm_sdei.c | 242 +++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 228 insertions(+), 14 deletions(-)

diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
index 5656830efff4..3bee2719d6f4 100644
--- a/drivers/firmware/arm_sdei.c
+++ b/drivers/firmware/arm_sdei.c
@@ -20,9 +20,11 @@
 
 #include <linux/acpi.h>
 #include <linux/arm-smccc.h>
+#include <linux/atomic.h>
 #include <linux/bitops.h>
 #include <linux/compiler.h>
 #include <linux/cpuhotplug.h>
+#include <linux/cpu.h>
 #include <linux/cpu_pm.h>
 #include <linux/errno.h>
 #include <linux/hardirq.h>
@@ -65,12 +67,49 @@ struct sdei_event {
 	u8			priority;
 
 	/* This pointer is handed to firmware as the event argument. */
-	struct sdei_registered_event *registered;
+	union {
+		/* Shared events */
+		struct sdei_registered_event *registered;
+
+		/* CPU private events */
+		struct sdei_registered_event __percpu *private_registered;
+	};
 };
 
 static LIST_HEAD(sdei_events);
 static DEFINE_SPINLOCK(sdei_events_lock);
 
+/* When frozen, cpu-hotplug notifiers shouldn't unregister/re-register events */
+static bool frozen;
+
+/* Private events are registered/enabled via IPI passing one of these */
+struct sdei_crosscall_args {
+	struct sdei_event *event;
+	atomic_t errors;
+	int first_error;
+};
+
+#define CROSSCALL_INIT(arg, event)	(arg.event = event, \
+					 arg.first_error = 0, \
+					 atomic_set(&arg.errors, 0))
+
+static inline int sdei_do_cross_call(void *fn, struct sdei_event * event)
+{
+	struct sdei_crosscall_args arg;
+
+	CROSSCALL_INIT(arg, event);
+	on_each_cpu(fn, &arg, true);
+
+	return arg.first_error;
+}
+
+static inline void
+sdei_cross_call_return(struct sdei_crosscall_args *arg, int err)
+{
+	if (err && (atomic_inc_return(&arg->errors) == 1))
+		arg->first_error = err;
+}
+
 static int sdei_to_linux_errno(unsigned long sdei_err)
 {
 	switch (sdei_err) {
@@ -214,6 +253,26 @@ static struct sdei_event *sdei_event_create(u32 event_num,
 		reg->callback = cb;
 		reg->callback_arg = cb_arg;
 		event->registered = reg;
+	} else {
+		int cpu;
+		struct sdei_registered_event __percpu *regs;
+
+		regs = alloc_percpu(struct sdei_registered_event);
+		if (!regs) {
+			kfree(event);
+			return ERR_PTR(-ENOMEM);
+		}
+
+		for_each_possible_cpu(cpu) {
+			reg = per_cpu_ptr(regs, cpu);
+
+			reg->event_num = event->event_num;
+			reg->priority = event->priority;
+			reg->callback = cb;
+			reg->callback_arg = cb_arg;
+		}
+
+		event->private_registered = regs;
 	}
 
 	if (sdei_event_find(event_num)) {
@@ -235,6 +294,8 @@ static void sdei_event_destroy(struct sdei_event *event)
 
 	if (event->type == SDEI_EVENT_TYPE_SHARED)
 		kfree(event->registered);
+	else
+		free_percpu(event->private_registered);
 
 	kfree(event);
 }
@@ -265,11 +326,6 @@ static void _ipi_mask_cpu(void *ignored)
 	sdei_mask_local_cpu();
 }
 
-static int sdei_cpuhp_down(unsigned int ignored)
-{
-	return sdei_mask_local_cpu();
-}
-
 int sdei_unmask_local_cpu(void)
 {
 	int err;
@@ -291,11 +347,6 @@ static void _ipi_unmask_cpu(void *ignored)
 	sdei_unmask_local_cpu();
 }
 
-static int sdei_cpuhp_up(unsigned int ignored)
-{
-	return sdei_unmask_local_cpu();
-}
-
 static void _ipi_private_reset(void *ignored)
 {
 	int err;
@@ -346,6 +397,16 @@ static int sdei_api_event_enable(u32 event_num)
 			      0, NULL);
 }
 
+static void _ipi_event_enable(void *data)
+{
+	int err;
+	struct sdei_crosscall_args *arg = data;
+
+	err = sdei_api_event_enable(arg->event->event_num);
+
+	sdei_cross_call_return(arg, err);
+}
+
 int sdei_event_enable(u32 event_num)
 {
 	int err = -EINVAL;
@@ -357,6 +418,8 @@ int sdei_event_enable(u32 event_num)
 		err = -ENOENT;
 	else if (event->type == SDEI_EVENT_TYPE_SHARED)
 		err = sdei_api_event_enable(event->event_num);
+	else
+		err = sdei_do_cross_call(_ipi_event_enable, event);
 	spin_unlock(&sdei_events_lock);
 
 	return err;
@@ -369,6 +432,16 @@ static int sdei_api_event_disable(u32 event_num)
 			      0, 0, NULL);
 }
 
+static void _ipi_event_disable(void *data)
+{
+	int err;
+	struct sdei_crosscall_args *arg = data;
+
+	err = sdei_api_event_disable(arg->event->event_num);
+
+	sdei_cross_call_return(arg, err);
+}
+
 int sdei_event_disable(u32 event_num)
 {
 	int err = -EINVAL;
@@ -380,6 +453,9 @@ int sdei_event_disable(u32 event_num)
 		err = -ENOENT;
 	else if (event->type == SDEI_EVENT_TYPE_SHARED)
 		err = sdei_api_event_disable(event->event_num);
+	else
+		err = sdei_do_cross_call(_ipi_event_disable, event);
+
 	spin_unlock(&sdei_events_lock);
 
 	return err;
@@ -392,6 +468,27 @@ static int sdei_api_event_unregister(u32 event_num)
 			      0, 0, 0, NULL);
 }
 
+static void _local_event_unregister(void *data)
+{
+	int err;
+	u64 result;
+	struct sdei_registered_event *reg;
+	struct sdei_crosscall_args *arg = data;
+
+	WARN_ON_ONCE(preemptible());
+	lockdep_assert_held(&sdei_events_lock);
+
+	reg = per_cpu_ptr(arg->event->private_registered, smp_processor_id());
+	err = sdei_api_event_status(reg->event_num, &result);
+	if (!err) {
+		reg->was_enabled = !!(result & BIT(SDEI_EVENT_STATUS_ENABLED));
+
+		err = sdei_api_event_unregister(arg->event->event_num);
+	}
+
+	sdei_cross_call_return(arg, err);
+}
+
 static int _sdei_event_unregister(struct sdei_event *event)
 {
 	int err;
@@ -412,7 +509,7 @@ static int _sdei_event_unregister(struct sdei_event *event)
 		return sdei_api_event_unregister(event->event_num);
 	}
 
-	return -EINVAL;
+	return sdei_do_cross_call(_local_event_unregister, event);
 }
 
 int sdei_event_unregister(u32 event_num)
@@ -471,15 +568,41 @@ static int sdei_api_event_register(u32 event_num, unsigned long entry_point,
 			      flags, affinity, NULL);
 }
 
+static void _local_event_register(void *data)
+{
+	int err;
+	struct sdei_registered_event *reg;
+	struct sdei_crosscall_args *arg = data;
+
+	WARN_ON(preemptible());
+	lockdep_assert_held(&sdei_events_lock);
+
+	reg = per_cpu_ptr(arg->event->private_registered, smp_processor_id());
+	err = sdei_api_event_register(arg->event->event_num, sdei_entry_point,
+				      reg, 0, 0);
+
+	sdei_cross_call_return(arg, err);
+}
+
 static int _sdei_event_register(struct sdei_event *event)
 {
+	int err;
+
 	if (event->type == SDEI_EVENT_TYPE_SHARED)
 		return sdei_api_event_register(event->event_num,
 					       sdei_entry_point,
 					       event->registered,
 					       SDEI_EVENT_REGISTER_RM_ANY, 0);
 
-	return -EINVAL;
+	get_online_cpus();
+
+	err = sdei_do_cross_call(_local_event_register, event);
+	if (err)
+		sdei_do_cross_call(_local_event_unregister, event);
+
+	put_online_cpus();
+
+	return err;
 }
 
 int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
@@ -519,6 +642,22 @@ int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
 }
 EXPORT_SYMBOL(sdei_event_register);
 
+static void _local_event_reenable(void *data)
+{
+	int err = 0;
+	struct sdei_registered_event *reg;
+	struct sdei_crosscall_args *arg = data;
+
+	WARN_ON_ONCE(preemptible());
+	lockdep_assert_held(&sdei_events_lock);
+
+	reg = per_cpu_ptr(arg->event->private_registered, smp_processor_id());
+	if (reg->was_enabled)
+		err = sdei_api_event_enable(arg->event->event_num);
+
+	sdei_cross_call_return(arg, err);
+}
+
 static int sdei_reregister_event(struct sdei_event *event)
 {
 	int err;
@@ -535,6 +674,8 @@ static int sdei_reregister_event(struct sdei_event *event)
 	if (event->type == SDEI_EVENT_TYPE_SHARED) {
 		if (event->registered->was_enabled)
 			err = sdei_api_event_enable(event->event_num);
+	} else {
+		err = sdei_do_cross_call(_local_event_reenable, event);
 	}
 
 	if (err)
@@ -559,6 +700,68 @@ static int sdei_reregister_events(void)
 	return err;
 }
 
+static int sdei_cpuhp_down(unsigned int cpu)
+{
+	struct sdei_event *event;
+	struct sdei_crosscall_args arg;
+
+	if (frozen) {
+		/* All events unregistered  */
+		return sdei_mask_local_cpu();
+	}
+
+	/* un-register private events */
+	spin_lock(&sdei_events_lock);
+	list_for_each_entry(event, &sdei_events, list) {
+		if (event->type == SDEI_EVENT_TYPE_SHARED)
+			continue;
+
+		CROSSCALL_INIT(arg, event);
+		/* call the cross-call function locally... */
+		_local_event_unregister(&arg);
+		if (arg.first_error)
+			pr_err("Failed to unregister event %u: %d\n",
+			       event->event_num, arg.first_error);
+	}
+	spin_unlock(&sdei_events_lock);
+
+	return sdei_mask_local_cpu();
+}
+
+static int sdei_cpuhp_up(unsigned int cpu)
+{
+	struct sdei_event *event;
+	struct sdei_crosscall_args arg;
+
+	if (frozen) {
+		/* Events will be re-registered when we thaw. */
+		return sdei_unmask_local_cpu();
+	}
+
+	/* re-register/enable private events */
+	spin_lock(&sdei_events_lock);
+	list_for_each_entry(event, &sdei_events, list) {
+		if (event->type == SDEI_EVENT_TYPE_SHARED)
+			continue;
+
+		CROSSCALL_INIT(arg, event);
+		/* call the cross-call function locally... */
+		_local_event_register(&arg);
+		if (arg.first_error)
+			pr_err("Failed to re-register event %u: %d\n",
+			       event->event_num, arg.first_error);
+
+		CROSSCALL_INIT(arg, event);
+		_local_event_reenable(&arg);
+		if (arg.first_error)
+			pr_err("Failed to re-enable event %u: %d\n",
+			       event->event_num, arg.first_error);
+	}
+	spin_unlock(&sdei_events_lock);
+
+	return sdei_unmask_local_cpu();
+}
+
 /* When entering idle, mask/unmask events for this cpu */
 static int sdei_pm_notifier(struct notifier_block *nb, unsigned long action,
 			    void *data)
@@ -611,6 +814,7 @@ static int sdei_device_freeze(struct device *dev)
 {
 	int err;
 
+	frozen = true;
 	err = sdei_event_unregister_all();
 	if (err)
 		return err;
@@ -620,9 +824,13 @@ static int sdei_device_freeze(struct device *dev)
 
 static int sdei_device_thaw(struct device *dev)
 {
+	int err;
+
 	sdei_device_resume(dev);
 
-	return sdei_reregister_events();
+	err = sdei_reregister_events();
+	frozen = false;
+	return err;
 }
 
 static int sdei_device_restore(struct device *dev)
@@ -654,6 +862,12 @@ static int sdei_reboot_notifier(struct notifier_block *nb, unsigned long action,
 
 	sdei_platform_reset();
 
+	/*
+	 * There is now no point trying to unregister private events if we go on
+	 * to take CPUs offline.
+	 */
+	frozen = true;
+
 	return NOTIFY_OK;
 }
 
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 11/13] firmware: arm_sdei: add support for CPU private events

[James Morse]

Private SDE events are per-cpu, and need to be registered and enabled
on each CPU.

