driverdev-devel.linuxdriverproject.org archive mirror
 help / color / mirror / Atom feed
From: "Jérôme Pouiller" <Jerome.Pouiller@silabs.com>
To: "devel@driverdev.osuosl.org" <devel@driverdev.osuosl.org>,
	"linux-wireless@vger.kernel.org" <linux-wireless@vger.kernel.org>
Cc: "netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	"David S . Miller" <davem@davemloft.net>,
	Kalle Valo <kvalo@codeaurora.org>
Subject: [PATCH 51/55] staging: wfx: workaround bug with "iw scan"
Date: Mon, 16 Dec 2019 17:03:59 +0000	[thread overview]
Message-ID: <20191216170302.29543-52-Jerome.Pouiller@silabs.com> (raw)
In-Reply-To: <20191216170302.29543-1-Jerome.Pouiller@silabs.com>

From: Jérôme Pouiller <jerome.pouiller@silabs.com>

mac80211 specification does not forbid hw_scan() to call
ieee80211_scan_completed(). However, from userspace point of view, not
all applications support this behavior. In particular, the code of iw
contains a big fat warning:

   /*
    * This code has a bug, which requires creating a separate
    * nl80211 socket to fix:
    * It is possible for a NL80211_CMD_NEW_SCAN_RESULTS or
    * NL80211_CMD_SCAN_ABORTED message to be sent by the kernel
    * before (!) we listen to it, because we only start listening
    * after we send our scan request.
    [...]
    * Alas, the kernel doesn't do that (yet).
    */

So, we have to avoid to call ieee80211_scan_completed() from hw_scan()
(it's a kind of unwritten rule).

This patch relocates the hw_scan() process to a work_struct to fix the
problem.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
---
 drivers/staging/wfx/scan.c | 47 ++++++++++++++++++++++++--------------
 drivers/staging/wfx/scan.h |  1 +
 drivers/staging/wfx/sta.c  |  1 +
 drivers/staging/wfx/wfx.h  |  2 ++
 4 files changed, 34 insertions(+), 17 deletions(-)

diff --git a/drivers/staging/wfx/scan.c b/drivers/staging/wfx/scan.c
index b73e61e8da46..540009b72240 100644
--- a/drivers/staging/wfx/scan.c
+++ b/drivers/staging/wfx/scan.c
@@ -71,23 +71,19 @@ static int send_scan_req(struct wfx_vif *wvif,
 	return i - start_idx;
 }
 
-int wfx_hw_scan(struct ieee80211_hw *hw, struct ieee80211_vif *vif,
-		struct ieee80211_scan_request *hw_req)
+/*
+ * It is not really necessary to run scan request asynchronously. However,
+ * there is a bug in "iw scan" when ieee80211_scan_completed() is called before
+ * wfx_hw_scan() return
+ */
+void wfx_hw_scan_work(struct work_struct *work)
 {
-	struct wfx_dev *wdev = hw->priv;
-	struct wfx_vif *wvif = (struct wfx_vif *) vif->drv_priv;
+	struct wfx_vif *wvif = container_of(work, struct wfx_vif, scan_work);
+	struct ieee80211_scan_request *hw_req = wvif->scan_req;
 	int chan_cur, ret;
 
