Karmic linux-2.6.31-10-generic (and later) breaks hostap

[QPrime]

Works correctly with linux-2.6.31-9-generic

*** uname -a ***
Linux ubuntu 2.6.31-10-generic #34-Ubuntu SMP Wed Sep 16 00:23:19 UTC 2009 i686 GNU/Linux

*** dmesg ***
[ 7.934484] hostap_pci 0000:02:09.0: enabling device (0014 -> 0016)
[ 7.934505] hostap_pci 0000:02:09.0: PCI INT A -> GSI 21 (level, low) -> IRQ 21
[ 7.936221] hostap_pci: Registered netdevice wifi0
[ 10.141676] hostap_pci: Initialization failed
[ 10.141680] hostap_pci: hardware initialization failed
[ 10.141717] hostap_pci 0000:02:09.0: PCI INT A disabled

*** lspci ***
00:00.0 Host bridge: Intel Corporation 82845 845 [Brookdale] Chipset Host Bridge (rev 03)
00:01.0 PCI bridge: Intel Corporation 82845 845 [Brookdale] Chipset AGP Bridge (rev 03)
00:1e.0 PCI bridge: Intel Corporation 82801 PCI Bridge (rev 12)
00:1f.0 ISA bridge: Intel Corporation 82801BA ISA Bridge (LPC) (rev 12)
00:1f.1 IDE interface: Intel Corporation 82801BA IDE U100 Controller (rev 12)
00:1f.2 USB Controller: Intel Corporation 82801BA/BAM USB Controller #1 (rev 12)
00:1f.3 SMBus: Intel Corporation 82801BA/BAM SMBus Controller (rev 12)
00:1f.4 USB Controller: Intel Corporation 82801BA/BAM USB Controller #1 (rev 12)
00:1f.5 Multimedia audio controller: Intel Corporation 82801BA/BAM AC'97 Audio Controller (rev 12)
01:00.0 VGA compatible controller: nVidia Corporation NV11DDR [GeForce2 MX200] (rev b2)
02:09.0 Network controller: Intersil Corporation Prism 2.5 Wavelan chipset (rev 01)
02:0b.0 Ethernet controller: Intel Corporation 82557/8/9/0/1 Ethernet Pro 100 (rev 0c)
02:0d.0 PCI bridge: Hint Corp HB6 Universal PCI-PCI bridge (non-transparent mode) (rev 15)
03:08.0 USB Controller: NEC Corporation USB (rev 41)
03:08.1 USB Controller: NEC Corporation USB (rev 41)
03:08.2 USB Controller: NEC Corporation USB 2.0 (rev 02)
03:0c.0 FireWire (IEEE 1394): Texas Instruments TSB43AB23 IEEE-1394a-2000 Controller (PHY/Link)

[arky (arky)]

Assigning to linux package

[PhracturedBlue (pblue)]

This bug seems to be very similar to #487978.
In my testing I found:
2.6.28-14 has working hostap for prism2.5 (but is unusable for me on karmic because the synaptic driver doesn't work with karmic's Xorg)
2.6.28-15 has broken hostap
2.6.31-9 has working hostap (and synaptic) and I'm using it now
2.6.31-10 (and all further versions) have broken hostap
2.6.32-7 has broken hostap

I got 2.6.31-9 from here (it was somewhat troublesome to find):
https://launchpad.net/ubuntu/karmic/i386/linux-image-2.6.31-9-386/2.6.31-9.29

The only difference I've noticed is that in the cases where hostap works I get this message:
ACPI: BIOS age (1997) fails cutoff (2000), acpi=force is required to enable ACPI
I don't know if it is a red-herring or not yet, but this was consistent in both 2.6.28 and 2.6.31. When I saw this message, hostap worked, when I didn't it didn't.

FYI, my log messages match those in #487978 on all failing kernels, specifically this:
[ 24.300028] wifi0: hfa384x_cmd: entry still in list? (entry=ed752680, type=0, res=0)
[ 24.300042] wifi0: hfa384x_cmd: command was not completed (res=0, entry=ed752680, type=0, cmd=0x0021, param0=0xfd0b, EVSTAT=8010 INTEN=0010)
[ 24.300048] wifi0: interrupt delivery does not seem to work
[ 24.300054] wifi0: hfa384x_get_rid: CMDCODE_ACCESS failed (res=-110, rid=fd0b, len=8)
[ 24.300058] Could not get RID for component NIC
[ 24.300063] hostap_pci: Initialization failed
[ 24.300068] hostap_pci: hardware initialization failed

[PhracturedBlue (pblue)]

Update: It looks like I was wrong about the ACPI thing. it is in my dmesg in both cases, but only displayed to the screen in working kernels. attaching dmesg for 2.6.31-9 (working) and 2.6.31-16 (broken)

[robyuan (bertoro1)]

I have same problem with a Thinkpad T30.
Kernel 2.6.31-9 works, other kernels not.

Hoping broken hostap will be solved in new kernel release.

[ktp420 (ktp420)]

Same issue here...going to have to try the -9 kernel to get wireless and system working.

[mabawsa (mabawsa)]

This bug affects me running kernel -18 on an thinkpad X23

[pullmandave (dpitts)]

I'm running on an IBM A30p and it affects me. In fact, loading the hostap_pci module causes my system to "freeze" up with a CPU hung message.

