[PATCH] wl12xx: fix DMA-API-related warnings

[PATCH] wl12xx: fix DMA-API-related warnings

[Mircea Gherzan]

On the PandaBoard (omap_hsmmc + wl12xx_sdio) with DMA_API_DEBUG:

 WARNING: at lib/dma-debug.c:930 check_for_stack.part.8+0x7c/0xe0()
 omap_hsmmc omap_hsmmc.4: DMA-API: device driver maps memory fromstack

Signed-off-by: Mircea Gherzan <mgherzan@gmail.com>
---
 drivers/net/wireless/wl12xx/boot.c  |   14 +++++++++++---
 drivers/net/wireless/wl12xx/cmd.c   |   25 ++++++++++++++++---------
 drivers/net/wireless/wl12xx/event.c |   18 +++++++++++-------
 3 files changed, 38 insertions(+), 19 deletions(-)

diff --git a/drivers/net/wireless/wl12xx/boot.c b/drivers/net/wireless/wl12xx/boot.c
index 8f9cf5a..89c78d1 100644
--- a/drivers/net/wireless/wl12xx/boot.c
+++ b/drivers/net/wireless/wl12xx/boot.c
@@ -142,14 +142,22 @@ static void wl1271_parse_fw_ver(struct wl1271 *wl)
 
 static void wl1271_boot_fw_version(struct wl1271 *wl)
 {
-	struct wl1271_static_data static_data;
+	struct wl1271_static_data *static_data;
 
-	wl1271_read(wl, wl->cmd_box_addr, &static_data, sizeof(static_data),
+	static_data = kmalloc(sizeof(*static_data), GFP_DMA);
+	if (!static_data) {
+		__WARN();
+		return;
+	}
+
+	wl1271_read(wl, wl->cmd_box_addr, static_data, sizeof(*static_data),
 		    false);
 
-	strncpy(wl->chip.fw_ver_str, static_data.fw_version,
+	strncpy(wl->chip.fw_ver_str, static_data->fw_version,
 		sizeof(wl->chip.fw_ver_str));
 
+	kfree(static_data);
+
 	/* make sure the string is NULL-terminated */
 	wl->chip.fw_ver_str[sizeof(wl->chip.fw_ver_str) - 1] = '\0';
 
diff --git a/drivers/net/wireless/wl12xx/cmd.c b/drivers/net/wireless/wl12xx/cmd.c
index 25990bd..a5c8800 100644
--- a/drivers/net/wireless/wl12xx/cmd.c
+++ b/drivers/net/wireless/wl12xx/cmd.c
@@ -342,8 +342,12 @@ int wl1271_cmd_ext_radio_parms(struct wl1271 *wl)
  */
 static int wl1271_cmd_wait_for_event_or_timeout(struct wl1271 *wl, u32 mask)
 {
-	u32 events_vector, event;
+	u32 *events_vector;
+	u32 event;
 	unsigned long timeout;
+	int ret = 0;
+
+	events_vector = kmalloc(sizeof(*events_vector), GFP_DMA);
 
 	timeout = jiffies + msecs_to_jiffies(WL1271_EVENT_TIMEOUT);
 
@@ -351,21 +355,24 @@ static int wl1271_cmd_wait_for_event_or_timeout(struct wl1271 *wl, u32 mask)
 		if (time_after(jiffies, timeout)) {
 			wl1271_debug(DEBUG_CMD, "timeout waiting for event %d",
 				     (int)mask);
-			return -ETIMEDOUT;
+			ret = -ETIMEDOUT;
+			goto out;
 		}
 
 		msleep(1);
 
 		/* read from both event fields */
-		wl1271_read(wl, wl->mbox_ptr[0], &events_vector,
-			    sizeof(events_vector), false);
-		event = events_vector & mask;
-		wl1271_read(wl, wl->mbox_ptr[1], &events_vector,
-			    sizeof(events_vector), false);
-		event |= events_vector & mask;
+		wl1271_read(wl, wl->mbox_ptr[0], events_vector,
+			    sizeof(*events_vector), false);
+		event = *events_vector & mask;
+		wl1271_read(wl, wl->mbox_ptr[1], events_vector,
+			    sizeof(*events_vector), false);
+		event |= *events_vector & mask;
 	} while (!event);
 