-	WARN_ON(hw_req->req.n_channels > HIF_API_MAX_NB_CHANNELS);
-
-	if (vif->type == NL80211_IFTYPE_AP)
-		return -EOPNOTSUPP;
-
-	if (wvif->state == WFX_STATE_PRE_STA)
-		return -EBUSY;
-
 	mutex_lock(&wvif->scan_lock);
-	mutex_lock(&wdev->conf_mutex);
+	mutex_lock(&wvif->wdev->conf_mutex);
 	update_probe_tmpl(wvif, &hw_req->req);
 	wfx_fwd_probe_req(wvif, true);
 	chan_cur = 0;
@@ -96,18 +92,35 @@ int wfx_hw_scan(struct ieee80211_hw *hw, struct ieee80211_vif *vif,
 		if (ret > 0)
 			chan_cur += ret;
 	} while (ret > 0 && chan_cur < hw_req->req.n_channels);
-	__ieee80211_scan_completed_compat(hw, ret < 0);
-	mutex_unlock(&wdev->conf_mutex);
+	mutex_unlock(&wvif->wdev->conf_mutex);
 	mutex_unlock(&wvif->scan_lock);
+	__ieee80211_scan_completed_compat(wvif->wdev->hw, ret < 0);
 	if (wvif->delayed_unjoin) {
 		wvif->delayed_unjoin = false;
-		wfx_tx_lock(wdev);
+		wfx_tx_lock(wvif->wdev);
 		if (!schedule_work(&wvif->unjoin_work))
-			wfx_tx_unlock(wdev);
+			wfx_tx_unlock(wvif->wdev);
 	} else if (wvif->delayed_link_loss) {
 		wvif->delayed_link_loss = false;
 		wfx_cqm_bssloss_sm(wvif, 1, 0, 0);
 	}
+}
+
+int wfx_hw_scan(struct ieee80211_hw *hw, struct ieee80211_vif *vif,
+		struct ieee80211_scan_request *hw_req)
+{
+	struct wfx_vif *wvif = (struct wfx_vif *)vif->drv_priv;
+
+	WARN_ON(hw_req->req.n_channels > HIF_API_MAX_NB_CHANNELS);
+
+	if (vif->type == NL80211_IFTYPE_AP)
+		return -EOPNOTSUPP;
+
+	if (wvif->state == WFX_STATE_PRE_STA)
+		return -EBUSY;
+
+	wvif->scan_req = hw_req;
+	schedule_work(&wvif->scan_work);
 	return 0;
 }
 
diff --git a/drivers/staging/wfx/scan.h b/drivers/staging/wfx/scan.h
index 03bc6c7e562d..b547f1927d72 100644
--- a/drivers/staging/wfx/scan.h
+++ b/drivers/staging/wfx/scan.h
@@ -15,6 +15,7 @@
 struct wfx_dev;
 struct wfx_vif;
 
+void wfx_hw_scan_work(struct work_struct *work);
 int wfx_hw_scan(struct ieee80211_hw *hw, struct ieee80211_vif *vif,
 		struct ieee80211_scan_request *req);
 void wfx_scan_complete(struct wfx_vif *wvif,
diff --git a/drivers/staging/wfx/sta.c b/drivers/staging/wfx/sta.c
index 16f5db873275..4354bb8081c5 100644
--- a/drivers/staging/wfx/sta.c
+++ b/drivers/staging/wfx/sta.c
@@ -1427,6 +1427,7 @@ int wfx_add_interface(struct ieee80211_hw *hw, struct ieee80211_vif *vif)
 
 	mutex_init(&wvif->scan_lock);
 	init_completion(&wvif->scan_complete);
+	INIT_WORK(&wvif->scan_work, wfx_hw_scan_work);
 
 	mutex_unlock(&wdev->conf_mutex);
 
diff --git a/drivers/staging/wfx/wfx.h b/drivers/staging/wfx/wfx.h
index 3356d0cbf7af..b5f763c3fac7 100644
--- a/drivers/staging/wfx/wfx.h
+++ b/drivers/staging/wfx/wfx.h
@@ -127,7 +127,9 @@ struct wfx_vif {
 
 	/* avoid some operations in parallel with scan */
 	struct mutex		scan_lock;
+	struct work_struct	scan_work;
 	struct completion	scan_complete;
+	struct ieee80211_scan_request *scan_req;
 
 	struct completion	set_pm_mode_complete;
 
-- 
2.20.1
_______________________________________________
devel mailing list
devel@linuxdriverproject.org
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel

  parent reply	other threads:[~2019-12-16 17:40 UTC|newest]

