Latest working kernel version: 2.6.26 Earliest failing kernel version: 2.6.27 Distribution: Archlinux Hardware Environment: Netgear WG111v2 Problem Description: If the upload rate reaches a threshold ( > 2-3 KB/s) the connection hangs and I can't reconnect. Only after a rrmod/modprobe of rtl8187 I can connect. I installed also the lastest compat-wireless (updated to gen 7, 2009) but the problem remains.
What are the USB ID's for this device? The output of lsusb will show that info.
2-3KB/s is quite sad - are you sure about the "Latest working kernel version: 2.6.26" part - did it work better before?
I used the 2.6.26 for a long time and it worked. The problem appeared after theupgrade to 2.6.27. Also 2.6.28 is affected (compiled by me). I'll downgrade the kernel in these days to test the behaviour. $ lsusb | grep WG Bus 002 Device 005: ID 0846:6a00 NetGear, Inc. WG111 WiFi (v2)
Ah, your device is an RTL8187, not an RTL8187B. I believe the developers all have RTL8187B's, In a test made since you posted the bug, I ran a Ktorrent test and was uploading 16 KB/s without trouble. I'll review the differences between the 8187 and 8187B code. If there are patches, could you test them?
Of course I can do it ;)
Created attachment 19800 [details] Patch 1 of 3
Created attachment 19801 [details] Patch 2 of 3
Created attachment 19802 [details] Patch 3 of 3
(In reply to comment #8) > Created an attachment (id=19802) [details] > Patch 3 of 3 > Am a bit worry about the fact that this patch isn't symmetrical (with rtl8187b)? On the rtl8187b, I regularly do uploads (to sourceforge) and 20-40kB/s is quite normal and limited by my upstream bandwidth. 2-3kB/s is quite sad.
As Herton noted, that patch may not be correct; however, without it there is no reporting back of the TX count. The asymmetry is due to a status endpoint in the 8187B that is probably not present in the 8187. Using tcpperf, I can push 3.2 Mb/s through my RTL8187B to another host on my LAN.
Davide, Patch 3 is no good. Ignore it. What does your device show with the command? lsusb -v -d0846:6a00 Thanks,
[dvd@abulafia ~]$ sudo lsusb -v -d0846:6a00 Bus 002 Device 005: ID 0846:6a00 NetGear, Inc. WG111 WiFi (v2) Device Descriptor: bLength 18 bDescriptorType 1 bcdUSB 2.00 bDeviceClass 0 (Defined at Interface level) bDeviceSubClass 0 bDeviceProtocol 0 bMaxPacketSize0 64 idVendor 0x0846 NetGear, Inc. idProduct 0x6a00 WG111 WiFi (v2) bcdDevice 1.00 iManufacturer 1 NETGEAR WG111v2 iProduct 2 NETGEAR WG111v2 iSerial 3 001B2F91F8F1 bNumConfigurations 1 Configuration Descriptor: bLength 9 bDescriptorType 2 wTotalLength 39 bNumInterfaces 1 bConfigurationValue 1 iConfiguration 4 Wireless Network Card bmAttributes 0x80 (Bus Powered) MaxPower 500mA Interface Descriptor: bLength 9 bDescriptorType 4 bInterfaceNumber 0 bAlternateSetting 0 bNumEndpoints 3 bInterfaceClass 0 (Defined at Interface level) bInterfaceSubClass 0 bInterfaceProtocol 0 iInterface 5 Bulk-IN,Bulk-OUT,Bulk-OUT Endpoint Descriptor: bLength 7 bDescriptorType 5 bEndpointAddress 0x81 EP 1 IN bmAttributes 2 Transfer Type Bulk Synch Type None Usage Type Data wMaxPacketSize 0x0200 1x 512 bytes bInterval 0 Endpoint Descriptor: bLength 7 bDescriptorType 5 bEndpointAddress 0x02 EP 2 OUT bmAttributes 2 Transfer Type Bulk Synch Type None Usage Type Data wMaxPacketSize 0x0200 1x 512 bytes bInterval 0 Endpoint Descriptor: bLength 7 bDescriptorType 5 bEndpointAddress 0x03 EP 3 OUT bmAttributes 2 Transfer Type Bulk Synch Type None Usage Type Data wMaxPacketSize 0x0200 1x 512 bytes bInterval 0 Device Qualifier (for other device speed): bLength 10 bDescriptorType 6 bcdUSB 2.00 bDeviceClass 0 (Defined at Interface level) bDeviceSubClass 0 bDeviceProtocol 0 bMaxPacketSize0 64 bNumConfigurations 1 Device Status: 0x0000 (Bus Powered) This is the output. (OT: have I to report the outputs as attachments or is ok like this?) Ok, I'm going to apply patch 1 and 2 to a vanilla 2.6.28.
