Stuck at "Welcome. Powering up..." after 6.2 slice upgrade

[IlliniFan]

I have a Philips DSR7000 that I had previously upgraded/hacked to 4.0.1 a couple of years ago. Today I decided to try upgrading to 6.2, following the steps outlined in this guide (http://www.dellanave.com/projects/tivo/62slices.html).

The dbload went smoothly, as did the installSw.itcl step. I didn't have a 3.1.5 kernel handy, so I thought I'd use the 62_init_framework_ddn method of setting up a monte configuration instead. After doing this, the TiVo just sits at "Welcome. Powering up..." indefinitely when I plug it back in.

Thinking that since I didn't use a monte config previously the new one was probably the problem, I searched a little harder and found a 3.1.5 kernel with killhdinitrd already applied. I removed test.conf and /init from the hard drive (the stuff that was put there by the init_framework), then dd'd the 3.1.5 kernel to both hda3 and hda6, but I still get the same thing--upon plugging in the TiVo, all I get on the screen is "Welcome. Powering up..."

I have a serial cable hooked up, and am getting absolutely no output whatsoever on it. I've tried removing the hard drive, and mounting partitions 7 and 9 in a Linux machine, and I can verify that there is still data there so I don't think I screwed up the drive. I looked through the logs but I don't see anything in them--after several attempts at booting, /var/log/kernel and /var/log/messages have not been changed.

It seems to me that the TiVo is never even getting to the point where it starts booting from my hard drive, but I'm unsure what to try next. Any suggestions would be greatly appreciated. If I've failed to provide sufficient information, please let me know. Thanks in advance!

[PlainBill]

You wouldn't believe the number of people who have problems with that guide. I usually give them all the same advice - ask the guy who wrote the procedure.

With the drive in a computer (I'll assume secondary master) and booting from a tools cd (MFS Live or PTVupgrade), what is the output from:

pdisk -l /dev/hdc

bootpage -p /dev/hdc

bootpage -b /dev/hdc

With the drive in the DirecTiVo and the serial cable hooked up, terminal program running, etc., what is the output when you hit 'Enter' several times IMMEADIATELY after plugging the power cord in?

PlainBill

[IlliniFan]

You wouldn't believe the number of people who have problems with that guide.It doesn't surprise me--there were a few tell-tale signs in it that the guy didn't really know what he was doing, but I figured as a basic list of the necessary commands to run, it should be okay. I did do a lot of reading on here before doing anything. Unfortunately, it appears as though the part that I needed further explanation on--the monte setup--is the part that he glossed over. Oh well, I'm where I'm at now, and need to figure out what went wrong and how to fix it.
I usually give them all the same advice - ask the guy who wrote the procedure.Well, look where following his advice got me. :(

Anyway, here is the output you requested:
pdisk -l /dev/hdbPartition map (with 512 byte blocks) on '/dev/hdb'
#: type name length base ( size )
1: Apple_partition_map Apple 63 @ 1
2: Image Bootstrap 1 1 @ 68353088
3: Image Kernel 1 8192 @ 68353089 ( 4.0M)
4: Ext2 Root 1 262144 @ 68361281 (128.0M)
5: Image Bootstrap 2 4096 @ 68623425 ( 2.0M)
6: Image Kernel 2 4096 @ 68627521 ( 2.0M)
7: Ext2 Root 2 262144 @ 68631617 (128.0M)
8: Swap Linux swap 260096 @ 68893761 (127.0M)
9: Ext2 /var 262144 @ 69153857 (128.0M)
10: MFS MFS application region 524288 @ 69416001 (256.0M)
11: MFS MFS media region 68353024 @ 64 ( 32.6G)
12: MFS Second MFS application region 524288 @ 69940289 (256.0M)
13: MFS Second MFS media region 85964800 @ 70464577 ( 41.0G)
14: MFS New MFS Application 1024 @ 156429377
15: MFS New MFS Media 78004224 @ 156430401 ( 37.2G)
16: Apple_Free Extra 7023 @ 234434625 ( 3.4M)

Device block size=512, Number of Blocks=234441648 (111.8G)
DeviceType=0x0, DeviceId=0x0

bootpage -p /dev/hdbroot=/dev/hda7 dsscon=true console=2,115200 upgradesoftware=false

bootpage -b /dev/hdbBoot Partition: 6

With the drive in the DirecTiVo and the serial cable hooked up, terminal program running, etc., what is the output when you hit 'Enter' several times IMMEADIATELY after plugging the power cord in?Absolutely nothing. I'm starting to think that my serial cable doesn't work, although it did work the last time I used it (about 2 years ago). I bought it from 9thtee.com.

