Hi,
The wireless connection keeps failing shortly after being established. Up to now, I've tracked it down to a DMA transmission sequence error in ring 3. Beyond that, I cannot say...
Happy to provide further information and to test any potential fixes.
System:
Dell Studio 17 running Fedora 18
Linux wks001.ise.net 3.8.2-206.fc18.x86_64 #1 SMP Fri Mar 8 15:03:34 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
Broadcom BCM 4312 wireless card:
08:00.0 Network controller [0280]: Broadcom Corporation BCM4312 802.11b/g LP-PHY [14e4:4315] (rev 01)
08:00.0 0280: 14e4:4315 (rev 01)
Subsystem: 1028:000c
Control: I/O- Mem+ BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
Interrupt: pin A routed to IRQ 17
Region 0: Memory at f8000000 (64-bit, non-prefetchable) [size=16K]
Capabilities: [40] Power Management version 3
Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
Firmware was obtained as per b43 wiki page, namely http://www.lwfinger.com/b43-firmware/broadcom-wl-5.100.138.tar.bz2.
In both examples below, I've added a single b43dbg statement to 'b43_dma_handle_txstatus()' which prints out the DMA ring and slot number after the out-of-order sanity check.
I've tried using the driver from Fedora 18 kernel-3.8.2-206.fc18.x86_64 (which is practically the same as the current driver on torvalds/linux.git) and from the latest commit on linville/wireless-testing.git. In both cases, I end up with an 'out of order TX status on DMA ring 3' on slot 138 which leads to an AP disconnect after a while due to inactivity timeout.
In the first case, I get a 'TX-status contains invalid cookie: 0x0000' when slot 138 is the first slot in ring 3. In the second case, it's 'Out of order TX status report on DMA ring 3. Expected 138, but got 126'. After the error, status is given for what would have been the correct slot (i.e. 140, 142, etc...) but obviously slot 138 was not processed so all subsequent status reports get dropped.
The error, at least on my laptop, is easy to reproduce - it happens within a couple of minutes (or less) every time I start using the b43 driver.
The same laptop running Fedora 16 (3.6.11) with the broadcom-wl driver or running Windows 7 works perfectly (at least, no noticeable disconnects). The b43 driver on Fedora 16 (3.6.11) and Fedora 17 (3.7.9) also disconnected but I don't have any debug info to provide for those cases. The broadcom-wl driver on Fedora 18 also works fine until it causes a panic - but before then, no disconnects
I've tried against two different AP with same results.
Using the driver from Fedora 18 kernel-3.8.2-206.fc18.x86_64, the sequence leading up to the failure is:
[ 7813.857650] cfg80211: World regulatory domain updated:
[ 7813.857658] cfg80211: (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[ 7813.857665] cfg80211: (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[ 7813.857672] cfg80211: (2457000 KHz - 2482000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[ 7813.857678] cfg80211: (2474000 KHz - 2494000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[ 7813.857685] cfg80211: (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[ 7813.857691] cfg80211: (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[ 7813.857978] cfg80211: Calling CRDA for country: IT
[ 7813.872325] cfg80211: Regulatory domain changed to country: IT
[ 7813.872334] cfg80211: (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[ 7813.872341] cfg80211: (2402000 KHz - 2482000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[ 7813.872347] cfg80211: (5170000 KHz - 5250000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[ 7813.872353] cfg80211: (5250000 KHz - 5330000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[ 7813.872359] cfg80211: (5490000 KHz - 5710000 KHz @ 40000 KHz), (N/A, 2700 mBm)
[ 7813.872365] cfg80211: (57240000 KHz - 65880000 KHz @ 2160000 KHz), (N/A, 4000 mBm)
[10226.231819] ssb: Found chip with id 0x4312, rev 0x01 and package 0x00
[10226.231959] ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x16, vendor 0x4243)
[10226.232093] ssb: Core 1 found: IEEE 802.11 (cc 0x812, rev 0x0F, vendor 0x4243)
[10226.232230] ssb: Core 2 found: PCMCIA (cc 0x80D, rev 0x0A, vendor 0x4243)
[10226.232364] ssb: Core 3 found: PCI-E (cc 0x820, rev 0x09, vendor 0x4243)
[10226.273825] ssb: Sonics Silicon Backplane found on PCI device 0000:08:00.0
[10226.302682] cfg80211: Calling CRDA to update world regulatory domain
[10226.309553] cfg80211: World regulatory domain updated:
[10226.309560] cfg80211: (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[10226.309565] cfg80211: (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[10226.309569] cfg80211: (2457000 KHz - 2482000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[10226.309573] cfg80211: (2474000 KHz - 2494000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[10226.309576] cfg80211: (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[10226.309580] cfg80211: (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[10226.351857] b43-phy0: Broadcom 4312 WLAN found (core revision 15)
[10226.366408] b43-phy0: Found PHY: Analog 6, Type 5 (LP), Revision 1
[10226.366458] b43-phy0 debug: Found Radio: Manuf 0x17F, Version 0x2062, Revision 2
[10226.373908] Broadcom 43xx driver loaded [ Features: PMNLS ]
[10226.376913] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[10226.386903] CE: hpet increased min_delta_ns to 30169 nsec
[10226.407881] cfg80211: Calling CRDA for country: IT
[10226.410574] cfg80211: Regulatory domain changed to country: IT
[10226.410577] cfg80211: (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[10226.410580] cfg80211: (2402000 KHz - 2482000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[10226.410582] cfg80211: (5170000 KHz - 5250000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[10226.410584] cfg80211: (5250000 KHz - 5330000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[10226.410586] cfg80211: (5490000 KHz - 5710000 KHz @ 40000 KHz), (N/A, 2700 mBm)
[10226.410588] cfg80211: (57240000 KHz - 65880000 KHz @ 2160000 KHz), (N/A, 4000 mBm)
[10226.571407] b43-phy0: Loading firmware version 666.2 (2011-02-23 01:15:07)
[10226.575295] b43-phy0 debug: b2062: Using crystal tab entry 19200 kHz.
