brcmfmac: Add timestamp to scan results report

Message ID 00000000000080af470567f3ef29@google.com
State New
Headers show
Series
  • brcmfmac: Add timestamp to scan results report
Related show

Commit Message

Dmitry Shmidt March 21, 2018, 10:44 p.m.
From 01da9ee11ce6e0973972b2ffd37eef660714f71a Mon Sep 17 00:00:00 2001
From: Dmitry Shmidt <dimitrysh@google.com>

Date: Wed, 21 Mar 2018 15:35:45 -0700
Subject: [PATCH] brcmfmac: Add timestamp to scan results report

Monotonic time from boot is good timestamp indicator
and it can be used by kernel and userspace.

Change-Id: I122374fa79c6dbd7967b2c30553cb04dee032ebc
Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

---
  drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c | 6 +++++-
  1 file changed, 5 insertions(+), 1 deletion(-)

  	u16 notify_capability;
@@ -2885,10 +2887,12 @@ static s32 brcmf_inform_single_bss(struct  
brcmf_cfg80211_info *cfg,
  	brcmf_dbg(CONN, "Beacon interval: %d\n", notify_interval);
  	brcmf_dbg(CONN, "Signal: %d\n", notify_signal);

+	get_monotonic_boottime(&ts);
+	timestamp = (u64)ts.tv_sec * 1000000 + ts.tv_nsec / 1000;
  	bss = cfg80211_inform_bss(wiphy, notify_channel,
  				  CFG80211_BSS_FTYPE_UNKNOWN,
  				  (const u8 *)bi->BSSID,
-				  0, notify_capability,
+				  timestamp, notify_capability,
  				  notify_interval, notify_ie,
  				  notify_ielen, notify_signal,
  				  GFP_KERNEL);
-- 
2.17.0.rc0.231.g781580f067-goog

Comments

Arend van Spriel March 22, 2018, 10:44 a.m. | #1
On 3/21/2018 11:44 PM, dimitrysh@google.com wrote:
>  From 01da9ee11ce6e0973972b2ffd37eef660714f71a Mon Sep 17 00:00:00 2001

> From: Dmitry Shmidt <dimitrysh@google.com>

> Date: Wed, 21 Mar 2018 15:35:45 -0700

> Subject: [PATCH] brcmfmac: Add timestamp to scan results report

>

> Monotonic time from boot is good timestamp indicator

> and it can be used by kernel and userspace.


Thanks, Dmitry

I recall a similar attempt for another driver, but I think the timestamp 
should really be the TSF of the beacon/proberesp received. As you 
already Cc-ed Johannes explicitly maybe he can clarify what is needed 
here as I do not recall the details. ;-)

Regards,
Arend

> Change-Id: I122374fa79c6dbd7967b2c30553cb04dee032ebc

> Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

> ---

>   drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c | 6 +++++-

>   1 file changed, 5 insertions(+), 1 deletion(-)
Johannes Berg March 22, 2018, 11:48 a.m. | #2
On Thu, 2018-03-22 at 11:44 +0100, Arend van Spriel wrote:
> I recall a similar attempt for another driver, but I think the timestamp 

> should really be the TSF of the beacon/proberesp received. As you 

> already Cc-ed Johannes explicitly maybe he can clarify what is needed 

> here as I do not recall the details. ;-)


Yes, this should be "tsf", it's actually labelled explicitly as such. I
know Android used to get this wrong and *even test* it wrongly, but
that's no excuse.

However, should switch this to cfg80211_inform_bss_data() and use the
timestamp set up here to fill the cfg80211_inform_bss::boottime_ns
value. That's also sent to userspace in the proper way for this.

johannes

Patch

diff --git a/drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c  
b/drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c
index 4157c90ad973..76d6ed553b7c 100644
--- a/drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c
+++ b/drivers/net/wireless/broadcom/brcm80211/brcmfmac/cfg80211.c
@@ -2845,6 +2845,8 @@  static s32 brcmf_inform_single_bss(struct  
brcmf_cfg80211_info *cfg,
  	struct cfg80211_bss *bss;
  	struct ieee80211_supported_band *band;
  	struct brcmu_chan ch;
+	struct timespec ts;
+	u64 timestamp;
  	u16 channel;
  	u32 freq;