[PATCH v2 5/7] arm64/efi: Use a semaphore to protect the EFI stack and FP/SIMD state

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

From: Ard Biesheuvel <ardb@kernel.org>

Replace the spinlock in the arm64 glue code with a semaphore, so that
the CPU can preempted while running the EFI runtime service.

Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
---
 arch/arm64/kernel/efi.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

diff --git a/arch/arm64/kernel/efi.c b/arch/arm64/kernel/efi.c
index 9b03f3d77a25..8b999c07c7d1 100644
--- a/arch/arm64/kernel/efi.c
+++ b/arch/arm64/kernel/efi.c
@@ -165,12 +165,19 @@ asmlinkage efi_status_t efi_handle_corrupted_x18(efi_status_t s, const char *f)
 	return s;
 }
 
-static DEFINE_RAW_SPINLOCK(efi_rt_lock);
+static DEFINE_SEMAPHORE(efi_rt_lock, 1);
 
 bool arch_efi_call_virt_setup(void)
 {
+	/*
+	 * This might be called from a non-sleepable context so try to take the
+	 * lock but don't block on it. This should never occur in practice, as
+	 * all EFI runtime calls are serialized under the efi_runtime_lock.
+	 */
+	if (WARN_ON(down_trylock(&efi_rt_lock)))
+		return false;
+
 	efi_virtmap_load();
-	raw_spin_lock(&efi_rt_lock);
 	__efi_fpsimd_begin();
 	return true;
 }
@@ -178,8 +185,8 @@ bool arch_efi_call_virt_setup(void)
 void arch_efi_call_virt_teardown(void)
 {
 	__efi_fpsimd_end();
-	raw_spin_unlock(&efi_rt_lock);
 	efi_virtmap_unload();
+	up(&efi_rt_lock);
 }
 
 asmlinkage u64 *efi_rt_stack_top __ro_after_init;
-- 
2.51.0.355.g5224444f11-goog