[10227.963115] b43-phy0 debug: Chip initialized
[10227.963578] b43-phy0 debug: 64-bit DMA initialized
[10227.963689] b43-phy0 debug: QoS enabled
[10227.975034] b43-phy0 debug: Wireless interface started
[10227.979312] b43-phy0 debug: Adding Interface type 2
[10227.980304] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[10228.052855] b43-phy0 debug: DMA ring 3 slot 0.
[10228.121660] b43-phy0 ERROR: PHY transmission error
[10228.121739] b43-phy0 debug: DMA ring 3 slot 2.
[10228.190536] b43-phy0 debug: DMA ring 3 slot 4.
[10228.259123] b43-phy0 debug: DMA ring 3 slot 6.
[10228.328973] b43-phy0 debug: DMA ring 3 slot 8.
[10228.398307] b43-phy0 debug: DMA ring 3 slot 10.
[10228.467265] b43-phy0 debug: DMA ring 3 slot 12.
[10228.535794] b43-phy0 debug: DMA ring 3 slot 14.
[10228.604772] b43-phy0 debug: DMA ring 3 slot 16.
[10228.677115] b43-phy0 debug: DMA ring 3 slot 18.
[10228.744996] b43-phy0 debug: DMA ring 3 slot 20.
[10228.818880] b43-phy0 debug: DMA ring 3 slot 22.
[10228.887719] b43-phy0 debug: DMA ring 3 slot 24.
[10251.086064] b43-phy0 debug: DMA ring 3 slot 26.
[10251.155056] b43-phy0 debug: DMA ring 3 slot 28.
[10251.224319] b43-phy0 debug: DMA ring 3 slot 30.
[10251.293053] b43-phy0 debug: DMA ring 3 slot 32.
[10251.362314] b43-phy0 debug: DMA ring 3 slot 34.
[10251.431257] b43-phy0 debug: DMA ring 3 slot 36.
[10251.501387] b43-phy0 debug: DMA ring 3 slot 38.
[10251.568804] b43-phy0 debug: DMA ring 3 slot 40.
[10251.637499] b43-phy0 debug: DMA ring 3 slot 42.
[10251.706885] b43-phy0 debug: DMA ring 3 slot 44.
[10251.777551] b43-phy0 debug: DMA ring 3 slot 46.
[10251.846896] b43-phy0 debug: DMA ring 3 slot 48.
[10251.916025] b43-phy0 debug: DMA ring 3 slot 50.
[10252.007155] b43-phy0 debug: DMA ring 3 slot 52.
[10252.007842] b43-phy0 debug: DMA ring 3 slot 54.
[10252.076407] b43-phy0 debug: DMA ring 3 slot 56.
[10252.077438] b43-phy0 debug: DMA ring 3 slot 58.
[10252.145070] b43-phy0 debug: DMA ring 3 slot 60.
[10252.146155] b43-phy0 debug: DMA ring 3 slot 62.
[10252.214077] b43-phy0 ERROR: PHY transmission error
[10252.214115] b43-phy0 debug: DMA ring 3 slot 64.
[10252.214180] b43-phy0 debug: DMA ring 3 slot 66.
[10252.283054] b43-phy0 debug: DMA ring 3 slot 68.
[10252.284047] b43-phy0 debug: DMA ring 3 slot 70.
[10252.351809] b43-phy0 debug: DMA ring 3 slot 72.
[10252.352947] b43-phy0 debug: DMA ring 3 slot 74.
[10252.420822] b43-phy0 debug: DMA ring 3 slot 76.
[10252.421505] b43-phy0 debug: DMA ring 3 slot 78.
[10252.489874] b43-phy0 debug: DMA ring 3 slot 80.
[10252.490817] b43-phy0 debug: DMA ring 3 slot 82.
[10252.559210] b43-phy0 debug: DMA ring 3 slot 84.
[10252.559916] b43-phy0 debug: DMA ring 3 slot 86.
[10252.627740] b43-phy0 debug: DMA ring 3 slot 88.
[10252.628684] b43-phy0 debug: DMA ring 3 slot 90.
