From: "Luis R. Rodriguez" <lrodriguez@atheros.com>
To: <johannes@sipsolutions.net>, <johannes@sipsolutions.net>,
<linville@tuxdriver.com>
Cc: "Luis R. Rodriguez" <lrodriguez@atheros.com>,
<linux-wireless@vger.kernel.org>
Subject: [PATCH 6/6] cfg80211: move regulatory hints to workqueue
Date: Thu, 12 Feb 2009 21:36:03 -0800 [thread overview]
Message-ID: <1234503363-11014-7-git-send-email-lrodriguez@atheros.com> (raw)
In-Reply-To: <1234503363-11014-6-git-send-email-lrodriguez@atheros.com>
Driver and userspace regulatory hints are now processed in
a workqueue. This should fix some lockdep warnings which complain
about possible circular locking dependencies such as:
1:
=======================================================
[ INFO: possible circular locking dependency detected ]
2.6.29-rc4-wl-13559-g0416c58-dirty #40
-------------------------------------------------------
iw/4598 is trying to acquire lock:
(rtnl_mutex){--..}, at: [<ffffffff805013e2>] rtnl_lock+0x12/0x20
but task is already holding lock:
(&drv->mtx){--..}, at: [<ffffffffa0352991>] cfg80211_get_dev_from_ifindex+0x61/0xa0 [cfg80211]
2:
=======================================================
[ INFO: possible circular locking dependency detected ]
2.6.29-rc4-wl #7
-------------------------------------------------------
wpa_supplicant/3333 is trying to acquire lock:
(&ifsta->work){--..}, at: [<ffffffff8025555d>] __cancel_work_timer+0x5d/0x1f0
but task is already holding lock:
(rtnl_mutex){--..}, at: [<ffffffff804997a6>] devinet_ioctl+0x136/0x7a0
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
---
drivers/net/wireless/ath9k/main.c | 8 ++-
drivers/net/wireless/zd1211rw/zd_mac.c | 6 +-
include/net/cfg80211.h | 2 +
include/net/wireless.h | 9 ++-
net/wireless/nl80211.c | 10 +--
net/wireless/reg.c | 140 +++++++++++++++++++++++++++++---
net/wireless/reg.h | 2 +
7 files changed, 153 insertions(+), 24 deletions(-)
diff --git a/drivers/net/wireless/ath9k/main.c b/drivers/net/wireless/ath9k/main.c
index fc3460f..f51e7c8 100644
--- a/drivers/net/wireless/ath9k/main.c
+++ b/drivers/net/wireless/ath9k/main.c
@@ -1656,8 +1656,12 @@ int ath_attach(u16 devid, struct ath_softc *sc)
error = ieee80211_register_hw(hw);
- if (!ath9k_is_world_regd(sc->sc_ah))
- regulatory_hint(hw->wiphy, sc->sc_ah->regulatory.alpha2);
+ if (!ath9k_is_world_regd(sc->sc_ah)) {
+ error = regulatory_hint(hw->wiphy,
+ sc->sc_ah->regulatory.alpha2);
+ if (error)
+ goto detach;
+ }
/* Initialize LED control */
ath_init_leds(sc);
diff --git a/drivers/net/wireless/zd1211rw/zd_mac.c b/drivers/net/wireless/zd1211rw/zd_mac.c
index 7579af2..da9214e 100644
--- a/drivers/net/wireless/zd1211rw/zd_mac.c
+++ b/drivers/net/wireless/zd1211rw/zd_mac.c
@@ -170,10 +170,10 @@ int zd_mac_init_hw(struct ieee80211_hw *hw)
goto disable_int;
r = zd_reg2alpha2(mac->regdomain, alpha2);
- if (!r)
- regulatory_hint(hw->wiphy, alpha2);
+ if (r)
+ goto disable_int;
- r = 0;
+ r = regulatory_hint(hw->wiphy, alpha2);
disable_int:
zd_chip_disable_int(chip);
out:
diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h
index 2a83510..ec29843 100644
--- a/include/net/cfg80211.h
+++ b/include/net/cfg80211.h
@@ -396,6 +396,7 @@ enum environment_cap {
* country IE
* @country_ie_env: lets us know if the AP is telling us we are outdoor,
* indoor, or if it doesn't matter
+ * @list: used to insert into a linked list
*/
struct regulatory_request {
int wiphy_idx;
@@ -404,6 +405,7 @@ struct regulatory_request {
bool intersect;
u32 country_ie_checksum;
enum environment_cap country_ie_env;
+ struct list_head list;
};
struct ieee80211_freq_range {
diff --git a/include/net/wireless.h b/include/net/wireless.h
index 1c6285e..afeade9 100644
--- a/include/net/wireless.h
+++ b/include/net/wireless.h
@@ -398,8 +398,15 @@ ieee80211_get_response_rate(struct ieee80211_supported_band *sband,
* domain should be in or by providing a completely build regulatory domain.
