* [PATCH v3 01/13] KVM: arm64: Store vcpu on the stack during __guest_enter()
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
@ 2017-09-22 18:26 ` James Morse
2017-09-22 18:26 ` [PATCH v3 02/13] KVM: arm/arm64: Convert kvm_host_cpu_state to a static per-cpu allocation James Morse
` (10 subsequent siblings)
11 siblings, 0 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 02/13] KVM: arm/arm64: Convert kvm_host_cpu_state to a static per-cpu allocation
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
2017-09-22 18:26 ` [PATCH v3 01/13] KVM: arm64: Store vcpu on the stack during __guest_enter() James Morse
@ 2017-09-22 18:26 ` James Morse
2017-09-22 18:26 ` [PATCH v3 03/13] KVM: arm64: Change hyp_panic()s dependency on tpidr_el2 James Morse
` (9 subsequent siblings)
11 siblings, 0 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 03/13] KVM: arm64: Change hyp_panic()s dependency on tpidr_el2
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
2017-09-22 18:26 ` [PATCH v3 01/13] KVM: arm64: Store vcpu on the stack during __guest_enter() James Morse
2017-09-22 18:26 ` [PATCH v3 02/13] KVM: arm/arm64: Convert kvm_host_cpu_state to a static per-cpu allocation James Morse
@ 2017-09-22 18:26 ` James Morse
2017-09-22 18:26 ` [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts James Morse
` (8 subsequent siblings)
11 siblings, 0 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (2 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 03/13] KVM: arm64: Change hyp_panic()s dependency on tpidr_el2 James Morse
@ 2017-09-22 18:26 ` James Morse
[not found] ` <20170922182614.27885-5-james.morse-5wv7dgnIgG8@public.gmane.org>
2017-10-16 10:17 ` Catalin Marinas
2017-09-22 18:26 ` [PATCH v3 05/13] KVM: arm64: Stop save/restoring host tpidr_el1 on VHE James Morse
` (7 subsequent siblings)
11 siblings, 2 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread[parent not found: <20170922182614.27885-5-james.morse-5wv7dgnIgG8@public.gmane.org>]
* Re: [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts
[not found] ` <20170922182614.27885-5-james.morse-5wv7dgnIgG8@public.gmane.org>
@ 2017-10-13 15:31 ` Catalin Marinas
[not found] ` <20171013153148.dnejsvhxeui6opfw-+1aNUgJU5qkijLcmloz0ER/iLCjYCKR+VpNB7YpNyf8@public.gmane.org>
0 siblings, 1 reply; 27+ messages in thread
From: Catalin Marinas @ 2017-10-13 15:31 UTC (permalink / raw)
To: James Morse
Cc: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r, Mark Rutland,
devicetree-u79uwXL29TY76Z2rM5mHXA, Lorenzo Pieralisi,
Marc Zyngier, Will Deacon, Rob Herring, Loc Ho,
kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg, Christoffer Dall
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
^ permalink raw reply [flat|nested] 27+ messages in thread
* Re: [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts
2017-09-22 18:26 ` [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts James Morse
[not found] ` <20170922182614.27885-5-james.morse-5wv7dgnIgG8@public.gmane.org>
@ 2017-10-16 10:17 ` Catalin Marinas
1 sibling, 0 replies; 27+ messages in thread
From: Catalin Marinas @ 2017-10-16 10:17 UTC (permalink / raw)
To: James Morse
Cc: devicetree, Lorenzo Pieralisi, Marc Zyngier, Will Deacon,
Rob Herring, Loc Ho, kvmarm, linux-arm-kernel
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
^ permalink raw reply [flat|nested] 27+ messages in thread
* [PATCH v3 05/13] KVM: arm64: Stop save/restoring host tpidr_el1 on VHE
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (3 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 04/13] arm64: alternatives: use tpidr_el2 on VHE hosts James Morse
@ 2017-09-22 18:26 ` James Morse
2017-09-22 18:26 ` [PATCH v3 06/13] Docs: dt: add devicetree binding for describing arm64 SDEI firmware James Morse
` (6 subsequent siblings)
11 siblings, 0 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 06/13] Docs: dt: add devicetree binding for describing arm64 SDEI firmware
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (4 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 05/13] KVM: arm64: Stop save/restoring host tpidr_el1 on VHE James Morse
