nl80211 wlcore regression in next

nl80211 wlcore regression in next

[Tony Lindgren]

Hi,

Looks like at least drivers/net/wireless/ti wlcore driver has stopped
working in Linux next with commit 901bb9891855 ("nl80211: require and
validate vendor command policy"). Reverting the commit above makes it
work again.

It fails with the warning below, any ideas what goes wrong?

Regards,

Tony

8< ----------------
WARNING: CPU: 0 PID: 21 at net/wireless/core.c:868 wiphy_register+0x85c/0xbd4 [cfg80211]
...
[<bf05f570>] (wiphy_register [cfg80211]) from [<bf121e08>] (ieee80211_register_hw+0x4e4/0xcd8 [mac80211])
[<bf121e08>] (ieee80211_register_hw [mac80211]) from [<bf33135c>] (wlcore_nvs_cb+0x758/0xabc [wlcore])
[<bf33135c>] (wlcore_nvs_cb [wlcore]) from [<c05c3770>] (request_firmware_work_func+0x50/0x8c)
[<c05c3770>] (request_firmware_work_func) from [<c0154bb8>] (process_one_work+0x20c/0x504)
...

Re: nl80211 wlcore regression in next

[Johannes Berg]

On Tue, 2019-06-25 at 00:38 -0700, Tony Lindgren wrote:
> Hi,
> 
> Looks like at least drivers/net/wireless/ti wlcore driver has stopped
> working in Linux next with commit 901bb9891855 ("nl80211: require and
> validate vendor command policy"). Reverting the commit above makes it
> work again.
> 
> It fails with the warning below, any ideas what goes wrong?

Oops. For some reason, I neglected to check the vendor command usage
beyond hwsim.

The patch below should work?

johannes

diff --git a/drivers/net/wireless/ath/wil6210/cfg80211.c b/drivers/net/wireless/ath/wil6210/cfg80211.c
index 804955d24b30..37ac95940c22 100644
--- a/drivers/net/wireless/ath/wil6210/cfg80211.c
+++ b/drivers/net/wireless/ath/wil6210/cfg80211.c
@@ -177,6 +177,7 @@ static const struct wiphy_vendor_command wil_nl80211_vendor_commands[] = {
 		.info.subcmd = QCA_NL80211_VENDOR_SUBCMD_DMG_RF_GET_SECTOR_CFG,
 		.flags = WIPHY_VENDOR_CMD_NEED_WDEV |
 			 WIPHY_VENDOR_CMD_NEED_RUNNING,
+		.policy = wil_rf_sector_policy,
 		.doit = wil_rf_sector_get_cfg
 	},
 	{
@@ -184,6 +185,7 @@ static const struct wiphy_vendor_command wil_nl80211_vendor_commands[] = {
 		.info.subcmd = QCA_NL80211_VENDOR_SUBCMD_DMG_RF_SET_SECTOR_CFG,
 		.flags = WIPHY_VENDOR_CMD_NEED_WDEV |
 			 WIPHY_VENDOR_CMD_NEED_RUNNING,
+		.policy = wil_rf_sector_policy,
 		.doit = wil_rf_sector_set_cfg
 	},
 	{
@@ -192,6 +194,7 @@ static const struct wiphy_vendor_command wil_nl80211_vendor_commands[] = {
 			QCA_NL80211_VENDOR_SUBCMD_DMG_RF_GET_SELECTED_SECTOR,
 		.flags = WIPHY_VENDOR_CMD_NEED_WDEV |
 			 WIPHY_VENDOR_CMD_NEED_RUNNING,
+		.policy = wil_rf_sector_policy,
 		.doit = wil_rf_sector_get_selected
 	},
 	{
@@ -200,6 +203,7 @@ static const struct wiphy_vendor_command wil_nl80211_vendor_commands[] = {
 			QCA_NL80211_VENDOR_SUBCMD_DMG_RF_SET_SELECTED_SECTOR,
 		.flags = WIPHY_VENDOR_CMD_NEED_WDEV |
 			 WIPHY_VENDOR_CMD_NEED_RUNNING,
+		.policy = wil_rf_sector_policy,
 		.doit = wil_rf_sector_set_selected
 	},
 };
diff --git a/drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.c b/drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.c
index d493021f6031..30ebadc5e5bb 100644
--- a/drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.c
+++ b/drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.c
@@ -123,6 +123,7 @@ const struct wiphy_vendor_command brcmf_vendor_cmds[] = {
 		},
 		.flags = WIPHY_VENDOR_CMD_NEED_WDEV |
 			 WIPHY_VENDOR_CMD_NEED_NETDEV,
+		.policy = VENDOR_CMD_RAW_DATA,
 		.doit = brcmf_cfg80211_vndr_cmds_dcmd_handler
 	},
 };
diff --git a/drivers/net/wireless/ti/wlcore/vendor_cmd.c b/drivers/net/wireless/ti/wlcore/vendor_cmd.c
index 75756fb8e7b0..cd82d3580e87 100644
--- a/drivers/net/wireless/ti/wlcore/vendor_cmd.c
+++ b/drivers/net/wireless/ti/wlcore/vendor_cmd.c
@@ -166,6 +166,7 @@ static const struct wiphy_vendor_command wlcore_vendor_commands[] = {
 		.flags = WIPHY_VENDOR_CMD_NEED_NETDEV |
 			 WIPHY_VENDOR_CMD_NEED_RUNNING,
 		.doit = wlcore_vendor_cmd_smart_config_start,
+		.policy = wlcore_vendor_attr_policy,
 	},
 	{
 		.info = {
@@ -175,6 +176,7 @@ static const struct wiphy_vendor_command wlcore_vendor_commands[] = {
 		.flags = WIPHY_VENDOR_CMD_NEED_NETDEV |
 			 WIPHY_VENDOR_CMD_NEED_RUNNING,
 		.doit = wlcore_vendor_cmd_smart_config_stop,
+		.policy = wlcore_vendor_attr_policy,
 	},
 	{
 		.info = {
@@ -184,6 +186,7 @@ static const struct wiphy_vendor_command wlcore_vendor_commands[] = {
 		.flags = WIPHY_VENDOR_CMD_NEED_NETDEV |
 			 WIPHY_VENDOR_CMD_NEED_RUNNING,
 		.doit = wlcore_vendor_cmd_smart_config_set_group_key,
+		.policy = wlcore_vendor_attr_policy,
 	},
 };
 

