[PATCH] ath10k: add extra check for frame tracing

[PATCH] ath10k: add extra check for frame tracing

[Michal Kazior]

Frames are logged via tracing in two slices:
header and payload, separately. This is done for
performance reasons when one wants to, e.g.
analyse metadata only of frames only.

If for some reason device delivered a frame buffer
which was sized below what 802.11 header implied
tracing logic would blow doing an invalid memory
accesses.

I've hit this problem when running IBSS on QCA988X
with 999.999.0.636 and tracing at the same time.

Fixes: 5ce8e7fdcc7a ("ath10k: handle ieee80211 header and payload tracing separately")
Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
---
 drivers/net/wireless/ath/ath10k/trace.h | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/drivers/net/wireless/ath/ath10k/trace.h b/drivers/net/wireless/ath/ath10k/trace.h
index 71dfcd96354b..71bdb368813d 100644
--- a/drivers/net/wireless/ath/ath10k/trace.h
+++ b/drivers/net/wireless/ath/ath10k/trace.h
@@ -21,11 +21,16 @@
 #include "core.h"
 
 #if !defined(_TRACE_H_)
-static inline u32 ath10k_frm_hdr_len(const void *buf)
+static inline u32 ath10k_frm_hdr_len(const void *buf, size_t len)
 {
 	const struct ieee80211_hdr *hdr = buf;
 
-	return ieee80211_hdrlen(hdr->frame_control);
+	/* In some rare cases (e.g. fcs error) device reports frame buffer
+	 * shorter than what frame header implies (e.g. len = 0). The buffer
+	 * can still be accessed so do a simple min() to guarantee caller
+	 * doesn't get value greater than len.
+	 */
+	return min_t(u32, len, ieee80211_hdrlen(hdr->frame_control));
 }
 #endif
 
@@ -360,13 +365,13 @@ DECLARE_EVENT_CLASS(ath10k_hdr_event,
 		__string(device, dev_name(ar->dev))
 		__string(driver, dev_driver_string(ar->dev))
 		__field(size_t, len)
-		__dynamic_array(u8, data, ath10k_frm_hdr_len(data))
+		__dynamic_array(u8, data, ath10k_frm_hdr_len(data, len))
 	),
 
 	TP_fast_assign(
 		__assign_str(device, dev_name(ar->dev));
 		__assign_str(driver, dev_driver_string(ar->dev));
-		__entry->len = ath10k_frm_hdr_len(data);
+		__entry->len = ath10k_frm_hdr_len(data, len);
 		memcpy(__get_dynamic_array(data), data, __entry->len);
 	),
 
@@ -387,15 +392,16 @@ DECLARE_EVENT_CLASS(ath10k_payload_event,
 		__string(device, dev_name(ar->dev))
 		__string(driver, dev_driver_string(ar->dev))
 		__field(size_t, len)
-		__dynamic_array(u8, payload, (len - ath10k_frm_hdr_len(data)))
+		__dynamic_array(u8, payload, (len -
+					      ath10k_frm_hdr_len(data, len)))
 	),
 
 	TP_fast_assign(
 		__assign_str(device, dev_name(ar->dev));
 		__assign_str(driver, dev_driver_string(ar->dev));
-		__entry->len = len - ath10k_frm_hdr_len(data);
+		__entry->len = len - ath10k_frm_hdr_len(data, len);
 		memcpy(__get_dynamic_array(payload),
-		       data + ath10k_frm_hdr_len(data), __entry->len);
+		       data + ath10k_frm_hdr_len(data, len), __entry->len);
 	),
 
 	TP_printk(
-- 
2.1.4


_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k

Re: [PATCH] ath10k: add extra check for frame tracing

[Kalle Valo]

Michal Kazior <michal.kazior@tieto.com> writes:

> Frames are logged via tracing in two slices:
> header and payload, separately. This is done for
> performance reasons when one wants to, e.g.
> analyse metadata only of frames only.
>
> If for some reason device delivered a frame buffer
> which was sized below what 802.11 header implied
> tracing logic would blow doing an invalid memory
> accesses.
>
> I've hit this problem when running IBSS on QCA988X
> with 999.999.0.636 and tracing at the same time.
>
> Fixes: 5ce8e7fdcc7a ("ath10k: handle ieee80211 header and payload tracing separately")
> Signed-off-by: Michal Kazior <michal.kazior@tieto.com>

Thanks, applied.

-- 
Kalle Valo

_______________________________________________
ath10k mailing list
ath10k@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/ath10k