Skip to contentSkip to conversation

wifi: brcmfmac: cyw: clean up PMKID and cookie code

PATCH In review Bogdan Nicolae 3 participants 5 messages Jul 9 – Jul 14 Broadcom BRCM80211 IEEE802.11 wireless drivers

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 wifi: brcmfmac: cyw: clean up PMKID and cookie code

Avoid setting packet_id to cookie, which is always 0. Instead, use an increasing atomic counter. Avoids mismatches of completion events later in brcmf_notify_mgmt_tx_status, where packet_id != vif->mgmt_tx_id is checked.

Also, zero out auth_status on initialization. Otherwise, garbage will leak from the stack to the firmware (when bssid is less than 32 bytes and/or when params->pmkid is set). Then, pass the params->pmkid to the firmware (without it, the firmware caches a garbage PMKID on successful authentication and denies a subsequent association request that includes the PMKID).

---
 .../net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c   | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +6 -2
index ce09d44fa..cca53ff19 100644
@@ -23,6 +23,8 @@
 #define MGMT_AUTH_FRAME_DWELL_TIME	4000
 #define MGMT_AUTH_FRAME_WAIT_TIME	(MGMT_AUTH_FRAME_DWELL_TIME + 100)
 
+static atomic_t brcmf_cyw_mgmt_tx_id = ATOMIC_INIT(0);
+
 static int brcmf_cyw_set_sae_pwd(struct brcmf_if *ifp,
 				 struct cfg80211_crypto_settings *crypto)
 {
@@ -155,7 +157,7 @@ int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev,
 
 	memcpy(&mf_params->da[0], &mgmt->da[0], ETH_ALEN);
 	memcpy(&mf_params->bssid[0], &mgmt->bssid[0], ETH_ALEN);
-	mf_params->packet_id = cpu_to_le32(*cookie);
+	mf_params->packet_id = cpu_to_le32(atomic_inc_return(&brcmf_cyw_mgmt_tx_id));
 	memcpy(mf_params->data, &buf[DOT11_MGMT_HDR_LEN],
 	       le16_to_cpu(mf_params->len));
 
@@ -200,7 +202,7 @@ brcmf_cyw_external_auth(struct wiphy *wiphy, struct net_device *dev,
 {
 	struct brcmf_if *ifp;
 	struct brcmf_pub *drvr;
-	struct brcmf_auth_req_status_le auth_status;
+	struct brcmf_auth_req_status_le auth_status = {};
 	int ret = 0;
 
 	brcmf_dbg(TRACE, "Enter\n");
@@ -208,6 +210,8 @@ brcmf_cyw_external_auth(struct wiphy *wiphy, struct net_device *dev,
 	ifp = netdev_priv(dev);
 	drvr = ifp->drvr;
 	if (params->status == WLAN_STATUS_SUCCESS) {
+		if (params->pmkid)
+			memcpy(auth_status.pmkid, params->pmkid, WLAN_PMKID_LEN);
 		auth_status.flags = cpu_to_le16(BRCMF_EXTAUTH_SUCCESS);
 	} else {
 		bphy_err(drvr, "External authentication failed: status=%d\n",
Signature · 1 line
2.55.0
maintainer
#1 lore raw
Quoted 5 lines · Bogdan Nicolae

Avoid setting packet_id to cookie, which is always 0. Instead, use an increasing atomic counter. Avoids mismatches of completion events later in brcmf_notify_mgmt_tx_status, where packet_id != vif->mgmt_tx_id is checked.

Quoted 8 lines

Also, zero out auth_status on initialization. Otherwise, garbage will leak from the stack to the firmware (when bssid is less than 32 bytes and/or when params->pmkid is set). Then, pass the params->pmkid to the firmware (without it, the firmware caches a garbage PMKID on successful authentication and denies a subsequent association request that includes the PMKID).

I always get a bit confused when people use different email addresses to send and sign-off patches.

