Larry, since you've been so helpful here's another one for you.  (A victim of your own success.)  Have you seen this issue before?  Seems to be a regression from 3.1 to 3.2?

Peter, did the 3.1.6, 3.1.7, 3.1.8 kernels work?  Those are using compat-wireless drivers from the 3.2-rc6 timeframe, so if this is really a regression in 3.2 it's either in a narrow window or something else is weird.

What is the PCI ID of that device?

(In reply to comment #2)
> What is the PCI ID of that device?

01:00.0 Network controller: Ralink corp. RT2860
        Subsystem: Ralink corp. Device 2790
        Physical Slot: eeepc-wifi
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
        Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
        Latency: 0, Cache Line Size: 32 bytes
        Interrupt: pin A routed to IRQ 19
        Region 0: Memory at f7ff0000 (32-bit, non-prefetchable) [size=64K]
        Capabilities: [40] Power Management version 3
                Flags: PMEClk- DSI- D1- D2- AuxCurrent=375mA PME(D0+,D1-,D2-,D3hot+,D3cold-)
                Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=0 PME+
        Capabilities: [50] MSI: Enable- Count=1/32 Maskable- 64bit+
                Address: 0000000000000000  Data: 0000
        Capabilities: [70] Express (v1) Endpoint, MSI 00
                DevCap: MaxPayload 128 bytes, PhantFunc 0, Latency L0s <128ns, L1 <2us
                        ExtTag- AttnBtn- AttnInd- PwrInd- RBE+ FLReset-
                DevCtl: Report errors: Correctable- Non-Fatal- Fatal- Unsupported-
                        RlxdOrd+ ExtTag- PhantFunc- AuxPwr- NoSnoop-
                        MaxPayload 128 bytes, MaxReadReq 512 bytes
                DevSta: CorrErr- UncorrErr- FatalErr- UnsuppReq- AuxPwr- TransPend-
                LnkCap: Port #0, Speed 2.5GT/s, Width x1, ASPM L0s L1, Latency L0 <512ns, L1 <64us
                        ClockPM+ Surprise- LLActRep- BwNot-
                LnkCtl: ASPM Disabled; RCB 128 bytes Disabled- Retrain- CommClk+
                        ExtSynch- ClockPM- AutWidDis- BWInt- AutBWInt-
                LnkSta: Speed 2.5GT/s, Width x1, TrErr- Train- SlotClk+ DLActive- BWMgmt- ABWMgmt-
        Capabilities: [100 v1] Advanced Error Reporting
                UESta:  DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
                UEMsk:  DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
                UESvrt: DLP+ SDES- TLP- FCP+ CmpltTO- CmpltAbrt- UnxCmplt- RxOF+ MalfTLP+ ECRC- UnsupReq- ACSViol-
                CESta:  RxErr- BadTLP- BadDLLP- Rollover- Timeout- NonFatalErr-
                CEMsk:  RxErr- BadTLP- BadDLLP- Rollover- Timeout- NonFatalErr+
                AERCap: First Error Pointer: 00, GenCap+ CGenEn- ChkCap+ ChkEn-
        Kernel driver in use: rt2800pci
        Kernel modules: rt2800pci

> Peter, did the 3.1.6, 3.1.7, 3.1.8 kernels work?  Those are using
> compat-wireless drivers from the 3.2-rc6 timeframe, so if this is really a
> regression in 3.2 it's either in a narrow window or something else is weird.

I've not tried them, the last 3.1.x kernel I had around was 3.1.2-1.fc16 from prior to "yum upgrade" to rawhide. I'll try one of the later ones now.

> Peter, did the 3.1.6, 3.1.7, 3.1.8 kernels work?  Those are using
> compat-wireless drivers from the 3.2-rc6 timeframe, so if this is really a
> regression in 3.2 it's either in a narrow window or something else is weird.

Just tested 3.1.8-2.fc16 and it works fine.

(In reply to comment #5)
> > Peter, did the 3.1.6, 3.1.7, 3.1.8 kernels work?  Those are using
> > compat-wireless drivers from the 3.2-rc6 timeframe, so if this is really a
> > regression in 3.2 it's either in a narrow window or something else is weird.
> 
> Just tested 3.1.8-2.fc16 and it works fine.

Hm.  Odd.  It might not be in the driver itself then, but something else in the 80211 stack?

