Bug 934 - firmware error when transfer mode b/g to a mode in ad-hoc mode
: firmware error when transfer mode b/g to a mode in ad-hoc mode
Status: VERIFIED FIXED
: IPW3945
firmware error
: 0.0.70
: All All
: P1 normal
Assigned To:
:
:
:
:
:
  Show dependency treegraph
 
Reported: 2006-03-05 22:25 by
Modified: 2006-04-25 20:19 (History)


Attachments
dmesg caught by "./load debug=0x43fff" (267.65 KB, text/plain)
2006-03-05 22:50, Wei Wang
Details
Modified ad-hoc channel selection to pick a valid channel (3.15 KB, patch)
2006-03-09 12:54, James Ketrenos
Details | Diff
fix this bug and more (62.30 KB, patch)
2006-03-17 16:02, Mohamed Abbas
Details | Diff
fix adhoc switch to a-band channel (3.65 KB, patch)
2006-04-12 14:33, Mohamed Abbas
Details | Diff
debug=0x43fff (57.44 KB, text/plain)
2006-04-25 18:35, zhanyuanhai
Details


Note

You need to log in before you can comment on or make changes to this bug.


Description From 2006-03-05 22:25:22
In ad-hoc mode, first set channel to 4 or another b or g channel. then set to 
a band, it will cause firmware error and the driver could never be loaded.

ipw3945: 3946ABG card ucode DOWNLOAD FAILED 
ipw3945: Unable to load firmware: -110
ipw3945: I ipw_net_hard_start_xmit Tx attempt while not associated.
ipw3945: 3946ABG card ucode DOWNLOAD FAILED 
ipw3945: Unable to load firmware: -110
ipw3945: 3946ABG card ucode DOWNLOAD FAILED 
ipw3945: Unable to load firmware: -110
ipw3945: I ipw_net_hard_start_xmit Tx attempt while not associated

Reproduce steps:
#./load
#iwconfig eth1 mode ad-hoc essid ipwadhoc channel 4
#iwconfig eth1 channel 48
#iwconfig  =>shows nothing for eth1.
#dmesg => find Microcode SW error detected.  Restarting. and Unable to load 
firmware: -110.

Then need to reload the driver manually.
------- Comment #1 From 2006-03-05 22:50:01 -------
Created an attachment (id=701) [details]
dmesg caught by "./load debug=0x43fff"
------- Comment #2 From 2006-03-06 07:56:31 -------
The SYSASSERT that happens repeatedly on ucode line 1211 is a result of trying 
to *modify* a non-existant entry in the ucode's STA table.  There's been some 
recent driver work in this regard ... I suggest assigning this bug to James or 
Mohammed.
------- Comment #3 From 2006-03-06 12:30:25 -------
I guess the problem we did reset the stations between resets which caused to 
insert the adhoc station in the  wrong slot. I am currently testing a patch 
for adhoc enhancement that will take care of this, you can assigned this to me
------- Comment #4 From 2006-03-09 12:54:31 -------
Created an attachment (id=705) [details]
Modified ad-hoc channel selection to pick a valid channel

This patch is actually from Mohamed, but he didn't attach it so I am :)
------- Comment #5 From 2006-03-09 13:38:39 -------
Bumped to P1.
------- Comment #6 From 2006-03-17 12:23:22 -------
(In reply to comment #4)
> Created an attachment (id=705) [edit] [details]
> Modified ad-hoc channel selection to pick a valid channel
> 
> This patch is actually from Mohamed, but he didn't attach it so I am :)

This attachment was actually for a different ad-hoc related problem seen when
switching from BG mode to A mode via 'iwpriv set_mode' when the first available
A channel was not a valid channel for an ad-hoc network.

------- Comment #7 From 2006-03-17 16:02:21 -------
Created an attachment (id=721) [details]
fix this bug and more

apply against 0.70. this patch include what the previous patch has and more
feature added to ad-hoc mode please try it out and let me know if does fix your
problem.
------- Comment #8 From 2006-03-17 16:03:13 -------
fix and add new ad-hoc enhancement
------- Comment #9 From 2006-03-27 13:33:29 -------
(In reply to comment #8)
> fix and add new ad-hoc enhancement

Can you provide a patch that *just* fixes the problem causing the microcode
error and a separate patch for the per station rate scaling (as a new bug filed
as a feature enhancement?)  We need to have the problem isolated in its code
change so we can apply that to a stable tree.  The per-station rate scaling
needs to go into our unstable tree.

Thanks,
James
------- Comment #10 From 2006-04-12 14:33:37 -------
Created an attachment (id=764) [details]
fix adhoc switch to a-band channel

apply against 0.74
------- Comment #11 From 2006-04-12 14:34:42 -------
this patch only take care of fixing the Firmware issues instead of the huge
patch 
------- Comment #12 From 2006-04-17 14:15:50 -------
Merged for 1.0.1.  Setting to Fixed.
------- Comment #13 From 2006-04-25 18:35:18 -------
Created an attachment (id=784) [details]
debug=0x43fff

Environment:
ipw3945-1.0.2 ieee-1.1.13,ucode-1.13,deamon-1.7.18
kernel-2.6.15.3
When switched the channel from 4 to 48 dmesg log showed:
"ipw_download_ucode 3945ABG card ucode DOWNLOAD is good" and no other FW error 

occurrd.
But ad-hoc host can't be set up with a-band though 48 is an active channel .
------- Comment #14 From 2006-04-25 20:19:18 -------
In comment #13, 
>But ad-hoc host can't be set up with a-band though 48 is an active channel .

I double checked 48 is not active in IBSS mode.

verify this bug.

In ad-hoc mode, there is no firmware error when transfer mode b/g to a mode. 
And the ad-hoc network can setup with a band such as band 149.