On Tue, 20 Apr 2004, Matt Brown wrote:
> I have been testing the latest release of the driver with 3 cards
> - Orinoco Gold Card PCMCIA (Hermes)
> - EnGenius NL-2511CD+ EXT2 PCMCIA (Prism 2.5)
> - EnGenius 2511MP+ miniPCI (Prism 2.5)
>
> On the whole the driver works really well, having monitor mode and
> scanning support is really nice. Thanks for all the hard work that has
> been put into this release.
>
> These cards were tested in a soekris (www.soekris.net) net 4521 using an
> Orinoco AP-500. The miniPCI card was configured as eth2, the orinoco
> PCMCIA card as eth3 and the prism 2.5 PCMCIA card as eth4. Logs of the
> insertion of each card can be found at the end of the email.
>
> - Scanning support appears to work perfectly on all of the cards.
> - Throughput is reasonable with all of the cards
> - There were no problems with general configuration tasks such as
> connecting to an AP, configuring ESSID etc.
> - In monitor mode using the Prism 2.5 based cards I didn't see any data
> frames, only beacons and acknowledgements, perhaps I did something
> wrong? The orinoco card was able to see both data, beacons and
> acknowledgement frames (despite the errors described below).
I'm not aware of this problem. Maybe it's specific to 1.4.9 firmware.
> I encountered the following problems:
>
> 1) Configuring an interface that uses a Prism based card results in
> "Error -110 setting multicast list." (also reported by Stephen Scholz)
I have seen it. Prism cards are not quite ready for 5-10 milliseconds
after port 0 is enabled. That explains may failures immediately after
the interface is brought up. It needs to be fixed.
> 2) Enabling Monitor Mode on Prism 2.5 PCMCIA card gives a BAP offset
> error, followed by a funny link status error
>
> root-X6RhF4x9mZo@xxxxxxxxxxxxxxxx:~# iwconfig eth4 mode monitor
> hermes @ IO 0x100: BAP offset error: reg=0x8000 offset=0x4
> eth4: New link status: UNKNOWN (c325)
This looks like another timeout. The return value of hermes_bap_pread()
in __orinoco_ev_info() is not always checked. Needs to be fixed.
> 3) Removing Prism 2.5 PCMCIA card results in lots (approx 50)
> orinoco_lock() called with hw_unavailable (dev=c3255800)
> messages followed by a single
> dv_mc_discard: multicast leakage! dmi_users=1
Messages about orinoco_lock() come from the interrupt handler. They are
harmless. Perhaps they should be disabled.
"multicast leakage" is likely caused by the fact that the multicast list
could not be set on startup.
> 4) Orinoco card in monitor mode results in lots of the following
> messages
>
> root-X6RhF4x9mZo@xxxxxxxxxxxxxxxx:~# iwconfig eth3 mode monitor
> eth3: New link status: Connected (0001)
> eth3: IRQ handler is looping too much! Resetting.
Monitor mode is not working correctly in 8.72 firmware. Perhaps it should
be disabled for this firmware version. I still hope to find a less
radical solution. Disabling fast channel change helps somewhat, but not
much - the card still goes crazy after some time.
> I'm not sure if monitor mode is meant to work correctly with Orinoco
> cards or not? Ignoring the errors above which sometimes stop after a few
> seconds I can tcpdump on the orinoco card with sensible looking results.
Monitor mode is working fine with 6.16 firmware.
> Hope this helps. If there are any other scenarios that I can help
> testing or you have questions about these results please let me know.
Yes, it's helpful. I was aware of most problems you are describing but
didn't consider them serious enough to fix. You report helps define
priorities for further work.
--
Regards,
Pavel Roskin
-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
|