Quoted 18 lines
---
  .../net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c   | 8 ++++++--
  1 file changed, 6 insertions(+), 2 deletions(-)
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +2 -0
index ce09d44fa..cca53ff19 100644
@@ -23,6 +23,8 @@
  #define MGMT_AUTH_FRAME_DWELL_TIME	4000
  #define MGMT_AUTH_FRAME_WAIT_TIME	(MGMT_AUTH_FRAME_DWELL_TIME + 100)
 
+static atomic_t brcmf_cyw_mgmt_tx_id = ATOMIC_INIT(0);
+
  static int brcmf_cyw_set_sae_pwd(struct brcmf_if *ifp,
  				 struct cfg80211_crypto_settings *crypto)
  {

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev, · line 160

Quoted 5 lines
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +1 -1
@@ -155,7 +157,7 @@ int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev,
 
  	memcpy(&mf_params->da[0], &mgmt->da[0], ETH_ALEN);
  	memcpy(&mf_params->bssid[0], &mgmt->bssid[0], ETH_ALEN);
-	mf_params->packet_id = cpu_to_le32(*cookie);
+	mf_params->packet_id = cpu_to_le32(atomic_inc_return(&brcmf_cyw_mgmt_tx_id));

As I understand things the cookie value here is an output parameter. The driver should assign it:

+	*cookie = atomic_inc_return(&brcmf_cyw_mgmt_tx_id);
	mf_params->packet_id = cpu_to_le32(*cookie);

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev, · line 163

Quoted 3 lines
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +0 -0
  	memcpy(mf_params->data, &buf[DOT11_MGMT_HDR_LEN],
  	       le16_to_cpu(mf_params->len));
 

Regards,
Arend

#2 lore raw

Hi Arend,

The code already sets *cookie=0 in the beginning, which according to my understanding tells the driver it does not require async tracking upstream. This is correct, because the function calls wait_for_completion_timeout, so mgmt tx completes by the time we exit the function and doesn't need to be tracked upstream. However, while we are waiting for mgmt tx to finish, the firmware may generate non-matching completion events captured by notify_mgmt_tx_status. That's why setting a non-zero packed_id (sent to and returned by firmware as part of a completion event) and checking against vif->mgmt_tx_id is important.

Cheers,
Bogdan

Quoted 53 lines · Arend van Spriel
Quoted 5 lines · Bogdan Nicolae

Avoid setting packet_id to cookie, which is always 0. Instead, use an increasing atomic counter. Avoids mismatches of completion events later in brcmf_notify_mgmt_tx_status, where packet_id != vif->mgmt_tx_id is checked.

Quoted 8 lines

Also, zero out auth_status on initialization. Otherwise, garbage will leak from the stack to the firmware (when bssid is less than 32 bytes and/or when params->pmkid is set). Then, pass the params->pmkid to the firmware (without it, the firmware caches a garbage PMKID on successful authentication and denies a subsequent association request that includes the PMKID).

I always get a bit confused when people use different email addresses to send and sign-off patches.

Quoted 18 lines
---
  .../net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c   | 8 ++++++--
  1 file changed, 6 insertions(+), 2 deletions(-)
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +2 -0
index ce09d44fa..cca53ff19 100644
@@ -23,6 +23,8 @@
  #define MGMT_AUTH_FRAME_DWELL_TIME  4000
  #define MGMT_AUTH_FRAME_WAIT_TIME   (MGMT_AUTH_FRAME_DWELL_TIME + 100)
 
+static atomic_t brcmf_cyw_mgmt_tx_id = ATOMIC_INIT(0);
+
  static int brcmf_cyw_set_sae_pwd(struct brcmf_if *ifp,
                               struct cfg80211_crypto_settings *crypto)
  {

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev, · line 160

Quoted 5 lines
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +1 -1
@@ -155,7 +157,7 @@ int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev,
 
      memcpy(&mf_params->da[0], &mgmt->da[0], ETH_ALEN);
      memcpy(&mf_params->bssid[0], &mgmt->bssid[0], ETH_ALEN);
-     mf_params->packet_id = cpu_to_le32(*cookie);
+     mf_params->packet_id = cpu_to_le32(atomic_inc_return(&brcmf_cyw_mgmt_tx_id));

As I understand things the cookie value here is an output parameter. The driver should assign it:

