2013-12-13 09:48:29

by Michal Kazior

Subject: [PATCH] ath10k: fix band reporting for mgmt CCK frames on 5GHz

Although CCK modulation isn't expected for 11a if
it happened it made ath10k report wrong band
(2GHz) for a mgmt frame that were actually
received on 5Ghz band. Frequency would also
decoded incorrectly too.

In case of 5GHz-only devices this triggered
mac80211 WARN_ON because there was no according
sband pointer to be found.

The patch should fix delivery of such frames by
using different means to acquire band parameter.

Signed-off-by: Michal Kazior <[email protected]>
---

One could argue whether such frames should be even
delievered to mac80211. I decided it's better to
deliver to increase interoperability with silly
devices.

If anyone disagrees I can change the patch to
simply drop such frames in ath10k.

drivers/net/wireless/ath/ath10k/wmi.c | 21 ++++++++++++++++++++-
1 file changed, 20 insertions(+), 1 deletion(-)

diff --git a/drivers/net/wireless/ath/ath10k/wmi.c b/drivers/net/wireless/ath/ath10k/wmi.c
index 9ea333a..3ec6c9a 100644
--- a/drivers/net/wireless/ath/ath10k/wmi.c
+++ b/drivers/net/wireless/ath/ath10k/wmi.c
@@ -875,6 +875,7 @@ static int ath10k_wmi_event_mgmt_rx(struct ath10k *ar, struct sk_buff *skb)
struct wmi_mgmt_rx_event_v2 *ev_v2;
struct wmi_mgmt_rx_hdr_v1 *ev_hdr;
struct ieee80211_rx_status *status = IEEE80211_SKB_RXCB(skb);
+ struct ieee80211_channel *ch;
struct ieee80211_hdr *hdr;
u32 rx_status;
u32 channel;
@@ -927,7 +928,25 @@ static int ath10k_wmi_event_mgmt_rx(struct ath10k *ar, struct sk_buff *skb)
if (rx_status & WMI_RX_STATUS_ERR_MIC)
status->flag |= RX_FLAG_MMIC_ERROR;

- status->band = phy_mode_to_band(phy_mode);
+ /* HW can Rx CCK rates on 5GHz. In that case phy_mode is set to
+ * MODE_11B. This means phy_mode is not a reliable source for the band
+ * of mgmt rx. */
+
+ ch = ar->scan_channel;
+ if (!ch)
+ ch = ar->rx_channel;
+
+ if (ch) {
+ status->band = ch->band;
+
+ if (phy_mode == MODE_11B &&
+ status->band == IEEE80211_BAND_5GHZ)
+ ath10k_dbg(ATH10K_DBG_MGMT, "wmi mgmt rx 11b (CCK) on 5GHz\n");
+ } else {
+ ath10k_warn("using (unreliable) phy_mode to extract band for mgmt rx\n");
+ status->band = phy_mode_to_band(phy_mode);
+ }
+
status->freq = ieee80211_channel_to_frequency(channel, status->band);
status->signal = snr + ATH10K_DEFAULT_NOISE_FLOOR;
status->rate_idx = get_rate_idx(rate, status->band);
--
1.8.4.rc3

2013-12-16 13:19:18

by Kalle Valo

Subject: Re: [PATCH] ath10k: fix band reporting for mgmt CCK frames on 5GHz

Michal Kazior <[email protected]> writes:

> Although CCK modulation isn't expected for 11a if
> it happened it made ath10k report wrong band
> (2GHz) for a mgmt frame that were actually
> received on 5Ghz band. Frequency would also
> decoded incorrectly too.
>
> In case of 5GHz-only devices this triggered
> mac80211 WARN_ON because there was no according
> sband pointer to be found.
>
> The patch should fix delivery of such frames by
> using different means to acquire band parameter.
>
> Signed-off-by: Michal Kazior <[email protected]>

Thanks, applied.

--
Kalle Valo

2013-12-16 07:29:42

by Alex Hacker

Subject: Re: [PATCH] ath10k: fix band reporting for mgmt CCK frames on 5GHz

Hello,
If QCA98xx can receive CCK on 5GHz band (like all ath9k supported chips), I
want to ask: Is it should do that? I think no... It looks like FW bug.

Regards,
Alex.