@ 2017-09-22 18:26 ` James Morse
2017-09-22 18:26 ` [PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions James Morse
` (5 subsequent siblings)
11 siblings, 0 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (5 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 06/13] Docs: dt: add devicetree binding for describing arm64 SDEI firmware James Morse
@ 2017-09-22 18:26 ` James Morse
2017-10-13 15:49 ` Catalin Marinas
2017-09-22 18:26 ` [PATCH v3 08/13] arm64: Add vmap_stack header file James Morse
` (4 subsequent siblings)
11 siblings, 1 reply; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* Re: [PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions
2017-09-22 18:26 ` [PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions James Morse
@ 2017-10-13 15:49 ` Catalin Marinas
0 siblings, 0 replies; 27+ messages in thread
From: Catalin Marinas @ 2017-10-13 15:49 UTC (permalink / raw)
To: James Morse
Cc: devicetree, Lorenzo Pieralisi, Marc Zyngier, Will Deacon,
Rob Herring, Loc Ho, kvmarm, linux-arm-kernel
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
^ permalink raw reply [flat|nested] 27+ messages in thread
* [PATCH v3 08/13] arm64: Add vmap_stack header file
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (6 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 07/13] firmware: arm_sdei: Add driver for Software Delegated Exceptions James Morse
@ 2017-09-22 18:26 ` James Morse
[not found] ` <20170922182614.27885-9-james.morse-5wv7dgnIgG8@public.gmane.org>
2017-09-22 18:26 ` [PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states James Morse
` (3 subsequent siblings)
11 siblings, 1 reply; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (7 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 08/13] arm64: Add vmap_stack header file James Morse
@ 2017-09-22 18:26 ` James Morse
[not found] ` <20170922182614.27885-11-james.morse-5wv7dgnIgG8@public.gmane.org>
2017-09-22 18:26 ` [PATCH v3 11/13] firmware: arm_sdei: add support for CPU private events James Morse
` (2 subsequent siblings)
11 siblings, 1 reply; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 11/13] firmware: arm_sdei: add support for CPU private events
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (8 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 10/13] firmware: arm_sdei: Add support for CPU and system power states James Morse
@ 2017-09-22 18:26 ` James Morse
2017-09-22 18:26 ` [PATCH v3 12/13] arm64: acpi: Remove __init from acpi_psci_use_hvc() for use by SDEI James Morse
2017-09-22 18:26 ` [PATCH v3 13/13] firmware: arm_sdei: Discover SDEI support via ACPI James Morse
11 siblings, 0 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 12/13] arm64: acpi: Remove __init from acpi_psci_use_hvc() for use by SDEI
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (9 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 11/13] firmware: arm_sdei: add support for CPU private events James Morse
@ 2017-09-22 18:26 ` James Morse
2017-09-22 18:26 ` [PATCH v3 13/13] firmware: arm_sdei: Discover SDEI support via ACPI James Morse
11 siblings, 0 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread* [PATCH v3 13/13] firmware: arm_sdei: Discover SDEI support via ACPI
[not found] ` <20170922182614.27885-1-james.morse-5wv7dgnIgG8@public.gmane.org>
` (10 preceding siblings ...)
2017-09-22 18:26 ` [PATCH v3 12/13] arm64: acpi: Remove __init from acpi_psci_use_hvc() for use by SDEI James Morse
@ 2017-09-22 18:26 ` James Morse
11 siblings, 0 replies; 27+ messages in thread
From: James Morse @ 2017-09-22 18:26 UTC (permalink / raw)
To: linux-arm-kernel-IAPFreCvJWM7uuMidbF8XUB+6BGkLq7r
Cc: kvmarm-FPEHb7Xf0XXUo1n7N8X6UoWGPAHP3yOg,
devicetree-u79uwXL29TY76Z2rM5mHXA, Will Deacon, Catalin Marinas,
Mark Rutland, Rob Herring, Marc Zyngier, Christoffer Dall,
Lorenzo Pieralisi, Loc Ho
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
^ permalink raw reply related [flat|nested] 27+ messages in thread