* If the driver provides an ISO/IEC 3166 alpha2 userspace will be queried
* for a regulatory domain structure for the respective country.
+ *
+ * The wiphy must have been registered to cfg80211 prior to this call.
+ * For cfg80211 drivers this means you must first use wiphy_register(),
+ * for mac80211 drivers you must first use ieee80211_register_hw().
+ *
+ * Drivers should check the return value, its possible you can get
+ * an -ENOMEM.
*/
-extern void regulatory_hint(struct wiphy *wiphy, const char *alpha2);
+extern int regulatory_hint(struct wiphy *wiphy, const char *alpha2);
/**
* regulatory_hint_11d - hints a country IE as a regulatory domain
diff --git a/net/wireless/nl80211.c b/net/wireless/nl80211.c
index 843d26d..fb1c99e 100644
--- a/net/wireless/nl80211.c
+++ b/net/wireless/nl80211.c
@@ -1911,14 +1911,8 @@ static int nl80211_req_set_reg(struct sk_buff *skb, struct genl_info *info)
if (is_world_regdom(data))
return -EINVAL;
#endif
- mutex_lock(&cfg80211_mutex);
- r = __regulatory_hint(NULL, REGDOM_SET_BY_USER, data, 0, ENVIRON_ANY);
- mutex_unlock(&cfg80211_mutex);
- /* This means the regulatory domain was already set, however
- * we don't want to confuse userspace with a "successful error"
- * message so lets just treat it as a success */
- if (r == -EALREADY)
- r = 0;
+ r = regulatory_hint_user(data);
+
return r;
}
diff --git a/net/wireless/reg.c b/net/wireless/reg.c
index 316b2e4..78ff554 100644
--- a/net/wireless/reg.c
+++ b/net/wireless/reg.c
@@ -64,6 +64,18 @@ const struct ieee80211_regdomain *cfg80211_regdomain;
* what it thinks should apply for the same country */
static const struct ieee80211_regdomain *country_ie_regdomain;
+static LIST_HEAD(reg_requests_list);
+static DEFINE_MUTEX(reg_mutex);
+
+static void reg_process_pending_hints(void);
+
+static void reg_todo(struct work_struct *work)
+{
+ reg_process_pending_hints();
+}
+
+static DECLARE_WORK(reg_work, reg_todo);
+
/* We keep a static world regulatory domain in case of the absence of CRDA */
static const struct ieee80211_regdomain world_regdom = {
.n_reg_rules = 1,
@@ -829,7 +841,7 @@ static void handle_channel(struct wiphy *wiphy, enum ieee80211_band band,
const struct ieee80211_power_rule *power_rule = NULL;
struct ieee80211_supported_band *sband;
struct ieee80211_channel *chan;
- struct wiphy *request_wiphy;
+ struct wiphy *request_wiphy = NULL;
assert_cfg80211_lock();
@@ -1206,21 +1218,66 @@ new_request:
return call_crda(alpha2);
}
-void regulatory_hint(struct wiphy *wiphy, const char *alpha2)
+/* This currently queues driver and user hints */
+static void queue_regulatory_request(struct regulatory_request *request)
{
- int r;
+ mutex_lock(®_mutex);
+ list_add_tail(&request->list, ®_requests_list);
+ mutex_unlock(®_mutex);
+
+ schedule_work(®_work);
+}
+
+/* Driver hints */
+int regulatory_hint(struct wiphy *wiphy, const char *alpha2)
+{
+ struct regulatory_request *request;
+
BUG_ON(!alpha2);
+ BUG_ON(!wiphy);