Re: nl80211 wlcore regression in next

[Tony Lindgren]

Hi,

* Johannes Berg <johannes-cdvu00un1VgdHxzADdlk8Q@public.gmane.org> [190625 07:47]:
> On Tue, 2019-06-25 at 00:38 -0700, Tony Lindgren wrote:
> > Hi,
> > 
> > Looks like at least drivers/net/wireless/ti wlcore driver has stopped
> > working in Linux next with commit 901bb9891855 ("nl80211: require and
> > validate vendor command policy"). Reverting the commit above makes it
> > work again.
> > 
> > It fails with the warning below, any ideas what goes wrong?
> 
> Oops. For some reason, I neglected to check the vendor command usage
> beyond hwsim.
> 
> The patch below should work?

Yeah thanks that fixes the issue for me:

Tested-by: Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org>

Re: nl80211 wlcore regression in next

[Johannes Berg]

On Tue, 2019-06-25 at 01:00 -0700, Tony Lindgren wrote:
> Hi,
> 
> * Johannes Berg <johannes@sipsolutions.net> [190625 07:47]:
> > On Tue, 2019-06-25 at 00:38 -0700, Tony Lindgren wrote:
> > > Hi,
> > > 
> > > Looks like at least drivers/net/wireless/ti wlcore driver has stopped
> > > working in Linux next with commit 901bb9891855 ("nl80211: require and
> > > validate vendor command policy"). Reverting the commit above makes it
> > > work again.
> > > 
> > > It fails with the warning below, any ideas what goes wrong?
> > 
> > Oops. For some reason, I neglected to check the vendor command usage
> > beyond hwsim.
> > 
> > The patch below should work?
> 
> Yeah thanks that fixes the issue for me:
> 
> Tested-by: Tony Lindgren <tony@atomide.com>

Thanks, I'll drop that into my tree and hopefully will remember to send
it on soon.

johannes

Re: nl80211 wlcore regression in next

[Arend Van Spriel]

On 6/25/2019 10:02 AM, Johannes Berg wrote:
> On Tue, 2019-06-25 at 01:00 -0700, Tony Lindgren wrote:
>> Hi,
>>
>> * Johannes Berg <johannes@sipsolutions.net> [190625 07:47]:
>>> On Tue, 2019-06-25 at 00:38 -0700, Tony Lindgren wrote:
>>>> Hi,
>>>>
>>>> Looks like at least drivers/net/wireless/ti wlcore driver has stopped
>>>> working in Linux next with commit 901bb9891855 ("nl80211: require and
>>>> validate vendor command policy"). Reverting the commit above makes it
>>>> work again.
>>>>
>>>> It fails with the warning below, any ideas what goes wrong?
>>>
>>> Oops. For some reason, I neglected to check the vendor command usage
>>> beyond hwsim.
>>>
>>> The patch below should work?
>>
>> Yeah thanks that fixes the issue for me:
>>
>> Tested-by: Tony Lindgren <tony@atomide.com>
> 
> Thanks, I'll drop that into my tree and hopefully will remember to send
> it on soon.

Hi Johannes,

By chance noticed the patch included brcmfmac. So I tried, but I get 
compile issue below. It is because ERR_PTR really is an inline function 
so that is not working. So also need to patch that. I left the extra 
braces around the error code although not strictly necessary.