Thread overview: 68+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-12-16 17:03 [PATCH 00/55] Improve wfx driver Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 01/55] staging: wfx: fix the cache of rate policies on interface reset Jérôme Pouiller
2019-12-17 11:52   ` Greg Kroah-Hartman
2019-12-17 14:35     ` Jérôme Pouiller
2019-12-17 14:49       ` Greg Kroah-Hartman
2019-12-16 17:03 ` [PATCH 03/55] staging: wfx: fix counter overflow Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 02/55] staging: wfx: fix case of lack of tx_retry_policies Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 05/55] staging: wfx: firmware does not support more than 32 total retries Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 04/55] staging: wfx: use boolean appropriately Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 07/55] staging: wfx: ensure that retry policy always fallbacks to MCS0 / 1Mbps Jérôme Pouiller
2019-12-16 18:08   ` Felix Fietkau
2019-12-17 11:01     ` Jérôme Pouiller
2019-12-17 11:20       ` Felix Fietkau
2019-12-17 15:15         ` Jérôme Pouiller
2019-12-18 11:25           ` Felix Fietkau
2019-12-16 17:03 ` [PATCH 06/55] staging: wfx: fix rate control handling Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 08/55] staging: wfx: detect race condition in WEP authentication Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 09/55] staging: wfx: fix hif_set_mfp() with big endian hosts Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 10/55] staging: wfx: fix wrong error message Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 11/55] staging: wfx: increase SPI bus frequency limit Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 13/55] staging: wfx: avoid double warning when no more tx policy are available Jérôme Pouiller
2020-01-03  9:21   ` Dan Carpenter
2020-01-03 10:39     ` Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 14/55] staging: wfx: improve error message on unexpected confirmation Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 12/55] staging: wfx: don't print useless error messages Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 15/55] staging: wfx: take advantage of IS_ERR_OR_NULL() Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 16/55] staging: wfx: uniformize naming rule Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 17/55] staging: wfx: use meaningful names for CFG_BYTE_ORDER_* Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 18/55] staging: wfx: remove useless include Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 20/55] staging: wfx: make conditions easier to read Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 19/55] staging: wfx: simplify variable assignment Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 21/55] staging: wfx: ensure that traces never modify arguments Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 22/55] staging: wfx: ensure that received hif messages are never modified Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 23/55] staging: wfx: fix typo in "num_of_ssi_ds" Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 25/55] staging: wfx: fix name of struct hif_req_start_scan_alt Jérôme Pouiller
2020-01-03  9:27   ` Dan Carpenter
2020-01-03 10:30     ` Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 24/55] staging: wfx: fix typo in "num_i_es" Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 26/55] staging: wfx: improve API of hif_req_join->infrastructure_bss_mode Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 27/55] staging: wfx: better naming for hif_req_join->short_preamble Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 28/55] staging: wfx: better naming for hif_mib_set_association_mode->greenfield Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 29/55] staging: wfx: simplify handling of tx_lock in wfx_do_join() Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 31/55] staging: wfx: declare wfx_set_pm() static Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 30/55] staging: wfx: firmware already handle powersave mode during scan Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 33/55] staging: wfx: remove redundant test while calling wfx_update_pm() Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 32/55] staging: wfx: drop useless argument from wfx_set_pm() Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 34/55] staging: wfx: drop unnecessary wvif->powersave_mode Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 35/55] staging: wfx: do not try to save call to hif_set_pm() Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 37/55] staging: wfx: simplify wfx_conf_tx() Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 36/55] staging: wfx: fix pm_mode timeout Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 38/55] staging: wfx: prefer a bitmask instead of an array of boolean Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 39/55] staging: wfx: simplify hif_set_uapsd_info() usage Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 40/55] staging: wfx: simplify hif_set_pm() usage Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 41/55] staging: wfx: drop struct wfx_edca_params Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 43/55] staging: wfx: simplify hif_set_edca_queue_params() usage Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 42/55] staging: wfx: remove unnecessary EDCA initialisation Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 45/55] staging: wfx: device already handle sleep mode during scan Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 44/55] staging: wfx: hif_scan() never fails Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 46/55] staging: wfx: drop useless wfx_scan_complete() Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 47/55] staging: wfx: simplify hif_scan() usage Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 48/55] staging: wfx: introduce update_probe_tmpl() Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 49/55] staging: wfx: simplify hif_set_template_frame() usage Jérôme Pouiller
2019-12-16 17:03 ` [PATCH 50/55] staging: wfx: rewrite wfx_hw_scan() Jérôme Pouiller
2019-12-16 17:03 ` Jérôme Pouiller [this message]
2019-12-16 17:03 ` [PATCH 52/55] staging: wfx: delayed_unjoin cannot happen Jérôme Pouiller
2019-12-16 17:04 ` [PATCH 53/55] staging: wfx: delayed_link_loss " Jérôme Pouiller
2019-12-16 17:04 ` [PATCH 54/55] staging: wfx: implement cancel_hw_scan() Jérôme Pouiller
2019-12-16 17:04 ` [PATCH 55/55] staging: wfx: update TODO Jérôme Pouiller

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20191216170302.29543-52-Jerome.Pouiller@silabs.com \
    --to=jerome.pouiller@silabs.com \
    --cc=davem@davemloft.net \
    --cc=devel@driverdev.osuosl.org \
    --cc=gregkh@linuxfoundation.org \
    --cc=kvalo@codeaurora.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-wireless@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).