| 1 |
From 521f2d87160344647a245ea77dca8e782cbc92da Mon Sep 17 00:00:00 2001 |
| 2 |
From: Emmanuel Grumbach <emmanuel.grumbach@intel.com> |
| 3 |
Date: Wed, 13 Jun 2018 11:49:20 +0300 |
| 4 |
Subject: [PATCH 028/145] iwlwifi: mvm: clear HW_RESTART_REQUESTED when |
| 5 |
stopping the interface |
| 6 |
|
| 7 |
[ Upstream commit 155f7e0441cd121b1e673d465a35e99f4b9b2f0b ] |
| 8 |
|
| 9 |
Fix a bug that happens in the following scenario: |
| 10 |
1) suspend without WoWLAN |
| 11 |
2) mac80211 calls drv_stop because of the suspend |
| 12 |
3) __iwl_mvm_mac_stop deallocates the aux station |
| 13 |
4) during drv_stop the firmware crashes |
| 14 |
5) iwlmvm: |
| 15 |
* sets IWL_MVM_STATUS_HW_RESTART_REQUESTED |
| 16 |
* asks mac80211 to kick the restart flow |
| 17 |
6) mac80211 puts the restart worker into a freezable |
| 18 |
queue which means that the worker will not run for now |
| 19 |
since the workqueue is already frozen |
| 20 |
7) ... |
| 21 |
8) resume |
| 22 |
9) mac80211 runs ieee80211_reconfig as part of the resume |
| 23 |
10) mac80211 detects that a restart flow has been requested |
| 24 |
and that we are now resuming from suspend and cancels |
| 25 |
the restart worker |
| 26 |
11) mac80211 calls drv_start() |
| 27 |
12) __iwl_mvm_mac_start checks that IWL_MVM_STATUS_HW_RESTART_REQUESTED |
| 28 |
clears it, sets IWL_MVM_STATUS_IN_HW_RESTART and calls |
| 29 |
iwl_mvm_restart_cleanup() |
| 30 |
13) iwl_fw_error_dump gets called and accesses the device |
| 31 |
to get debug data |
| 32 |
14) iwl_mvm_up adds the aux station |
| 33 |
15) iwl_mvm_add_aux_sta() allocates an internal station for |
| 34 |
the aux station |
| 35 |
16) iwl_mvm_allocate_int_sta() tests IWL_MVM_STATUS_IN_HW_RESTART |
| 36 |
and doesn't really allocate a station ID for the aux |
| 37 |
station |
| 38 |
17) a new queue is added for the aux station |
| 39 |
|
| 40 |
Note that steps from 5 to 9 aren't really part of the |
| 41 |
problem but were described for the sake of completeness. |
| 42 |
|
| 43 |
Once the iwl_mvm_mac_stop() is called, the device is not |
| 44 |
accessible, meaning that step 12) can't succeed and we'll |
| 45 |
see the following: |
| 46 |
|
| 47 |
drivers/net/wireless/intel/iwlwifi/pcie/trans.c:2122 iwl_trans_pcie_grab_nic_access+0xc0/0x1d6 [iwlwifi]() |
| 48 |
Timeout waiting for hardware access (CSR_GP_CNTRL 0x080403d8) |
| 49 |
Call Trace: |
| 50 |
[<ffffffffc03e6ad3>] iwl_trans_pcie_grab_nic_access+0xc0/0x1d6 [iwlwifi] |
| 51 |
[<ffffffffc03e6a13>] iwl_trans_pcie_dump_regs+0x3fd/0x3fd [iwlwifi] |
| 52 |
[<ffffffffc03dad42>] iwl_fw_error_dump+0x4f5/0xe8b [iwlwifi] |
| 53 |
[<ffffffffc04bd43e>] __iwl_mvm_mac_start+0x5a/0x21a [iwlmvm] |
| 54 |
[<ffffffffc04bd6d2>] iwl_mvm_mac_start+0xd4/0x103 [iwlmvm] |
| 55 |
[<ffffffffc042d378>] drv_start+0xa1/0xc5 [iwl7000_mac80211] |
| 56 |
[<ffffffffc045a339>] ieee80211_reconfig+0x145/0xf50 [mac80211] |
| 57 |
[<ffffffffc044788b>] ieee80211_resume+0x62/0x66 [mac80211] |
| 58 |
[<ffffffffc0366c5b>] wiphy_resume+0xa9/0xc6 [cfg80211] |