Hide this detail from the caller by adapting our {,un}register and
{en,dis}able calls to send an IPI to each CPU if the event is private.

CPU private events are unregistered when the CPU is powered-off, and
re-registered when the CPU is brought back online. This saves bringing
secondary cores back online to call private_reset() on shutdown, kexec
and resume from hibernate.

Signed-off-by: James Morse <james.morse@arm.com>
---
 drivers/firmware/arm_sdei.c | 242 +++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 228 insertions(+), 14 deletions(-)

diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
index 5656830efff4..3bee2719d6f4 100644
--- a/drivers/firmware/arm_sdei.c
+++ b/drivers/firmware/arm_sdei.c
@@ -20,9 +20,11 @@
 
 #include <linux/acpi.h>
 #include <linux/arm-smccc.h>
+#include <linux/atomic.h>
 #include <linux/bitops.h>
 #include <linux/compiler.h>
 #include <linux/cpuhotplug.h>
+#include <linux/cpu.h>
 #include <linux/cpu_pm.h>
 #include <linux/errno.h>
 #include <linux/hardirq.h>
@@ -65,12 +67,49 @@ struct sdei_event {
 	u8			priority;
 
 	/* This pointer is handed to firmware as the event argument. */
-	struct sdei_registered_event *registered;
+	union {
+		/* Shared events */
+		struct sdei_registered_event *registered;
+
+		/* CPU private events */
+		struct sdei_registered_event __percpu *private_registered;
+	};
 };
 
 static LIST_HEAD(sdei_events);
 static DEFINE_SPINLOCK(sdei_events_lock);
 
+/* When frozen, cpu-hotplug notifiers shouldn't unregister/re-register events */
+static bool frozen;
+
+/* Private events are registered/enabled via IPI passing one of these */
+struct sdei_crosscall_args {
+	struct sdei_event *event;
+	atomic_t errors;
+	int first_error;
+};
+
+#define CROSSCALL_INIT(arg, event)	(arg.event = event, \
+					 arg.first_error = 0, \
+					 atomic_set(&arg.errors, 0))
+
+static inline int sdei_do_cross_call(void *fn, struct sdei_event * event)
+{
+	struct sdei_crosscall_args arg;
+
+	CROSSCALL_INIT(arg, event);
+	on_each_cpu(fn, &arg, true);
+
+	return arg.first_error;
+}
+
+static inline void
+sdei_cross_call_return(struct sdei_crosscall_args *arg, int err)
+{
+	if (err && (atomic_inc_return(&arg->errors) == 1))
+		arg->first_error = err;
+}
+
 static int sdei_to_linux_errno(unsigned long sdei_err)
 {
 	switch (sdei_err) {
@@ -214,6 +253,26 @@ static struct sdei_event *sdei_event_create(u32 event_num,
 		reg->callback = cb;
 		reg->callback_arg = cb_arg;
 		event->registered = reg;
+	} else {
+		int cpu;
+		struct sdei_registered_event __percpu *regs;
+
+		regs = alloc_percpu(struct sdei_registered_event);
+		if (!regs) {
+			kfree(event);
+			return ERR_PTR(-ENOMEM);
+		}
+
+		for_each_possible_cpu(cpu) {
+			reg = per_cpu_ptr(regs, cpu);
+
+			reg->event_num = event->event_num;
+			reg->priority = event->priority;
+			reg->callback = cb;
+			reg->callback_arg = cb_arg;
+		}
+
+		event->private_registered = regs;
 	}
 
 	if (sdei_event_find(event_num)) {
@@ -235,6 +294,8 @@ static void sdei_event_destroy(struct sdei_event *event)
 
 	if (event->type == SDEI_EVENT_TYPE_SHARED)
 		kfree(event->registered);
+	else
+		free_percpu(event->private_registered);
 
 	kfree(event);
 }
@@ -265,11 +326,6 @@ static void _ipi_mask_cpu(void *ignored)
 	sdei_mask_local_cpu();
 }
 
-static int sdei_cpuhp_down(unsigned int ignored)
-{
-	return sdei_mask_local_cpu();
-}
-
 int sdei_unmask_local_cpu(void)
 {
 	int err;
@@ -291,11 +347,6 @@ static void _ipi_unmask_cpu(void *ignored)
 	sdei_unmask_local_cpu();
 }
 
-static int sdei_cpuhp_up(unsigned int ignored)
-{
-	return sdei_unmask_local_cpu();
-}
-
 static void _ipi_private_reset(void *ignored)
 {
 	int err;
@@ -346,6 +397,16 @@ static int sdei_api_event_enable(u32 event_num)
 			      0, NULL);
 }
 
+static void _ipi_event_enable(void *data)
+{
+	int err;
+	struct sdei_crosscall_args *arg = data;
+
+	err = sdei_api_event_enable(arg->event->event_num);
+
+	sdei_cross_call_return(arg, err);
+}
+
 int sdei_event_enable(u32 event_num)
 {
 	int err = -EINVAL;
@@ -357,6 +418,8 @@ int sdei_event_enable(u32 event_num)
 		err = -ENOENT;
 	else if (event->type == SDEI_EVENT_TYPE_SHARED)
 		err = sdei_api_event_enable(event->event_num);
+	else
+		err = sdei_do_cross_call(_ipi_event_enable, event);
 	spin_unlock(&sdei_events_lock);
 
 	return err;
@@ -369,6 +432,16 @@ static int sdei_api_event_disable(u32 event_num)
 			      0, 0, NULL);
 }
 
+static void _ipi_event_disable(void *data)
+{
+	int err;
+	struct sdei_crosscall_args *arg = data;
+
+	err = sdei_api_event_disable(arg->event->event_num);
+
+	sdei_cross_call_return(arg, err);
+}
+
 int sdei_event_disable(u32 event_num)
 {
 	int err = -EINVAL;
@@ -380,6 +453,9 @@ int sdei_event_disable(u32 event_num)
 		err = -ENOENT;
 	else if (event->type == SDEI_EVENT_TYPE_SHARED)
 		err = sdei_api_event_disable(event->event_num);
+	else
+		err = sdei_do_cross_call(_ipi_event_disable, event);
+
 	spin_unlock(&sdei_events_lock);
 
 	return err;
@@ -392,6 +468,27 @@ static int sdei_api_event_unregister(u32 event_num)
 			      0, 0, 0, NULL);
 }
 
+static void _local_event_unregister(void *data)
+{
+	int err;
+	u64 result;
+	struct sdei_registered_event *reg;
+	struct sdei_crosscall_args *arg = data;
+
+	WARN_ON_ONCE(preemptible());
+	lockdep_assert_held(&sdei_events_lock);
+
+	reg = per_cpu_ptr(arg->event->private_registered, smp_processor_id());
+	err = sdei_api_event_status(reg->event_num, &result);
+	if (!err) {
+		reg->was_enabled = !!(result & BIT(SDEI_EVENT_STATUS_ENABLED));
+
+		err = sdei_api_event_unregister(arg->event->event_num);
+	}
+
+	sdei_cross_call_return(arg, err);
+}
+
 static int _sdei_event_unregister(struct sdei_event *event)
 {
 	int err;
@@ -412,7 +509,7 @@ static int _sdei_event_unregister(struct sdei_event *event)
 		return sdei_api_event_unregister(event->event_num);
 	}
 
-	return -EINVAL;
+	return sdei_do_cross_call(_local_event_unregister, event);
 }
 
 int sdei_event_unregister(u32 event_num)
@@ -471,15 +568,41 @@ static int sdei_api_event_register(u32 event_num, unsigned long entry_point,
 			      flags, affinity, NULL);
 }
 
+static void _local_event_register(void *data)
+{
+	int err;
+	struct sdei_registered_event *reg;
+	struct sdei_crosscall_args *arg = data;
+
+	WARN_ON(preemptible());
+	lockdep_assert_held(&sdei_events_lock);
+
+	reg = per_cpu_ptr(arg->event->private_registered, smp_processor_id());
+	err = sdei_api_event_register(arg->event->event_num, sdei_entry_point,
+				      reg, 0, 0);
+
+	sdei_cross_call_return(arg, err);
+}
+
 static int _sdei_event_register(struct sdei_event *event)
 {
+	int err;
+
 	if (event->type == SDEI_EVENT_TYPE_SHARED)
 		return sdei_api_event_register(event->event_num,
 					       sdei_entry_point,
 					       event->registered,
 					       SDEI_EVENT_REGISTER_RM_ANY, 0);
 
-	return -EINVAL;
+	get_online_cpus();
+
+	err = sdei_do_cross_call(_local_event_register, event);
+	if (err)
+		sdei_do_cross_call(_local_event_unregister, event);
+
+	put_online_cpus();
+
+	return err;
 }
 
 int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
@@ -519,6 +642,22 @@ int sdei_event_register(u32 event_num, sdei_event_callback *cb, void *arg,
 }
 EXPORT_SYMBOL(sdei_event_register);
 
+static void _local_event_reenable(void *data)
+{
+	int err = 0;
+	struct sdei_registered_event *reg;
+	struct sdei_crosscall_args *arg = data;
+
+	WARN_ON_ONCE(preemptible());
+	lockdep_assert_held(&sdei_events_lock);
+
+	reg = per_cpu_ptr(arg->event->private_registered, smp_processor_id());
+	if (reg->was_enabled)
+		err = sdei_api_event_enable(arg->event->event_num);
+
+	sdei_cross_call_return(arg, err);
+}
+
 static int sdei_reregister_event(struct sdei_event *event)
 {
 	int err;
@@ -535,6 +674,8 @@ static int sdei_reregister_event(struct sdei_event *event)
 	if (event->type == SDEI_EVENT_TYPE_SHARED) {
 		if (event->registered->was_enabled)
 			err = sdei_api_event_enable(event->event_num);
+	} else {
+		err = sdei_do_cross_call(_local_event_reenable, event);
 	}
 
 	if (err)
@@ -559,6 +700,68 @@ static int sdei_reregister_events(void)
 	return err;
 }
 
+static int sdei_cpuhp_down(unsigned int cpu)
+{
+	struct sdei_event *event;
+	struct sdei_crosscall_args arg;
+
+	if (frozen) {
+		/* All events unregistered  */
+		return sdei_mask_local_cpu();
+	}
+
+	/* un-register private events */
+	spin_lock(&sdei_events_lock);
+	list_for_each_entry(event, &sdei_events, list) {
+		if (event->type == SDEI_EVENT_TYPE_SHARED)
+			continue;
+
+		CROSSCALL_INIT(arg, event);
+		/* call the cross-call function locally... */
+		_local_event_unregister(&arg);
+		if (arg.first_error)
+			pr_err("Failed to unregister event %u: %d\n",
+			       event->event_num, arg.first_error);
+	}
+	spin_unlock(&sdei_events_lock);
+
+	return sdei_mask_local_cpu();
+}
+
+static int sdei_cpuhp_up(unsigned int cpu)
+{
+	struct sdei_event *event;
+	struct sdei_crosscall_args arg;
+
+	if (frozen) {
+		/* Events will be re-registered when we thaw. */
+		return sdei_unmask_local_cpu();
+	}
+
+	/* re-register/enable private events */
+	spin_lock(&sdei_events_lock);
+	list_for_each_entry(event, &sdei_events, list) {
+		if (event->type == SDEI_EVENT_TYPE_SHARED)
+			continue;
+
+		CROSSCALL_INIT(arg, event);
+		/* call the cross-call function locally... */
+		_local_event_register(&arg);
+		if (arg.first_error)
+			pr_err("Failed to re-register event %u: %d\n",
+			       event->event_num, arg.first_error);
+
+		CROSSCALL_INIT(arg, event);
+		_local_event_reenable(&arg);
+		if (arg.first_error)
+			pr_err("Failed to re-enable event %u: %d\n",
+			       event->event_num, arg.first_error);
+	}
+	spin_unlock(&sdei_events_lock);
+
+	return sdei_unmask_local_cpu();
+}
+
 /* When entering idle, mask/unmask events for this cpu */
 static int sdei_pm_notifier(struct notifier_block *nb, unsigned long action,
 			    void *data)
@@ -611,6 +814,7 @@ static int sdei_device_freeze(struct device *dev)
 {
 	int err;
 
+	frozen = true;
 	err = sdei_event_unregister_all();
 	if (err)
 		return err;
@@ -620,9 +824,13 @@ static int sdei_device_freeze(struct device *dev)
 
 static int sdei_device_thaw(struct device *dev)
 {
+	int err;
+
 	sdei_device_resume(dev);
 
-	return sdei_reregister_events();
+	err = sdei_reregister_events();
+	frozen = false;
+	return err;
 }
 
 static int sdei_device_restore(struct device *dev)
@@ -654,6 +862,12 @@ static int sdei_reboot_notifier(struct notifier_block *nb, unsigned long action,
 
 	sdei_platform_reset();
 
+	/*
+	 * There is now no point trying to unregister private events if we go on
+	 * to take CPUs offline.
+	 */
+	frozen = true;
+
 	return NOTIFY_OK;
 }
 
-- 
2.13.3

[PATCH v3 12/13] arm64: acpi: Remove __init from acpi_psci_use_hvc() for use by SDEI

[James Morse]

SDEI inherits the 'use hvc' bit that is also used by PSCI. PSCI does all
its initialisation early, SDEI does its late.