-	return 0;
+out:
+	kfree(events_vector);
+	return ret;
 }
 
 static int wl1271_cmd_wait_for_event(struct wl1271 *wl, u32 mask)
diff --git a/drivers/net/wireless/wl12xx/event.c b/drivers/net/wireless/wl12xx/event.c
index d3280df68..bcd0c93 100644
--- a/drivers/net/wireless/wl12xx/event.c
+++ b/drivers/net/wireless/wl12xx/event.c
@@ -439,25 +439,29 @@ void wl1271_event_mbox_config(struct wl1271 *wl)
 
 int wl1271_event_handle(struct wl1271 *wl, u8 mbox_num)
 {
-	struct event_mailbox mbox;
-	int ret;
+	struct event_mailbox *mbox;
+	int ret = 0;
 
 	wl1271_debug(DEBUG_EVENT, "EVENT on mbox %d", mbox_num);
 
 	if (mbox_num > 1)
 		return -EINVAL;
 
+	/* no GFP_ATOMIC: we're called from the threaded IRQ handler */
+	mbox = kmalloc(sizeof(*mbox), GFP_DMA);
+
 	/* first we read the mbox descriptor */
-	wl1271_read(wl, wl->mbox_ptr[mbox_num], &mbox,
-		    sizeof(struct event_mailbox), false);
+	wl1271_read(wl, wl->mbox_ptr[mbox_num], mbox, sizeof(*mbox), false);
 
 	/* process the descriptor */
-	ret = wl1271_event_process(wl, &mbox);
+	ret = wl1271_event_process(wl, mbox);
 	if (ret < 0)
-		return ret;
+		goto out;
 
 	/* then we let the firmware know it can go on...*/
 	wl1271_write32(wl, ACX_REG_INTERRUPT_TRIG, INTR_TRIG_EVENT_ACK);
 
-	return 0;
+out:
+	kfree(mbox);
+	return ret;
 }
-- 
1.7.9.1

Re: [PATCH] wl12xx: fix DMA-API-related warnings

[Arnd Bergmann]

On Saturday 17 March 2012, Mircea Gherzan wrote:
>  int wl1271_event_handle(struct wl1271 *wl, u8 mbox_num)
>  {
> -       struct event_mailbox mbox;
> -       int ret;
> +       struct event_mailbox *mbox;
> +       int ret = 0;
>  
>         wl1271_debug(DEBUG_EVENT, "EVENT on mbox %d", mbox_num);
>  
>         if (mbox_num > 1)
>                 return -EINVAL;
>  
> +       /* no GFP_ATOMIC: we're called from the threaded IRQ handler */
> +       mbox = kmalloc(sizeof(*mbox), GFP_DMA);
> +
>         /* first we read the mbox descriptor */
> -       wl1271_read(wl, wl->mbox_ptr[mbox_num], &mbox,
> -                   sizeof(struct event_mailbox), false);
> +       wl1271_read(wl, wl->mbox_ptr[mbox_num], mbox, sizeof(*mbox), false);
>  
>         /* process the descriptor */
> -       ret = wl1271_event_process(wl, &mbox);
> +       ret = wl1271_event_process(wl, mbox);
>         if (ret < 0)
> -               return ret;
> +               goto out;
>  
>         /* then we let the firmware know it can go on...*/
>         wl1271_write32(wl, ACX_REG_INTERRUPT_TRIG, INTR_TRIG_EVENT_ACK);
>  
> -       return 0;
> +out:
> +       kfree(mbox);
> +       return ret;
>  }

I think it would be better here to put another field into struct wl1271 to hold
the mailbox. There is no point allocating and freeing the field every time
you get into the interrupt handler.

	Arnd
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html