S2 SA loose network after update?

[ercdvs]

I did an in place update of my series 2 stand alone from 7.x to the latest 8.1a-x software, all via the bash prompt.

I applied the new software, dd'ed the hacked kernel in, removed my iptables, set my rc.sysinit.author file, all well and good.

I rebooted and found that I could not connect to the tivo remotely via the network. I am using a Belkin USB->Eithernet adaptor, which until the reboot, worked just fine. I see lights on the adaptor, so the net connection is fine, but the tivo network setup shows no connection.

Pulling the USB while the tivo is on causes a reboot .. but plugging it in while the tivo is on is fine..

Any thoughts on what is going on? Is my only resort to pull the drive and see what is going on in the logs ?


edit: dammit can't edit my title

[ScanMan]

You may have fallen victim to an issue with network drivers with the 8.1 upgrade; check out the "8.1 and my NIC" thread. Short answer for your solution is to use the backport drivers or copy the old (7.x) drivers in from your alternate partition.

[ercdvs]

ahh poop... i guess i need to pull the drive then.

I read that thread a while ago, and I thought it was just a Dtivo issue. Guess I was wrong.


What was the 'final' method? I don't really see a consensus (or at least one valid for a Stand alone S2)

[ScanMan]

What was the 'final' method? I don't really see a consensus (or at least one valid for a Stand alone S2)
I don't think there was one. I personally use the backport drivers, they give a little better performance. Others had success with the 7.x drivers. If you use the backport, make sure you copy all of the "hosts" files (usbcore.o, etc.) as well as the particular driver for your NIC.

[ercdvs]

I think I may have used an old kernel to write to the /dev/hda6 (my kernel) partition...

My working tivo with identical hardware is still running 7.3.1-01-2-140 with 8.1a pending.. Is there a way I can get the kernel image from here or do I need to resort to the image begging thread? uname -a shows 2.4.20 running .. I pulled the latest killhdinitrd so i should be able to patch when the other tivo is ready for upgrade..

[ScanMan]

I think I may have used an old kernel to write to the /dev/hda6 (my kernel) partition...

My working tivo with identical hardware is still running 7.3.1-01-2-140 with 8.1a pending.. Is there a way I can get the kernel image from here or do I need to resort to the image begging thread? uname -a shows 2.4.20 running .. I pulled the latest killhdinitrd so i should be able to patch when the other tivo is ready for upgrade..
Which killhdinitrd kernel are you using? The 3.1.5? You can get the kernel from your other tivo by determining the active kernel partition and using dd.
dd if=/dev/hda3 of=MyKernel.pxThat will write the disk image from /dev/hda3 to a file named MyKernel.px. Adjust to taste. There is also a "7.2.2-oth-K1" kernel in the image begging thread; you will need to apply killhdinitrd to it first though.

[ercdvs]

I thought I got the correct kernel using uname -a

i seem to be running 2.4.20 ... if i am pulling this from a hacked tivo, I *should* be able to throw it right on ... right ? But I am thinking that its really a usb driver issue at this point... I could do all this once I get access to the tivo ... not sure what my next step should be.


I am pulling the drive, and throwing on the backport drivers to /lib/modules .. lets see where that gets me

[ScanMan]

Well sort of. 2.4.20 is the linux kernel version, 3.1.5 is the tivo software version that the kernel was released with. The 3.1.5 is a linux 2.4.20 so that's probably it. However, for example the 7.2.2-oth-K1 is also a 2.4.20. The reason the software version is distinguished is because only certain tivo kernels are susceptible to the killhdinitrd exploit. And yes, that hacked kernel should drop right in to your other tivo. Good luck!

[ercdvs]

this is really wierd .. I have my drive pulled (hdc in this box) and although I have a var directory .. its not pointing to a new partition like my original side does.

I have no logs at all when the tivo was plugged in ... i have no way to look at any info at all... hmm. I'm getting to my hack knowledge limit ... at least I have logs to troubleshoot linux problems..

[cheer]

There is no new partition for /var -- /var is always on /dev/hdx9. /var doesn't change during an update.

[ercdvs]

Ok.. my bad .. I assumed since I saw var explicitly mapped to /dev/hda9 on my original partition and mapped to /var on the new side, that there was a problem.

my logs are not updated in either case... is it possible to rollback the update? to go back to dev/hda4 from /dev/hda7 ?

[cheer]

Probably not -- upgrading from one major rev to another usually makes changes to MFS that aren't undoable.

[ScanMan]

my logs are not updated in either case... is it possible to rollback the update? to go back to dev/hda4 from /dev/hda7 ?The logs should still be available when you mount the drive in your PC; as Cheer said mount /dev/hdx9 at some mount point (mkdir /tivovar) and look for a file named "kernel" in the log directory.

What problem are you having? Reboots? Pulling the USB NIC and booting without it should stop the rebooting. If not, something else is going wrong. You might to try replacing the killhdinitrd kernel again. You could also try to 'dd' the alternate kernel partition to the active one. (i.e., 3->6 or 6->3)

[ercdvs]

on my pc, the /dev/hdx9 partition doesn't seem to contain updated log files.

I *could* have confused the UTC timestamps, but I seem to remember seeing a date of the 20th as a last modifies when i expected 21/22 at the very least. I'll check again.

The problem I am having is that although the 8.1a update *seemed* to run sucessfully, I have no network connectivity. My usb adaptor (wired) for both the netgear FA100 and Belkin (model # escapes me at the moment) both show lights (and blinking with activity!), but are unrecognized to the tivo.