Remove the __init annotation from acpi_psci_use_hvc().

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
---
The function name is unchanged as this bit is named 'PSCI_USE_HVC'
in table 5-37 of ACPIv6.2.

 arch/arm64/kernel/acpi.c | 2 +-
 include/linux/psci.h     | 3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kernel/acpi.c b/arch/arm64/kernel/acpi.c
index b3162715ed78..252396a96c78 100644
--- a/arch/arm64/kernel/acpi.c
+++ b/arch/arm64/kernel/acpi.c
@@ -117,7 +117,7 @@ bool __init acpi_psci_present(void)
 }
 
 /* Whether HVC must be used instead of SMC as the PSCI conduit */
-bool __init acpi_psci_use_hvc(void)
+bool acpi_psci_use_hvc(void)
 {
 	return acpi_gbl_FADT.arm_boot_flags & ACPI_FADT_PSCI_USE_HVC;
 }
diff --git a/include/linux/psci.h b/include/linux/psci.h
index bdea1cb5e1db..e25a99992a82 100644
--- a/include/linux/psci.h
+++ b/include/linux/psci.h
@@ -46,10 +46,11 @@ static inline int psci_dt_init(void) { return 0; }
 #if defined(CONFIG_ARM_PSCI_FW) && defined(CONFIG_ACPI)
 int __init psci_acpi_init(void);
 bool __init acpi_psci_present(void);
-bool __init acpi_psci_use_hvc(void);
+bool acpi_psci_use_hvc(void);
 #else
 static inline int psci_acpi_init(void) { return 0; }
 static inline bool acpi_psci_present(void) { return false; }
+static inline bool acpi_psci_use_hvc(void) {return false; }
 #endif
 
 #endif /* __LINUX_PSCI_H */
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 12/13] arm64: acpi: Remove __init from acpi_psci_use_hvc() for use by SDEI

[James Morse]

SDEI inherits the 'use hvc' bit that is also used by PSCI. PSCI does all
its initialisation early, SDEI does its late.

Remove the __init annotation from acpi_psci_use_hvc().

Signed-off-by: James Morse <james.morse@arm.com>
---
The function name is unchanged as this bit is named 'PSCI_USE_HVC'
in table 5-37 of ACPIv6.2.

 arch/arm64/kernel/acpi.c | 2 +-
 include/linux/psci.h     | 3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/arch/arm64/kernel/acpi.c b/arch/arm64/kernel/acpi.c
index b3162715ed78..252396a96c78 100644
--- a/arch/arm64/kernel/acpi.c
+++ b/arch/arm64/kernel/acpi.c
@@ -117,7 +117,7 @@ bool __init acpi_psci_present(void)
 }
 
 /* Whether HVC must be used instead of SMC as the PSCI conduit */
-bool __init acpi_psci_use_hvc(void)
+bool acpi_psci_use_hvc(void)
 {
 	return acpi_gbl_FADT.arm_boot_flags & ACPI_FADT_PSCI_USE_HVC;
 }
diff --git a/include/linux/psci.h b/include/linux/psci.h
index bdea1cb5e1db..e25a99992a82 100644
--- a/include/linux/psci.h
+++ b/include/linux/psci.h
@@ -46,10 +46,11 @@ static inline int psci_dt_init(void) { return 0; }
 #if defined(CONFIG_ARM_PSCI_FW) && defined(CONFIG_ACPI)
 int __init psci_acpi_init(void);
 bool __init acpi_psci_present(void);
-bool __init acpi_psci_use_hvc(void);
+bool acpi_psci_use_hvc(void);
 #else
 static inline int psci_acpi_init(void) { return 0; }
 static inline bool acpi_psci_present(void) { return false; }
+static inline bool acpi_psci_use_hvc(void) {return false; }
 #endif
 
 #endif /* __LINUX_PSCI_H */
-- 
2.13.3

[PATCH v3 13/13] firmware: arm_sdei: Discover SDEI support via ACPI

[James Morse]

SDEI defines a new ACPI table to indicate the presence of the interface.
The conduit is discovered in the same way as PSCI.

For ACPI we need to create the platform device ourselves as SDEI doesn't
have an entry in the DSDT.

The SDEI platform device should be created after ACPI has been initialised
so that we can parse the table, but before GHES devices are created, which
may register SDE events if they use SDEI as their notification type.

Signed-off-by: James Morse <james.morse-5wv7dgnIgG8@public.gmane.org>
---
 drivers/firmware/arm_sdei.c | 41 ++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 40 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
index 3bee2719d6f4..50b6b1d0fe72 100644
--- a/drivers/firmware/arm_sdei.c
+++ b/drivers/firmware/arm_sdei.c
@@ -912,6 +912,14 @@ static int sdei_get_conduit(struct platform_device *pdev)
 		}
 
 		pr_warn("invalid \"method\" property: %s\n", method);
+	} else if (IS_ENABLED(CONFIG_ACPI) && !acpi_disabled) {
+		if (acpi_psci_use_hvc()) {
+			sdei_firmware_call = &sdei_smccc_hvc;
+			return CONDUIT_HVC;
+		} else {
+			sdei_firmware_call = &sdei_smccc_smc;
+			return CONDUIT_SMC;
+		}
 	}
 
 	return CONDUIT_INVALID;
@@ -1025,14 +1033,45 @@ static bool __init sdei_present_dt(void)
 	return true;
 }
 
+static bool __init sdei_present_acpi(void)
+{
+	acpi_status status;
+	struct platform_device *pdev;
+	struct acpi_table_header *sdei_table_header;
+
+	if (acpi_disabled)
+		return false;
+
+	status = acpi_get_table(ACPI_SIG_SDEI, 0, &sdei_table_header);
+	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
+		const char *msg = acpi_format_exception(status);
+
+		pr_info("Failed to get ACPI:SDEI table, %s\n", msg);
+	}
+	if (ACPI_FAILURE(status))
+		return false;
+
+	pdev = platform_device_register_simple(sdei_driver.driver.name, 0, NULL,
+					       0);
+	if (IS_ERR(pdev))
+		return false;
+
+	return true;
+}
+
 static int __init sdei_init(void)
 {
-	if (sdei_present_dt())
+	if (sdei_present_dt() || sdei_present_acpi())
 		platform_driver_register(&sdei_driver);
 
 	return 0;
 }
 
+/*
+ * On an ACPI system SDEI needs to be ready before HEST:GHES tries to register
+ * its events. ACPI is initialised from a subsys_initcall(), GHES is initialised
+ * by device_initcall(). We want to be called in the middle.
+ */
 subsys_initcall_sync(sdei_init);
 
 int sdei_event_handler(struct pt_regs *regs,
-- 
2.13.3

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 13/13] firmware: arm_sdei: Discover SDEI support via ACPI

[James Morse]

SDEI defines a new ACPI table to indicate the presence of the interface.
The conduit is discovered in the same way as PSCI.

For ACPI we need to create the platform device ourselves as SDEI doesn't
have an entry in the DSDT.

The SDEI platform device should be created after ACPI has been initialised
so that we can parse the table, but before GHES devices are created, which
may register SDE events if they use SDEI as their notification type.

Signed-off-by: James Morse <james.morse@arm.com>
---
 drivers/firmware/arm_sdei.c | 41 ++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 40 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
index 3bee2719d6f4..50b6b1d0fe72 100644
--- a/drivers/firmware/arm_sdei.c
+++ b/drivers/firmware/arm_sdei.c
@@ -912,6 +912,14 @@ static int sdei_get_conduit(struct platform_device *pdev)
 		}
 
 		pr_warn("invalid \"method\" property: %s\n", method);
+	} else if (IS_ENABLED(CONFIG_ACPI) && !acpi_disabled) {
+		if (acpi_psci_use_hvc()) {
+			sdei_firmware_call = &sdei_smccc_hvc;
+			return CONDUIT_HVC;
+		} else {
+			sdei_firmware_call = &sdei_smccc_smc;
+			return CONDUIT_SMC;
+		}
 	}
 
 	return CONDUIT_INVALID;
@@ -1025,14 +1033,45 @@ static bool __init sdei_present_dt(void)
 	return true;
 }
 
+static bool __init sdei_present_acpi(void)
+{
+	acpi_status status;
+	struct platform_device *pdev;
+	struct acpi_table_header *sdei_table_header;
+
+	if (acpi_disabled)
+		return false;
+
+	status = acpi_get_table(ACPI_SIG_SDEI, 0, &sdei_table_header);
+	if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
+		const char *msg = acpi_format_exception(status);
+
+		pr_info("Failed to get ACPI:SDEI table, %s\n", msg);
+	}
+	if (ACPI_FAILURE(status))
+		return false;
+
+	pdev = platform_device_register_simple(sdei_driver.driver.name, 0, NULL,
+					       0);
+	if (IS_ERR(pdev))
+		return false;
+
+	return true;
+}
+
 static int __init sdei_init(void)
 {
-	if (sdei_present_dt())
+	if (sdei_present_dt() || sdei_present_acpi())
 		platform_driver_register(&sdei_driver);
 
 	return 0;
 }
 
+/*
+ * On an ACPI system SDEI needs to be ready before HEST:GHES tries to register
+ * its events. ACPI is initialised from a subsys_initcall(), GHES is initialised
+ * by device_initcall(). We want to be called in the middle.
+ */
 subsys_initcall_sync(sdei_init);
 
 int sdei_event_handler(struct pt_regs *regs,
-- 
2.13.3

Re: [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> index cd52d365d1f0..8e4c7da2b126 100644
> --- a/arch/arm64/kernel/cpufeature.c
> +++ b/arch/arm64/kernel/cpufeature.c
> @@ -865,6 +865,7 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
>  		.capability = ARM64_HAS_VIRT_HOST_EXTN,
>  		.def_scope = SCOPE_SYSTEM,
>  		.matches = runs_at_el2,
> +		.enable = cpu_copy_el2regs,
>  	},
>  	{
>  		.desc = "32-bit EL0 Support",
> @@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
>  }
>  
>  late_initcall(enable_mrs_emulation);
> +
> +int cpu_copy_el2regs(void *__unused)
> +{
> +	int do_copyregs = 0;
> +
> +	/*
> +	 * Copy register values that aren't redirected by hardware.
> +	 *
> +	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
> +	 * this value to tpidr_el2 before we patch the code. Once we've done
> +	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
> +	 * do anything here.
> +	 */
> +	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
> +				 ARM64_HAS_VIRT_HOST_EXTN)
> +		    : "=r" (do_copyregs) : : );

Can you just do:

	if (cpu_have_const_cap(ARM64_HAS_VIRT_HOST_EXTN))
		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);

At this point the capability bits should be set and the jump labels
enabled.

Otherwise:

Reviewed-by: Catalin Marinas <catalin.marinas-5wv7dgnIgG8@public.gmane.org>
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> index cd52d365d1f0..8e4c7da2b126 100644
> --- a/arch/arm64/kernel/cpufeature.c
> +++ b/arch/arm64/kernel/cpufeature.c
> @@ -865,6 +865,7 @@ static const struct arm64_cpu_capabilities arm64_features[] = {
>  		.capability = ARM64_HAS_VIRT_HOST_EXTN,
>  		.def_scope = SCOPE_SYSTEM,
>  		.matches = runs_at_el2,
> +		.enable = cpu_copy_el2regs,
>  	},
>  	{
>  		.desc = "32-bit EL0 Support",
> @@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
>  }
>  
>  late_initcall(enable_mrs_emulation);
> +
> +int cpu_copy_el2regs(void *__unused)
> +{
> +	int do_copyregs = 0;
> +
> +	/*
> +	 * Copy register values that aren't redirected by hardware.
> +	 *
> +	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
> +	 * this value to tpidr_el2 before we patch the code. Once we've done
> +	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
> +	 * do anything here.
> +	 */
> +	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
> +				 ARM64_HAS_VIRT_HOST_EXTN)
> +		    : "=r" (do_copyregs) : : );

Can you just do:

	if (cpu_have_const_cap(ARM64_HAS_VIRT_HOST_EXTN))
		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);

At this point the capability bits should be set and the jump labels
enabled.