[10252.698563] b43-phy0 debug: DMA ring 3 slot 92.
[10252.699431] b43-phy0 debug: DMA ring 3 slot 94.
[10252.767375] b43-phy0 debug: DMA ring 3 slot 96.
[10252.768623] b43-phy0 debug: DMA ring 3 slot 98.
[10252.836728] b43-phy0 ERROR: PHY transmission error
[10252.836783] b43-phy0 debug: DMA ring 3 slot 100.
[10252.837246] b43-phy0 debug: DMA ring 3 slot 102.
[10252.867641] wlan0: authenticate with e0:91:53:56:f0:ad
[10252.875267] wlan0: capabilities/regulatory prevented using AP HT/VHT configuration, downgraded
[10252.884491] wlan0: send auth to e0:91:53:56:f0:ad (try 1/3)
[10252.885652] b43-phy0 debug: DMA ring 3 slot 104.
[10252.886988] wlan0: authenticated
[10252.887954] wlan0: associate with e0:91:53:56:f0:ad (try 1/3)
[10252.890166] b43-phy0 debug: DMA ring 3 slot 106.
[10252.891817] wlan0: RX AssocResp from e0:91:53:56:f0:ad (capab=0x431 status=0 aid=1)
[10252.892904] wlan0: associated
[10252.892950] IPv6: ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
[10252.893027] cfg80211: Calling CRDA for country: IT
[10252.898442] cfg80211: Regulatory domain changed to country: IT
[10252.898449] cfg80211: (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[10252.898456] cfg80211: (2402000 KHz - 2482000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[10252.898462] cfg80211: (5170000 KHz - 5250000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[10252.898468] cfg80211: (5250000 KHz - 5330000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[10252.898473] cfg80211: (5490000 KHz - 5710000 KHz @ 40000 KHz), (N/A, 2700 mBm)
[10252.898479] cfg80211: (57240000 KHz - 65880000 KHz @ 2160000 KHz), (N/A, 4000 mBm)
[10252.901143] b43-phy0 debug: DMA ring 1 slot 0.
[10252.909940] b43-phy0 debug: Using hardware based encryption for keyidx: 0, mac: e0:91:53:56:f0:ad
[10252.910308] b43-phy0 debug: Using hardware based encryption for keyidx: 2, mac: ff:ff:ff:ff:ff:ff
[10252.911624] b43-phy0 debug: DMA ring 1 slot 2.
[10253.204622] b43-phy0 debug: DMA ring 1 slot 4.
[10254.204703] b43-phy0 debug: DMA ring 1 slot 6.
[10256.758088] b43-phy0 debug: DMA ring 3 slot 108.
[10258.212861] b43-phy0 debug: DMA ring 1 slot 8.
[10259.747293] b43-phy0 debug: DMA ring 3 slot 110.
[10261.330670] b43-phy0 debug: DMA ring 1 slot 10.
[10262.217511] b43-phy0 debug: DMA ring 1 slot 12.
[10262.744251] b43-phy0 debug: DMA ring 3 slot 112.
[10265.742042] b43-phy0 debug: DMA ring 3 slot 114.
[10268.738872] b43-phy0 debug: DMA ring 3 slot 116.
[10271.736376] b43-phy0 debug: DMA ring 3 slot 118.
[10274.733344] b43-phy0 debug: DMA ring 3 slot 120.
[10277.730636] b43-phy0 debug: DMA ring 3 slot 122.
[10280.728237] b43-phy0 debug: DMA ring 3 slot 124.
[10283.725525] b43-phy0 debug: DMA ring 3 slot 126.
[10284.031830] b43-phy0 debug: DMA ring 3 slot 128.
[10284.061704] b43-phy0 debug: DMA ring 3 slot 130.
[10284.130623] b43-phy0 debug: DMA ring 3 slot 132.
[10284.199890] b43-phy0 debug: DMA ring 3 slot 134.
[10284.269002] b43-phy0 debug: DMA ring 3 slot 136.