The network setup claims no network cable is plugged in. Inserting or removing the usb causes a reboot.. but i have *not* noticed random restarts in the timeframe I was watching the tivo (20-60min) so I think i am not in those woods. At this point, it seems to be a USB issue.

I for the life of me could not mount a USB stick or SATA drive or SATA cdrom from the PTVUpgrade boot disk (latest version) I have .. so I am going the route of playing with my serial cable connection to see startup errors and hopefully transfer via that method.

Else I need to pull my old computer with pata optical drives. The serial cable that comes with the S2SA tivos will work as a null modem cable, right? or is there something else special about the cables people are using ?

I *do* have a copy of all hacks & binaries .. is it possible to re-run the SwSoftware.tcl to blow away the partitions and restore it to a 'pure' state? I would rather get it all working again, then hack away.. At least this way i can see what went wrong for sure..

[Narf54321]

You can use something like the MFS Live CD which seems to work better with SATA and USB. You can also use something like Ubuntu and use tivopart to force it to see the Tivo partition table.

If you want to reset everything back to stock software again, you can put the drive back in your tivo and try a kickstart 52 code as it boots up.

[ScanMan]

The serial cable that comes with the S2SA tivos will work as a null modem cable, right? or is there something else special about the cables people are using ?If you're just using the serial cable that comes with the tivo, you'll need a null modem adapter and a gender changer.

I *do* have a copy of all hacks & binaries .. is it possible to re-run the SwSoftware.tcl to blow away the partitions and restore it to a 'pure' state? I would rather get it all working again, then hack away.. At least this way i can see what went wrong for sure..You can't re-run 'installSw.itcl' because the upgrade is already active so it won't run. (I suppose you could hack your s/w version in MFS to 'trick' the script into running but that's beyond the scope here...and it might not work correctly considering the MFS database has already been updated)

I would try Narf's suggestion and do a kickstart 52.

[ercdvs]

Kickstart 52 done ... and no change...

So, assuming this did run properly, the included tivo software / kernel is not compatable with my current usb adaptor ?

I'll order a cable from PTVUpgrade, all the ebay ones seem to be gone. I realize I don't even have a serial port on this box .. so I need a usb->serial anyway.

I'll pull the drive in the meantime .. the bootpage should have switched on a kickstart 52 right ? i'm not sure what can be done if a full restore to stock SW still doesn't work.

I'll be leaving my other tivo go for now until I can find out what the hell is going on... Thanks for all the insight so far

[ScanMan]

Are you able to mount your tivo drive in your PC? Figure out which partition is at 8.1a and if the other is still 7.3.1. One way you can tell is take the md5sum of the tivoapp.fc22299984f6101ab797325e21e061b9 /tvbin/tivoapp81a.originalFind the tivoapp that matches that md5sum, that's your 8.1a partition. If the tivoapp on both partitions match, then both partitions are at 8.1a (probably due to the kickstart 52). Make sure you take the md5sum of an unpatched tivoapp to compare apples-2-apples. If the other partition doesn't match, then that's probably the 7.3.1. Also check filesizes and dates for a comparison.

Assuming you have 7.3.1 still on a partition, copy all of the files from:
/lib/modules
/platform/lib/modules
on the 7.3.1 partition to the 8.1a partition. That should get you the 7.x drivers installed in 8.1a and might solve your problem.

[ercdvs]

interesting ..i'll check it out now... the only hard part is getting the files over, as I can't get this live cd to find my sata drive or usb key ...

i have to drag the old comp out and ide it all the way ..


EDIT.. I am kind off a dumbass. of course a '52' reset would not work, because the other partition is already hacked as well.. there is no clean restore point.

just looking through the kernel messages of the drive, and I can see that it properly identifies the belkin usb device, and loads the pegasys drivers for it... I really need to see this thing boot up live to see what eth0 is reporting..

Hmmm... booting with no usb adaptor installed, I get no network option in the setup .. it says 'No adaptor found' whereas before, it was claiming that the cable was unplugged. So it *is* recognizing the adaptor.. just not a network on the other end?

What is the default network schema of the tivo? My home gateway is setup on 10.1.1.1 .. with the tivo originally having a 10.1.1.2 address ... do i need to pop back to a 192.168.xx schema ?? I can confirm that all plugs are tight and working ... so .. is it not seeing the gateway? i would think that would generate a new message


EDIT 2: having the netgear fa-120 causes a reboot loop .. pulling that out and trying belkin agian .. then pulling drive for messages

EDIT 3: Looking at the kernel logs, i can see entries for the Belkin adaptor, and that its loading the pegasys drivers ... but i also see lines saying adaptor 0x## / #### etc not recognized. I looked at these particular driver id's in /etc/hotplug/usb.map and they correspond to the FA-120 and Belkin ... *sigh*

Connecting via phone for now.. and the serial cable + usb->serial is on the way .. its easier to update via serial then drag out an old comp to transfer... unless you suggest maybe i pull a virgin image and just wipe everything out.....

Actually: What kernel are people using with Series 2 stand alone .. the 'original' series 2 boxes .. I wonder if I should bump it up

[ercdvs]

Just as an update... I connected the tivo drive to an old box still using ide drives, and mounted the backprt drivers cd..

made a copy of my existing drivers in /lib/modules and platfor/lib/modules and overwrote the originals..

Ethernet is up & running ...doing some testing on benchmarks now. Going to do the same thing to the other tivo i have .. its starting to refuse service connectons until I update software