Page 3 of 50 FirstFirst 1234513 ... LastLast
Results 31 to 45 of 736

Thread: USB2 backport from 2.4.27 to 2.4.4/2.4.18/2.4.20 SUPPORT

  1. #31
    Join Date
    Jun 2004
    Posts
    191
    wbgolden, I too have got the ASOHOUSB and AGIGAUSB adapters both running on my 6.2 directivos. So it can be done, it was very simple for me... it's good you have serial bash setup otherwise it's a pain to do (unless you script it)

    For those that need it...
    In the /etc/hotplug/usb.map file you need to add these two lines to the
    device ax8817x section:
    Code:
        product 0b95 7720 Airlink ASOHOUSB
        product 0b95 1780 Airlink AGIGAUSB

  2. #32
    Join Date
    Aug 2004
    Posts
    4,087
    Quote Originally Posted by Jeff D
    wbgolden, I too have got the ASOHOUSB and AGIGAUSB adapters both running on my 6.2 directivos. So it can be done, it was very simple for me... it's good you have serial bash setup otherwise it's a pain to do (unless you script it)
    Glad you got it working. Did you sort out the mysteries you reported in post #20?

  3. #33
    Join Date
    Jun 2004
    Posts
    191
    Quote Originally Posted by Jamie
    Glad you got it working. Did you sort out the mysteries you reported in post #20?
    Nope, hotswapping the dlink adapter out kills something. I can swap the airlink adapters all day. If you see in my code section of post #20 that unplugging the airlink adapters correctly "frees" and dec's the use counter and and then plugging in inc's the use count. The dlink just wipes out the module competely when removed. There was the one odd time where the dlink adapter wasn't identified when plugged in, but I'm going to say that was user error.

    As you suggested... how many times are you really going to be hotswapping this thing?!?! Other than testing I don't know why either. So, I've just let it go. I have been looking at the links you provided, but I suspect I won't find time to get into it. The ramp up on linux development would be enough to turn me off to the whole thing...

    BTW, I haven't tried the GIGA at giga speeds, still just using it at the 100 speed.

  4. #34
    Join Date
    Jun 2004
    Posts
    191
    wbgolden, if it helps, I've documented my steps, I can send them to you if you want to see how someone else has done it...

  5. #35
    Join Date
    Nov 2004
    Posts
    10
    Quote Originally Posted by Jeff D
    wbgolden, if it helps, I've documented my steps, I can send them to you if you want to see how someone else has done it...
    That would be cool. I could compare what you've done to get them working with that I am doing to not get them working. I've PM'ed you my email address. Thanks
    Imagine what I could do with a Cowboy HatCam

  6. #36
    Join Date
    Jan 2004
    Posts
    110

    Prolific USB Host-to-Host cable

    Any idea what driver might work with this? Is it proprietary?

    http://www.prolific.com.tw/eng/Products.asp?ID=18
    Last edited by hxmiller; 07-04-2005 at 09:55 PM.

  7. #37
    Join Date
    Jan 2004
    Posts
    6

    transmit errors

    Hi,

    It looks like I am getting a significant number of transmit errors / dropped xmit packets using the usbnet driver on a hr10-250. I have usbcore.o, ehci-dummy.o and usbnet.o loaded, and nothing else other than the default modules. The network card is a netgear FA-120. Any ideas what is causing this or if there is a fix for it? As you can see below there are well over 2000 of them in 3 seconds while the box is pretty much idle from a network standpoint

    bash-2.02# ifconfig -a
    lo Link encap:Local Loopback
    inet addr:127.0.0.1 Bcast:0.0.0.0 Mask:255.0.0.0
    UP LOOPBACK RUNNING MTU:16436 Metric:1
    RX packets:1 errors:0 dropped:0 overruns:0 frame:0
    TX packets:0 errors:98 dropped:1 overruns:0 carrier:0 coll:0

    eth0 Link encap:Ethernet HWaddr 00:0F:B5:84:CCD
    inet addr:192.168.1.100 Bcast:192.168.1.255 Mask:255.255.255.0
    UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
    RX packets:31727 errors:0 dropped:0 overruns:0 frame:0
    TX packets:0 errors:80886700 dropped:62678 overruns:0 carrier:0 coll:0