Note that I've given up on the monte setup, and have tried dd'ing a killhdinitrd'd 3.1.5 kernel onto /dev/hda3 and /dev/hda6.

Thanks again for any suggestions you can provide.

[PlainBill]

It certainly appears there is a problem with the serial cable. It's late, but at first glance all the boot parameters look good, as does the partition table. No changes to the logs indicates it never mounts partition 9.

Hitting 'Enter' repeatedly should bring up the prom menu, the fact that it doesn't is significant. This would also explain the lack of any error message.

I'd suggest the following steps: Hyperterm should work, Teraterm is a superior product. Make sure it is set up for 115200 baud, 8,n,1, no flow control. Short the tip and ring on the serial cable together and make sure typed characters are echoed back to the screen (this is NOT a definitive test). And my favorite, make sure the serial cable is plugged in completely. And my other favorite, make sure you choose the correct com port.

You could also try troubleshooting it from the other direction: Are you sure you fully unpacked the killhdinitrd kernel?

PlainBill

[IlliniFan]

I guess my serial cable doesn't work at all--I double-checked everything you mentioned, and it all looked correct. I tried shorting the ring and tip with a paper clip, and still got nothing in Tera Term when I pounded on the keyboard. So I'm going to assume that I need a new serial cable, which is unfortunate because tomorrow is Sunday. I'll try my local Radio Shack to see what they have tomorrow.

I also re-downloaded the 3.1.5 kernel (using wget), gunzip'd it, and re-dd'd it to both partitions again. Still no luck.

I'm calling it a night for now. Thanks for the suggestions--I'm always open to more. I'm hoping that if I get a working serial cable it'll make the error readily apparent.

[drez]

Dellanave's 62_init_framework_ddn.tgz has a kmonte.o for 2.4.4 kernels. There's your problem. The 3.1.5 kernel is a 2.4.20 kernel, so it needs a 2.4.20 kmonte.o.

You don't need a monte with 6.2x. You can simply delete test.conf and the init folder that Dellanave's .tgz made and your Tivo should boot.


...

If you are still interested in a monte for significantly faster network transfers, I've been planning on releasing my own monte setup with Jamie's custom 7.2 kernel (http://www.dealdatabase.com/forum/showthread.php?p=244219#post244219) as the to-kernel.

I've tested it with 6.2x & 6.3x (thanks dburckh) with the 3.1.5 & 7.2.2 killhdinitrd kernels as the from-kernels. (My monte setup uses a 2.4.20 kmonte.o.)

...I'll make a thread for it one of these days. Until then, I've attached it for you.

[IlliniFan]

You can simply delete test.conf and the init folder that Dellanave's .tgz made and your Tivo should boot.I've done this, and am still experiencing the same symptoms. Any other ideas? I'd like to get it booting before trying to switch to a monte setup, but once I do I'd be interested in trying out yours.

I think that my question has morphed into, simply enough, this: Is it correct to say that there is nothing I can do to the hard drive that would result in absolutely zero output on the serial port? If that's the case, then I need to fix my serial cable before proceeding any further.

[PlainBill]

I've done this, and am still experiencing the same symptoms. Any other ideas? I'd like to get it booting before trying to switch to a monte setup, but once I do I'd be interested in trying out yours.

I think that my question has morphed into, simply enough, this: Is it correct to say that there is nothing I can do to the hard drive that would result in absolutely zero output on the serial port? If that's the case, then I need to fix my serial cable before proceeding any further.

Exactly correct. I just tried my spare system and got the following error message with the hard drive disconnected.

Partition 0 signature is false, trying alternate
diskboot found partition 0 signature is false.
Ram size = 64
Service number is 3810000E0XXXXXX.
What is password?

I also verified you DON'T get the message if the serial cable is not connected to the computer :o or if the baud rate is not set to 115200.

PlainBill

[IlliniFan]

Okay, finally I'm getting somewhere. I made myself a new serial cable and it works. Here's what I get when I power on the DirecTiVo:
Boot failed reason=53
Ram size = 64
Service number is xxxxxxxxxxxxxxx.
What is password?
I've done some searching, and it looks like that means that the kernel I'm using is bad. I got the 3.1.5 kernel from here (ftp://tivok:tivokernel@ftp.tivosupplies.com/tivo_kernels/s2_kernels/), and the README at that site says that killhdinitrd has already been applied to it. Should I try a different kernel? Which one?