Your posting in comment #3 shows everything about the PCI interface *EXCEPT* the PCI ID. If you run the command 'lspci -nn' and post the single line comparable to 

06:00.0 Network controller [0280]: Realtek Semiconductor Co., Ltd. RTL8192SE Wireless LAN Controller [10ec:8174] (rev 10)

The numbers in the place where mine has [10ec:8174] is what I need.

(In reply to comment #7)
> Your posting in comment #3 shows everything about the PCI interface *EXCEPT*
> the PCI ID. If you run the command 'lspci -nn' and post the single line
> comparable to 

Sorry, would have thought lspci -vv would have provided it

01:00.0 Network controller [0280]: Ralink corp. RT2860 [1814:0781]

ralink != realtek

(In reply to comment #9)
> ralink != realtek

I don't think anyone had those confused with each other... Larry was just pointing out an example of the lspci output he was expecting in comment #7

This looks is problem with PCI subsystem. I had already the report for 3945 where PCI bridge was disabled, hence kernel can not communicate with the wifi device:
https://lkml.org/lkml/2012/1/5/168 what looks very similar like this case.
Except that bug should be fixed in the final 3.2 release.

Peter, please attach full dmesg on from 3.1 and 3.2. There should be difference on PCI subsystem initialization, what should be good hint for finding where the problem is.

Could we see dmesg's requested in comment 11?

> Peter, please attach full dmesg on from 3.1 and 3.2. There should be difference
> on PCI subsystem initialization, what should be good hint for finding where the
> problem is.

Sorry, I was travelling for FUDCon and didn't have the device with me or remotely accessible. Attachments incoming.

Created attachment 555826 [details]
3.1.8 working wifi kernel

This is the output of the kernel that works with the wifi

Created attachment 555827 [details]
3.2.0 non working wifi kernel

The 3.2.0 non working kernel.

Hmm, unfortunately I do not see any difference between PCI initialization in both dmesg. Let's try to attack problem from different site. I'll prepered two patches that reverted latest 10 and 20 pci commits (there are overall 30 PCI commits  between 3.1 and 3.2). Please test these patches, respective kernel build are here (currently compiling):
http://koji.fedoraproject.org/koji/taskinfo?taskID=3711292
http://koji.fedoraproject.org/koji/taskinfo?taskID=3711296
This will narrow problem to 10 suspicious PCI patches. If problem is really in the PCI subsystem, but I'm not sure what else that could be (maybe ACPI ?).

neither pci10 or pci20 works (not the mainline 3.2.1-4)

I'm going to prepare test kernel, which revert all 30 PCI patches, but last 10 patches are not so easy to revert, hence it'll take some time. On the meantime can you check if one of that boot kernel parameters helps?

pcie_aspm=off
pcie_aspm=force
pci=nomsi
pci=noacpi
pci=nocsr
pci=use_csr

I'll test some of these over the next couple of days but if it helps the 3.3rc1 kernel is back to working. I rebooted between 3.1.8. 3.2.1-8 and the 3.3rc1 and the first and third work, the 3.2.1 doesn't

That's good news! Let's wait with testing boot parameters for now, we have various pending PCI, x86, ACPI patches in 3.2 queue backported from 3.3-rc1, which possibly could fix this problem. If 3.3.2 will not fix, we will need to work more on that bug.

(In reply to comment #20)
> which possibly could fix this problem. If 3.3.2 will not fix, we will need to
s/3.3.2/3.2.2/

I have an Eee Box B202 which is exhibiting the same problem.  I tried all of the kernel options suggested.  None of them help.  That last two are invalid options according to the kernel diagnostic messages.

I tried the latest 3.3rc1 kernels for F17 from koji.  But it panics on an Atom cpu like 3.2.1 did, apparently due to https://bugzilla.redhat.com/show_bug.cgi?id=784404.

It does not work with 3.2.2-1 kernel, the latest provided by fedora-updates.

lspci -nn reports:
03:06.0 Network controller [0280]: Ralink corp. RT2760 Wireless 802.11n 1T/2R Cardbus [1814:0701]

the errors on dmesg:
[   12.170095] rt2800pci 0000:03:06.0: PCI INT A -> GSI 21 (level, low) -> IRQ 21
[   12.180965] phy0 -> rt2800_init_eeprom: Error - Invalid RF chipset 0x0 detected.
[   12.180974] phy0 -> rt2x00lib_probe_dev: Error - Failed to allocate device.
[   12.181059] rt2800pci 0000:03:06.0: PCI INT A disabled

The device fails to boot...