[10284.337647] b43-phy0 debug: TX-status contains invalid cookie: 0x0000
[10284.406370] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 140
[10284.475308] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 142
[10284.544219] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 144
[10284.613465] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 146
[10284.682521] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 148
Using the b43 driver compiled from commit d41d9c7419e3ac9c81841f43bbd7639dd0a5819e (Wed Mar 13 14:46:11 2013 -0400) from git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-testing.git:
[11716.327942] ssb: Found chip with id 0x4312, rev 0x01 and package 0x00
[11716.328082] ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x16, vendor 0x4243)
[11716.328216] ssb: Core 1 found: IEEE 802.11 (cc 0x812, rev 0x0F, vendor 0x4243)
[11716.328353] ssb: Core 2 found: PCMCIA (cc 0x80D, rev 0x0A, vendor 0x4243)
[11716.328487] ssb: Core 3 found: PCI-E (cc 0x820, rev 0x09, vendor 0x4243)
[11716.365095] ssb: Sonics Silicon Backplane found on PCI device 0000:08:00.0
[11716.384429] cfg80211: Calling CRDA to update world regulatory domain
[11716.391467] cfg80211: World regulatory domain updated:
[11716.391473] cfg80211: (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[11716.391478] cfg80211: (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[11716.391482] cfg80211: (2457000 KHz - 2482000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[11716.391486] cfg80211: (2474000 KHz - 2494000 KHz @ 20000 KHz), (300 mBi, 2000 mBm)
[11716.391490] cfg80211: (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[11716.391528] cfg80211: (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[11716.422638] b43-phy0: Broadcom 4312 WLAN found (core revision 15)
[11716.437604] b43-phy0: Found PHY: Analog 6, Type 5 (LP), Revision 1
[11716.437651] b43-phy0 debug: Found Radio: Manuf 0x17F, Version 0x2062, Revision 2
[11716.445008] Broadcom 43xx driver loaded [ Features: PMNLS ]
[11716.447753] ieee80211 phy0: Selected rate control algorithm 'minstrel_ht'
[11716.471947] cfg80211: Calling CRDA for country: IT
[11716.478615] cfg80211: Regulatory domain changed to country: IT
[11716.478622] cfg80211: (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[11716.478629] cfg80211: (2402000 KHz - 2482000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[11716.478635] cfg80211: (5170000 KHz - 5250000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[11716.478641] cfg80211: (5250000 KHz - 5330000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[11716.478647] cfg80211: (5490000 KHz - 5710000 KHz @ 40000 KHz), (N/A, 2700 mBm)
[11716.478653] cfg80211: (57240000 KHz - 65880000 KHz @ 2160000 KHz), (N/A, 4000 mBm)
[11716.633541] b43-phy0: Loading firmware version 666.2 (2011-02-23 01:15:07)
[11716.637589] b43-phy0 debug: b2062: Using crystal tab entry 19200 kHz.
[11718.028304] b43-phy0 debug: Chip initialized
[11718.028694] b43-phy0 debug: 64-bit DMA initialized
[11718.028808] b43-phy0 debug: QoS enabled
[11718.038689] b43-phy0 debug: Wireless interface started
[11718.043523] b43-phy0 debug: Adding Interface type 2
[11718.044504] IPv6: ADDRCONF(NETDEV_UP): wlan0: link is not ready
[11718.114760] b43-phy0 debug: DMA ring 3 slot 0.
[11718.183468] b43-phy0 ERROR: PHY transmission error
[11718.183502] b43-phy0 debug: DMA ring 3 slot 2.
[11718.252602] b43-phy0 debug: DMA ring 3 slot 4.
[11718.321559] b43-phy0 debug: DMA ring 3 slot 6.
[11718.390009] b43-phy0 debug: DMA ring 3 slot 8.
[11718.459025] b43-phy0 debug: DMA ring 3 slot 10.
[11718.528598] b43-phy0 debug: DMA ring 3 slot 12.
[11718.596907] b43-phy0 debug: DMA ring 3 slot 14.
[11718.665827] b43-phy0 debug: DMA ring 3 slot 16.
[11718.736213] b43-phy0 debug: DMA ring 3 slot 18.
[11718.805424] b43-phy0 debug: DMA ring 3 slot 20.
[11718.873616] b43-phy0 debug: DMA ring 3 slot 22.
[11718.943168] b43-phy0 debug: DMA ring 3 slot 24.
[11741.713698] b43-phy0 debug: DMA ring 3 slot 26.
[11741.782711] b43-phy0 debug: DMA ring 3 slot 28.
[11741.852316] b43-phy0 debug: DMA ring 3 slot 30.
[11741.920798] b43-phy0 debug: DMA ring 3 slot 32.
[11741.989955] b43-phy0 debug: DMA ring 3 slot 34.
[11742.058667] b43-phy0 debug: DMA ring 3 slot 36.
[11742.127301] b43-phy0 debug: DMA ring 3 slot 38.
[11742.196276] b43-phy0 debug: DMA ring 3 slot 40.
[11742.265185] b43-phy0 debug: DMA ring 3 slot 42.
[11742.334446] b43-phy0 debug: DMA ring 3 slot 44.
[11742.403438] b43-phy0 debug: DMA ring 3 slot 46.
[11742.472509] b43-phy0 debug: DMA ring 3 slot 48.
[11742.541014] b43-phy0 debug: DMA ring 3 slot 50.
[11742.639980] b43-phy0 debug: DMA ring 3 slot 52.
[11742.640725] b43-phy0 debug: DMA ring 3 slot 54.
[11742.708748] b43-phy0 debug: DMA ring 3 slot 56.
[11742.709563] b43-phy0 debug: DMA ring 3 slot 58.
[11742.777698] b43-phy0 debug: DMA ring 3 slot 60.
[11742.778736] b43-phy0 debug: DMA ring 3 slot 62.
[11742.846647] b43-phy0 debug: DMA ring 3 slot 64.
[11742.847371] b43-phy0 debug: DMA ring 3 slot 66.
[11742.916310] b43-phy0 debug: DMA ring 3 slot 68.
[11742.916910] b43-phy0 debug: DMA ring 3 slot 70.
