[Skiboot] [RFC PATCH 3/6] core/flash: Don't hold flash_lock for the entirety of an opal_flash_op()

[Stewart Smith]

From: Cyril Bur <cyril.bur at au1.ibm.com>

It doesn't make sense to hold the lock to the flash for an entire flash
op. The flash_lock provides mutual exclusion to the flashes structure
and each flash element has a busy boolean which ensures that mutual
exclusion on access of the flash.

Signed-off-by: Cyril Bur <cyril.bur at au1.ibm.com>
Reviewed-By: Alistair Popple <alistair at popple.id.au>
Signed-off-by: Stewart Smith <stewart at linux.ibm.com>
---
 core/flash.c | 19 ++++++-------------
 1 file changed, 6 insertions(+), 13 deletions(-)

diff --git a/core/flash.c b/core/flash.c
index f468b4263c79..69d157afc16d 100644
--- a/core/flash.c
+++ b/core/flash.c
@@ -352,23 +352,16 @@ static int64_t opal_flash_op(enum flash_op op, uint64_t id, uint64_t offset,
 	struct flash *flash = NULL;
 	int rc;
 
-	if (!try_lock(&flash_lock))
-		return OPAL_BUSY;
-
 	list_for_each(&flashes, flash, list)
 		if (flash->id == id)
 			break;
 
-	if (flash->id != id) {
+	if (flash->id != id)
 		/* Couldn't find the flash */
-		rc = OPAL_PARAMETER;
-		goto err;
-	}
+		return OPAL_PARAMETER;
 
-	if (flash->busy) {
-		rc = OPAL_BUSY;
-		goto err;
-	}
+	if (!flash_reserve(flash))
+		return OPAL_BUSY;
 
 	if (size > flash->size || offset >= flash->size
 			|| offset + size > flash->size) {
@@ -404,13 +397,13 @@ static int64_t opal_flash_op(enum flash_op op, uint64_t id, uint64_t offset,
 		goto err;
 	}
 
-	unlock(&flash_lock);
+	flash_release(flash);
 
 	opal_queue_msg(OPAL_MSG_ASYNC_COMP, NULL, NULL, token, rc);
 	return OPAL_ASYNC_COMPLETION;
 
 err:
-	unlock(&flash_lock);
+	flash_release(flash);
 	return rc;
 }
 
-- 
2.20.1