+       *cookie = atomic_inc_return(&brcmf_cyw_mgmt_tx_id);
        mf_params->packet_id = cpu_to_le32(*cookie);

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev, · line 163

Quoted 3 lines
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +0 -0
      memcpy(mf_params->data, &buf[DOT11_MGMT_HDR_LEN],
             le16_to_cpu(mf_params->len));
 

Regards,
Arend

#3 lore raw
Quoted 12 lines · Bogdan Nicolae

Hi Arend,

The code already sets *cookie=0 in the beginning, which according to my understanding tells the driver it does not require async tracking upstream. This is correct, because the function calls wait_for_completion_timeout, so mgmt tx completes by the time we exit the function and doesn't need to be tracked upstream. However, while we are waiting for mgmt tx to finish, the firmware may generate non-matching completion events captured by notify_mgmt_tx_status. That's why setting a non-zero packed_id (sent to and returned by firmware as part of a completion event) and checking against vif->mgmt_tx_id is important.

Avoid top-posting. Need to use interleaved style.
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

Quoted 14 lines

Cheers,
Bogdan

Quoted 9 lines · Arend van Spriel
Quoted 7 lines · Bogdan Nicolae

Avoid setting packet_id to cookie, which is always 0. Instead, use an increasing atomic counter. Avoids mismatches of completion events later in brcmf_notify_mgmt_tx_status, where packet_id != vif->mgmt_tx_id is checked.

Also, zero out auth_status on initialization. Otherwise, garbage will leak from the stack to the firmware (when bssid is less than 32 bytes

BSSID will always be less than 32, you must have meant to say "ssid" here instead of "bssid". Also the auth_req_status struct does not have any "bssid" member, it only has an "ssid" struct member. Need to fix in commit message.

Quoted 36 lines
Quoted 36 lines
Quoted 6 lines

and/or when params->pmkid is set). Then, pass the params->pmkid to the firmware (without it, the firmware caches a garbage PMKID on successful authentication and denies a subsequent association request that includes the PMKID).

I always get a bit confused when people use different email addresses to send and sign-off patches.

Quoted 18 lines
---
  .../net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c   | 8 ++++++--
  1 file changed, 6 insertions(+), 2 deletions(-)
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +2 -0
index ce09d44fa..cca53ff19 100644
@@ -23,6 +23,8 @@
  #define MGMT_AUTH_FRAME_DWELL_TIME  4000
  #define MGMT_AUTH_FRAME_WAIT_TIME   (MGMT_AUTH_FRAME_DWELL_TIME + 100)
 
+static atomic_t brcmf_cyw_mgmt_tx_id = ATOMIC_INIT(0);
+
  static int brcmf_cyw_set_sae_pwd(struct brcmf_if *ifp,
                               struct cfg80211_crypto_settings *crypto)
  {

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev, · line 160

Quoted 5 lines
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +1 -1
@@ -155,7 +157,7 @@ int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev,
 
      memcpy(&mf_params->da[0], &mgmt->da[0], ETH_ALEN);
      memcpy(&mf_params->bssid[0], &mgmt->bssid[0], ETH_ALEN);
-     mf_params->packet_id = cpu_to_le32(*cookie);
+     mf_params->packet_id = cpu_to_le32(atomic_inc_return(&brcmf_cyw_mgmt_tx_id));

As I understand things the cookie value here is an output parameter. The driver should assign it:

Quoted 2 lines
Quoted 2 lines
+       *cookie = atomic_inc_return(&brcmf_cyw_mgmt_tx_id);
        mf_params->packet_id = cpu_to_le32(*cookie);

Agree with Arend's suggestion here. It is not correct to leave *cookie = 0 as it is and only assign a different packet_id while sending requst to firmware, because in that case the driver will not return the real cookie for this TX to cfg80211. Driver should assign the generated ID to *cookie and then copy that into packet_id.

