Bug 659415
Summary: | 802.11n connections drop sporadically | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Joshua Boyd <boydjd> |
Component: | kernel | Assignee: | Stanislaw Gruszka <sgruszka> |
Status: | CLOSED DUPLICATE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
Severity: | medium | Docs Contact: | |
Priority: | low | ||
Version: | 13 | CC: | boydjd, dougsland, gansalmon, itamar, jonathan, kernel-maint, kmcmartin, linville, madhu.chinakonda, steinpilz, wey-yi.w.guy |
Target Milestone: | --- | ||
Target Release: | --- | ||
Hardware: | x86_64 | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2011-01-21 08:05:27 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Joshua Boyd
2010-12-02 18:37:26 UTC
[root@carbine ~]# iw dev wlan1 station dump Station c0:3f:0e:5e:91:0e (on wlan1) inactive time: 1750 ms rx bytes: 576228 rx packets: 931 tx bytes: 66884 tx packets: 541 signal: -45 dBm tx bitrate: 270.0 MBit/s MCS 15 40Mhz I am also seeing these errors now after doing a factory restore of my router: iwlagn 0000:03:00.0: PCI INT A disabled iwlagn: Intel(R) Wireless WiFi Link AGN driver for Linux, in-tree:d iwlagn: Copyright(c) 2003-2010 Intel Corporation iwlagn 0000:03:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17 iwlagn 0000:03:00.0: setting latency timer to 64 iwlagn 0000:03:00.0: Detected Intel Wireless WiFi Link 5300AGN REV=0x24 iwlagn 0000:03:00.0: Tunable channels: 13 802.11bg, 24 802.11a channels iwlagn 0000:03:00.0: irq 36 for MSI/MSI-X iwlagn 0000:03:00.0: firmware: requesting iwlwifi-5000-2.ucode iwlagn 0000:03:00.0: loaded firmware version 8.24.2.12 phy0: Selected rate control algorithm 'iwl-agn-rs' iwlagn 0000:03:00.0: iwl_tx_agg_start on ra = c0:3f:0e:5e:91:0e tid = 0 iwlagn 0000:03:00.0: iwl_tx_agg_start on ra = c0:3f:0e:5e:91:0e tid = 0 iwlagn 0000:03:00.0: low ack count detected, restart firmware iwlagn 0000:03:00.0: On demand firmware reload iwlagn 0000:03:00.0: Stopping AGG while state not ON or starting iwlagn 0000:03:00.0: queue number out of range: 0, must be 10 to 19 iwlagn 0000:03:00.0: iwl_tx_agg_start on ra = c0:3f:0e:5e:91:0e tid = 0 iwlagn 0000:03:00.0: iwl_tx_agg_start on ra = c0:3f:0e:5e:91:0e tid = 0 iwlagn 0000:03:00.0: low ack count detected, restart firmware iwlagn 0000:03:00.0: On demand firmware reload iwlagn 0000:03:00.0: Stopping AGG while state not ON or starting iwlagn 0000:03:00.0: queue number out of range: 0, must be 10 to 19 iwlagn 0000:03:00.0: PCI INT A disabled I think this is known problem reported here: https://bugzilla.kernel.org/show_bug.cgi?id=16691 , and here: http://bugzilla.intellinuxwireless.org/show_bug.cgi?id=2275 There is patch you can test: https://bugzilla.kernel.org/attachment.cgi?id=38632 it disable check for "low ack count" so firmware resets do not happen. However this is rather workaround not right fix. If you are not familiar with kernel compilation, I can prepare test kernel for you. Eee, above is wrong patch, this one is correct: https://bugzilla.kernel.org/attachment.cgi?id=38502 Joshua, please test this kernel http://koji.fedoraproject.org/koji/taskinfo?taskID=2697824. Does it solve the problem for you and not cause any other issues? I will check this evening and report back. I've got the kernel installed and am testing over an 802.11n connection now. No drops so far, but it's only been about 10 minutes. The machine hard locked and was unresponsive after about 5 minutes of testing network throughput over 802.11n against my local NFS server. I did not install the debug kernel. If you'd like me to do that and attempt to reproduce, let me know what steps I need to take to get you any usable data. I now have a whole bunch of these showing up without stressing the connection: [ 2993.938818] iwlagn 0000:03:00.0: BA scd_flow 0 does not match txq_id 10 This is on an N only connection. And, the same problem still occurs. So actually, this kernel is worse, because it'll lock up under high throughput. Re comment 8: I think this can be NFS client problem, can you reproduce using ftp or scp ? Regarding steps to get useful data. I'm planing to prepare some text on fedora wiki describing how to get debug info when kernel hung. Unfortunately there is no one easy way, as long you do not have serial cable connected to other computer. Re comment 9: I think someone report this is fixed in 2.6.36, I will provide you packages to test. I can reproduce using netperf as well, so it's not an NFS problem specifically. I wonder if a serial port on a docking station or the SmartBay would work. If so, I could try to find one of those, as both are available for my laptop. Serial console from docking station should work (if not, perhaps console is disabled in BIOS). You have to add "console=ttyS0,115200n8" to kernel boot parameters in /boot/grub/grub.conf , and configure same boud/parity on other machine. Here are 2.6.36 wireless drivers/stack + test patch compiled for 2.6.35.10-74.fc14 kernel: http://koji.fedoraproject.org/koji/taskinfo?taskID=2700598 . Please test. Short description how to use package is here: http://people.redhat.com/sgruszka/compact_wireless.html Regarding hard lock from comment 8 and 12, we probably have problem mac80211 in 2.6.35.10 kernel, it was reported in bug 667459. I'll prepare a new test kernel. Here is a new kernel to test: http://koji.fedoraproject.org/koji/taskinfo?taskID=2706682 Please give it a try. Joshua, any news about kernel from comment 15? Stanislaw, does the new kernel (comment 15) need the drivers from comment 13 installed as well? I have the same wireless card, an Intel 5300, in a Dell Latitude 6400, and I am running into the same problems as the OP when connecting to my 802.11n network. The kernel from comment 15 (64bit) alone does not completely fix it, and I'd have to downgrade the kernel to install the compat-wireless files. I am still getting lots of iwlagn 0000:0c:00.0: BA scd_flow 0 does not match txq_id 10 and a few iwlagn 0000:0c:00.0: Fail finding valid aggregation tid: 1 in /var/log/messages. It looks like the firmware resets iwlagn 0000:0c:00.0: low ack count detected, restart firmware might be gone, but I'll have to test it a bit more for that. (In reply to comment #17) > Stanislaw, does the new kernel (comment 15) need the drivers from comment 13 No, kernel include patched drivers, it generally was intended to fix kernel crash from bug 667459. > not completely fix it, and I'd have to downgrade the kernel to install the > compat-wireless files. > > I am still getting lots of > > iwlagn 0000:0c:00.0: BA scd_flow 0 does not match txq_id 10 > Intel wifi definitely have problems with 11n, this seems to be a firmware problems. You can check experimental firmware, see bug 648732 . If this not help, only advice is to disable 11n. I'm closing this bug as duplicate of firmware bug, since the real issue is in the firmware. *** This bug has been marked as a duplicate of bug 648732 *** |