[PATCH v4 4/7] arm64/fpsimd: Permit kernel mode NEON with IRQs off

[Ard Biesheuvel <ardb+git@google.com>]

From: Ard Biesheuvel <ardb@kernel.org>

Currently, may_use_simd() will return false when called from a context
where IRQs are disabled. One notable case where this happens is when
calling the ResetSystem() EFI runtime service from the reboot/poweroff
code path. For this case alone, there is a substantial amount of FP/SIMD
support code to handle the corner case where a EFI runtime service is
invoked with IRQs disabled.

The only reason kernel mode SIMD is not allowed when IRQs are disabled
is that re-enabling softirqs in this case produces a noisy diagnostic
when lockdep is enabled. The warning is valid, in the sense that
delivering pending softirqs over the back of the call to
local_bh_enable() is problematic when IRQs are disabled.

While the API lacks a facility to simply mask and unmask softirqs
without triggering their delivery, disabling softirqs is not needed to
begin with when IRQs are disabled, given that softirqs are only every
taken asynchronously over the back of a hard IRQ.

So dis/enable softirq processing conditionally, based on whether IRQs
are enabled, and relax the check in may_use_simd().

Acked-by: Will Deacon <will@kernel.org>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
 arch/arm64/include/asm/simd.h |  2 +-
 arch/arm64/kernel/fpsimd.c    | 25 +++++++++++++++-----
 2 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/arch/arm64/include/asm/simd.h b/arch/arm64/include/asm/simd.h
index 8e86c9e70e48..abd642c92f86 100644
--- a/arch/arm64/include/asm/simd.h
+++ b/arch/arm64/include/asm/simd.h
@@ -29,7 +29,7 @@ static __must_check inline bool may_use_simd(void)
 	 */
 	return !WARN_ON(!system_capabilities_finalized()) &&
 	       system_supports_fpsimd() &&
-	       !in_hardirq() && !irqs_disabled() && !in_nmi();
+	       !in_hardirq() && !in_nmi();
 }
 
 #else /* ! CONFIG_KERNEL_MODE_NEON */
diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c
index d26a02ea2bb9..5af3043fb0ca 100644
--- a/arch/arm64/kernel/fpsimd.c
+++ b/arch/arm64/kernel/fpsimd.c
@@ -225,10 +225,21 @@ static void fpsimd_bind_task_to_cpu(void);
  */
 static void get_cpu_fpsimd_context(void)
 {
-	if (!IS_ENABLED(CONFIG_PREEMPT_RT))
-		local_bh_disable();
-	else
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT)) {
+		/*
+		 * The softirq subsystem lacks a true unmask/mask API, and
+		 * re-enabling softirq processing using local_bh_enable() will
+		 * not only unmask softirqs, it will also result in immediate
+		 * delivery of any pending softirqs.
+		 * This is undesirable when running with IRQs disabled, but in
+		 * that case, there is no need to mask softirqs in the first
+		 * place, so only bother doing so when IRQs are enabled.
+		 */
+		if (!irqs_disabled())
+			local_bh_disable();
+	} else {
 		preempt_disable();
+	}
 }
 
 /*
@@ -240,10 +251,12 @@ static void get_cpu_fpsimd_context(void)
  */
 static void put_cpu_fpsimd_context(void)
 {
-	if (!IS_ENABLED(CONFIG_PREEMPT_RT))
-		local_bh_enable();
-	else
+	if (!IS_ENABLED(CONFIG_PREEMPT_RT)) {
+		if (!irqs_disabled())
+			local_bh_enable();
+	} else {
 		preempt_enable();
+	}
 }
 
 unsigned int task_get_vl(const struct task_struct *task, enum vec_type type)
-- 
2.51.0.534.gc79095c0ca-goog