Gokul

Quoted 6 lines
Quoted 5 lines

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev, · line 163

Quoted 3 lines
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +0 -0
      memcpy(mf_params->data, &buf[DOT11_MGMT_HDR_LEN],
             le16_to_cpu(mf_params->len));
 

Regards,
Arend

#4 lore raw
Quoted 30 lines · Gokul Sivakumar
Quoted 12 lines · Bogdan Nicolae

Hi Arend,

The code already sets *cookie=0 in the beginning, which according to my understanding tells the driver it does not require async tracking upstream. This is correct, because the function calls wait_for_completion_timeout, so mgmt tx completes by the time we exit the function and doesn't need to be tracked upstream. However, while we are waiting for mgmt tx to finish, the firmware may generate non-matching completion events captured by notify_mgmt_tx_status. That's why setting a non-zero packed_id (sent to and returned by firmware as part of a completion event) and checking against vif->mgmt_tx_id is important.

Avoid top-posting. Need to use interleaved style.
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

Quoted 12 lines

Cheers,
Bogdan

Quoted 7 lines · Arend van Spriel
Quoted 5 lines · Bogdan Nicolae

Avoid setting packet_id to cookie, which is always 0. Instead, use an increasing atomic counter. Avoids mismatches of completion events later in brcmf_notify_mgmt_tx_status, where packet_id != vif->mgmt_tx_id is checked.

Quoted 6 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Also, zero out auth_status on initialization. Otherwise, garbage will leak from the stack to the firmware (when bssid is less than 32 bytes

BSSID will always be less than 32, you must have meant to say "ssid" here instead of "bssid". Also the auth_req_status struct does not have any "bssid" member, it only has an "ssid" struct member. Need to fix in commit message.

Fixed

Quoted 36 lines
Quoted 36 lines
Quoted 36 lines
Quoted 6 lines

and/or when params->pmkid is set). Then, pass the params->pmkid to the firmware (without it, the firmware caches a garbage PMKID on successful authentication and denies a subsequent association request that includes the PMKID).

I always get a bit confused when people use different email addresses to send and sign-off patches.

Quoted 18 lines
---
  .../net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c   | 8 ++++++--
  1 file changed, 6 insertions(+), 2 deletions(-)
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +2 -0
index ce09d44fa..cca53ff19 100644
@@ -23,6 +23,8 @@
  #define MGMT_AUTH_FRAME_DWELL_TIME  4000
  #define MGMT_AUTH_FRAME_WAIT_TIME   (MGMT_AUTH_FRAME_DWELL_TIME + 100)
 
+static atomic_t brcmf_cyw_mgmt_tx_id = ATOMIC_INIT(0);
+
  static int brcmf_cyw_set_sae_pwd(struct brcmf_if *ifp,
                               struct cfg80211_crypto_settings *crypto)
  {

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev, · line 160

Quoted 5 lines
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +1 -1
@@ -155,7 +157,7 @@ int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev,
 
      memcpy(&mf_params->da[0], &mgmt->da[0], ETH_ALEN);
      memcpy(&mf_params->bssid[0], &mgmt->bssid[0], ETH_ALEN);
-     mf_params->packet_id = cpu_to_le32(*cookie);
+     mf_params->packet_id = cpu_to_le32(atomic_inc_return(&brcmf_cyw_mgmt_tx_id));

As I understand things the cookie value here is an output parameter. The driver should assign it:

Quoted 7 lines
Quoted 2 lines
Quoted 2 lines
+       *cookie = atomic_inc_return(&brcmf_cyw_mgmt_tx_id);
        mf_params->packet_id = cpu_to_le32(*cookie);

Agree with Arend's suggestion here. It is not correct to leave *cookie = 0 as it is and only assign a different packet_id while sending requst to firmware, because in that case the driver will not return the real cookie for this TX to cfg80211. Driver should assign the generated ID to *cookie and then copy that into packet_id.

Fixed

Quoted 9 lines

Gokul

Quoted 5 lines
Quoted 5 lines

drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c int brcmf_cyw_mgmt_tx(struct wiphy *wiphy, struct wireless_dev *wdev, · line 163

Quoted 3 lines
drivers/net/wireless/broadcom/brcm80211/brcmfmac/cyw/core.c +0 -0
      memcpy(mf_params->data, &buf[DOT11_MGMT_HDR_LEN],
             le16_to_cpu(mf_params->len));
 

Regards,
Arend

Regards,
Bogdan