| Message ID | 1421612283-27012-1-git-send-email-emmanuel.grumbach@intel.com (mailing list archive) |
|---|---|
| State | Accepted |
| Delegated to: | Johannes Berg |
| Headers | show |
On 01/18/15 21:18, Emmanuel Grumbach wrote: > When we have an active scheduled scan, and the RFKILL > interrupt kicks in, the stack will cancel the scheduled > scan as part of the down flow. But cancelling scheduled > scan usually implies sending a command to the firwmare > which has been killed as part of the RFKILL interrupt > handling. > Because of that, we returned an error to mac80211 when > it asked to stop the scheduled scan and didn't notify the > end of the scheduled scan. Besides a fat warning, this led > to a situation in which cfg80211 would refuse any new scan > request. > > To disentangle this, fake that the scheduled scan has been > stopped without sending the command to the firwmare, return > 0 after having properly let cfg80211 know that the scan > has been cancelled. > > This is basically the same as: > commit 9b520d84957d63348e87c0f2cbd21d86e1e8f2f2 > Author: Emmanuel Grumbach<emmanuel.grumbach@intel.com> > Date: Tue Nov 4 15:54:11 2014 +0200 > > iwlwifi: mvm: abort scan upon RFKILL > > This code existed but not for all the different FW APIs > we support. > Fix this. > > but for the scheduled scan case. So looking at the patch I guess cfg80211 will still issues a WARN_ON, right? Regards, Arend > Link: http://permalink.gmane.org/gmane.linux.kernel.wireless.general/133232 > Reported-by: Linus Torvalds<torvalds@linux-foundation.org> > Signed-off-by: Emmanuel Grumbach<emmanuel.grumbach@intel.com> > --- > drivers/net/wireless/iwlwifi/mvm/scan.c | 7 +++++++ > 1 file changed, 7 insertions(+) > > diff --git a/drivers/net/wireless/iwlwifi/mvm/scan.c b/drivers/net/wireless/iwlwifi/mvm/scan.c > index 3fbba4b..844bf7c 100644 > --- a/drivers/net/wireless/iwlwifi/mvm/scan.c > +++ b/drivers/net/wireless/iwlwifi/mvm/scan.c > @@ -1107,6 +1107,12 @@ int iwl_mvm_scan_offload_stop(struct iwl_mvm *mvm, bool notify) > return iwl_umac_scan_stop(mvm, IWL_UMAC_SCAN_UID_SCHED_SCAN, > notify); > > + if (mvm->scan_status == IWL_MVM_SCAN_NONE) > + return 0; > + > + if (iwl_mvm_is_radio_killed(mvm)) > + goto out; > + > if (mvm->scan_status != IWL_MVM_SCAN_SCHED&& > (!(mvm->fw->ucode_capa.api[0]& IWL_UCODE_TLV_API_LMAC_SCAN) || > mvm->scan_status != IWL_MVM_SCAN_OS)) { > @@ -1143,6 +1149,7 @@ int iwl_mvm_scan_offload_stop(struct iwl_mvm *mvm, bool notify) > if (mvm->scan_status == IWL_MVM_SCAN_OS) > iwl_mvm_unref(mvm, IWL_MVM_REF_SCAN); > > +out: > mvm->scan_status = IWL_MVM_SCAN_NONE; > > if (notify) { -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sun, 2015-01-18 at 23:14 +0100, Arend van Spriel wrote: > On 01/18/15 21:18, Emmanuel Grumbach wrote: > > When we have an active scheduled scan, and the RFKILL > > interrupt kicks in, the stack will cancel the scheduled > > scan as part of the down flow. But cancelling scheduled > > scan usually implies sending a command to the firwmare > > which has been killed as part of the RFKILL interrupt > > handling. > > Because of that, we returned an error to mac80211 when > > it asked to stop the scheduled scan and didn't notify the > > end of the scheduled scan. Besides a fat warning, this led > > to a situation in which cfg80211 would refuse any new scan > > request. > > > > To disentangle this, fake that the scheduled scan has been > > stopped without sending the command to the firwmare, return > > 0 after having properly let cfg80211 know that the scan > > has been cancelled. > > > > This is basically the same as: > > commit 9b520d84957d63348e87c0f2cbd21d86e1e8f2f2 > > Author: Emmanuel Grumbach<emmanuel.grumbach@intel.com> > > Date: Tue Nov 4 15:54:11 2014 +0200 > > > > iwlwifi: mvm: abort scan upon RFKILL > > > > This code existed but not for all the different FW APIs > > we support. > > Fix this. > > > > but for the scheduled scan case. > > So looking at the patch I guess cfg80211 will still issues a WARN_ON, right? > Why? cfg80211 gets notified about the scan completion, so no scanning should remain active when the down flow is over. > > Link: http://permalink.gmane.org/gmane.linux.kernel.wireless.general/133232 > > Reported-by: Linus Torvalds<torvalds@linux-foundation.org> > > Signed-off-by: Emmanuel Grumbach<emmanuel.grumbach@intel.com> > > --- > > drivers/net/wireless/iwlwifi/mvm/scan.c | 7 +++++++ > > 1 file changed, 7 insertions(+) > > > > diff --git a/drivers/net/wireless/iwlwifi/mvm/scan.c b/drivers/net/wireless/iwlwifi/mvm/scan.c > > index 3fbba4b..844bf7c 100644 > > --- a/drivers/net/wireless/iwlwifi/mvm/scan.c > > +++ b/drivers/net/wireless/iwlwifi/mvm/scan.c > > @@ -1107,6 +1107,12 @@ int iwl_mvm_scan_offload_stop(struct iwl_mvm *mvm, bool notify) > > return iwl_umac_scan_stop(mvm, IWL_UMAC_SCAN_UID_SCHED_SCAN, > > notify); > > > > + if (mvm->scan_status == IWL_MVM_SCAN_NONE) > > + return 0; > > + > > + if (iwl_mvm_is_radio_killed(mvm)) > > + goto out; > > + > > if (mvm->scan_status != IWL_MVM_SCAN_SCHED&& > > (!(mvm->fw->ucode_capa.api[0]& IWL_UCODE_TLV_API_LMAC_SCAN) || > > mvm->scan_status != IWL_MVM_SCAN_OS)) { > > @@ -1143,6 +1149,7 @@ int iwl_mvm_scan_offload_stop(struct iwl_mvm *mvm, bool notify) > > if (mvm->scan_status == IWL_MVM_SCAN_OS) > > iwl_mvm_unref(mvm, IWL_MVM_REF_SCAN); > > > > +out: > > mvm->scan_status = IWL_MVM_SCAN_NONE; > > > > if (notify) { >
diff --git a/drivers/net/wireless/iwlwifi/mvm/scan.c b/drivers/net/wireless/iwlwifi/mvm/scan.c index 3fbba4b..844bf7c 100644 --- a/drivers/net/wireless/iwlwifi/mvm/scan.c +++ b/drivers/net/wireless/iwlwifi/mvm/scan.c @@ -1107,6 +1107,12 @@ int iwl_mvm_scan_offload_stop(struct iwl_mvm *mvm, bool notify) return iwl_umac_scan_stop(mvm, IWL_UMAC_SCAN_UID_SCHED_SCAN, notify); + if (mvm->scan_status == IWL_MVM_SCAN_NONE) + return 0; + + if (iwl_mvm_is_radio_killed(mvm)) + goto out; + if (mvm->scan_status != IWL_MVM_SCAN_SCHED && (!(mvm->fw->ucode_capa.api[0] & IWL_UCODE_TLV_API_LMAC_SCAN) || mvm->scan_status != IWL_MVM_SCAN_OS)) { @@ -1143,6 +1149,7 @@ int iwl_mvm_scan_offload_stop(struct iwl_mvm *mvm, bool notify) if (mvm->scan_status == IWL_MVM_SCAN_OS) iwl_mvm_unref(mvm, IWL_MVM_REF_SCAN); +out: mvm->scan_status = IWL_MVM_SCAN_NONE; if (notify) {
When we have an active scheduled scan, and the RFKILL interrupt kicks in, the stack will cancel the scheduled scan as part of the down flow. But cancelling scheduled scan usually implies sending a command to the firwmare which has been killed as part of the RFKILL interrupt handling. Because of that, we returned an error to mac80211 when it asked to stop the scheduled scan and didn't notify the end of the scheduled scan. Besides a fat warning, this led to a situation in which cfg80211 would refuse any new scan request. To disentangle this, fake that the scheduled scan has been stopped without sending the command to the firwmare, return 0 after having properly let cfg80211 know that the scan has been cancelled. This is basically the same as: commit 9b520d84957d63348e87c0f2cbd21d86e1e8f2f2 Author: Emmanuel Grumbach <emmanuel.grumbach@intel.com> Date: Tue Nov 4 15:54:11 2014 +0200 iwlwifi: mvm: abort scan upon RFKILL This code existed but not for all the different FW APIs we support. Fix this. but for the scheduled scan case. Link: http://permalink.gmane.org/gmane.linux.kernel.wireless.general/133232 Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> --- drivers/net/wireless/iwlwifi/mvm/scan.c | 7 +++++++ 1 file changed, 7 insertions(+)