    bash-2.02# date
    Sun Jul 10 17:31:16 UTC 2005
    bash-2.02# ifconfig -a
    lo Link encap:Local Loopback
    inet addr:127.0.0.1 Bcast:0.0.0.0 Mask:255.0.0.0
    UP LOOPBACK RUNNING MTU:16436 Metric:1
    RX packets:1 errors:0 dropped:0 overruns:0 frame:0
    TX packets:0 errors:98 dropped:1 overruns:0 carrier:0 coll:0

    eth0 Link encap:Ethernet HWaddr 00:0F:B5:84:CCD
    inet addr:192.168.1.100 Bcast:192.168.1.255 Mask:255.255.255.0
    UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
    RX packets:31754 errors:0 dropped:0 overruns:0 frame:0
    TX packets:0 errors:80888433 dropped:62697 overruns:0 carrier:0 coll:0

    bash-2.02#

    Thanks,
    James
    Last edited by ochs38578; 07-10-2005 at 01:47 PM. Reason: forgot to put in the network card type

  8. #38
    Join Date
    Apr 2003
    Posts
    2,402
    Quote Originally Posted by ochs38578
    Hi,

    It looks like I am getting a significant number of transmit errors / dropped xmit packets using the usbnet driver on a hr10-250. I have usbcore.o, ehci-dummy.o and usbnet.o loaded, and nothing else other than the default modules. The network card is a netgear FA-120. Any ideas what is causing this or if there is a fix for it? As you can see below there are well over 2000 of them in 3 seconds while the box is pretty much idle from a network standpoint

    ..snip..

    Thanks,
    James
    Might be a bad copy of ifconfig. I think I remember reading about one that falsely reports a lot of errors. Mine is version
    Code:
    [bash DTiVo(7): ~] $ifconfig --version
    net-tools 1.432
    ifconfig 1.29 (1997-09-23)
    and it reports a lot of errors also.

    ew

    EDIT:
    Just installed the newer verions and it reports no errors:
    Code:
    [bash DTiVo(7): ~] $ifconfig --version
    net-tools 1.60
    ifconfig 1.42 (2001-04-13)
    This is the one in usbobj2.4.27-20050619.tar.bz2.
    Last edited by eastwind; 07-10-2005 at 02:20 PM. Reason: New ifconfig info

  9. #39
    Join Date
    Aug 2004
    Posts
    59

    Off by 1 error?

    I just finished moving my systems over to pure gigabit, with the
    SMC switches.

    I suspect a problem with my Windows box as all ftp/http sessions
    to JF enabled systems hangs. If I lower the MTU on the servers
    to 8000 it works.

    I also noticed the high errors on the AirLink, but realized something
    else-
    eth0 Link encap:Ethernet HWaddr 00:12:0E:02:C7:66
    inet addr:199.254.167.50 Bcast:199.254.167.255 Mask:255.255.255.0
    UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
    RX packets:43394 errors:0 dropped:0 overruns:0 frame:0
    TX packets:0 errors:147761691 dropped:97897 overruns:0 carrier:0 coll:
    0

    Note the TX Packet count. It looks like the counters are 'shifted'

    Either way I still get my best performance using MTU=1500.
    Specs-
    HR10-250
    AirLink GB USB 2.0
    SMC 8508T switches (2)
    Tytool (HD Beta3)
    Broadcom 5700 series nic in Windows PC

    Dan

  10. #40
    Join Date
    Apr 2003
    Posts
    2,402
    Quote Originally Posted by JustDan
    I also noticed the high errors on the AirLink, but realized something
    else-Dan
    Looks like the same problem as ochs38578. Use the ifconfig from the usb drivers package (in nonusb/user). It is the counters in a different format than expected.

    ew

  11. #41
    Join Date
    Aug 2004
    Posts
    4,087
    Quote Originally Posted by hxmiller
    Any idea what driver might work with this? Is it proprietary?

    http://www.prolific.com.tw/eng/Products.asp?ID=18
    I see that BSD supports it through the cdce module, and all CDC (Communction Device Class) support in linux is in the usbnet driver. So, if anything, I'd expect in the usbnet module in linux to suport it. If it does not, it might just be a matter of adding the vendor/product id (that was the case for the ALI host-to-host cables). If you have such a cable, send me the vendor/product id's and we can try. The vendor has a data sheet, and bsd supports it, so it might not be hard to add linux support if it isn't already there.