Inline like this is fine. I attached the patches because you would need them as files.
Ok, I git-clone from wireless-testing, applied both patches and I compiled. Now upload works but it's capped to ~100KB/s (ftp in a LAN). Browsing is very variable: sometimes sloooow, sometimes ok. Opening emesene or amule causes the hang and needs a rmmod/modprobe!!
If possible, I would like you to back the patches back out one at a time to see if either, or both helped your upload, or if it was something else in wireless-testing.
Argh, there is no need to git-clone the whole wireless-testing kernel - if your (possibly vendor) kernel and headers are in place, and most bits are as kernel modules, it is a bit quicker for trying out new changes with compat-wireless, which can swap modules with newer versions a bit quicker. Now you have git-clone'ed wireless-testing, you might as well keep it and use it to update compat-wireless...
Just looked a the lsusb in detail - it only has 3 endpoints and quite different from 8187b. (in the back of my head I knew this as one of the main differences, but seeing it feels different...)
In addition, the current driver only uses 1 of the OUT endpoints. The vendor driver distinguishes them with ep 2 for low- and ep 3 for normal-priority transmits. I have a patch to put the management frames on ep2 and other frames on ep3, but will wait until I get a chance to test here. I bought a Netgear WG111v2 on E-bay yesterday. I should have it within a week.
I used the git because the patch was for the wireless-testing folder and it isn't present in the compat-wireless package. Sorry for the mistake but I don't knew where to apply the patch. I'll try the patch one at a time this evening if it's possible (It's 9 pm here!).
It was not a mistake. What Hin-Tak was saying is that you could have just downloaded compat-wireless. Any patch for wireless-testing will also apply to that code. On the other hand, he just asked about a "bug" that turned out to be an incompatibility with his underlying kernel. Now that you have the git tree, it is trivial to keep up with wireless-testing.
Probably confused by campat-wireless git repository vs compat-wireless tar balls. compat-wireless git repository contains a bunch of scripts and some patches which massages wireless-testing into a form suitable for building against released kernel headers - the massaged result is snapshott'ed daily into tar balls. i.e. you can apply the patch against the daily snapshot compat-wireless tar balls; or since you already have wireless-testing git and compat-wireless git, you can generate the snapshot by running inside compat-wireless git, export GIT_TREE=/where/wireless-testing-git-is ./script/admin-refresh.sh then apply patches to the resulting directory.
I'm sorry, but at this time I'm VERY busy and I can't do any test.. This may be the same bug: http://bbs.archlinux.org/viewtopic.php?id=63274
(In reply to comment #22) > I'm sorry, but at this time I'm VERY busy and I can't do any test.. > This may be the same bug: http://bbs.archlinux.org/viewtopic.php?id=63274 That report is unrelated - it is about the Realtek gigabit (wired) ethernet driver, not any of the wireless drivers. If you can find the time to try things out eventually and report back it would be good.
Created attachment 19939 [details] Patch for V2.6.28 and before This patch, which is for kernels 2.6.28 and before, adds a short packet following transmission of the data packet. It helped my RTL8187 device. The problem showed up when doing a git pull on a tree located on an NFS volume.
Created attachment 19940 [details] Patch for V2.6.29-rcX This patch, which is for 2.6.29-rcX, adds a short packet after the data transaction.
I am just finally getting around to debugging my RTL8187B from a Il1 netbook. It has the same problem exhibited above. The most recent snapshot compat-wireles-1-24 does seem to be helping a lot with reliability. I am definitely seeing problems with rate control, and there seems to be problems with small packets transfers like UDP dns lookups. I can get you any information that will help in debugging. Just ask.
Created attachment 19988 [details] Patch for power setting on RTL8187L Please try this patch on top of current wireless-testing. Larry
Created attachment 19991 [details] Patch for power setting on RTL8187L The original version of this patch will work for the Netgear WG111V2, but would not help all versions of the RTL8187B. The new one will help will all units.
This definitely has my rlt8187b running the best it has been. Been connected for a couple of hours now without dropouts. The range of the chipset is still a bit limited but that may just be the hardware. Thanks
commit eb83bbf57429ab80f49b413e3e44d3b19c3fdc5a