| 59 |
|
| 60 |
The station id of the aux station is set to 0xff in step 3 |
| 61 |
and because we don't really allocate a new station id for |
| 62 |
the auxliary station (as explained in 16), we end up sending |
| 63 |
a command to the firmware asking to connect the queue |
| 64 |
to station id 0xff. This makes the firmware crash with the |
| 65 |
following information: |
| 66 |
|
| 67 |
0x00002093 | ADVANCED_SYSASSERT |
| 68 |
0x000002F0 | trm_hw_status0 |
| 69 |
0x00000000 | trm_hw_status1 |
| 70 |
0x00000B38 | branchlink2 |
| 71 |
0x0001978C | interruptlink1 |
| 72 |
0x00000000 | interruptlink2 |
| 73 |
0xFF080501 | data1 |
| 74 |
0xDEADBEEF | data2 |
| 75 |
0xDEADBEEF | data3 |
| 76 |
Firmware error during reconfiguration - reprobe! |
| 77 |
FW error in SYNC CMD SCD_QUEUE_CFG |
| 78 |
|
| 79 |
Fix this by clearing IWL_MVM_STATUS_HW_RESTART_REQUESTED |
| 80 |
in iwl_mvm_mac_stop(). We won't be able to collect debug |
| 81 |
data anyway and when we will brought up again, we will |
| 82 |
have a clean state from the firmware perspective. |
| 83 |
Since we won't have IWL_MVM_STATUS_IN_HW_RESTART set in |
| 84 |
step 12) we won't get to the 2093 ASSERT either. |
| 85 |
|
| 86 |
Fixes: bf8b286f86fc ("iwlwifi: mvm: defer setting IWL_MVM_STATUS_IN_HW_RESTART") |
| 87 |
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com> |
| 88 |
Signed-off-by: Luca Coelho <luciano.coelho@intel.com> |
| 89 |
Signed-off-by: Sasha Levin <sashal@kernel.org> |
| 90 |
--- |
| 91 |
drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c | 9 ++++++--- |
| 92 |
1 file changed, 6 insertions(+), 3 deletions(-) |
| 93 |
|
| 94 |
diff --git a/drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c b/drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c |
| 95 |
index b15b0d84bb7e..155cc2ac0120 100644 |
| 96 |
--- a/drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c |
| 97 |
+++ b/drivers/net/wireless/intel/iwlwifi/mvm/mac80211.c |
| 98 |
@@ -1233,12 +1233,15 @@ void __iwl_mvm_mac_stop(struct iwl_mvm *mvm) |
| 99 |
iwl_mvm_del_aux_sta(mvm); |
| 100 |
|
| 101 |
/* |
| 102 |
- * Clear IN_HW_RESTART flag when stopping the hw (as restart_complete() |
| 103 |
- * won't be called in this case). |
| 104 |
+ * Clear IN_HW_RESTART and HW_RESTART_REQUESTED flag when stopping the |
| 105 |
+ * hw (as restart_complete() won't be called in this case) and mac80211 |
| 106 |
+ * won't execute the restart. |
| 107 |
* But make sure to cleanup interfaces that have gone down before/during |
| 108 |
* HW restart was requested. |
| 109 |
*/ |
| 110 |
- if (test_and_clear_bit(IWL_MVM_STATUS_IN_HW_RESTART, &mvm->status)) |
| 111 |
+ if (test_and_clear_bit(IWL_MVM_STATUS_IN_HW_RESTART, &mvm->status) || |
| 112 |
+ test_and_clear_bit(IWL_MVM_STATUS_HW_RESTART_REQUESTED, |
| 113 |
+ &mvm->status)) |
| 114 |
ieee80211_iterate_interfaces(mvm->hw, 0, |
| 115 |
iwl_mvm_cleanup_iterator, mvm); |
| 116 |
|
| 117 |
-- |
| 118 |
2.19.1 |
| 119 |
|
Parent Directory
| 
Revision Log