[PATCH 2/2] efi/runtime-wrappers: disable EFI runtime services after a hang

[Breno Leitao]

Once a runtime call has tripped EFI_RTS_TIMEOUT the firmware is
wedged and any subsequent call would also hang for 120 seconds
before timing out, blocking userspace each time.

Introduce a one-shot efi_rts_dead flag set on timeout, and check it
at the entry of __efi_queue_work() so further calls fail fast with
EFI_DEVICE_ERROR.

Beyond the wall-clock saving, the flag is required for correctness:
without it the next __efi_queue_work() caller after a timeout walks
into INIT_WORK() and init_completion() on the work_struct and
completion that the leaked worker still owns, which can corrupt the
workqueue's bookkeeping for the leaked work and let the next caller
observe the leaked call's status as if it were its own.

The flag is intentionally never cleared: the worker that wedged is
leaked inside firmware, so reusing efi_rts_work is unsafe.

Signed-off-by: Breno Leitao <leitao@xxxxxxxxxx>
---
 drivers/firmware/efi/runtime-wrappers.c | 17 ++++++++++++++++-
 1 file changed, 16 insertions(+), 1 deletion(-)

diff --git a/drivers/firmware/efi/runtime-wrappers.c b/drivers/firmware/efi/runtime-wrappers.c
index 6ce6d094066e..a76ba819855f 100644
--- a/drivers/firmware/efi/runtime-wrappers.c
+++ b/drivers/firmware/efi/runtime-wrappers.c
@@ -126,6 +126,16 @@ struct efi_runtime_work efi_rts_work;
  */
 #define EFI_RTS_TIMEOUT		(120 * HZ)
 
+/*
+ * Set once an EFI runtime service call has hung past EFI_RTS_TIMEOUT.
+ * Subsequent __efi_queue_work() callers fail fast: the wedged worker
+ * is still inside firmware, owns efi_rts_work, and any reuse of the
+ * shared work_struct or completion would corrupt the workqueue's
+ * bookkeeping for the leaked work. The flag is intentionally never
+ * cleared - recovery requires reboot.
+ */
+static bool efi_rts_dead;
+
 /*
  * efi_queue_work:	Queue EFI runtime service call and wait for completion
  * @_rts:		EFI runtime service function identifier
@@ -328,6 +338,9 @@ static void __nocfi efi_call_rts(struct work_struct *work)
 static efi_status_t __efi_queue_work(enum efi_rts_ids id,
 				     union efi_rts_args *args)
 {
+	if (READ_ONCE(efi_rts_dead))
+		return EFI_DEVICE_ERROR;
+
 	efi_rts_work.efi_rts_id = id;
 	efi_rts_work.args = args;
 	efi_rts_work.caller = __builtin_return_address(0);
@@ -353,7 +366,9 @@ static efi_status_t __efi_queue_work(enum efi_rts_ids id,
 
 	if (!wait_for_completion_timeout(&efi_rts_work.efi_rts_comp,
 					 EFI_RTS_TIMEOUT)) {
-		pr_err("EFI runtime service %d wedged in firmware\n", id);
+		WRITE_ONCE(efi_rts_dead, true);
+		pr_err("EFI runtime service %d wedged in firmware; disabling EFI runtime services\n",
+		       id);
 		return EFI_TIMEOUT;
 	}
 

-- 
2.53.0-Meta