  12. #42
    Join Date
    Aug 2004
    Posts
    4,087
    Quote Originally Posted by JustDan
    I just finished moving my systems over to pure gigabit, with the
    SMC switches. ...
    Are you sure your windows systems are using jumbo frames? I suspect not if you get better performance with an mtu of 1500. The device does support jumbo frames. I think with my intel GigE devices I had to muck with some settings in the device manager to get XP to use jumbo frames. Examine all the options for that device carefully in the windows device manager.

    As EW noted, the ifconfig issue is well known and has been covered before: use the ifconfig included in the backport package.

  13. #43
    Join Date
    Aug 2004
    Posts
    59
    Quote Originally Posted by Jamie
    Are you sure your windows systems are using jumbo frames? I suspect not if you get better performance with an mtu of 1500. The device does support jumbo frames. I think with my intel GigE devices I had to muck with some settings in the device manager to get XP to use jumbo frames. Examine all the options for that device carefully in the windows device manager.
    I believe it is using Jumbo Frames. The embedded Gb card didn't
    support them, so I dug out an older 3Com card I had which claims to.
    I plan to try an Intel card to make sure though.

    I should have been clearer, I was getting close to 3Mb/s with the
    USB200M, but around 2Mb/s using the Airlink and 8k jumbo frames.
    I only get around 1Mb/s using the 1500 MTU.

    Dan

  14. #44
    Join Date
    Aug 2004
    Posts
    4,087
    I've attached some tools that may help identify whether you are really getting jumbo frames end-to-end. These are the ping and tracepath compiled for a series 2 from the Fedora Core 3 iputils-20020927-16 sources.

    This ping supports a 'no fragment' hint that can be used to see if packets are getting fragmented. If you send too large a packet, instead of fragmenting it, you'll get a message back that the message was too long:
    Code:
    bash-2.02# ping -M do -s 8973 192.168.1.100
    PING 192.168.1.100 (192.168.1.100) 8973(9001) bytes of data.
    ping: local error: Message too long, mtu=9000
    This message is one byte too long for my mtu.

    Note that some versions of ping (for example, the windows versions) may use -f to indicate don't fragment. In the ping I'm posting here, -f means "flood ping" which you probably only want to use when testing for packet loss.

    Here's the -M options from the man page:
    Code:
           -M hint
                  Select Path MTU Discovery strategy.   hint  may  be  either  do
                  (prohibit fragmentation, even local one), want (do PMTU discov-
                  ery, fragment locally when packet size is large), or  dont  (do
                  not set DF flag).
    I also included tracepath, which does path mtu discovery and displays the lowest mtu along the path:
    Code:
    bash-2.02# tracepath 192.168.1.100
     1?: [LOCALHOST]     pmtu 9000
     1:  fedora.home.lan (192.168.1.100)                        2.709ms reached
         Resume: pmtu 9000 hops 1 back 1
    Last edited by Jamie; 07-11-2005 at 01:02 PM.

  15. #45
    Join Date
    Aug 2004
    Posts
    59

    Talking Very Cool, Thanks.

    It appeared I was half JF enabled.

    Using the tools provided I can confirm that I cannot use anything
    larger than 1500 to connect to my desktop. Oddly when using the
    windows ping with -f -s I can send JF up to 28 bytes under the
    configured MTU.

    So the NIC appears to be sending JFs, but cannot receive them.
    Since that doesn't make sense, I started turning off the protection
    tools required by work (McAfee, Windows Firewall, etc...)

    And the culprit is- Windows Firewall. With it off I am seeing over
    5Mb/s.

    Thanks for the tools and help. Time for some serious extration

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •