USB2 backport from 2.4.27 to 2.4.4/2.4.18/2.4.20 SUPPORT

[Jamie]

Originally Posted by pmoore

Ok, I ended up running into a reboot loop. Console showed it was rebooting as the adapter was starting up. I now have a monte setup - 7.2.2 to a stock kernel, using stock drivers. Same thing is happening though. If I unplug the adapter, TiVo comes up, and I can plug the adapter in. It will eventually reboot again though.

Just as a sanity check, uname -a returns:

Linux (none) 2.4.20 #1 Tue Dec 2 20:54:47 PST 2008 mips unknown

Have I got the correct kernel? Any thoughts? I'm starting to consider the possibility that my adapter has a bad component...

I guess I'd try a completely stock image before I'd be ready to blame the hardware. If you've made any modifications, there is a chance you left an old driver in place somewhere.

Here is the "Linux version" string from the 9.3.2a Series2 kernel:

Linux version 2.4.20 (build@buildmaster83) (gcc version 3.3.4) #1 Tue Dec 2 20:54:47 PST 2008

You get it from /proc/version. It matches the time stamp of yours, so I think it is right. So perhaps you have one or more non-stock modules still in place? The kernel/messages logs might give some clues.

[pmoore]

So, I extracted the entire (?) root file system from GZhpk-Series2 and GZcore. I restored my "monte" rc.sysinit. Rebooted, and same thing. Kernel logs are attached. Always, prior to system crash, are hub.c log entries. You can find these by searching for the first Jan 2 entry in the sequence, and then looking at the line prior. The kernel.txt file encompasses 3 reboots. The Okernel.txt has 1 (at the very end).

This is the line that appears frequently just prior to the reboot (I don't see it every time, but often do):

Feb 9 03:32:15 (none) kernel: hub.c: new USB device 00:0d.0-2, assigned address 3

[Jamie]

I see a bunch of stuff in the logs that makes me think this isn't a completely stock image. For example,

sp="true" must be second pass
...
 /lib/modules/ehci-hcd.o: init_module: No such device
...
/gotomydvr/etc/rc.gotomydvr: .: filename argument required

With all this non-stock stuff running, I'd still be hesitant to blame the hardware.

I don't see any sort of crash message between reboots. Perhaps it is in /var/log/messages. Anything useful in /var/log/tverr?

Again, I'd go back to a completely stock image and see if you have a failure there too, to remove any possibility of misapplied hacks.

[pmoore]

Thanks. I was trying to avoid having to reapply the hacks. I was thinking that restoring the stock root files would get me back to original drivers. I'll go the completely stock image route. I will check out those other logs, as well.

Q1: Can I simply do this by (a) mounting my alternate root (hda4), wiping it, and extracting the root files from GZhpk-Series2 and GZcore, (b) dd'ing the stock kernel to hda3, and (c) flip over to these partitions? The reason I ask, is because I want to avoid first time setup and having to get two TiVo software updates.

Q2: Why does the ehci-hcd.o log entry indicate a non-stock driver?

[Jamie]

There's an emergency reinstall kickstart code that will install stock software in the alternate partitions then "flip": link. You can flip back to your original partitions when you are done.

Looking back, I see you have a 140 model. Those didn't have a USB2 capable host controller, IIRC. So maybe the "No such device" error from the ehci-hcd.o init_module is normal.

I suspect there aren't that many of the 140 models around, so maybe there is a compatibility problem with the TiVo wireless adapter on that model that hasn't been noticed before. TiVo does listed it as a supported adapter for that model though.

[pmoore]

So, I tried the kickstart, and it sort of worked, but it never flipped over. Looks like I screwed up my install by copying over fstab at some point (and who knows what else!) to the other partition. No big deal, I just pulled the drive and dropped a clean 9.3.1 image on it. After the initial setup, and subsequent s/w drops, I plugged in and set up the adapter. Rebooted, and same thing. Sigh. Gonna borrow a friend's TiVo Wireless Adapter to see if it's the adapter itself, or an onboard component. At this point, I may just leave the damn thing as is and use the phone line... ugh.

[dossy]

Has anyone backported the drivers for the TiVo USB Wireless Adapter?

[jt1134]

Originally Posted by dossy

Has anyone backported the drivers for the TiVo USB Wireless Adapter?

it's a closed source driver, so no. judging from your previous inquiries in the killhdinitrd thread, I would recommend that you use monte to chainload a 9.x kernel and just use the stock drivers already on your system.

[djl]

I've been using AGIGAUSB adapters on my 6.2a systems for ages, and jumbo frames have worked really well. I recently picked up an hr10-250 and did a slice upgrade to 6.4a on it, and have not been able to get jumbo frames back. The adapter works fine with an mtu of 1500, but anything higher makes it unreachable.