[11742.984535] b43-phy0 debug: DMA ring 3 slot 72.
[11742.985450] b43-phy0 debug: DMA ring 3 slot 74.
[11743.055580] b43-phy0 debug: DMA ring 3 slot 76.
[11743.056451] b43-phy0 debug: DMA ring 3 slot 78.
[11743.124278] b43-phy0 debug: DMA ring 3 slot 80.
[11743.125413] b43-phy0 debug: DMA ring 3 slot 82.
[11743.194425] b43-phy0 debug: DMA ring 3 slot 84.
[11743.195400] b43-phy0 debug: DMA ring 3 slot 86.
[11743.263413] b43-phy0 debug: DMA ring 3 slot 88.
[11743.264128] b43-phy0 debug: DMA ring 3 slot 90.
[11743.332371] b43-phy0 debug: DMA ring 3 slot 92.
[11743.333341] b43-phy0 debug: DMA ring 3 slot 94.
[11743.401073] b43-phy0 debug: DMA ring 3 slot 96.
[11743.402389] b43-phy0 debug: DMA ring 3 slot 98.
[11743.470221] b43-phy0 debug: DMA ring 3 slot 100.
[11743.471055] b43-phy0 debug: DMA ring 3 slot 102.
[11743.501443] wlan0: authenticate with e0:91:53:56:f0:ad
[11743.509894] wlan0: capabilities/regulatory prevented using AP HT/VHT configuration, downgraded
[11743.519126] wlan0: send auth to e0:91:53:56:f0:ad (try 1/3)
[11743.520581] b43-phy0 debug: DMA ring 3 slot 104.
[11743.521910] wlan0: authenticated
[11743.522623] wlan0: associate with e0:91:53:56:f0:ad (try 1/3)
[11743.524228] b43-phy0 debug: DMA ring 3 slot 106.
[11743.525891] wlan0: RX AssocResp from e0:91:53:56:f0:ad (capab=0x431 status=0 aid=1)
[11743.526656] wlan0: associated
[11743.526709] IPv6: ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
[11743.526820] cfg80211: Calling CRDA for country: IT
[11743.531765] cfg80211: Regulatory domain changed to country: IT
[11743.531771] cfg80211: (start_freq - end_freq @ bandwidth), (max_antenna_gain, max_eirp)
[11743.531776] cfg80211: (2402000 KHz - 2482000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[11743.531780] cfg80211: (5170000 KHz - 5250000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[11743.531783] cfg80211: (5250000 KHz - 5330000 KHz @ 40000 KHz), (N/A, 2000 mBm)
[11743.531787] cfg80211: (5490000 KHz - 5710000 KHz @ 40000 KHz), (N/A, 2700 mBm)
[11743.531790] cfg80211: (57240000 KHz - 65880000 KHz @ 2160000 KHz), (N/A, 4000 mBm)
[11743.534289] b43-phy0 debug: DMA ring 1 slot 0.
[11743.543874] b43-phy0 debug: Using hardware based encryption for keyidx: 0, mac: e0:91:53:56:f0:ad
[11743.544297] b43-phy0 debug: Using hardware based encryption for keyidx: 1, mac: ff:ff:ff:ff:ff:ff
[11743.545278] b43-phy0 debug: DMA ring 1 slot 2.
[11744.476694] b43-phy0 debug: DMA ring 1 slot 4.
[11745.478607] b43-phy0 debug: DMA ring 1 slot 6.
[11747.385740] b43-phy0 debug: DMA ring 3 slot 108.
[11749.480238] b43-phy0 debug: DMA ring 1 slot 8.
[11750.375256] b43-phy0 debug: DMA ring 3 slot 110.
[11751.030677] b43-phy0 debug: DMA ring 1 slot 10.
[11753.372513] b43-phy0 debug: DMA ring 3 slot 112.
[11753.484286] b43-phy0 debug: DMA ring 1 slot 12.
[11756.369324] b43-phy0 debug: DMA ring 3 slot 114.
[11759.366577] b43-phy0 debug: DMA ring 3 slot 116.
[11762.363796] b43-phy0 debug: DMA ring 3 slot 118.
[11765.361195] b43-phy0 debug: DMA ring 3 slot 120.
[11768.358493] b43-phy0 debug: DMA ring 3 slot 122.
[11771.355663] b43-phy0 debug: DMA ring 3 slot 124.
[11774.352940] b43-phy0 debug: DMA ring 3 slot 126.
[11774.660317] b43-phy0 debug: DMA ring 3 slot 128.
[11774.690273] b43-phy0 debug: DMA ring 3 slot 130.
[11774.759303] b43-phy0 debug: DMA ring 3 slot 132.
[11774.828665] b43-phy0 debug: DMA ring 3 slot 134.
[11774.897446] b43-phy0 debug: DMA ring 3 slot 136.
[11774.966354] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 126
[11775.035204] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 140
[11775.104113] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 142
[11775.173070] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 144
[11775.241811] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 146
[11775.311094] b43-phy0 debug: Out of order TX status report on DMA ring 3. Expected 138, but got 148
--
-- isedev
--
-- isedev
On 03/13/2013 08:06 PM, ISE Development wrote:
Please do not drop the mailing lists from the Cc list. That should always be
true, but very important here as there are other developers that should see and
comment on this change.
> On Wednesday 13 Mar 2013 19:22:16 you wrote:
>> On 03/13/2013 05:31 PM, ISE Development wrote:
>>> Hi,
>>>
>>> The wireless connection keeps failing shortly after being established. Up to now, I've tracked it down to a DMA transmission sequence error in ring 3. Beyond that, I cannot say...
>>>
>>> Happy to provide further information and to test any potential fixes.
>>
>> Have you applied the patch that increases the number of RX ring slots? It is
>> commit ccae0e50c16a7f and was committed on Feb 17, 2013. That fixed a lot of
>> problems on the BCM4312. It seems that the firmware fails to check for overflow
>> of the RX ring and blindly corrupts things.
>>
>> Larry
>>
>>
>
> Hi,
>
> I am afraid so. The problem occurs with the latest code from linux-wireless (which includes the patch).
>
> The failures do not appear all that random, though. Some observations:
>
> 1. Failures (so far) only occur on two slots (138 and 208) on TX rings 1 and 3.
> 2. This can happen even without a preceeding 'invalid cookie' error.
> 3. There is _always_ an invalid cookie '0x0000' for ring 3 slot 138 when starting the driver and establishing the link.
> 4. After that, the invalid cookie has a different value every time.
> 5. The invalid cookie occurs far less frequently than the out of order error.
>
> This is an example of reported errors during a ~3.8Mb download at 640KB/s:
>
> [22803.075774] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 208, but got 210
> [22825.120354] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 208, but got 210
> [22826.752905] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 208, but got 210
> [22827.156976] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 138, but got 140
> [22827.302840] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 208, but got 210
> [22827.699721] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 138, but got 140
> [22827.855924] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 208, but got 210
> [22828.244330] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 138, but got 140
> [22828.385746] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 208, but got 210
> [22828.777014] b43-phy0 debug: TX-status contains invalid cookie: 0x2A0A
> [22828.781611] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 138, but got 140
> [22828.935206] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 208, but got 210
> [22829.330987] b43-phy0 debug: TX-status contains invalid cookie: 0xC90A
> [22829.334384] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 138, but got 140
> [22829.472203] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 208, but got 210
> [22829.856035] b43-phy0 debug: TX-status contains invalid cookie: 0x770C
> [22829.860472] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 138, but got 140
> [22830.416641] b43-phy0 debug: Out of order TX status report on DMA ring 1. Expected 138, but got 140
>
> I've hacked the driver to 'skip' one header and data frame if receiving an interrupt for the first slot + 2. It's not pretty and I have literally no idea if it will causes other problems, but it has allowed me to keep the Wifi connection up for a little over 3 hours now (as compared to the 45 seconds previously). It does not appear to be corrupting the data stream (checked by download large signed binaries and verifying the signature) and as far as my limited knowledge can tell, it should not be causing a memory leak.
>
> The patch is listed below, for reference. However, I do not claim that it is valid, safe or even reasonsable. It does provide me with much needed relief though.
>
> The diff is against the current head of linville/wireless-testing.git (d41d9c7419e3ac9c81841f43bbd7639dd0a5819e).
>
> diff --git a/drivers/net/wireless/b43/dma.c b/drivers/net/wireless/b43/dma.c
> index 38bc5a7..a3f787b 100644
> --- a/drivers/net/wireless/b43/dma.c
> +++ b/drivers/net/wireless/b43/dma.c
> @@ -1489,6 +1489,7 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
> struct b43_dmadesc_meta *meta;
> int slot, firstused;
> bool frame_succeed;
> + int skip;
>
> ring = parse_cookie(dev, status->cookie, &slot);
> if (unlikely(!ring))
> @@ -1501,15 +1502,24 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
> firstused = ring->current_slot - ring->used_slots + 1;
> if (firstused < 0)
> firstused = ring->nr_slots + firstused;
> +
> + skip = 0;
> if (unlikely(slot != firstused)) {
> /* This possibly is a firmware bug and will result in
> * malfunction, memory leaks and/or stall of DMA functionality. */
> b43dbg(dev->wl, "Out of order TX status report on DMA ring %d. "
> "Expected %d, but got %d\n",
> ring->index, firstused, slot);
> - return;
> + if(slot == firstused + 2) {
> + slot = firstused;
> + skip = 2;
> + } else {
> + return;
> + }
> }
>
> + b43dbg(dev->wl, "DMA ring %d slot %d.\n", ring->index, slot);
> +
> ops = ring->ops;
> while (1) {
> B43_WARN_ON(slot < 0 || slot >= ring->nr_slots);
> @@ -1522,6 +1532,7 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
> slot, firstused, ring->index);
> break;
> }
> +
> if (meta->skb) {
> struct b43_private_tx_info *priv_info =
> b43_get_priv_tx_info(IEEE80211_SKB_CB(meta->skb));
> @@ -1552,7 +1563,15 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
> * Call back to inform the ieee80211 subsystem about
> * the status of the transmission.
