[PATCHv4] arm64/cpufeature: don't use mutex in bringup path

From: Mark Rutland
Date: Fri May 12 2017 - 13:50:59 EST


Currently, cpus_set_cap() calls static_branch_enable_cpuslocked(), which
must take the jump_label mutex.

We call cpus_set_cap() in the secondary bringup path, from the idle
thread where interrupts are disabled. Taking a mutex in this path "is a
NONO" regardless of whether it's contended, and something we must avoid.
Additionally, the secondary CPU doesn't hold the percpu rwsem (as this
is held by the primary CPU), so this triggers a lockdep splat.

This patch fixes both issues. The poking of static keys is deferred
until enable_cpu_capabilities(), which runs in a suitable context on the
boot CPU. To account for the static keys being set later,
cpus_have_const_cap() is updated to use another static key to check
whether the const cap keys have been initialised, falling back to the
caps bitmap until this is the case.

This means that users of cpus_have_const_cap() gain should only gain a
single additional NOP in the fast path once the const caps are
initialised, but should always see the current cap value.

The hyp code should never dereference the caps array, since the caps are
initialized before we run the module initcall to initialise hyp. A check
is added to the hyp init code to document this requirement.

This rework means that we can remove the *_cpuslocked() helpers added in
commit d54bb72551b999dd ("arm64/cpufeature: Use
static_branch_enable_cpuslocked()").

Signed-off-by: Mark Rutland <mark.rutland@xxxxxxx>
Reviewed-by: Marc Zyniger <marc.zyngier@xxxxxxx>
Reviewed-by: Suzuki Poulose <suzuki.poulose@xxxxxxx>
Acked-by: Will Deacon <will.deacon@xxxxxxx>
Cc: Catalin Marinas <catalin.marinas@xxxxxxx>
Cc: Christoffer Dall <christoffer.dall@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Sebastian Sewior <bigeasy@xxxxxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
---
arch/arm64/include/asm/cpufeature.h | 13 ++++++++++---
arch/arm64/include/asm/kvm_host.h | 8 ++++++--
arch/arm64/kernel/cpu_errata.c | 9 +--------
arch/arm64/kernel/cpufeature.c | 25 ++++++++++++++++++++++---
4 files changed, 39 insertions(+), 16 deletions(-)

Thomas, are you happy to pick this up?

Will's happy for this to go via tip. I've checked that this builds and boots
atop of the tip smp/hotplug branch.

Thanks,
Mark.

Since v1 [1]:
* Kill redundant update_cpu_errata_workarounds() prototype
* Introduce arm64_const_caps_ready
Since v2 [2]:
* Add hyp init check
* Clean up commit message
* Drop fixes tag
Since v3 [3]:
* Fix typos
* Accumulate tags

[1] http://lists.infradead.org/pipermail/linux-arm-kernel/2017-May/505731.html
[2] http://lists.infradead.org/pipermail/linux-arm-kernel/2017-May/505763.html
[3] http://lists.infradead.org/pipermail/linux-arm-kernel/2017-May/505887.html

diff --git a/arch/arm64/include/asm/cpufeature.h b/arch/arm64/include/asm/cpufeature.h
index a8d1b5b..aba574c 100644
--- a/arch/arm64/include/asm/cpufeature.h
+++ b/arch/arm64/include/asm/cpufeature.h
@@ -115,6 +115,7 @@ struct arm64_cpu_capabilities {

extern DECLARE_BITMAP(cpu_hwcaps, ARM64_NCAPS);
extern struct static_key_false cpu_hwcap_keys[ARM64_NCAPS];
+extern struct static_key_false arm64_const_caps_ready;

bool this_cpu_has_cap(unsigned int cap);

@@ -124,7 +125,7 @@ static inline bool cpu_have_feature(unsigned int num)
}

/* System capability check for constant caps */
-static inline bool cpus_have_const_cap(int num)
+static inline bool __cpus_have_const_cap(int num)
{
if (num >= ARM64_NCAPS)
return false;
@@ -138,6 +139,14 @@ static inline bool cpus_have_cap(unsigned int num)
return test_bit(num, cpu_hwcaps);
}

+static inline bool cpus_have_const_cap(int num)
+{
+ if (static_branch_likely(&arm64_const_caps_ready))
+ return __cpus_have_const_cap(num);
+ else
+ return cpus_have_cap(num);
+}
+
static inline void cpus_set_cap(unsigned int num)
{
if (num >= ARM64_NCAPS) {
@@ -145,7 +154,6 @@ static inline void cpus_set_cap(unsigned int num)
num, ARM64_NCAPS);
} else {
__set_bit(num, cpu_hwcaps);
- static_branch_enable_cpuslocked(&cpu_hwcap_keys[num]);
}
}