• I installed the backport drivers by copying over ehci-dummy.o, usb-ohci.o, usbcore.o to /lib/modules (I got rid of the symlinks from /platform)
• Renamed the dummy file to ehci-hcd.o, copied usbnet.o from drivers and symlinked it to ax8817x.o.
• Gave it a static address with netconfig.tcl.
• Made additions to usb.map.

My router is an RSV4000, and as I said has worked fine with this setup with my other Tivos. Any idea what I'm missing?
Here's some kernel log. You can see where I'm plugging/unplugging and messing with the mtu values.

Apr 26 14:58:28 (none) kernel: hub.c: new USB device 00:0d.2-1, assigned address 2 
Apr 26 14:58:28 (none) kernel: usb.c: USB device 2 (vend/prod 0xb95/0x1780) is not claimed by any active driver. 
Apr 26 14:58:29 (none) kernel: usb.c: registered new driver usbnet 
Apr 26 14:58:30 (none) kernel: eth0: register usbnet usb-00:0d.2-1, ASIX AX88178 USB 2.0 GIGE Ethernet 
Apr 26 14:58:31 (none) kernel: eth0: open: enable queueing (rx 73, tx 73) mtu 1500 ASIX framing 
Apr 26 14:59:01 (none) kernel: ApgManager Transition from state ACQUIRE_MARKER to ACQUIRE_MARKER 
Apr 26 14:59:23 (none) kernel: eth0: stop stats: rx/tx 0/2, errs 0/0 
Apr 26 14:59:23 (none) kernel: unlinking :849119e0 
Apr 26 14:59:23 (none) kernel: eth0: open: enable queueing (rx 73, tx 73) mtu 1500 ASIX framing 
Apr 26 15:00:26 (none) kernel: eth0: stop stats: rx/tx 0/12, errs 0/0 
Apr 26 15:00:26 (none) kernel: unlinking :849119e0 
Apr 26 15:00:26 (none) kernel: eth0: open: enable queueing (rx 9, tx 9) mtu 1600 ASIX Jumbo framing 
Apr 26 15:04:55 (none) kernel: ApgManager Transition from state ACQUIRE_MARKER to ACQUIRE_MARKER 
Apr 26 15:09:14 (none) kernel:
Apr 26 15:09:14 (none) kernel: eth0: stop stats: rx/tx 16/44, errs 0/0 
Apr 26 15:09:14 (none) kernel: unlinking :849119e0 
Apr 26 15:09:14 (none) kernel: eth0: open: enable queueing (rx 73, tx 73) mtu 1500 ASIX framing 
Apr 26 15:10:48 (none) kernel: ApgManager Transition from state ACQUIRE_MARKER to ACQUIRE_MARKER 
Apr 26 15:16:42 (none) kernel: ApgManager Transition from state ACQUIRE_MARKER to ACQUIRE_MARKER 
Apr 26 15:22:35 (none) kernel: ApgManager Transition from state ACQUIRE_MARKER to ACQUIRE_MARKER 
Apr 26 15:24:42 (none) kernel: eth0: stop stats: rx/tx 99/223, errs 0/0 
Apr 26 15:24:42 (none) kernel: unlinking :849119e0 
Apr 26 15:24:42 (none) kernel: eth0: open: enable queueing (rx 9, tx 9) mtu 1900 ASIX Jumbo framing 
Apr 26 15:25:21 (none) kernel: usb.c: USB disconnect on device 00:0d.2-1 address 2 
Apr 26 15:25:21 (none) kernel: eth0: unregister usbnet usb-00:0d.2-1, ASIX AX88178 USB 2.0 GIGE Ethernet 
Apr 26 15:25:21 (none) kernel: eth0: stop stats: rx/tx 126/242, errs 14/0 
Apr 26 15:25:21 (none) kernel: unlinking :849119e0 
Apr 26 15:25:26 (none) kernel: usb.c: deregistering driver usbnet 
Apr 26 15:25:26 (none) kernel: hub.c: new USB device 00:0d.2-1, assigned address 3 
Apr 26 15:25:27 (none) kernel: usb.c: USB device 3 (vend/prod 0xb95/0x1780) is not claimed by any active driver. 
Apr 26 15:25:28 (none) kernel: usb.c: registered new driver usbnet 
Apr 26 15:25:29 (none) kernel: eth0: register usbnet usb-00:0d.2-1, ASIX AX88178 USB 2.0 GIGE Ethernet 
Apr 26 15:25:30 (none) kernel: eth0: open: enable queueing (rx 73, tx 73) mtu 1500 ASIX framing