Bug 317691

Summary:

orinoco driver sucks

Product:

[Fedora] Fedora

Reporter:

Matthew Miller <mattdm>

Component:

kernel

Assignee:

John W. Linville <linville>

Status:

CLOSED CURRENTRELEASE

QA Contact:

Fedora Extras Quality Assurance <extras-qa>

Severity:

low

Docs Contact:

Priority:

low

Version:

rawhide

CC:

linville, marmalodak

Target Milestone:

---

Target Release:

---

Hardware:

All

OS:

Linux

Whiteboard:

Fixed In Version:

kernel-2.6.23.1-49.fc8

Doc Type:

Bug Fix

Doc Text:

Story Points:

---

Clone Of:

Environment:

Last Closed:

2007-11-08 22:11:13 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:

Attachments:

Description	Flags
handle scan results better	none
(take 2) handle scan results better	none

Description Matthew Miller 2007-10-04 01:33:44 UTC

As discussed here:

https://www.redhat.com/archives/fedora-devel-list/2007-October/msg00065.html

On Tue, Oct 02, 2007 at 07:59:10AM -0400, Dan Williams wrote:
> I think there's a lot of room for improvement in the driver.  It's
> pretty racy with scan results.  First, reading of scan results should be
> decoupled from the actual process of scanning, and scan results should
> be aged before being dropped completely.  There's a fairly large window
> of time where scan results won't be available to the caller even though
> a successful scan could have happened a second ago.  The scan_inprogress
> flag shouldn't have any semantics leak out to userspace.

Having this improved would make me really happy. Thanks!

Comment 1 Dan Williams 2007-10-09 02:30:41 UTC

Created attachment 220531 [details]
handle scan results better

Candidate patch; will send to linux-wireless

Comment 2 Dan Williams 2007-10-09 02:32:59 UTC

Matt; any chance you could test the patch out?  If you'd rather get built
modules or a test kernel, I can poke linville to see if he can put one together.

Comment 3 Matthew Miller 2007-10-09 11:37:36 UTC

I have a busy week, so I can test it today or tomorrow if I have a built module
or kernel; by the end of the week building it myself.

Comment 4 John W. Linville 2007-10-10 18:34:30 UTC

Dan, are there some orinoco.h bits missing from the patch?  It seems to refer 
to orinoco_private fields that don't exist...?

Comment 5 Dan Williams 2007-10-11 03:53:31 UTC

Created attachment 223851 [details]
(take 2) handle scan results better

include the orinoco.h bits this time

Comment 6 John W. Linville 2007-10-11 18:27:11 UTC

Rawhide-based test kernels building now:

   http://koji.fedoraproject.org/koji/taskinfo?taskID=191632

Please post test results here...thanks!

Comment 7 Matthew Miller 2007-10-14 03:25:16 UTC

Much, much better. I can switch back and forth between networks with nm-applet
with no trouble, and iwlist scan hasn't failed on me yet. Anything more in
specific I should test?

Thanks!

Comment 8 Dan Williams 2007-10-14 17:24:55 UTC

Yeah, can you revert back to the older driver/kernel and verify that the problem
is still present one more time before I send the final bits upstream?

Thanks!

Comment 9 Matthew Miller 2007-10-14 18:31:36 UTC

Yeah, kernel-2.6.23-6.fc8 doesn't work.

Comment 10 Dan Williams 2007-11-06 16:11:42 UTC

Linville has queued this up for 2.6.25 FYI.  Note that both airo and orinoco
will apparently only show the connected AP in scan results when they are locked
on a particular BSSID, which is what wpa_supplicant tells them to do.  Not too
much we can do about that though.  So after you're connected for a while, you'll
only see one AP in the scan list until you disconnect.

Comment 11 Matthew Miller 2007-11-06 16:24:13 UTC

Might be nice, then, for nm-applet to have a "disconnect and rescan" menu
option. I suppose that's a different bug request. :)

Comment 12 John W. Linville 2007-11-08 22:11:13 UTC

This patch has been merged upstream.  It should appear in 
kernel-2.6.23.1-49.fc8 (or later) as well.