@@ -223,7 +231,6 @@ void update_cpu_capabilities(const struct arm64_cpu_capabilities *caps,
void check_local_cpu_capabilities(void);

void update_cpu_errata_workarounds(void);
-void update_cpu_errata_workarounds_cpuslocked(void);
void __init enable_errata_workarounds(void);
void verify_local_cpu_errata_workarounds(void);

diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index e7705e7..9973f0a 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -24,6 +24,7 @@

#include <linux/types.h>
#include <linux/kvm_types.h>
+#include <asm/cpufeature.h>
#include <asm/kvm.h>
#include <asm/kvm_asm.h>
#include <asm/kvm_mmio.h>
@@ -356,9 +357,12 @@ static inline void __cpu_init_hyp_mode(phys_addr_t pgd_ptr,
unsigned long vector_ptr)
{
/*
- * Call initialization code, and switch to the full blown
- * HYP code.
+ * Call initialization code, and switch to the full blown HYP code.
+ * If the cpucaps haven't been finalized yet, something has gone very
+ * wrong, and hyp will crash and burn when it uses any
+ * cpus_have_const_cap() wrapper.
*/
+ BUG_ON(!static_branch_likely(&arm64_const_caps_ready));
__kvm_call_hyp((void *)pgd_ptr, hyp_stack_ptr, vector_ptr);
}

diff --git a/arch/arm64/kernel/cpu_errata.c b/arch/arm64/kernel/cpu_errata.c
index e650cf8..f6cc67e 100644
--- a/arch/arm64/kernel/cpu_errata.c
+++ b/arch/arm64/kernel/cpu_errata.c
@@ -175,16 +175,9 @@ void verify_local_cpu_errata_workarounds(void)
}
}

-void update_cpu_errata_workarounds_cpuslocked(void)
-{
- update_cpu_capabilities(arm64_errata, "enabling workaround for");
-}
-
void update_cpu_errata_workarounds(void)
{
- get_online_cpus();
- update_cpu_errata_workarounds_cpuslocked();
- put_online_cpus();
+ update_cpu_capabilities(arm64_errata, "enabling workaround for");
}

void __init enable_errata_workarounds(void)
diff --git a/arch/arm64/kernel/cpufeature.c b/arch/arm64/kernel/cpufeature.c
index 113c06d..ebfd868 100644
--- a/arch/arm64/kernel/cpufeature.c
+++ b/arch/arm64/kernel/cpufeature.c
@@ -976,8 +976,16 @@ void update_cpu_capabilities(const struct arm64_cpu_capabilities *caps,
*/
void __init enable_cpu_capabilities(const struct arm64_cpu_capabilities *caps)
{
- for (; caps->matches; caps++)
- if (caps->enable && cpus_have_cap(caps->capability))
+ for (; caps->matches; caps++) {
+ unsigned int num = caps->capability;
+
+ if (!cpus_have_cap(num))
+ continue;
+
+ /* Ensure cpus_have_const_cap(num) works */
+ static_branch_enable(&cpu_hwcap_keys[num]);
+
+ if (caps->enable) {
/*
* Use stop_machine() as it schedules the work allowing
* us to modify PSTATE, instead of on_each_cpu() which
@@ -985,6 +993,8 @@ void __init enable_cpu_capabilities(const struct arm64_cpu_capabilities *caps)
* we return.
*/
stop_machine(caps->enable, NULL, cpu_online_mask);
+ }
+ }
}

/*
@@ -1076,7 +1086,7 @@ void check_local_cpu_capabilities(void)
* advertised capabilities.
*/
if (!sys_caps_initialised)
- update_cpu_errata_workarounds_cpuslocked();
+ update_cpu_errata_workarounds();
else
verify_local_cpu_capabilities();
}
@@ -1089,6 +1099,14 @@ static void __init setup_feature_capabilities(void)
enable_cpu_capabilities(arm64_features);
}

+DEFINE_STATIC_KEY_FALSE(arm64_const_caps_ready);
+EXPORT_SYMBOL(arm64_const_caps_ready);
+
+static void __init mark_const_caps_ready(void)
+{
+ static_branch_enable(&arm64_const_caps_ready);
+}
+
/*
* Check if the current CPU has a given feature capability.
* Should be called from non-preemptible context.
@@ -1115,6 +1133,7 @@ void __init setup_cpu_features(void)
/* Set the CPU feature capabilies */
setup_feature_capabilities();
enable_errata_workarounds();
+ mark_const_caps_ready();
setup_elf_hwcaps(arm64_elf_hwcaps);

if (system_supports_32bit_el0())
--
1.9.1