(In reply to comment #24)
> It does not work with 3.2.2-1 kernel, the latest provided by fedora-updates.
> 
> lspci -nn reports:
> 03:06.0 Network controller [0280]: Ralink corp. RT2760 Wireless 802.11n 1T/2R
> Cardbus [1814:0701]
> 
> the errors on dmesg:
> [   12.170095] rt2800pci 0000:03:06.0: PCI INT A -> GSI 21 (level, low) -> IRQ
> 21
> [   12.180965] phy0 -> rt2800_init_eeprom: Error - Invalid RF chipset 0x0
> detected.
> [   12.180974] phy0 -> rt2x00lib_probe_dev: Error - Failed to allocate device.
> [   12.181059] rt2800pci 0000:03:06.0: PCI INT A disabled
> 
> The device fails to boot...

Forgot to add: this is a stock card, connected to the pci bus of my desktop (not laptop) machine, if this helps. Also, none of the mentioned boot options help for resolving the situation.

Same problem here on eeePC 1000h with rt2860 wifi
Works on Kernel 3.1.0.7, after upgrade to 3.2.2 wireless is broken.

Perhaps this is same problem (at least for some of you) like in bug 785393, wrong eeprom_93cx6.ko module used. This could be workaround by removing "backports" from /etc/depmod.d/dist.conf, and restart the system, please check that.

(In reply to comment #27)
> This could be workaround by removing
> "backports" from /etc/depmod.d/dist.conf, and restart the system, please check
> that.
Actually it can not, because we do not install vanilla wireless modules. Ehh, so workaround would be install own compiled compat-wireless-3.3-rc1-2 , I'll also prepare scratch build with c-w disabled.

Here is the 3.2.2 build with compat-wireless disabled:
http://koji.fedoraproject.org/koji/taskinfo?taskID=3745583

On my eee-pc wifi works with kernel-3.2.2-1.without_cw.fc16.i686

Created attachment 558425 [details]
dmesg.log on Eee PC 1000H and kernel-3.2.2-1.without_cw.fc16.i686

Experiencing same problem on eee901 with stock 3.2.2-1.fc16.i686 kernel; problem did not occur with 3.2.1-3, problem introduced somewhere between the two. Attaching dmesg log from working kernel. Interesting that I still see:

[   40.305981] phy0 -> rt2800pci_mcu_status: Error - MCU request failed, no response from hardware

and 

[   54.123487] eeepc_laptop: BIOS says wireless lan is unblocked, but the pci device is absent

yet the wireless DOES work on that kernel, while it doesn't (and I get the same 

phy0 -> rt2800_init_eeprom: Error - Invalid RF chipset 0x0 detected.
phy0 -> rt2x00lib_probe_dev: Error - Failed to allocate device.
rt2800pci 0000:01:00.0: PCI INT A disabled

messages on the 3.2.2-1 kernel as noted above).

Created attachment 558612 [details]
dmesg log from working 3.2.1-3 kernel

I can confirm it works with the provided kernel, without any additional boot options.

This issue will be resolved in the next f16 build by enabling the eeprom_93cx6 module from compat-wireless.

Please try the test kernels here:

http://koji.fedoraproject.org/koji/taskinfo?taskID=3749626

Does that restore the driver?

The provided kernel does not work for me, at least with the default settings.
Attaching:
* dmesg.log - the dmesg output. Things to note: 
- The kernel even sets the tainted bit to true while running hostapd, which is incorrect.
- A warn_slowpath appears.

* lspci_nn.log - the output of `lspci -nn`.

Created attachment 558690 [details]
dmesg output of gadnio, kernel 3.2.2-1.linville.1.fc16.x86_64

Created attachment 558691 [details]
`lspci -nn` output of the 3.2.2-1.linville.1.fc16.x86_64 kernel

3.2.2-1.linvulle.1.fc16.i686 worked fine on my eee901 host with no modifications. Didn't see any of the above errors in dmesg. Thanks a million, John!!!

Indeed dmesg from comment 38 looks different than eeprom messages. Looks like update to 3.3-rc1 code causing that. It interesting why this problem does not show up on 3.2 kernel. However we have some patches posed on rt2x00 mailing list that should address this problem, I'll apply the patches and prepare test build.

Created attachment 558848 [details]
rt2x00_fix_MCU_request_failures.patch

Patch from Jakub Kicinski fixing MCU requests failed. Kernel build with the patch is here:
http://koji.fedoraproject.org/koji/taskinfo?taskID=3751893

hristo, please test.