Regards,
Arend
---
diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h
index 2d17e32..da8249b 100644
--- a/include/net/cfg80211.h
+++ b/include/net/cfg80211.h
@@ -4172,7 +4172,7 @@ struct sta_opmode_info {
         u8 rx_nss;
  };

-#define VENDOR_CMD_RAW_DATA ((const struct nla_policy *)ERR_PTR(-ENODATA))
+#define VENDOR_CMD_RAW_DATA ((const struct nla_policy *)(-ENODATA))

  /**
   * struct wiphy_vendor_command - vendor command definition


---8<--------------------------------------------------------------------
   CC [M]  drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.o
In file included from 
drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.c:18:0:
./include/net/cfg80211.h:4175:29: error: initializer element is not constant
  #define VENDOR_CMD_RAW_DATA ((const struct nla_policy *)ERR_PTR(-ENODATA))
                              ^
drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.c:126:13: note: 
in expansion of macro \u2018VENDOR_CMD_RAW_DATA\u2019
    .policy = VENDOR_CMD_RAW_DATA,
              ^
./include/net/cfg80211.h:4175:29: note: (near initialization for 
\u2018brcmf_vendor_cmds[0].policy\u2019)
  #define VENDOR_CMD_RAW_DATA ((const struct nla_policy *)ERR_PTR(-ENODATA))
                              ^
drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.c:126:13: note: 
in expansion of macro \u2018VENDOR_CMD_RAW_DATA\u2019
    .policy = VENDOR_CMD_RAW_DATA,
              ^
make[3]: *** [drivers/net/wireless/broadcom/brcm80211/brcmfmac/vendor.o] 
Error 1

Re: nl80211 wlcore regression in next

[Johannes Berg]

Hi Arend,


> By chance noticed the patch included brcmfmac. So I tried, but I get 
> compile issue below. It is because ERR_PTR really is an inline function 
> so that is not working. So also need to patch that. I left the extra 
> braces around the error code although not strictly necessary.

D'oh. I'll check that out, thanks for the report/patch!

johannes

Re: nl80211 wlcore regression in next

[Tony Lindgren]

Hi,

* Johannes Berg <johannes@sipsolutions.net> [190625 08:03]:
> On Tue, 2019-06-25 at 01:00 -0700, Tony Lindgren wrote:
> > Hi,
> > 
> > * Johannes Berg <johannes@sipsolutions.net> [190625 07:47]:
> > > On Tue, 2019-06-25 at 00:38 -0700, Tony Lindgren wrote:
> > > > Hi,
> > > > 
> > > > Looks like at least drivers/net/wireless/ti wlcore driver has stopped
> > > > working in Linux next with commit 901bb9891855 ("nl80211: require and
> > > > validate vendor command policy"). Reverting the commit above makes it
> > > > work again.
> > > > 
> > > > It fails with the warning below, any ideas what goes wrong?
> > > 
> > > Oops. For some reason, I neglected to check the vendor command usage
> > > beyond hwsim.
> > > 
> > > The patch below should work?
> > 
> > Yeah thanks that fixes the issue for me:
> > 
> > Tested-by: Tony Lindgren <tony@atomide.com>
> 
> Thanks, I'll drop that into my tree and hopefully will remember to send
> it on soon.

Looks like this one crept back as the fix is missing from v5.3-rc1.

Forgot to include in the pull request?

Regards,

Tony

Re: nl80211 wlcore regression in next

[Johannes Berg]

Hi,

> Looks like this one crept back as the fix is missing from v5.3-rc1.
> 
> Forgot to include in the pull request?

More like forgot to send the pull request, my bad. I eventually realized
a couple of days ago and it'll be coming upstream soon. Sorry about
that.

johannes

Re: nl80211 wlcore regression in next

[Tony Lindgren]

* Johannes Berg <johannes@sipsolutions.net> [190723 05:08]:
> Hi,
> 
> > Looks like this one crept back as the fix is missing from v5.3-rc1.
> > 
> > Forgot to include in the pull request?
> 
> More like forgot to send the pull request, my bad. I eventually realized
> a couple of days ago and it'll be coming upstream soon. Sorry about
> that.

OK thanks,

Tony

Re: nl80211 wlcore regression in next

[Kalle Valo]

Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org> writes:

> Hi,
>
> * Johannes Berg <johannes-cdvu00un1VgdHxzADdlk8Q@public.gmane.org> [190625 07:47]:
>> On Tue, 2019-06-25 at 00:38 -0700, Tony Lindgren wrote:
>> > Hi,
>> > 
>> > Looks like at least drivers/net/wireless/ti wlcore driver has stopped
>> > working in Linux next with commit 901bb9891855 ("nl80211: require and
>> > validate vendor command policy"). Reverting the commit above makes it
>> > work again.
>> > 
>> > It fails with the warning below, any ideas what goes wrong?
>> 
>> Oops. For some reason, I neglected to check the vendor command usage
>> beyond hwsim.
>> 
>> The patch below should work?
>
> Yeah thanks that fixes the issue for me:
>
> Tested-by: Tony Lindgren <tony-4v6yS6AI5VpBDgjK7y7TUQ@public.gmane.org>

Acked-by: Kalle Valo <kvalo-sgV2jX0FEOL9JmXXK+q4OQ@public.gmane.org>

-- 
Kalle Valo