> */
> - frame_succeed = b43_fill_txstatus_report(dev, info, status);
> + if(!skip)
> + {
> + frame_succeed = b43_fill_txstatus_report(dev, info, status);
> + }
> + else
> + {
> + struct b43_txstatus fake = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
> + frame_succeed = b43_fill_txstatus_report(dev, info, &fake);
> + }
> #ifdef CONFIG_B43_DEBUG
> if (frame_succeed)
> ring->nr_succeed_tx_packets++;
> @@ -1580,12 +1599,14 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
> /* Everything unmapped and free'd. So it's not used anymore. */
> ring->used_slots--;
>
> - if (meta->is_last_fragment) {
> + if (meta->is_last_fragment && !skip) {
> /* This is the last scatter-gather
> * fragment of the frame. We are done. */
> break;
> }
> slot = next_slot(ring, slot);
> + if(skip > 0)
> + --skip;
> }
> if (ring->stopped) {
> B43_WARN_ON(free_slots(ring) < TX_SLOTS_PER_FRAME);
I am testing the patch on BCM4312 and other cards.
Larry
On Thursday 14 Mar 2013 21:39:56 Chris Vine wrote:
> On Thu, 14 Mar 2013 14:08:33 +0100
>
> ISE Development <[email protected]> wrote:
> > On Wednesday 13 Mar 2013 21:37:52 Larry Finger wrote:
> > > On 03/13/2013 08:06 PM, ISE Development wrote:
> > > > I've hacked the driver to 'skip' one header and data frame if
> > > > receiving an interrupt for the first slot + 2. It's not pretty
> > > > and I have literally no idea if it will causes other problems,
> > > > but it has allowed me to keep the Wifi connection up for a little
> > > > over 3 hours now (as compared to the 45 seconds previously). It
> > > > does not appear to be corrupting the data stream (checked by
> > > > download large signed binaries and verifying the signature) and
> > > > as far as my limited knowledge can tell, it should not be causing
> > > > a memory leak.
> > > >
> > > > The patch is listed below, for reference. However, I do not claim
> > > > that it is valid, safe or even reasonsable. It does provide me
> > > > with much needed relief though.
> > > >
> > > > The diff is against the current head of
> > > > linville/wireless-testing.git
> > > > (d41d9c7419e3ac9c81841f43bbd7639dd0a5819e).
> > >
> > > I am testing the patch on BCM4312 and other cards.
> > >
> > > Larry
> >
> > Here's a slighted cleaner version, with comments, in case you are
> > considering integrating it.
>
> [snip]
>
> It might look like a bit of a hack (probably one that broadcom have
> hidden away in their own wl driver if it is a firmware issue) but it is
> certainly effective.
>
> I have applied this patch to the 3.8.2 kernel and for the first time I
> get reliable performance from the bcm4312 card in my netbook using the
> b43 driver. I have banged about 5 GB through it and I am continuing to
> do so, but it is still up. I get I would say on average about one TX
> ring error (on ring 1 in the case of my card) per 500 MB of
> throughput and the frame skip resolves the problem for me.
>
> As this patch also avoid spamming the debug log with shed loads of
> error reports once an inconsistency arises, it also reveals that the
> inconsistency always begins with an expected status report of 138 and a
> report of 140 being received. Yours, however, may well be different,
> and this may be meaningless anyway.
>
> Chris
Same symptom here: starts with a miss on 138 (or less frequently on 208).
--
-- isedev
On Thu, 14 Mar 2013 14:08:33 +0100
ISE Development <[email protected]> wrote:
> On Wednesday 13 Mar 2013 21:37:52 Larry Finger wrote:
> > On 03/13/2013 08:06 PM, ISE Development wrote:
> >
> > >
> > > I've hacked the driver to 'skip' one header and data frame if
> > > receiving an interrupt for the first slot + 2. It's not pretty
> > > and I have literally no idea if it will causes other problems,
> > > but it has allowed me to keep the Wifi connection up for a little
> > > over 3 hours now (as compared to the 45 seconds previously). It
> > > does not appear to be corrupting the data stream (checked by
> > > download large signed binaries and verifying the signature) and
> > > as far as my limited knowledge can tell, it should not be causing
> > > a memory leak.
> > >
> > > The patch is listed below, for reference. However, I do not claim
> > > that it is valid, safe or even reasonsable. It does provide me
> > > with much needed relief though.
> > >
> > > The diff is against the current head of
> > > linville/wireless-testing.git
> > > (d41d9c7419e3ac9c81841f43bbd7639dd0a5819e).
> > >
> >
> > I am testing the patch on BCM4312 and other cards.
> >
> > Larry
> >
>
> Here's a slighted cleaner version, with comments, in case you are
> considering integrating it.
[snip]
It might look like a bit of a hack (probably one that broadcom have
hidden away in their own wl driver if it is a firmware issue) but it is
certainly effective.