Otherwise:

Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>

Re: [PATCH v3 08/13] arm64: Add vmap_stack header file

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:09PM +0100, James Morse wrote:
> diff --git a/arch/arm64/include/asm/vmap_stack.h b/arch/arm64/include/asm/vmap_stack.h
> new file mode 100644
> index 000000000000..f41d043cac31
> --- /dev/null
> +++ b/arch/arm64/include/asm/vmap_stack.h
> @@ -0,0 +1,41 @@
> +/*
> + * Copyright (C) 2017 ARM Ltd.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
> + */
> +#ifndef __ASM_VMAP_STACK_H
> +#define __ASM_VMAP_STACK_H
> +
> +#include <linux/vmalloc.h>
> +#include <asm/memory.h>
> +#include <asm/pgtable.h>
> +#include <asm/thread_info.h>
> +
> +#ifdef CONFIG_VMAP_STACK
> +/*
> + * To ensure that VMAP'd stack overflow detection works correctly, all VMAP'd
> + * stacks need to have the same alignment.
> + */
> +static inline unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node)
> +{
> +	return __vmalloc_node_range(stack_size, THREAD_ALIGN,
> +				    VMALLOC_START, VMALLOC_END,
> +				    THREADINFO_GFP, PAGE_KERNEL, 0, node,
> +				    __builtin_return_address(0));
> +}
> +
> +#else
> +unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node); // link error

Do you actually need this here? The calling site (at least in this
patch), is surrounded by CONFIG_VMAP_STACK. Removing the dummy
declaration should be fine (I haven't read the rest of the patches yet,
maybe it's needed later; would a BUILD_BUG do?).

-- 
Catalin
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 08/13] arm64: Add vmap_stack header file

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:09PM +0100, James Morse wrote:
> diff --git a/arch/arm64/include/asm/vmap_stack.h b/arch/arm64/include/asm/vmap_stack.h
> new file mode 100644
> index 000000000000..f41d043cac31
> --- /dev/null
> +++ b/arch/arm64/include/asm/vmap_stack.h
> @@ -0,0 +1,41 @@
> +/*
> + * Copyright (C) 2017 ARM Ltd.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
> + */
> +#ifndef __ASM_VMAP_STACK_H
> +#define __ASM_VMAP_STACK_H
> +
> +#include <linux/vmalloc.h>
> +#include <asm/memory.h>
> +#include <asm/pgtable.h>
> +#include <asm/thread_info.h>
> +
> +#ifdef CONFIG_VMAP_STACK
> +/*
> + * To ensure that VMAP'd stack overflow detection works correctly, all VMAP'd
> + * stacks need to have the same alignment.
> + */
> +static inline unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node)
> +{
> +	return __vmalloc_node_range(stack_size, THREAD_ALIGN,
> +				    VMALLOC_START, VMALLOC_END,
> +				    THREADINFO_GFP, PAGE_KERNEL, 0, node,
> +				    __builtin_return_address(0));
> +}
> +
> +#else
> +unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node); // link error

Do you actually need this here? The calling site (at least in this
patch), is surrounded by CONFIG_VMAP_STACK. Removing the dummy
declaration should be fine (I haven't read the rest of the patches yet,
maybe it's needed later; would a BUILD_BUG do?).

-- 
Catalin

Re: [PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:08PM +0100, James Morse wrote:
> The Software Delegated Exception Interface (SDEI) is an ARM standard
> for registering callbacks from the platform firmware into the OS.
> This is typically used to implement firmware notifications (such as
> firmware-first RAS) or promote an IRQ that has been promoted to a
> firmware-assisted NMI.
> 
> Add the code for detecting the SDEI version and the framework for
> registering and unregistering events. Subsequent patches will add the
> arch-specific backend code and the necessary power management hooks.
> 
> Only shared events are supported, power management, private events and
> discovery for ACPI systems will be added by later patches.
> 
> Signed-off-by: James Morse <james.morse@arm.com>
> ---
> Changes since v2:
>  * Copy the priority into the structure the arch asm handler gets. This
>    is used for VMAP stacks where we can't know if a critical event interrupted
>    a normal priority event, thus they need separate stacks.
> 
> Changes since v1:
>  * Changed entry point to unsigned long, if we support non-vhe systems this
>    won't be a valid pointer
>  * Made invoke_sdei_fn() pass the only register we are interested in, instead
>    of the whole arm_smccc_res
>  * Made all the locking WARN_ON()s lockdep_assert_held()s.
>  * Moved more messages from 'err' to 'warn'.
>  * Made IS_SDEI_CALL() not depend on whether the config option is selected.
>  * Made 'event failed' messages rate limited.
> 
>  drivers/firmware/Kconfig    |   7 +
>  drivers/firmware/Makefile   |   1 +
>  drivers/firmware/arm_sdei.c | 616 ++++++++++++++++++++++++++++++++++++++++++++
>  include/linux/sdei.h        | 100 +++++++
>  include/uapi/linux/sdei.h   |  91 +++++++
>  5 files changed, 815 insertions(+)
>  create mode 100644 drivers/firmware/arm_sdei.c
>  create mode 100644 include/linux/sdei.h
>  create mode 100644 include/uapi/linux/sdei.h

I haven't reviewed this patch (yet...). However, I think it's worth
adding a MAINTAINERS entry with your name on it (unless you really want
to distance yourself from this code once merged ;)).

Thanks.

-- 
Catalin

[PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:08PM +0100, James Morse wrote:
> The Software Delegated Exception Interface (SDEI) is an ARM standard
> for registering callbacks from the platform firmware into the OS.
> This is typically used to implement firmware notifications (such as
> firmware-first RAS) or promote an IRQ that has been promoted to a
> firmware-assisted NMI.
> 
> Add the code for detecting the SDEI version and the framework for
> registering and unregistering events. Subsequent patches will add the
> arch-specific backend code and the necessary power management hooks.
> 
> Only shared events are supported, power management, private events and
> discovery for ACPI systems will be added by later patches.
> 
> Signed-off-by: James Morse <james.morse@arm.com>
> ---
> Changes since v2:
>  * Copy the priority into the structure the arch asm handler gets. This
>    is used for VMAP stacks where we can't know if a critical event interrupted
>    a normal priority event, thus they need separate stacks.
> 
> Changes since v1:
>  * Changed entry point to unsigned long, if we support non-vhe systems this
>    won't be a valid pointer
>  * Made invoke_sdei_fn() pass the only register we are interested in, instead
>    of the whole arm_smccc_res
>  * Made all the locking WARN_ON()s lockdep_assert_held()s.
>  * Moved more messages from 'err' to 'warn'.
>  * Made IS_SDEI_CALL() not depend on whether the config option is selected.
>  * Made 'event failed' messages rate limited.
> 
>  drivers/firmware/Kconfig    |   7 +
>  drivers/firmware/Makefile   |   1 +
>  drivers/firmware/arm_sdei.c | 616 ++++++++++++++++++++++++++++++++++++++++++++
>  include/linux/sdei.h        | 100 +++++++
>  include/uapi/linux/sdei.h   |  91 +++++++
>  5 files changed, 815 insertions(+)
>  create mode 100644 drivers/firmware/arm_sdei.c
>  create mode 100644 include/linux/sdei.h
>  create mode 100644 include/uapi/linux/sdei.h

I haven't reviewed this patch (yet...). However, I think it's worth
adding a MAINTAINERS entry with your name on it (unless you really want
to distance yourself from this code once merged ;)).

Thanks.

-- 
Catalin

Re: [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[James Morse]

Hi Catalin,

On 13/10/17 16:31, Catalin Marinas wrote:
> On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
>> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
>> index cd52d365d1f0..8e4c7da2b126 100644
>> --- a/arch/arm64/kernel/cpufeature.c
>> +++ b/arch/arm64/kernel/cpufeature.c

>> @@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
>>  }
>>  
>>  late_initcall(enable_mrs_emulation);
>> +
>> +int cpu_copy_el2regs(void *__unused)
>> +{
>> +	int do_copyregs = 0;
>> +
>> +	/*
>> +	 * Copy register values that aren't redirected by hardware.
>> +	 *
>> +	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
>> +	 * this value to tpidr_el2 before we patch the code. Once we've done
>> +	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
>> +	 * do anything here.
>> +	 */
>> +	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
>> +				 ARM64_HAS_VIRT_HOST_EXTN)
>> +		    : "=r" (do_copyregs) : : );
> 
> Can you just do:
> 
> 	if (cpu_have_const_cap(ARM64_HAS_VIRT_HOST_EXTN))
> 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
> 
> At this point the capability bits should be set and the jump labels
> enabled.

These are enabled too early, we haven't done patching yet.

We need to copy tpidr_el1 -> tpidr_el2 on all CPUs that are online before code
patching.

After code patching new CPUs set tpidr_el2 when they come online, so we don't
need to do the copy... but this enable method is still called. There is nothing
for us to do, and tpidr_el1 now contains junk, so the copy


cpu_have_const_cap() is great for knowing if we have a feature, here we want to
know if we've done the patching for this feature.

I can wrap the ALTERNATIVE() into a helper, something like:
> arm64_alternatives_applied(ARM64_HAS_VIRT_HOST_EXTN)

which should make it clearer.

Christoffer had the same question at connect, so I evidently haven't found the
right way of describing this yet.


> Otherwise:
> 
> Reviewed-by: Catalin Marinas <catalin.marinas-5wv7dgnIgG8@public.gmane.org>

Thanks for taking a look, I'll leave this RB until your happy with the
ALTERNATIVE() hackery above.


James
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[James Morse]

Hi Catalin,

On 13/10/17 16:31, Catalin Marinas wrote:
> On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
>> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
>> index cd52d365d1f0..8e4c7da2b126 100644
>> --- a/arch/arm64/kernel/cpufeature.c
>> +++ b/arch/arm64/kernel/cpufeature.c

>> @@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
>>  }
>>  
>>  late_initcall(enable_mrs_emulation);
>> +
>> +int cpu_copy_el2regs(void *__unused)
>> +{
>> +	int do_copyregs = 0;
>> +
>> +	/*
>> +	 * Copy register values that aren't redirected by hardware.
>> +	 *
>> +	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
>> +	 * this value to tpidr_el2 before we patch the code. Once we've done
>> +	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
>> +	 * do anything here.
>> +	 */
>> +	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
>> +				 ARM64_HAS_VIRT_HOST_EXTN)
>> +		    : "=r" (do_copyregs) : : );
> 
> Can you just do:
> 
> 	if (cpu_have_const_cap(ARM64_HAS_VIRT_HOST_EXTN))
> 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
> 
> At this point the capability bits should be set and the jump labels
> enabled.

These are enabled too early, we haven't done patching yet.

We need to copy tpidr_el1 -> tpidr_el2 on all CPUs that are online before code
patching.

After code patching new CPUs set tpidr_el2 when they come online, so we don't
need to do the copy... but this enable method is still called. There is nothing
for us to do, and tpidr_el1 now contains junk, so the copy


cpu_have_const_cap() is great for knowing if we have a feature, here we want to
know if we've done the patching for this feature.

I can wrap the ALTERNATIVE() into a helper, something like:
> arm64_alternatives_applied(ARM64_HAS_VIRT_HOST_EXTN)

which should make it clearer.

Christoffer had the same question at connect, so I evidently haven't found the
right way of describing this yet.


> Otherwise:
> 
> Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>

Thanks for taking a look, I'll leave this RB until your happy with the
ALTERNATIVE() hackery above.


James

Re: [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[Catalin Marinas]

On Fri, Oct 13, 2017 at 05:50:45PM +0100, James Morse wrote:
> Hi Catalin,
> 
> On 13/10/17 16:31, Catalin Marinas wrote:
> > On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
> >> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> >> index cd52d365d1f0..8e4c7da2b126 100644
> >> --- a/arch/arm64/kernel/cpufeature.c
> >> +++ b/arch/arm64/kernel/cpufeature.c
> 
> >> @@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
> >>  }
> >>  
> >>  late_initcall(enable_mrs_emulation);
> >> +
> >> +int cpu_copy_el2regs(void *__unused)
> >> +{
> >> +	int do_copyregs = 0;
> >> +
> >> +	/*
> >> +	 * Copy register values that aren't redirected by hardware.
> >> +	 *
> >> +	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
> >> +	 * this value to tpidr_el2 before we patch the code. Once we've done
> >> +	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
> >> +	 * do anything here.
> >> +	 */
> >> +	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
> >> +				 ARM64_HAS_VIRT_HOST_EXTN)
> >> +		    : "=r" (do_copyregs) : : );
> > 
> > Can you just do:
> > 
> > 	if (cpu_have_const_cap(ARM64_HAS_VIRT_HOST_EXTN))
> > 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
> > 
> > At this point the capability bits should be set and the jump labels
> > enabled.
> 
> These are enabled too early, we haven't done patching yet.
> 
> We need to copy tpidr_el1 -> tpidr_el2 on all CPUs that are online before code
> patching.
> 
> After code patching new CPUs set tpidr_el2 when they come online, so we don't
> need to do the copy... but this enable method is still called. There is nothing
> for us to do, and tpidr_el1 now contains junk, so the copy

Ah, I get it now (should've read the comment but I usually expect the
code to be obvious; it wasn't, at least to me, in this case ;)). You
could have added the sysreg copying directly in the asm volatile.