[IlliniFan]

I tried it with drez's chainload framework, and re-dd'ing the 3.1.5 kernel from the link above (for about the 100th time), and still no luck. I also tried re-applying killhdinitrd to that kernel, then dd'ing it again, and I still get "reason=53" when I try to boot.

Is there any other kernel I can try? Any other ideas as to why this one is giving me so many problems?

[Jamie]

... re-dd'ing the 3.1.5 kernel from the link above (for about the 100th time), and still no luck. ...you are gunziping the .gz'd kernel before you dd into the kernel partition, right?

[IlliniFan]

you are gunziping the .gz'd kernel before you dd into the kernel partition, right?Yep. Here's a full transcript of what I'm doing. This is in a Linux machine, with my TiVo's hard drive hooked up as the primary slave:
localhost tivo # bin/tivopart.x86 r /dev/hdb
localhost tivo # wget 'ftp://tivok:tivokernel@ftp.tivosupplies.com/tivo_kernels/s2_kernels/3.1.5/vmlinux.px.gz'
--09:27:08-- ftp://tivok:*password*@ftp.tivosupplies.com/tivo_kernels/s2_kernels/3.1.5/vmlinux.px.gz
=> `vmlinux.px.gz'
Resolving ftp.tivosupplies.com... 64.40.149.70
Connecting to ftp.tivosupplies.com|64.40.149.70|:21... connected.
Logging in as tivok ... Logged in!
==> SYST ... done. ==> PWD ... done.
==> TYPE I ... done. ==> CWD /tivo_kernels/s2_kernels/3.1.5 ... done.
==> PASV ... done. ==> RETR vmlinux.px.gz ... done.
Length: 1,204,794 (1.1M) (unauthoritative)

100%[====================================================>] 1,204,794 138.55K/s ETA 00:00

09:27:16 (153.62 KB/s) - `vmlinux.px.gz' saved [1204794]

localhost tivo # gunzip vmlinux.px.gz
localhost tivo # dd if=/dev/zero of=/dev/hdb3
dd: writing to `/dev/hdb3': No space left on device
8193+0 records in
8192+0 records out
localhost tivo # dd if=vmlinux.px of=/dev/hdb3
3954+0 records in
3954+0 records out
localhost tivo # dd if=/dev/zero of=/dev/hdb6
dd: writing to `/dev/hdb6': No space left on device
4097+0 records in
4096+0 records out
localhost tivo # dd if=vmlinux.px of=/dev/hdb6
3954+0 records in
3954+0 records out
localhost tivo #


I then put the drive in the TiVo and boot, and get the "reason=53" error on the serial console.

[Jamie]

Yep. Here's a full transcript of what I'm doing. ...
It looks right to me. I looked at my copy of that kernel, and the md5sum of the uncompressed vmlinux.px is: f878a7900faf8edcc8345ba43710aaef, if you want to cross check. You might want to dd the kernel back out from the kernel partition and check the md5sum of that. Use the bs and count dd options to get just the kernel and not the junk after it.

I assume you are doing a clean shutdown and all disk caches etc are known to be flushed before you remove the drive from the linux PC.

[IlliniFan]

localhost tivo # ls -l vmlinux.px
-rw-r--r-- 1 root root 2024448 Apr 1 09:27 vmlinux.px
localhost tivo # dd if=/dev/hdb6 of=test bs=1 count=2024448
2024448+0 records in
2024448+0 records out
localhost tivo # md5sum vmlinux.px
f878a7900faf8edcc8345ba43710aaef vmlinux.px
localhost tivo # md5sum test
f878a7900faf8edcc8345ba43710aaef test

I assume you are doing a clean shutdown and all disk caches etc are known to be flushed before you remove the drive from the linux PC.I do sync a couple of times, and then 'shutdown -h now' and wait for it to power off before removing the power cable and then removing the disk.

There's gotta be something I'm overlooking here, but I'm just not sure what else to check at this point. Thanks again for your help, guys.

[pentium101]

Yep. Here's a full transcript of what I'm doing. This is in a Linux machine, with my TiVo's hard drive hooked up as the primary slave:Before putting the drive back into the Tivo did you make sure that the drive jumper was set to Master?

[PlainBill]

Before putting the drive back into the Tivo did you make sure that the drive jumper was set to Master?
Something to be aware of, but in this case...