- mutex_lock(&cfg80211_mutex);
- r = __regulatory_hint(wiphy, REGDOM_SET_BY_DRIVER,
- alpha2, 0, ENVIRON_ANY);
- /* This is required so that the orig_* parameters are saved */
- if (r == -EALREADY && wiphy->strict_regulatory)
- wiphy_update_regulatory(wiphy, REGDOM_SET_BY_DRIVER);
- mutex_unlock(&cfg80211_mutex);
+ request = kzalloc(sizeof(struct regulatory_request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+
+ request->wiphy_idx = wiphy_idx(wiphy);
+
+ /* Must have registered wiphy first */
+ BUG_ON(!wiphy_idx_valid(request->wiphy_idx));
+
+ request->alpha2[0] = alpha2[0];
+ request->alpha2[1] = alpha2[1];
+ request->initiator = REGDOM_SET_BY_DRIVER;
+ request->country_ie_env = ENVIRON_ANY;
+
+ queue_regulatory_request(request);
+
+ return 0;
}
EXPORT_SYMBOL(regulatory_hint);
+/* User hints */
+int regulatory_hint_user(const char *alpha2)
+{
+ struct regulatory_request *request;
+
+ BUG_ON(!alpha2);
+
+ request = kzalloc(sizeof(struct regulatory_request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+
+ request->wiphy_idx = WIPHY_IDX_STALE;
+ request->alpha2[0] = alpha2[0];
+ request->alpha2[1] = alpha2[1];
+ request->initiator = REGDOM_SET_BY_USER,
+ request->country_ie_env = ENVIRON_ANY;
+
+ queue_regulatory_request(request);
+
+ return 0;
+}
+
static bool reg_same_country_ie_hint(struct wiphy *wiphy,
u32 country_ie_checksum)
{
@@ -1620,6 +1677,58 @@ int set_regdom(const struct ieee80211_regdomain *rd)
return r;
}
+/* This currently only processes user and driver regulatory hints */
+static int reg_process_hint(struct regulatory_request *reg_request)
+{
+ int r = 0;
+ struct wiphy *wiphy = NULL;
+
+ BUG_ON(!reg_request->alpha2);
+
+ mutex_lock(&cfg80211_mutex);
+
+ if (wiphy_idx_valid(reg_request->wiphy_idx))
+ wiphy = wiphy_idx_to_wiphy(reg_request->wiphy_idx);
+
+ if (reg_request->initiator == REGDOM_SET_BY_DRIVER &&
+ !wiphy) {
+ r = -ENODEV;
+ goto out;
+ }
+
+ r = __regulatory_hint(wiphy, reg_request->initiator,
+ reg_request->alpha2, 0, reg_request->country_ie_env);
+ /* This is required so that the orig_* parameters are saved */
+ if (r == -EALREADY && wiphy->strict_regulatory)
+ wiphy_update_regulatory(wiphy, reg_request->initiator);
+out:
+ mutex_unlock(&cfg80211_mutex);
+
+ if (r == -EALREADY)
+ r = 0;
+
+ return r;
+}
+
+static void reg_process_pending_hints(void)
+ {
+ struct regulatory_request *reg_request, *tmp;
+ int r;
+
+ mutex_lock(®_mutex);
+ list_for_each_entry_safe(reg_request, tmp, ®_requests_list, list) {
+ r = reg_process_hint(reg_request);
+ if (r)
+ printk(KERN_ERR "cfg80211: wiphy_idx %d sent a "
+ "regulatory hint but now has gone fishing, "
+ "ignoring request\n", reg_request->wiphy_idx);
+ list_del(®_request->list);
+ kfree(reg_request);
+ }
+ mutex_unlock(®_mutex);
+}
+
+
/* Caller must hold cfg80211_mutex */
void reg_device_remove(struct wiphy *wiphy)
{
@@ -1673,6 +1782,11 @@ int regulatory_init(void)
void regulatory_exit(void)
{
+ struct regulatory_request *reg_request, *tmp;