Anyway, I think it's better if we keep it entirely in C with this hunk
(untested):

--------------8<------------------------------
diff --git a/arch/arm64/include/asm/alternative.h b/arch/arm64/include/asm/alternative.h
index 6e1cb8c5af4d..f9e2f69f296e 100644
--- a/arch/arm64/include/asm/alternative.h
+++ b/arch/arm64/include/asm/alternative.h
@@ -11,6 +11,8 @@
 #include <linux/stddef.h>
 #include <linux/stringify.h>
 
+extern int alternatives_applied;
+
 struct alt_instr {
 	s32 orig_offset;	/* offset to original instruction */
 	s32 alt_offset;		/* offset to replacement instruction */
diff --git a/arch/arm64/kernel/alternative.c b/arch/arm64/kernel/alternative.c
index 6dd0a3a3e5c9..414288a558c8 100644
--- a/arch/arm64/kernel/alternative.c
+++ b/arch/arm64/kernel/alternative.c
@@ -32,6 +32,8 @@
 #define ALT_ORIG_PTR(a)		__ALT_PTR(a, orig_offset)
 #define ALT_REPL_PTR(a)		__ALT_PTR(a, alt_offset)
 
+int alternatives_applied;
+
 struct alt_region {
 	struct alt_instr *begin;
 	struct alt_instr *end;
@@ -143,7 +145,6 @@ static void __apply_alternatives(void *alt_region, bool use_linear_alias)
  */
 static int __apply_alternatives_multi_stop(void *unused)
 {
-	static int patched = 0;
 	struct alt_region region = {
 		.begin	= (struct alt_instr *)__alt_instructions,
 		.end	= (struct alt_instr *)__alt_instructions_end,
@@ -151,14 +152,14 @@ static int __apply_alternatives_multi_stop(void *unused)
 
 	/* We always have a CPU 0 at this point (__init) */
 	if (smp_processor_id()) {
-		while (!READ_ONCE(patched))
+		while (!READ_ONCE(alternatives_applied))
 			cpu_relax();
 		isb();
 	} else {
-		BUG_ON(patched);
+		BUG_ON(alternatives_applied);
 		__apply_alternatives(&region, true);
 		/* Barriers provided by the cache flushing */
-		WRITE_ONCE(patched, 1);
+		WRITE_ONCE(alternatives_applied, 1);
 	}
 
 	return 0;
--------------8<------------------------------

and in the cpu_copy_el2regs():

	if (!READ_ONCE(alternatives_applied))
 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);

-- 
Catalin
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[Catalin Marinas]

On Fri, Oct 13, 2017 at 05:50:45PM +0100, James Morse wrote:
> Hi Catalin,
> 
> On 13/10/17 16:31, Catalin Marinas wrote:
> > On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
> >> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> >> index cd52d365d1f0..8e4c7da2b126 100644
> >> --- a/arch/arm64/kernel/cpufeature.c
> >> +++ b/arch/arm64/kernel/cpufeature.c
> 
> >> @@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
> >>  }
> >>  
> >>  late_initcall(enable_mrs_emulation);
> >> +
> >> +int cpu_copy_el2regs(void *__unused)
> >> +{
> >> +	int do_copyregs = 0;
> >> +
> >> +	/*
> >> +	 * Copy register values that aren't redirected by hardware.
> >> +	 *
> >> +	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
> >> +	 * this value to tpidr_el2 before we patch the code. Once we've done
> >> +	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
> >> +	 * do anything here.
> >> +	 */
> >> +	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
> >> +				 ARM64_HAS_VIRT_HOST_EXTN)
> >> +		    : "=r" (do_copyregs) : : );
> > 
> > Can you just do:
> > 
> > 	if (cpu_have_const_cap(ARM64_HAS_VIRT_HOST_EXTN))
> > 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
> > 
> > At this point the capability bits should be set and the jump labels
> > enabled.
> 
> These are enabled too early, we haven't done patching yet.
> 
> We need to copy tpidr_el1 -> tpidr_el2 on all CPUs that are online before code
> patching.
> 
> After code patching new CPUs set tpidr_el2 when they come online, so we don't
> need to do the copy... but this enable method is still called. There is nothing
> for us to do, and tpidr_el1 now contains junk, so the copy

Ah, I get it now (should've read the comment but I usually expect the
code to be obvious; it wasn't, at least to me, in this case ;)). You
could have added the sysreg copying directly in the asm volatile.

Anyway, I think it's better if we keep it entirely in C with this hunk
(untested):

--------------8<------------------------------
diff --git a/arch/arm64/include/asm/alternative.h b/arch/arm64/include/asm/alternative.h
index 6e1cb8c5af4d..f9e2f69f296e 100644
--- a/arch/arm64/include/asm/alternative.h
+++ b/arch/arm64/include/asm/alternative.h
@@ -11,6 +11,8 @@
 #include <linux/stddef.h>
 #include <linux/stringify.h>
 
+extern int alternatives_applied;
+
 struct alt_instr {
 	s32 orig_offset;	/* offset to original instruction */
 	s32 alt_offset;		/* offset to replacement instruction */
diff --git a/arch/arm64/kernel/alternative.c b/arch/arm64/kernel/alternative.c
index 6dd0a3a3e5c9..414288a558c8 100644
--- a/arch/arm64/kernel/alternative.c
+++ b/arch/arm64/kernel/alternative.c
@@ -32,6 +32,8 @@
 #define ALT_ORIG_PTR(a)		__ALT_PTR(a, orig_offset)
 #define ALT_REPL_PTR(a)		__ALT_PTR(a, alt_offset)
 
+int alternatives_applied;
+
 struct alt_region {
 	struct alt_instr *begin;
 	struct alt_instr *end;
@@ -143,7 +145,6 @@ static void __apply_alternatives(void *alt_region, bool use_linear_alias)
  */
 static int __apply_alternatives_multi_stop(void *unused)
 {
-	static int patched = 0;
 	struct alt_region region = {
 		.begin	= (struct alt_instr *)__alt_instructions,
 		.end	= (struct alt_instr *)__alt_instructions_end,
@@ -151,14 +152,14 @@ static int __apply_alternatives_multi_stop(void *unused)
 
 	/* We always have a CPU 0 at this point (__init) */
 	if (smp_processor_id()) {
-		while (!READ_ONCE(patched))
+		while (!READ_ONCE(alternatives_applied))
 			cpu_relax();
 		isb();
 	} else {
-		BUG_ON(patched);
+		BUG_ON(alternatives_applied);
 		__apply_alternatives(&region, true);
 		/* Barriers provided by the cache flushing */
-		WRITE_ONCE(patched, 1);
+		WRITE_ONCE(alternatives_applied, 1);
 	}
 
 	return 0;
--------------8<------------------------------

and in the cpu_copy_el2regs():

	if (!READ_ONCE(alternatives_applied))
 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);

-- 
Catalin

Re: [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
> diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h
> index 29adab8138c3..8f2d0f7d193b 100644
> --- a/arch/arm64/include/asm/processor.h
> +++ b/arch/arm64/include/asm/processor.h
> @@ -193,5 +193,6 @@ static inline void spin_lock_prefetch(const void *ptr)
>  
>  int cpu_enable_pan(void *__unused);
>  int cpu_enable_cache_maint_trap(void *__unused);
> +int cpu_copy_el2regs(void *__unused);
>  
>  #endif /* __ASM_PROCESSOR_H */
> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> index cd52d365d1f0..8e4c7da2b126 100644
> --- a/arch/arm64/kernel/cpufeature.c
> +++ b/arch/arm64/kernel/cpufeature.c
[...]
> +int cpu_copy_el2regs(void *__unused)

Can this be static? I couldn't find it used anywhere else apart from
cpufeature.c

-- 
Catalin

[PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
> diff --git a/arch/arm64/include/asm/processor.h b/arch/arm64/include/asm/processor.h
> index 29adab8138c3..8f2d0f7d193b 100644
> --- a/arch/arm64/include/asm/processor.h
> +++ b/arch/arm64/include/asm/processor.h
> @@ -193,5 +193,6 @@ static inline void spin_lock_prefetch(const void *ptr)
>  
>  int cpu_enable_pan(void *__unused);
>  int cpu_enable_cache_maint_trap(void *__unused);
> +int cpu_copy_el2regs(void *__unused);
>  
>  #endif /* __ASM_PROCESSOR_H */
> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
> index cd52d365d1f0..8e4c7da2b126 100644
> --- a/arch/arm64/kernel/cpufeature.c
> +++ b/arch/arm64/kernel/cpufeature.c
[...]
> +int cpu_copy_el2regs(void *__unused)

Can this be static? I couldn't find it used anywhere else apart from
cpufeature.c

-- 
Catalin

Re: [PATCH v3 09/13] arm64: kernel: Add arch-specific SDEI entry code and CPU masking

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:10PM +0100, James Morse wrote:
> diff --git a/arch/arm64/include/asm/sdei.h b/arch/arm64/include/asm/sdei.h
> new file mode 100644
> index 000000000000..ed329e01a301
> --- /dev/null
> +++ b/arch/arm64/include/asm/sdei.h
> @@ -0,0 +1,63 @@
> +/*
> + * Copyright (C) 2017 ARM Ltd.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
> + */
> +#ifndef __ASM_SDEI_H
> +#define __ASM_SDEI_H
> +
> +/* Values for sdei_exit_mode */
> +#define SDEI_EXIT_HVC  0
> +#define SDEI_EXIT_SMC  1
> +
> +#define SDEI_STACK_SIZE		IRQ_STACK_SIZE
> +
> +#ifndef __ASSEMBLY__
> +
> +#include <linux/linkage.h>
> +#include <linux/types.h>
> +
> +#include <asm/virt.h>
> +
> +extern unsigned long sdei_exit_mode;
> +
> +/* Software Delegated Exception entry point from firmware*/
> +asmlinkage void __sdei_asm_handler(unsigned long event_num, unsigned long arg,
> +				   unsigned long pc, unsigned long pstate);
> +
> +/*
> + * The above entry point does the minimum to call C code. This function does
> + * anything else, before calling the driver.
> + */
> +struct sdei_registered_event;
> +asmlinkage unsigned long __sdei_handler(struct pt_regs *regs,
> +					struct sdei_registered_event *arg);
> +
> +extern unsigned long sdei_arch_get_entry_point(int conduit);

Nitpick: drop the "extern" here.

> diff --git a/arch/arm64/kernel/sdei-entry.S b/arch/arm64/kernel/sdei-entry.S
> new file mode 100644
> index 000000000000..cf12f8da0789
> --- /dev/null
> +++ b/arch/arm64/kernel/sdei-entry.S
> @@ -0,0 +1,122 @@
> +/*
> + * Software Delegated Exception entry point from firmware to the kernel.
> + *
> + * Copyright (C) 2017 ARM Ltd.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
> + */
> +
> +#include <linux/linkage.h>
> +
> +#include <asm/alternative.h>
> +#include <asm/asm-offsets.h>
> +#include <asm/assembler.h>
> +#include <asm/memory.h>
> +#include <asm/ptrace.h>
> +#include <asm/sdei.h>
> +#include <uapi/linux/sdei.h>
> +
> +/*
> + * Software Delegated Exception entry point.
> + *
> + * x0: Event number

Currently the only event number is 0. Shall we plan for having other
events in the future or we just ignore this argument?

> + * x1: struct sdei_registered_event argument from registration time.
> + * x2: interrupted PC
> + * x3: interrupted PSTATE
[...]
> +/*
> + * __sdei_handler() returns one of:
> + *  SDEI_EV_HANDLED -  success, return to the interrupted context.
> + *  SDEI_EV_FAILED  -  failure, return this error code to firmare.
> + *  virtual-address -  success, return to this address.
> + */
> +static __kprobes unsigned long _sdei_handler(struct pt_regs *regs,
> +					     struct sdei_registered_event *arg)
> +{
> +	u32 mode;
> +	int i, err = 0;
> +	int clobbered_registers = 4;

Maybe const int here if you want to avoid magic numbers in the 'for'
loop below. Otherwise it looks like some variable you intend to modify
in this function.