I have applied this patch to the 3.8.2 kernel and for the first time I
get reliable performance from the bcm4312 card in my netbook using the
b43 driver. I have banged about 5 GB through it and I am continuing to
do so, but it is still up. I get I would say on average about one TX
ring error (on ring 1 in the case of my card) per 500 MB of
throughput and the frame skip resolves the problem for me.
As this patch also avoid spamming the debug log with shed loads of
error reports once an inconsistency arises, it also reveals that the
inconsistency always begins with an expected status report of 138 and a
report of 140 being received. Yours, however, may well be different,
and this may be meaningless anyway.
Chris
On Wednesday 13 Mar 2013 21:37:52 Larry Finger wrote:
> On 03/13/2013 08:06 PM, ISE Development wrote:
>
> >
> > I've hacked the driver to 'skip' one header and data frame if receiving an interrupt for the first slot + 2. It's not pretty and I have literally no idea if it will causes other problems, but it has allowed me to keep the Wifi connection up for a little over 3 hours now (as compared to the 45 seconds previously). It does not appear to be corrupting the data stream (checked by download large signed binaries and verifying the signature) and as far as my limited knowledge can tell, it should not be causing a memory leak.
> >
> > The patch is listed below, for reference. However, I do not claim that it is valid, safe or even reasonsable. It does provide me with much needed relief though.
> >
> > The diff is against the current head of linville/wireless-testing.git (d41d9c7419e3ac9c81841f43bbd7639dd0a5819e).
> >
>
> I am testing the patch on BCM4312 and other cards.
>
> Larry
>
Here's a slighted cleaner version, with comments, in case you are considering integrating it.
diff --git a/drivers/net/wireless/b43/dma.c b/drivers/net/wireless/b43/dma.c
index 38bc5a7..edc759d 100644
--- a/drivers/net/wireless/b43/dma.c
+++ b/drivers/net/wireless/b43/dma.c
@@ -1489,6 +1489,7 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
struct b43_dmadesc_meta *meta;
int slot, firstused;
bool frame_succeed;
+ int skip;
ring = parse_cookie(dev, status->cookie, &slot);
if (unlikely(!ring))
@@ -1501,13 +1502,22 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
firstused = ring->current_slot - ring->used_slots + 1;
if (firstused < 0)
firstused = ring->nr_slots + firstused;
+
+ skip = 0;
if (unlikely(slot != firstused)) {
/* This possibly is a firmware bug and will result in
* malfunction, memory leaks and/or stall of DMA functionality. */
b43dbg(dev->wl, "Out of order TX status report on DMA ring %d. "
"Expected %d, but got %d\n",
ring->index, firstused, slot);
- return;
+ if(slot == firstused + 2) {
+ /* If a single header/data pair was missed, skip over the first
+ * two slots in an attempt to recover. */
+ slot = firstused;
+ skip = 2;
+ } else {
+ return;
+ }
}
ops = ring->ops;
@@ -1522,6 +1532,7 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
slot, firstused, ring->index);
break;
}
+
if (meta->skb) {
struct b43_private_tx_info *priv_info =
b43_get_priv_tx_info(IEEE80211_SKB_CB(meta->skb));
@@ -1552,7 +1563,18 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
* Call back to inform the ieee80211 subsystem about
* the status of the transmission.
*/
- frame_succeed = b43_fill_txstatus_report(dev, info, status);
+ if(!skip)
+ {
+ frame_succeed = b43_fill_txstatus_report(dev, info, status);
+ }
+ else
+ {
+ /* When skipping over a missed TX status report, use a status
+ * structure which indicates that the frame was not sent
+ * (frame_count 0) and not acknowledged */
+ struct b43_txstatus fake = B43_FAKE_TXSTATUS;
+ frame_succeed = b43_fill_txstatus_report(dev, info, &fake);
+ }
#ifdef CONFIG_B43_DEBUG
if (frame_succeed)
ring->nr_succeed_tx_packets++;
@@ -1580,12 +1602,14 @@ void b43_dma_handle_txstatus(struct b43_wldev *dev,
/* Everything unmapped and free'd. So it's not used anymore. */
ring->used_slots--;
- if (meta->is_last_fragment) {
+ if (meta->is_last_fragment && !skip) {
/* This is the last scatter-gather
* fragment of the frame. We are done. */
break;
}
slot = next_slot(ring, slot);
+ if(skip > 0)
+ --skip;
}
if (ring->stopped) {
B43_WARN_ON(free_slots(ring) < TX_SLOTS_PER_FRAME);
diff --git a/drivers/net/wireless/b43/xmit.h b/drivers/net/wireless/b43/xmit.h
index 98d9074..eae730c 100644
--- a/drivers/net/wireless/b43/xmit.h
+++ b/drivers/net/wireless/b43/xmit.h
@@ -218,6 +218,9 @@ struct b43_txstatus {
u8 acked; /* Wireless ACK received */
};
+/* This needs to match the b43_txstatus structure above, all zeroed-out */
+#define B43_FAKE_TXSTATUS { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
+
/* txstatus supp_reason values */
enum {
B43_TXST_SUPP_NONE, /* Not suppressed */
--
-- isedev