SMP support committed to CVS

Live forum: http://rt2x00.serialmonkey.com/viewtopic.php?t=3675

Zi7

25-04-2007 05:51:19

I just committed to CVS a few patches that have resolved all SMP/preemption problems of legacy rt61 so far.

If i'm lucky, rt2500 might now support SMP kernels.
Or perhaps i just broke the whole thing. ;)

Guys, it's time to boot your SMP kernels again!
Tests and reports welcome...

Psycho Dad

25-04-2007 22:06:52

I dont have an SMP-System (a normal mobile AMD XP 2400+) but the new cvs-version of the rt2500-driver causes a kernel-panic for me.
It occurs while logging in.
I have an ASUS WL-107G running with Gentoo and kernel 2.6.18-gentoo-r7
cvs-Version from yesterday boots fine.
Log attached

[url124jc3b3]http://familieriedel.fa.funpic.de/panic.txt.gz[/url124jc3b3]

Zi7

25-04-2007 22:10:58

Ok, thanks for the report. )
I'll look into this today, however i don't have a rt2500 to test it myself. What i mean is that i'll need your kind help soon. ;)

Zi7

25-04-2007 23:23:48

Could you try this patch and report?

Psycho Dad

26-04-2007 12:33:14

Tested your patch against the cvs from today (including the change in mlme.c from 06.31).
The kernel panic is fixed now, either by the change in cvs from today or by the patch.txt.
The message

[codehvhwr6a8]
schleppi rt2500ERROR: empty Elem in MlmeQueue
[/codehvhwr6a8]

is already there, but has no influence on functionality.
(This message did not occur before the changes from yesterday).

Zi7

26-04-2007 15:39:45

Tested your patch against the cvs from today (including the change in mlme.c from 06.31).
The kernel panic is fixed now, either by the change in cvs from today or by the patch.txt.[/quote1q1rckj7]
Could you please try without the patch? I need to know whether it needs to be committed to CVS or not.

The message [code1q1rckj7]
schleppi rt2500ERROR: empty Elem in MlmeQueue
[/code1q1rckj7]
is already there, but has no influence on functionality.
(This message did not occur before the changes from yesterday).[/quote1q1rckj7]
Yes, this is just an artifact that will be corrected soon.

Psycho Dad

26-04-2007 18:32:48

With the current cvs without the patch the system boots fine.
So the change in mlme.c from today 06.31 fixes the issue, the patch is not neccessary.

easuter

01-06-2007 09:10:11

Did the SMP fixes also get commited to the RT61 tree by any chance?
Because I'm using an SMP kernel with the latest RT61 legacy drivers and have had no crashes.

Just asking to be sure...

Spy84464

01-06-2007 11:04:00

These changes have actually been committed to rt61 first, then ported to rt2500 ;).

easuter

13-06-2007 09:05:48

These changes have actually been committed to rt61 first, then ported to rt2500 ;).[/quote35g8ul19]

Cool! You guys rock!! D