Notice that when I tried with the drive disconnected (essentially the same as I would expect if the jumper were set to slave) the error message was

Partition 0 signature is false, trying alternate
diskboot found partition 0 signature is false.
Ram size = 64
Service number is 3810000E0XXXXXX.
What is password?

He's getting

Boot failed reason=53
Ram size = 64
Service number is xxxxxxxxxxxxxxx.
What is password?

Still, you're right, make sure the jumper is in the correct position, the ribbon cable is firmly seated, trays are stowed, and chairs are in the full upright position...

PlainBill

[IlliniFan]

PlainBill is correct--it's easy to tell if you have the jumper correct or not because you get a different error message. In the dozens of times I've removed and replaced the drive, I have forgotten to change the jumper once or twice, in which case I get the "Partition 0 signature is false" error. "Boot failed reason=53," however, appears to indicate that it can see the drive and read from it just fine, it just doesn't like what it sees for whatever reason.

[drez]

IlliniFan: check your PM

[PlainBill]

At this point I'm really grasping at straws. A few suggestions:
Try a different kernel - the 7.2.2-oth-k1 (http://dealdatabase.com/forum/showpost.php?p=270112&postcount=917) should be fully compatible. try specifying the block size bs=1024 in the dd command.
Try a different 'tools cd'. Try a different computer. Try a different ide port in the computer.

If nothing else works, try sacrificing a chicken, dancing in your back yard naked at midnight, or something equally bizarre.

PlainBill

[ScanMan]

After you try the 7.2.2-oth-k1 kernel, if still no joy, just for kicks and giggles, go back and dd in your old 4.01a (or whatever) killhdinitrd kernel. Perhaps something didn't take during the upgrade and your attempt to boot the 3.1.5 (a linux 2.4.20 kernel) is somehow incompatible with the current state of your system. I got a similar error when I accidentally booted a 3.1.5 kernel into a 4.0.1b filesystem. Like PlainBill, I'm grasping at straws too...

[Jamie]

More grasping at straws: It seems to me that we know a valid kernel is in the kernel partition, but the tivo isn't recognizing it as valid, perhaps due to an I/O error when the PROM code is reading the kernel. So: try a different ide cable in the tivo, and, if necessary, try a different disk.

Just for the record, can we enumerate the possible boot fail reasons we know about? PROM can't read partition table:Partition 0 signature is false, trying alternate diskboot found partition 0 signature is false. Partition table okay, but PROM can't read kernel:Boot failed reason=53 Kernel read okay, but fails signature check:
Boot failed reason=60Do these look right? Do we know any others?

[IlliniFan]

Okay, the 7.2.2-oth-k1 (http://dealdatabase.com/forum/showpost.php?p=270112&postcount=917) kernel appears to have worked for me. Thanks for your help, everybody!

Still not sure why the 3.1.5 kernel wouldn't work, but oh well, I've got a working TiVo now, so I'm happy. Hopefully this thread stands as yet another warning against guides.

[pentium101]

Something to be aware of, but in this case...

Notice that when I tried with the drive disconnected (essentially the same as I would expect if the jumper were set to slave) the error message was

Partition 0 signature is false, trying alternate
diskboot found partition 0 signature is false.
Ram size = 64
Service number is 3810000E0XXXXXX.
What is password?

He's getting

Boot failed reason=53
Ram size = 64
Service number is xxxxxxxxxxxxxxx.
What is password?I never change the drive jumper so that was a guess.

The only time that I've ever seen the "partition 0 signature is false" error was the one time I was in a hurry and forgot to re-attach the power cable when I re-installed the hard drive. It really sucked to see that message on the console when I plugged it in after having closed it up and connecting all of the cables back up. :rolleyes:

[clustermonkey]

I'm fighting with this exact thing right now. Same 3.1.5 image that's making me angry, same md5sums and double checking the write with a read, dd, md5sum. Well, except I had 1 successful boot after a couple rounds of "I'll just re-dd the image and verify again, I must have screwed something up". During the successful boot, I superpatched tivoapp and set up some other stuff (tivowebplus, etc) ... then power cycled and it went back to reason=53 :(

I suspect the IDE cable or disk. I'm going to try the 7.2.2-oth-k1 kernel tonight and cross my fingers. If that doesn't work I'll start swapping hardware. If that doesn't work, I'll have to execute the tivo "office space printer smackdown" style ;)