+
+ cancel_work_sync(®_work);
+
+ mutex_lock(®_mutex);
mutex_lock(&cfg80211_mutex);
reset_regdomains();
@@ -1684,5 +1798,11 @@ void regulatory_exit(void)
platform_device_unregister(reg_pdev);
+ list_for_each_entry_safe(reg_request, tmp, ®_requests_list, list) {
+ list_del(®_request->list);
+ kfree(reg_request);
+ }
+
mutex_unlock(&cfg80211_mutex);
+ mutex_unlock(®_mutex);
}
diff --git a/net/wireless/reg.h b/net/wireless/reg.h
index fe8c83f..4730def 100644
--- a/net/wireless/reg.h
+++ b/net/wireless/reg.h
@@ -6,6 +6,8 @@ extern const struct ieee80211_regdomain *cfg80211_regdomain;
bool is_world_regdom(const char *alpha2);
bool reg_is_valid_request(const char *alpha2);
+int regulatory_hint_user(const char *alpha2);
+
void reg_device_remove(struct wiphy *wiphy);
int regulatory_init(void);
--
1.6.1.2.253.ga34a
next prev parent reply other threads:[~2009-02-13 5:36 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-02-13 5:35 [PATCH 0/6] cfg80211: adds a workqueue for regulatory hints Luis R. Rodriguez
2009-02-13 5:35 ` [PATCH 1/6] cfg80211: rename cfg80211_registered_device's idx to wiphy_idx Luis R. Rodriguez
2009-02-13 5:35 ` [PATCH 2/6] cfg80211: add wiphy_idx_valid to check for wiphy_idx sanity Luis R. Rodriguez
2009-02-13 5:36 ` [PATCH 3/6] cfg80211: rename cfg80211_drv_mutex to cfg80211_mutex Luis R. Rodriguez
2009-02-13 5:36 ` [PATCH 4/6] cfg80211: add assert_cfg80211_lock() to ensure proper protection Luis R. Rodriguez
2009-02-13 5:36 ` [PATCH 5/6] cfg80211: make regulatory_request use wiphy_idx instead of wiphy Luis R. Rodriguez
2009-02-13 5:36 ` Luis R. Rodriguez [this message]
2009-02-13 6:56 ` [PATCH 6/6] cfg80211: move regulatory hints to workqueue Johannes Berg
2009-02-13 7:09 ` Luis R. Rodriguez
2009-02-13 7:14 ` Luis R. Rodriguez
2009-02-13 7:14 ` Luis R. Rodriguez
2009-02-13 7:17 ` Johannes Berg
2009-02-13 7:35 ` Luis R. Rodriguez
2009-02-13 7:48 ` Johannes Berg
2009-02-13 7:57 ` Luis R. Rodriguez
2009-02-13 8:08 ` Johannes Berg
2009-02-13 8:19 ` Luis R. Rodriguez
2009-02-13 11:04 ` [PATCH 5/6] cfg80211: make regulatory_request use wiphy_idx instead of wiphy Johannes Berg
2009-02-13 18:21 ` Luis R. Rodriguez
2009-02-13 21:24 ` Johannes Berg
2009-02-13 21:40 ` Luis R. Rodriguez
2009-02-13 6:54 ` [PATCH 4/6] cfg80211: add assert_cfg80211_lock() to ensure proper protection Johannes Berg
2009-02-13 7:08 ` Luis R. Rodriguez
2009-02-13 6:53 ` [PATCH 2/6] cfg80211: add wiphy_idx_valid to check for wiphy_idx sanity Johannes Berg
2009-02-13 7:08 ` Luis R. Rodriguez
2009-02-13 7:18 ` Johannes Berg
2009-02-13 9:13 ` [PATCH 0/6] cfg80211: adds a workqueue for regulatory hints Johannes Berg
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=1234503363-11014-7-git-send-email-lrodriguez@atheros.com \
--to=lrodriguez@atheros.com \
--cc=johannes@sipsolutions.net \
--cc=linux-wireless@vger.kernel.org \
--cc=linville@tuxdriver.com \
/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