[souplin (klage)]

Can confirm this bug. My hostap_pci card doesn't work with the upcoming Ubuntu Lucid Version.
2.6.32-16-generic #25-Ubuntu SMP Tue Mar 9 16:33:52 UTC 2010 i686 GNU/Linux

Hopefully it'll be fixed soon

[ 7558.572064] hostap_pci 0000:05:01.0: PCI INT A -> GSI 22 (level, low) -> IRQ 22
[ 7558.574541] hostap_pci: Registered netdevice wifi0
[ 7558.574553] wifi0: Original COR value: 0x0
[ 7558.773796] prism2_hw_init: initialized in 196 ms
[ 7558.774665] wifi0: Interrupt, but dev not configured
[ 7558.774672] wifi0: Interrupt, but dev not configured
[ 7558.774677] wifi0: Interrupt, but dev not configured
[ 7558.774682] wifi0: Interrupt, but dev not configured
[ 7558.774687] wifi0: Interrupt, but dev not configured
[ 7558.774692] wifi0: Interrupt, but dev not configured
[ 7558.774697] wifi0: Interrupt, but dev not configured
[ 7558.774702] wifi0: Interrupt, but dev not configured
[ 7558.774707] wifi0: Interrupt, but dev not configured
[ 7558.774712] wifi0: Interrupt, but dev not configured
[ 7560.772026] wifi0: hfa384x_cmd: entry still in list? (entry=f0ad1940, type=0, res=0)
[ 7560.772037] wifi0: hfa384x_cmd: command was not completed (res=0, entry=f0ad1940, type=0, cmd=0x0021, param0=0xfd0b, EVSTAT=8010 INTEN=0010)
[ 7560.772041] wifi0: interrupt delivery does not seem to work
[ 7560.772048] wifi0: hfa384x_get_rid: CMDCODE_ACCESS failed (res=-110, rid=fd0b, len=8)
[ 7560.772056] Could not get RID for component NIC
[ 7560.772059] hostap_pci: Initialization failed
[ 7560.772063] hostap_pci: hardware initialization failed
[ 7560.772097] hostap_pci 0000:05:01.0: PCI INT A disabled
[ 7562.772757] wifi0: hfa384x_cmd: entry still in list? (entry=f0ad1280, type=0, res=0)
[ 7562.772765] wifi0: hfa384x_cmd: command was not completed (res=0, entry=f0ad1280, type=0, cmd=0x0021, param0=0xfdc6, EVSTAT=8010 INTEN=0000)
[ 7562.772767] wifi0: interrupt delivery does not seem to work
[ 7562.772772] wifi0: hfa384x_get_rid: CMDCODE_ACCESS failed (res=-110, rid=fdc6, len=12)

[koshimazaki (kozza2)]

I confirm as well, tested today on 2.6.31-21-generic (karmic-proposed) and 2.6.31-20. I tried the proposed kernel because it contains a fix for another hostap-related bug, #444801, but it seems unrelated to this bug.

Specifically, my log contains all the messages mentioned by Geoffrey Hausheer in post #2 (apart from hostap_pci: 'hardware initialization failed') and I also get lots of 'wifi0: Interrupt, but dev not configured' lines.

After some searching, I found a possible clue in the changelog for 2.6.31-10, https://launchpad.net/ubuntu/+source/linux/2.6.31-10.30 . It says:
* SAUCE: wireless: hostap, fix oops due to early probing interrupt
The diff for that version (available from the above mentioned page) reveals that the following was added to hostap_hw.c:

+ /* Detect early interrupt before driver is fully configued */
+ if (!dev->base_addr) {
+ if (net_ratelimit()) {
+ printk(KERN_DEBUG "%s: Interrupt, but dev not configured\n",
+ dev->name);
+ }
+ return IRQ_HANDLED;
+ }
+

Note the printk message. Apparently the code was added to fix bug #254837. This fix was introduced in 2.6.31-10.30 and 2.6.28-15.52 which seems consistent with Geoffrey's report about these being the versions where hostap broke.
I suspect that this fix is what broke hostap for us and I'm tempted to build a kernel with these lines removed. I realise they where put there for a reason but I've used the hostap driver with Prism 2.5 successfully in the past, before they were introduced, and I can't use it now.

Note that I am not at all familiar with the linux or ubuntu kernel development and only have a vague idea of what I'm talking about so excuse me if it's all wrong, or obvious.

[koshimazaki (kozza2)]

Reverting the change that I mentioned in my previous post gave me back working wireless.

My steps were:
* download ubuntu kernel source for version 2.6.31-20.58 (current karmic kernel)
* revert change to hostap introduced in version 2.6.31-10.30: "SAUCE: wireless: hostap, fix oops due to early probing interrupt"
* apply change to hostap introduced in version 2.6.31-21.59: "hostap: Revert a toxic part of the conversion to net_device_ops"
* compile hostap driver

So far, with these changes, my wireless card seems to work without problems as an access point.

[QPrime (mwells)]

Maverick linux 2.6.35-12.17 and (possibly?) linux-backports-modules 2.6.35-11.3 appears to restore out-of-the-box functionally for this wireless adapter using card firmware v1.1.1 (primary) and v1.8.2 (secondary)

[souplin (klage)]

I'm running 2.6.35-14-generic #19~lucid1-Ubuntu SMP Tue Aug 3 08:57:49 UTC 2010 i686 GNU/Linux
without installing any linux-backports-modules

[ 154.422842] wifi0: PRI: id=0x15 v1.1.1
[ 154.422979] wifi0: STA: id=0x1f v1.7.4
[ 154.426361] wifi0: Intersil Prism2.5 PCI: mem=0xf6ffe000, irq=22

hostap_pci is loading without any problems.

[QPrime (mwells)]

Thanks souplin. Closing this.