Created attachment 558921 [details]
/var/log/messages output of gadnio, kernel 3.2.2-1.rt2x00.fc16.x86_64, 4 consequtive boots

I am reluctant to say this kernel works flawlessly:
- The first time it failed to boot. dracut couldn't launch systemd or whatever the first thing init launches is. This was not registered in the logs;
- The second time it incorrectly renamed my two wireless cards (ath5k switched places with the rt2800pci card) and neither of them worked with hostapd. At 22:18:50 it even segfaulted. During that time, I applied updates to the SELinux policy;
- The third and fourth time I booted it worked flawlessly.

I don't know what the problem is and why it behaves so unreliably during reboots, given the fact I am not changing my setup or configuration by any means.

Attached is the complete /var/log/messages output of my tests. The first boot gets some avc errors.

As a side note, Virtuoso failed to start the third time I booted the kernel, but I cannot tell for sure whether this was a kernel's fault or virtuoso's. Fourth time the charm, for me.

Ok, but none of these problems seems to be rt2x00 driver related.

Can we see the output of the other people who had problems with eeePC then?
If none of the problems were kernel-related, I'd say let's include the fixes in the next version.

Also, the "MCU request failed" problem does not appear with the new kernel build (for the first time in several releases).

(In reply to comment #42)
> Created attachment 558848 [details]
> rt2x00_fix_MCU_request_failures.patch
> 
> Patch from Jakub Kicinski fixing MCU requests failed. Kernel build with the
> patch is here:
> http://koji.fedoraproject.org/koji/taskinfo?taskID=3751893
> 
> hristo, please test.

This kernel fixes the problem for me. Sorry for the delayed update

Josh, please apply patch from comment 42 as fix for this bug for F-16 and rawhide (author of original promises it will post them soon upstream).

(In reply to comment #47)
> Josh, please apply patch from comment 42 as fix for this bug for F-16 and
> rawhide (author of original promises it will post them soon upstream).

Done.  Thank you!

Created attachment 559359 [details]
dmesg.log on EeePC1000h with kernel-3.2.2-1.rt2x00.fc16.i686

wifi works well with kernel-3.2.2-1.rt2x00.fc16.i686 on my eeepc1000h

Just updated to 3.2.2.1 kernel and wifi broke on my Asus EEE PC 1000 SD with the Ralink RT2860 [1814:0781] wireless. :(

(In reply to comment #50)
> Just updated to 3.2.2.1 kernel and wifi broke on my Asus EEE PC 1000 SD with
> the Ralink RT2860 [1814:0781] wireless. :(

try http://koji.fedoraproject.org/koji/getfile?taskID=3751898&name=kernel-3.2.2-1.rt2x00.fc16.i686.rpm

(In reply to comment #51)
> (In reply to comment #50)
> > Just updated to 3.2.2.1 kernel and wifi broke on my Asus EEE PC 1000 SD with
> > the Ralink RT2860 [1814:0781] wireless. :(
> 
> try
> http://koji.fedoraproject.org/koji/getfile?taskID=3751898&name=kernel-3.2.2-1.rt2x00.fc16.i686.rpm

Perfect!  Thanks.

When can we expect to see this in the update stream?

The next kernel version is available for testing at
http://koji.fedoraproject.org/koji/buildinfo?buildID=297275

http://kojipkgs.fedoraproject.org/packages/kernel/3.2.3/2.fc16/i686/kernel-3.2.3-2.fc16.i686.rpm
 


On my Eee PC 1000h wifi working properly with that kernel

(In reply to comment #53)
> The next kernel version is available for testing at
> http://koji.fedoraproject.org/koji/buildinfo?buildID=297275
> 
> http://kojipkgs.fedoraproject.org/packages/kernel/3.2.3/2.fc16/i686/kernel-3.2.3-2.fc16.i686.rpm
> 
> 
> 
> On my Eee PC 1000h wifi working properly with that kernel

Same with my EEE PC 1000 sd.  Perfect.

Quick note: I seem to be experiencing this issue on ZOTAC H55-ITX WiFi [H55ITX-C-E]. 

# uname -a
Linux htpc.php.net 3.2.2-1.fc16.x86_64 #1 SMP Thu Jan 26 03:21:58 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux

System runs x86_64, can we have build done for all platforms to test?

I have not tried previous kernels - it's a fresh netinstall from scratch using Fedora repo's

It should be fixed in build 3.2.3-1 and later. Try this one

http://koji.fedoraproject.org/koji/buildinfo?buildID=297275

If somebody has still problems with rt2800pci driver on kernel-3.2.3-2, plese open a saperate bug report.