> +	u64 elr = read_sysreg(elr_el1);
> +	u32 kernel_mode = read_sysreg(CurrentEL) | 1;	/* +SPSel */
> +	unsigned long vbar = read_sysreg(vbar_el1);
> +
> +	/* Retrieve the missing registers values */
> +	for (i = 0; i < clobbered_registers; i++) {
> +		/* from within the handler, this call always succeeds */
> +		sdei_api_event_context(i, &regs->regs[i]);
> +	}
> +
> +	/*
> +	 * We didn't take an exception to get here, set PAN. UAO will be cleared
> +	 * by sdei_event_handler()s set_fs(USER_DS) call.
> +	 */
> +	asm(ALTERNATIVE("nop", SET_PSTATE_PAN(1), ARM64_HAS_PAN,
> +			CONFIG_ARM64_PAN));

Can you use uaccess_disable() directly?

> diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
> index d8a9859f6102..1f6633b337aa 100644
> --- a/drivers/firmware/Kconfig
> +++ b/drivers/firmware/Kconfig
> @@ -50,6 +50,7 @@ config ARM_SCPI_POWER_DOMAIN
>  
>  config ARM_SDE_INTERFACE
>  	bool "ARM Software Delegated Exception Interface (SDEI)"
> +	depends on ARM64
>  	help
>  	  The Software Delegated Exception Interface (SDEI) is an ARM
>  	  standard for registering callbacks from the platform firmware
> diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
> index 4b3c7fd99c34..3ea6a19ae439 100644
> --- a/drivers/firmware/arm_sdei.c
> +++ b/drivers/firmware/arm_sdei.c
> @@ -154,6 +154,7 @@ int sdei_api_event_context(u32 query, u64 *result)
>  	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_CONTEXT, query, 0, 0, 0, 0,
>  			      result);
>  }
> +NOKPROBE_SYMBOL(sdei_api_event_context);

Should this be part of the patch introducing arm_sdei.c?

>  
>  static int sdei_api_event_get_info(u32 event, u32 info, u64 *result)
>  {
> diff --git a/include/linux/sdei.h b/include/linux/sdei.h
> index bb3dd000771e..df3fe6373e32 100644
> --- a/include/linux/sdei.h
> +++ b/include/linux/sdei.h
> @@ -32,6 +32,8 @@ enum sdei_conduit_types {
>  	CONDUIT_HVC,
>  };
>  
> +#include <asm/sdei.h>
> +
>  /* Arch code should override this to set the entry point from firmware... */
>  #ifndef sdei_arch_get_entry_point
>  #define sdei_arch_get_entry_point(conduit)	(0)

Same here. This should not be built before the Kconfig change anyway.

-- 
Catalin

[PATCH v3 09/13] arm64: kernel: Add arch-specific SDEI entry code and CPU masking

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:10PM +0100, James Morse wrote:
> diff --git a/arch/arm64/include/asm/sdei.h b/arch/arm64/include/asm/sdei.h
> new file mode 100644
> index 000000000000..ed329e01a301
> --- /dev/null
> +++ b/arch/arm64/include/asm/sdei.h
> @@ -0,0 +1,63 @@
> +/*
> + * Copyright (C) 2017 ARM Ltd.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
> + */
> +#ifndef __ASM_SDEI_H
> +#define __ASM_SDEI_H
> +
> +/* Values for sdei_exit_mode */
> +#define SDEI_EXIT_HVC  0
> +#define SDEI_EXIT_SMC  1
> +
> +#define SDEI_STACK_SIZE		IRQ_STACK_SIZE
> +
> +#ifndef __ASSEMBLY__
> +
> +#include <linux/linkage.h>
> +#include <linux/types.h>
> +
> +#include <asm/virt.h>
> +
> +extern unsigned long sdei_exit_mode;
> +
> +/* Software Delegated Exception entry point from firmware*/
> +asmlinkage void __sdei_asm_handler(unsigned long event_num, unsigned long arg,
> +				   unsigned long pc, unsigned long pstate);
> +
> +/*
> + * The above entry point does the minimum to call C code. This function does
> + * anything else, before calling the driver.
> + */
> +struct sdei_registered_event;
> +asmlinkage unsigned long __sdei_handler(struct pt_regs *regs,
> +					struct sdei_registered_event *arg);
> +
> +extern unsigned long sdei_arch_get_entry_point(int conduit);

Nitpick: drop the "extern" here.

> diff --git a/arch/arm64/kernel/sdei-entry.S b/arch/arm64/kernel/sdei-entry.S
> new file mode 100644
> index 000000000000..cf12f8da0789
> --- /dev/null
> +++ b/arch/arm64/kernel/sdei-entry.S
> @@ -0,0 +1,122 @@
> +/*
> + * Software Delegated Exception entry point from firmware to the kernel.
> + *
> + * Copyright (C) 2017 ARM Ltd.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License
> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
> + */
> +
> +#include <linux/linkage.h>
> +
> +#include <asm/alternative.h>
> +#include <asm/asm-offsets.h>
> +#include <asm/assembler.h>
> +#include <asm/memory.h>
> +#include <asm/ptrace.h>
> +#include <asm/sdei.h>
> +#include <uapi/linux/sdei.h>
> +
> +/*
> + * Software Delegated Exception entry point.
> + *
> + * x0: Event number

Currently the only event number is 0. Shall we plan for having other
events in the future or we just ignore this argument?

> + * x1: struct sdei_registered_event argument from registration time.
> + * x2: interrupted PC
> + * x3: interrupted PSTATE
[...]
> +/*
> + * __sdei_handler() returns one of:
> + *  SDEI_EV_HANDLED -  success, return to the interrupted context.
> + *  SDEI_EV_FAILED  -  failure, return this error code to firmare.
> + *  virtual-address -  success, return to this address.
> + */
> +static __kprobes unsigned long _sdei_handler(struct pt_regs *regs,
> +					     struct sdei_registered_event *arg)
> +{
> +	u32 mode;
> +	int i, err = 0;
> +	int clobbered_registers = 4;

Maybe const int here if you want to avoid magic numbers in the 'for'
loop below. Otherwise it looks like some variable you intend to modify
in this function.

> +	u64 elr = read_sysreg(elr_el1);
> +	u32 kernel_mode = read_sysreg(CurrentEL) | 1;	/* +SPSel */
> +	unsigned long vbar = read_sysreg(vbar_el1);
> +
> +	/* Retrieve the missing registers values */
> +	for (i = 0; i < clobbered_registers; i++) {
> +		/* from within the handler, this call always succeeds */
> +		sdei_api_event_context(i, &regs->regs[i]);
> +	}
> +
> +	/*
> +	 * We didn't take an exception to get here, set PAN. UAO will be cleared
> +	 * by sdei_event_handler()s set_fs(USER_DS) call.
> +	 */
> +	asm(ALTERNATIVE("nop", SET_PSTATE_PAN(1), ARM64_HAS_PAN,
> +			CONFIG_ARM64_PAN));

Can you use uaccess_disable() directly?

> diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
> index d8a9859f6102..1f6633b337aa 100644
> --- a/drivers/firmware/Kconfig
> +++ b/drivers/firmware/Kconfig
> @@ -50,6 +50,7 @@ config ARM_SCPI_POWER_DOMAIN
>  
>  config ARM_SDE_INTERFACE
>  	bool "ARM Software Delegated Exception Interface (SDEI)"
> +	depends on ARM64
>  	help
>  	  The Software Delegated Exception Interface (SDEI) is an ARM
>  	  standard for registering callbacks from the platform firmware
> diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
> index 4b3c7fd99c34..3ea6a19ae439 100644
> --- a/drivers/firmware/arm_sdei.c
> +++ b/drivers/firmware/arm_sdei.c
> @@ -154,6 +154,7 @@ int sdei_api_event_context(u32 query, u64 *result)
>  	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_CONTEXT, query, 0, 0, 0, 0,
>  			      result);
>  }
> +NOKPROBE_SYMBOL(sdei_api_event_context);

Should this be part of the patch introducing arm_sdei.c?

>  
>  static int sdei_api_event_get_info(u32 event, u32 info, u64 *result)
>  {
> diff --git a/include/linux/sdei.h b/include/linux/sdei.h
> index bb3dd000771e..df3fe6373e32 100644
> --- a/include/linux/sdei.h
> +++ b/include/linux/sdei.h
> @@ -32,6 +32,8 @@ enum sdei_conduit_types {
>  	CONDUIT_HVC,
>  };
>  
> +#include <asm/sdei.h>
> +
>  /* Arch code should override this to set the entry point from firmware... */
>  #ifndef sdei_arch_get_entry_point
>  #define sdei_arch_get_entry_point(conduit)	(0)

Same here. This should not be built before the Kconfig change anyway.

-- 
Catalin

Re: [PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:11PM +0100, James Morse wrote:
> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
> index f24bfb2b9a2d..466b949474df 100644
> --- a/include/linux/cpuhotplug.h
> +++ b/include/linux/cpuhotplug.h
> @@ -88,6 +88,7 @@ enum cpuhp_state {
>  	CPUHP_AP_PERF_XTENSA_STARTING,
>  	CPUHP_AP_PERF_METAG_STARTING,
>  	CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
> +	CPUHP_AP_SDEI_STARTING,

Nitpick: how generic is this as to apply to other architectures?
Probably not, so shall we prefix this with ARM_?

Not actually a strong preferences but similar question for the
definitions in the sdei.h files (and maybe the filenames themselves).

-- 
Catalin
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states

[Catalin Marinas]

On Fri, Sep 22, 2017 at 07:26:11PM +0100, James Morse wrote:
> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
> index f24bfb2b9a2d..466b949474df 100644
> --- a/include/linux/cpuhotplug.h
> +++ b/include/linux/cpuhotplug.h
> @@ -88,6 +88,7 @@ enum cpuhp_state {
>  	CPUHP_AP_PERF_XTENSA_STARTING,
>  	CPUHP_AP_PERF_METAG_STARTING,
>  	CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
> +	CPUHP_AP_SDEI_STARTING,

Nitpick: how generic is this as to apply to other architectures?
Probably not, so shall we prefix this with ARM_?

Not actually a strong preferences but similar question for the
definitions in the sdei.h files (and maybe the filenames themselves).

-- 
Catalin

Re: [PATCH v3 09/13] arm64: kernel: Add arch-specific SDEI entry code and CPU masking

[James Morse]

Hi Catalin,

On 16/10/17 14:41, Catalin Marinas wrote:
> On Fri, Sep 22, 2017 at 07:26:10PM +0100, James Morse wrote:
>> diff --git a/arch/arm64/kernel/sdei-entry.S b/arch/arm64/kernel/sdei-entry.S
>> new file mode 100644
>> index 000000000000..cf12f8da0789
>> --- /dev/null
>> +++ b/arch/arm64/kernel/sdei-entry.S
>> @@ -0,0 +1,122 @@
>> +/*
>> + * Software Delegated Exception entry point from firmware to the kernel.
>> + *
>> + * Copyright (C) 2017 ARM Ltd.
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License version 2 as
>> + * published by the Free Software Foundation.
>> + *
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> + * GNU General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU General Public License
>> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
>> + */
>> +
>> +#include <linux/linkage.h>
>> +
>> +#include <asm/alternative.h>
>> +#include <asm/asm-offsets.h>
>> +#include <asm/assembler.h>
>> +#include <asm/memory.h>
>> +#include <asm/ptrace.h>
>> +#include <asm/sdei.h>
>> +#include <uapi/linux/sdei.h>
>> +
>> +/*
>> + * Software Delegated Exception entry point.
>> + *
>> + * x0: Event number
> 
> Currently the only event number is 0. Shall we plan for having other
> events in the future or we just ignore this argument?

'0' is the only event number specified by the SDEI specification. For use with
RAS the event number is read from the HEST. (and there may me more than one of
them).



>> + * x1: struct sdei_registered_event argument from registration time.
>> + * x2: interrupted PC
>> + * x3: interrupted PSTATE
> [...]
>> +/*
>> + * __sdei_handler() returns one of:
>> + *  SDEI_EV_HANDLED -  success, return to the interrupted context.
>> + *  SDEI_EV_FAILED  -  failure, return this error code to firmare.
>> + *  virtual-address -  success, return to this address.
>> + */
>> +static __kprobes unsigned long _sdei_handler(struct pt_regs *regs,
>> +					     struct sdei_registered_event *arg)
>> +{
>> +	u32 mode;
>> +	int i, err = 0;
>> +	int clobbered_registers = 4;

> Maybe const int here if you want to avoid magic numbers in the 'for'
> loop below. Otherwise it looks like some variable you intend to modify
> in this function.

