From: Andi Kleen <andi@firstfloor.org>
To: linux-wireless@vger.kernel.org
Subject: rate instability in wireless stack
Date: Sun, 4 Jan 2009 17:46:49 +0100 [thread overview]
Message-ID: <20090104164649.GV496@one.firstfloor.org> (raw)
Hi,
I recently had to switch my desktop over to use wireless to connect
to my DSL line. Unfortunately it was not a entirely happy experience.
I have the following setup:
- One zd1211rw usb stick taped to a desk.
- A wireless antenna next to it, connected to PCI wireless cards
I'm using the zd1211rw stick, which works (54Mbit/s connection
with no apparent packet loss through two walls from a 16Mbit/s DSL line).
Unfortunately it has horrible queueing behaviour under load: when I have
multiple download streams going the connection gets extremly laggy.
The problem seems to be in the client here, when I do the same from a laptop
with Intel wireless there is no significant lag.
To avoid this problem I was trying to switch over to PCI cards.
I tried two different ones:
05:02.0 Network controller: RaLink RT2561/RT61 802.11g PCI
and
06:02.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-8185 IEEE 802.11a/b/g Wireless LAN Controller (rev 20)
in different machines. The reception should be about the same
or better than with the USB stick: they are connected to a real
antenna which is sitting next to the USB stick on a desk.
Unfortunately they both show the same problem. The connection
works, but quickly goes down to 1Mbit/s after setup and is then very slow.
I tried to manually lock it to 54Mbit/s, but then no data
is transferred. When it's set back to auto it quickly goes
down to 1Mbit/s again.
The USB stick happily stays at 54Mbit/s on its own (and works
well except for the lag behaviour). Both to the same AP.
Since I see this with two different chipsets I suppose it's
some general problem in the midlevel layer. There are no special
messages in dmesg. My laptops which have Intel wireless (ipw2200)
don't show this problem. So it must be in some part that
is used by the RTL and RT drivers, but not by Intel or zd1211rw.
Kernels I tried: 2.6.27, 2.6.28, current 2.6.29-git (now with Larry's
patch). All same behaviour.
Any ideas?
-Andi
--
ak@linux.intel.com
next reply other threads:[~2009-01-04 16:33 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2009-01-04 16:46 Andi Kleen [this message]
2009-01-04 16:36 ` rate instability in wireless stack Michael Buesch
2009-01-04 17:41 ` Andi Kleen
2009-01-05 0:29 ` Larry Finger
2009-01-05 2:53 ` Andi Kleen
2009-01-05 3:55 ` Larry Finger
2009-01-05 5:02 ` Andi Kleen
2009-01-05 4:59 ` Larry Finger
2009-01-05 5:48 ` Andi Kleen
2009-01-05 5:51 ` Larry Finger
2009-01-05 8:17 ` Johannes Berg
2009-01-05 12:54 ` Andi Kleen
2009-01-05 13:32 ` Felix Fietkau
2009-01-05 14:23 ` Andi Kleen
2009-01-05 14:41 ` Felix Fietkau
2009-01-05 15:24 ` Andi Kleen
2009-01-05 15:25 ` Felix Fietkau
2009-01-05 16:35 ` Andi Kleen
2009-01-05 16:36 ` Felix Fietkau
2009-01-05 17:11 ` Andi Kleen
2009-01-05 17:01 ` Felix Fietkau
2009-01-05 19:33 ` Hin-Tak Leung
2009-01-05 20:16 ` Larry Finger
2009-01-05 4:22 ` Larry Finger
-- strict thread matches above, loose matches on Subject: below --
2009-01-07 6:46 Martín Ernesto Barreyro
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20090104164649.GV496@one.firstfloor.org \
--to=andi@firstfloor.org \
--cc=linux-wireless@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).