Sure.

(this was modified in my various attempts to get this working on non-vhe systems
with KVM).


>> +	u64 elr = read_sysreg(elr_el1);
>> +	u32 kernel_mode = read_sysreg(CurrentEL) | 1;	/* +SPSel */
>> +	unsigned long vbar = read_sysreg(vbar_el1);
>> +
>> +	/* Retrieve the missing registers values */
>> +	for (i = 0; i < clobbered_registers; i++) {
>> +		/* from within the handler, this call always succeeds */
>> +		sdei_api_event_context(i, &regs->regs[i]);
>> +	}
>> +
>> +	/*
>> +	 * We didn't take an exception to get here, set PAN. UAO will be cleared
>> +	 * by sdei_event_handler()s set_fs(USER_DS) call.
>> +	 */
>> +	asm(ALTERNATIVE("nop", SET_PSTATE_PAN(1), ARM64_HAS_PAN,
>> +			CONFIG_ARM64_PAN));

> Can you use uaccess_disable() directly?

Wouldn't this invoke sw-pan's __uaccess_ttbr0_disable():
> write_sysreg(ttbr, ttbr0_el1);

Probing depends on VHE if booted at EL2, can we guarantee to have PAN (and
therefore not-SW-PAN) too?

(otherwise I can add 'depends on !ARM64_SW_TTBR0_PAN' to the Kconfig)


>> diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
>> index d8a9859f6102..1f6633b337aa 100644
>> --- a/drivers/firmware/Kconfig
>> +++ b/drivers/firmware/Kconfig
>> @@ -50,6 +50,7 @@ config ARM_SCPI_POWER_DOMAIN
>>  
>>  config ARM_SDE_INTERFACE
>>  	bool "ARM Software Delegated Exception Interface (SDEI)"
>> +	depends on ARM64
>>  	help
>>  	  The Software Delegated Exception Interface (SDEI) is an ARM
>>  	  standard for registering callbacks from the platform firmware
>> diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
>> index 4b3c7fd99c34..3ea6a19ae439 100644
>> --- a/drivers/firmware/arm_sdei.c
>> +++ b/drivers/firmware/arm_sdei.c
>> @@ -154,6 +154,7 @@ int sdei_api_event_context(u32 query, u64 *result)
>>  	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_CONTEXT, query, 0, 0, 0, 0,
>>  			      result);
>>  }
>> +NOKPROBE_SYMBOL(sdei_api_event_context);

> Should this be part of the patch introducing arm_sdei.c?

Yes.


>>  static int sdei_api_event_get_info(u32 event, u32 info, u64 *result)
>>  {
>> diff --git a/include/linux/sdei.h b/include/linux/sdei.h
>> index bb3dd000771e..df3fe6373e32 100644
>> --- a/include/linux/sdei.h
>> +++ b/include/linux/sdei.h
>> @@ -32,6 +32,8 @@ enum sdei_conduit_types {
>>  	CONDUIT_HVC,
>>  };
>>  
>> +#include <asm/sdei.h>
>> +
>>  /* Arch code should override this to set the entry point from firmware... */
>>  #ifndef sdei_arch_get_entry_point
>>  #define sdei_arch_get_entry_point(conduit)	(0)
> 
> Same here. This should not be built before the Kconfig change anyway.

(I've been building this on x86 to check I'd not done anything arm64 specific
outside the arch code...)

Sure, I'll move this hunk and the Kconfig create a place holder asm/sdei.h in
the earlier patch.


Thanks,

James

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 09/13] arm64: kernel: Add arch-specific SDEI entry code and CPU masking

[James Morse]

Hi Catalin,

On 16/10/17 14:41, Catalin Marinas wrote:
> On Fri, Sep 22, 2017 at 07:26:10PM +0100, James Morse wrote:
>> diff --git a/arch/arm64/kernel/sdei-entry.S b/arch/arm64/kernel/sdei-entry.S
>> new file mode 100644
>> index 000000000000..cf12f8da0789
>> --- /dev/null
>> +++ b/arch/arm64/kernel/sdei-entry.S
>> @@ -0,0 +1,122 @@
>> +/*
>> + * Software Delegated Exception entry point from firmware to the kernel.
>> + *
>> + * Copyright (C) 2017 ARM Ltd.
>> + *
>> + * This program is free software; you can redistribute it and/or modify
>> + * it under the terms of the GNU General Public License version 2 as
>> + * published by the Free Software Foundation.
>> + *
>> + * This program is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> + * GNU General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU General Public License
>> + * along with this program.  If not, see <http://www.gnu.org/licenses/>.
>> + */
>> +
>> +#include <linux/linkage.h>
>> +
>> +#include <asm/alternative.h>
>> +#include <asm/asm-offsets.h>
>> +#include <asm/assembler.h>
>> +#include <asm/memory.h>
>> +#include <asm/ptrace.h>
>> +#include <asm/sdei.h>
>> +#include <uapi/linux/sdei.h>
>> +
>> +/*
>> + * Software Delegated Exception entry point.
>> + *
>> + * x0: Event number
> 
> Currently the only event number is 0. Shall we plan for having other
> events in the future or we just ignore this argument?

'0' is the only event number specified by the SDEI specification. For use with
RAS the event number is read from the HEST. (and there may me more than one of
them).



>> + * x1: struct sdei_registered_event argument from registration time.
>> + * x2: interrupted PC
>> + * x3: interrupted PSTATE
> [...]
>> +/*
>> + * __sdei_handler() returns one of:
>> + *  SDEI_EV_HANDLED -  success, return to the interrupted context.
>> + *  SDEI_EV_FAILED  -  failure, return this error code to firmare.
>> + *  virtual-address -  success, return to this address.
>> + */
>> +static __kprobes unsigned long _sdei_handler(struct pt_regs *regs,
>> +					     struct sdei_registered_event *arg)
>> +{
>> +	u32 mode;
>> +	int i, err = 0;
>> +	int clobbered_registers = 4;

> Maybe const int here if you want to avoid magic numbers in the 'for'
> loop below. Otherwise it looks like some variable you intend to modify
> in this function.

Sure.

(this was modified in my various attempts to get this working on non-vhe systems
with KVM).


>> +	u64 elr = read_sysreg(elr_el1);
>> +	u32 kernel_mode = read_sysreg(CurrentEL) | 1;	/* +SPSel */
>> +	unsigned long vbar = read_sysreg(vbar_el1);
>> +
>> +	/* Retrieve the missing registers values */
>> +	for (i = 0; i < clobbered_registers; i++) {
>> +		/* from within the handler, this call always succeeds */
>> +		sdei_api_event_context(i, &regs->regs[i]);
>> +	}
>> +
>> +	/*
>> +	 * We didn't take an exception to get here, set PAN. UAO will be cleared
>> +	 * by sdei_event_handler()s set_fs(USER_DS) call.
>> +	 */
>> +	asm(ALTERNATIVE("nop", SET_PSTATE_PAN(1), ARM64_HAS_PAN,
>> +			CONFIG_ARM64_PAN));

> Can you use uaccess_disable() directly?

Wouldn't this invoke sw-pan's __uaccess_ttbr0_disable():
> write_sysreg(ttbr, ttbr0_el1);

Probing depends on VHE if booted at EL2, can we guarantee to have PAN (and
therefore not-SW-PAN) too?

(otherwise I can add 'depends on !ARM64_SW_TTBR0_PAN' to the Kconfig)


>> diff --git a/drivers/firmware/Kconfig b/drivers/firmware/Kconfig
>> index d8a9859f6102..1f6633b337aa 100644
>> --- a/drivers/firmware/Kconfig
>> +++ b/drivers/firmware/Kconfig
>> @@ -50,6 +50,7 @@ config ARM_SCPI_POWER_DOMAIN
>>  
>>  config ARM_SDE_INTERFACE
>>  	bool "ARM Software Delegated Exception Interface (SDEI)"
>> +	depends on ARM64
>>  	help
>>  	  The Software Delegated Exception Interface (SDEI) is an ARM
>>  	  standard for registering callbacks from the platform firmware
>> diff --git a/drivers/firmware/arm_sdei.c b/drivers/firmware/arm_sdei.c
>> index 4b3c7fd99c34..3ea6a19ae439 100644
>> --- a/drivers/firmware/arm_sdei.c
>> +++ b/drivers/firmware/arm_sdei.c
>> @@ -154,6 +154,7 @@ int sdei_api_event_context(u32 query, u64 *result)
>>  	return invoke_sdei_fn(SDEI_1_0_FN_SDEI_EVENT_CONTEXT, query, 0, 0, 0, 0,
>>  			      result);
>>  }
>> +NOKPROBE_SYMBOL(sdei_api_event_context);

> Should this be part of the patch introducing arm_sdei.c?

Yes.


>>  static int sdei_api_event_get_info(u32 event, u32 info, u64 *result)
>>  {
>> diff --git a/include/linux/sdei.h b/include/linux/sdei.h
>> index bb3dd000771e..df3fe6373e32 100644
>> --- a/include/linux/sdei.h
>> +++ b/include/linux/sdei.h
>> @@ -32,6 +32,8 @@ enum sdei_conduit_types {
>>  	CONDUIT_HVC,
>>  };
>>  
>> +#include <asm/sdei.h>
>> +
>>  /* Arch code should override this to set the entry point from firmware... */
>>  #ifndef sdei_arch_get_entry_point
>>  #define sdei_arch_get_entry_point(conduit)	(0)
> 
> Same here. This should not be built before the Kconfig change anyway.

(I've been building this on x86 to check I'd not done anything arm64 specific
outside the arch code...)

Sure, I'll move this hunk and the Kconfig create a place holder asm/sdei.h in
the earlier patch.


Thanks,

James

Re: [PATCH v3 08/13] arm64: Add vmap_stack header file

[James Morse]

Hi Catalin,

On 13/10/17 16:42, Catalin Marinas wrote:
> On Fri, Sep 22, 2017 at 07:26:09PM +0100, James Morse wrote:
>> diff --git a/arch/arm64/include/asm/vmap_stack.h b/arch/arm64/include/asm/vmap_stack.h
>> new file mode 100644
>> index 000000000000..f41d043cac31
>> --- /dev/null
>> +++ b/arch/arm64/include/asm/vmap_stack.h
>> @@ -0,0 +1,41 @@

>> +#ifndef __ASM_VMAP_STACK_H
>> +#define __ASM_VMAP_STACK_H
>> +
>> +#include <linux/vmalloc.h>
>> +#include <asm/memory.h>
>> +#include <asm/pgtable.h>
>> +#include <asm/thread_info.h>
>> +
>> +#ifdef CONFIG_VMAP_STACK
>> +/*
>> + * To ensure that VMAP'd stack overflow detection works correctly, all VMAP'd
>> + * stacks need to have the same alignment.
>> + */
>> +static inline unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node)
>> +{
>> +	return __vmalloc_node_range(stack_size, THREAD_ALIGN,
>> +				    VMALLOC_START, VMALLOC_END,
>> +				    THREADINFO_GFP, PAGE_KERNEL, 0, node,
>> +				    __builtin_return_address(0));
>> +}
>> +
>> +#else
>> +unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node); // link error

> Do you actually need this here? The calling site (at least in this
> patch), is surrounded by CONFIG_VMAP_STACK. Removing the dummy
> declaration should be fine (I haven't read the rest of the patches yet,
> maybe it's needed later;

This was to avoid having to #ifdef the calling site I add in a later patch.


> would a BUILD_BUG do?).

Ah, you can put those in header files, yes that would be clearer.


Thanks,

James





--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 08/13] arm64: Add vmap_stack header file

[James Morse]

Hi Catalin,

On 13/10/17 16:42, Catalin Marinas wrote:
> On Fri, Sep 22, 2017 at 07:26:09PM +0100, James Morse wrote:
>> diff --git a/arch/arm64/include/asm/vmap_stack.h b/arch/arm64/include/asm/vmap_stack.h
>> new file mode 100644
>> index 000000000000..f41d043cac31
>> --- /dev/null
>> +++ b/arch/arm64/include/asm/vmap_stack.h
>> @@ -0,0 +1,41 @@

>> +#ifndef __ASM_VMAP_STACK_H
>> +#define __ASM_VMAP_STACK_H
>> +
>> +#include <linux/vmalloc.h>
>> +#include <asm/memory.h>
>> +#include <asm/pgtable.h>
>> +#include <asm/thread_info.h>
>> +
>> +#ifdef CONFIG_VMAP_STACK
>> +/*
>> + * To ensure that VMAP'd stack overflow detection works correctly, all VMAP'd
>> + * stacks need to have the same alignment.
>> + */
>> +static inline unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node)
>> +{
>> +	return __vmalloc_node_range(stack_size, THREAD_ALIGN,
>> +				    VMALLOC_START, VMALLOC_END,
>> +				    THREADINFO_GFP, PAGE_KERNEL, 0, node,
>> +				    __builtin_return_address(0));
>> +}
>> +
>> +#else
>> +unsigned long *arch_alloc_vmap_stack(size_t stack_size, int node); // link error

> Do you actually need this here? The calling site (at least in this
> patch), is surrounded by CONFIG_VMAP_STACK. Removing the dummy
> declaration should be fine (I haven't read the rest of the patches yet,
> maybe it's needed later;

This was to avoid having to #ifdef the calling site I add in a later patch.


> would a BUILD_BUG do?).

Ah, you can put those in header files, yes that would be clearer.


Thanks,

James

Re: [PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states

[James Morse]

Hi Catalin,

On 16/10/17 14:52, Catalin Marinas wrote:
> On Fri, Sep 22, 2017 at 07:26:11PM +0100, James Morse wrote:
>> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
>> index f24bfb2b9a2d..466b949474df 100644
>> --- a/include/linux/cpuhotplug.h
>> +++ b/include/linux/cpuhotplug.h
>> @@ -88,6 +88,7 @@ enum cpuhp_state {
>>  	CPUHP_AP_PERF_XTENSA_STARTING,
>>  	CPUHP_AP_PERF_METAG_STARTING,
>>  	CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
>> +	CPUHP_AP_SDEI_STARTING,
> 
> Nitpick: how generic is this as to apply to other architectures?
> Probably not, so shall we prefix this with ARM_?

This may get used on 32bit ARM. I blindly copied PSCI, which doesn't have a
prefix. But it makes sense to have one.


> Not actually a strong preferences but similar question for the
> definitions in the sdei.h files (and maybe the filenames themselves).


Thanks,

James

[PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states

[James Morse]

Hi Catalin,

On 16/10/17 14:52, Catalin Marinas wrote:
> On Fri, Sep 22, 2017 at 07:26:11PM +0100, James Morse wrote:
>> diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h
>> index f24bfb2b9a2d..466b949474df 100644
>> --- a/include/linux/cpuhotplug.h
>> +++ b/include/linux/cpuhotplug.h
>> @@ -88,6 +88,7 @@ enum cpuhp_state {
>>  	CPUHP_AP_PERF_XTENSA_STARTING,
>>  	CPUHP_AP_PERF_METAG_STARTING,
>>  	CPUHP_AP_MIPS_OP_LOONGSON3_STARTING,
>> +	CPUHP_AP_SDEI_STARTING,
> 
> Nitpick: how generic is this as to apply to other architectures?
> Probably not, so shall we prefix this with ARM_?

This may get used on 32bit ARM. I blindly copied PSCI, which doesn't have a
prefix. But it makes sense to have one.


> Not actually a strong preferences but similar question for the
> definitions in the sdei.h files (and maybe the filenames themselves).


Thanks,

James

Re: [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[James Morse]

Hi Catalin,

On 16/10/17 10:58, Catalin Marinas wrote:
> On Fri, Oct 13, 2017 at 05:50:45PM +0100, James Morse wrote:
>> On 13/10/17 16:31, Catalin Marinas wrote:
>>> On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
>>>> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
>>>> index cd52d365d1f0..8e4c7da2b126 100644
>>>> --- a/arch/arm64/kernel/cpufeature.c
>>>> +++ b/arch/arm64/kernel/cpufeature.c
>>
>>>> @@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
>>>>  }
>>>>  
>>>>  late_initcall(enable_mrs_emulation);
>>>> +
>>>> +int cpu_copy_el2regs(void *__unused)
>>>> +{
>>>> +	int do_copyregs = 0;
>>>> +
>>>> +	/*
>>>> +	 * Copy register values that aren't redirected by hardware.
>>>> +	 *
>>>> +	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
>>>> +	 * this value to tpidr_el2 before we patch the code. Once we've done
>>>> +	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
>>>> +	 * do anything here.
>>>> +	 */
>>>> +	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
>>>> +				 ARM64_HAS_VIRT_HOST_EXTN)
>>>> +		    : "=r" (do_copyregs) : : );
>>>
>>> Can you just do:
>>>
>>> 	if (cpu_have_const_cap(ARM64_HAS_VIRT_HOST_EXTN))
>>> 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
>>>
>>> At this point the capability bits should be set and the jump labels
>>> enabled.
>>
>> These are enabled too early, we haven't done patching yet.
>>
>> We need to copy tpidr_el1 -> tpidr_el2 on all CPUs that are online before code
>> patching.
>>
>> After code patching new CPUs set tpidr_el2 when they come online, so we don't
>> need to do the copy... but this enable method is still called. There is nothing
>> for us to do, and tpidr_el1 now contains junk, so the copy

> Ah, I get it now (should've read the comment but I usually expect the
> code to be obvious; it wasn't, at least to me, in this case ;)).

> You could have added the sysreg copying directly in the asm volatile.

I was trying to stick to the sysreg C accessors, and thought there would be more
registers that needed copying. (I discovered this VHE doesn't remap all the _ELx
registers quite late.)


> Anyway, I think it's better if we keep it entirely in C with this hunk
> (untested):

[...]

Yes, that looks much better. I got tangled up in 'which alternative', but you're
right, they are all applied in one go so it doesn't matter.


>	if (!READ_ONCE(alternatives_applied))
> 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);

I don't think this READ_ONCE() is needed, that only matters within the
stop_machine()/alternatives-patching code that modifies the value on one CPU.


Thanks,

James

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts

[James Morse]

Hi Catalin,

On 16/10/17 10:58, Catalin Marinas wrote:
> On Fri, Oct 13, 2017 at 05:50:45PM +0100, James Morse wrote:
>> On 13/10/17 16:31, Catalin Marinas wrote:
>>> On Fri, Sep 22, 2017 at 07:26:05PM +0100, James Morse wrote:
>>>> diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
>>>> index cd52d365d1f0..8e4c7da2b126 100644
>>>> --- a/arch/arm64/kernel/cpufeature.c
>>>> +++ b/arch/arm64/kernel/cpufeature.c
>>
>>>> @@ -1308,3 +1309,25 @@ static int __init enable_mrs_emulation(void)
>>>>  }
>>>>  
>>>>  late_initcall(enable_mrs_emulation);
>>>> +
>>>> +int cpu_copy_el2regs(void *__unused)
>>>> +{
>>>> +	int do_copyregs = 0;
>>>> +
>>>> +	/*
>>>> +	 * Copy register values that aren't redirected by hardware.
>>>> +	 *
>>>> +	 * Before code patching, we only set tpidr_el1, all CPUs need to copy
>>>> +	 * this value to tpidr_el2 before we patch the code. Once we've done
>>>> +	 * that, freshly-onlined CPUs will set tpidr_el2, so we don't need to
>>>> +	 * do anything here.
>>>> +	 */
>>>> +	asm volatile(ALTERNATIVE("mov %0, #1", "mov %0, #0",
>>>> +				 ARM64_HAS_VIRT_HOST_EXTN)
>>>> +		    : "=r" (do_copyregs) : : );
>>>
>>> Can you just do:
>>>
>>> 	if (cpu_have_const_cap(ARM64_HAS_VIRT_HOST_EXTN))
>>> 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);
>>>
>>> At this point the capability bits should be set and the jump labels
>>> enabled.
>>
>> These are enabled too early, we haven't done patching yet.
>>
>> We need to copy tpidr_el1 -> tpidr_el2 on all CPUs that are online before code
>> patching.
>>
>> After code patching new CPUs set tpidr_el2 when they come online, so we don't
>> need to do the copy... but this enable method is still called. There is nothing
>> for us to do, and tpidr_el1 now contains junk, so the copy

> Ah, I get it now (should've read the comment but I usually expect the
> code to be obvious; it wasn't, at least to me, in this case ;)).

> You could have added the sysreg copying directly in the asm volatile.

I was trying to stick to the sysreg C accessors, and thought there would be more
registers that needed copying. (I discovered this VHE doesn't remap all the _ELx
registers quite late.)


> Anyway, I think it's better if we keep it entirely in C with this hunk
> (untested):

[...]

Yes, that looks much better. I got tangled up in 'which alternative', but you're
right, they are all applied in one go so it doesn't matter.


>	if (!READ_ONCE(alternatives_applied))
> 		write_sysreg(read_sysreg(tpidr_el1), tpidr_el2);

I don't think this READ_ONCE() is needed, that only matters within the
stop_machine()/alternatives-patching code that modifies the value on one CPU.


Thanks,

James

Re: [PATCH v3 09/13] arm64: kernel: Add arch-specific SDEI entry code and CPU masking

[Catalin Marinas]

On Tue, Oct 17, 2017 at 05:34:13PM +0100, James Morse wrote:
> On 16/10/17 14:41, Catalin Marinas wrote:
> > On Fri, Sep 22, 2017 at 07:26:10PM +0100, James Morse wrote:
> >> +	u64 elr = read_sysreg(elr_el1);
> >> +	u32 kernel_mode = read_sysreg(CurrentEL) | 1;	/* +SPSel */
> >> +	unsigned long vbar = read_sysreg(vbar_el1);
> >> +
> >> +	/* Retrieve the missing registers values */
> >> +	for (i = 0; i < clobbered_registers; i++) {
> >> +		/* from within the handler, this call always succeeds */
> >> +		sdei_api_event_context(i, &regs->regs[i]);
> >> +	}
> >> +
> >> +	/*
> >> +	 * We didn't take an exception to get here, set PAN. UAO will be cleared
> >> +	 * by sdei_event_handler()s set_fs(USER_DS) call.
> >> +	 */
> >> +	asm(ALTERNATIVE("nop", SET_PSTATE_PAN(1), ARM64_HAS_PAN,
> >> +			CONFIG_ARM64_PAN));
> 
> > Can you use uaccess_disable() directly?
> 
> Wouldn't this invoke sw-pan's __uaccess_ttbr0_disable():
> > write_sysreg(ttbr, ttbr0_el1);
> 
> Probing depends on VHE if booted at EL2, can we guarantee to have PAN (and
> therefore not-SW-PAN) too?
> 
> (otherwise I can add 'depends on !ARM64_SW_TTBR0_PAN' to the Kconfig)

We want the Kconfig to be able to include all features. What you can do
though is:

	select ARM64_PAN if ARM64_SW_TTBR0_PAN

With VHE (ARMv8.2) we can guarantee that hardware PAN is around.

-- 
Catalin
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v3 09/13] arm64: kernel: Add arch-specific SDEI entry code and CPU masking

[Catalin Marinas]

On Tue, Oct 17, 2017 at 05:34:13PM +0100, James Morse wrote:
> On 16/10/17 14:41, Catalin Marinas wrote:
> > On Fri, Sep 22, 2017 at 07:26:10PM +0100, James Morse wrote:
> >> +	u64 elr = read_sysreg(elr_el1);
> >> +	u32 kernel_mode = read_sysreg(CurrentEL) | 1;	/* +SPSel */
> >> +	unsigned long vbar = read_sysreg(vbar_el1);
> >> +
> >> +	/* Retrieve the missing registers values */
> >> +	for (i = 0; i < clobbered_registers; i++) {
> >> +		/* from within the handler, this call always succeeds */
> >> +		sdei_api_event_context(i, &regs->regs[i]);
> >> +	}
> >> +
> >> +	/*
> >> +	 * We didn't take an exception to get here, set PAN. UAO will be cleared
> >> +	 * by sdei_event_handler()s set_fs(USER_DS) call.
> >> +	 */
> >> +	asm(ALTERNATIVE("nop", SET_PSTATE_PAN(1), ARM64_HAS_PAN,
> >> +			CONFIG_ARM64_PAN));
> 
> > Can you use uaccess_disable() directly?
> 
> Wouldn't this invoke sw-pan's __uaccess_ttbr0_disable():
> > write_sysreg(ttbr, ttbr0_el1);
> 
> Probing depends on VHE if booted at EL2, can we guarantee to have PAN (and
> therefore not-SW-PAN) too?
> 
> (otherwise I can add 'depends on !ARM64_SW_TTBR0_PAN' to the Kconfig)

We want the Kconfig to be able to include all features. What you can do
though is:

	select ARM64_PAN if ARM64_SW_TTBR0_PAN

With VHE (ARMv8.2) we can guarantee that hardware PAN is around.

-- 
Catalin