Late feedback on July Update:SOLVED

OK, so I'm running a bit late.
Actually I don't usually update my OS while I'm developing, so I put it off for a while. But I read that DeluxeMusic works better under the update, so I took the plunge last night.
Oops, something is not quite right.
After installing, I reset the machine from the case reset button, as instructed. While starting up, the usual three second delay to select the boot source (the choice is in RED on the boot screen) seems to be taking longer. Yup, the machine locked up on me. That's new.
The next reboots came up OK, but locked up in the first three minutes every time.

I read the available comments from other users, I didn't see anything that resembles this. I want to provide useful feedback to the developers if I can. I can't provide a crashlog, as it never brought a reaper. It just froze up.

After about five repeats of this, I help CTRL down to stop the boot process, and I copied my system backup back over the SYS: partition. (of course I made a backup just before updating, doesn't everyone?) Now all is well again.

My machine has been very stable before the update, this really caught me by suprise. A few details about my machine may help:
Micro-AOne with a IBM 750GX processor, running at original clock rates, with the original 256Meg ram stick.
OS version before update was a clean install of "Final Update".
I have the most recent public versions of the SDK, PERL, SimpleMail, CAMD, all working fine. I had tried installing X11 at some point, but couldn't tell you if that worked.

My keyboard and mouse are both PS2.
Attached USB devices include a card reader and 2 MIDI devices, the MIDI stuff was still working fine after the update (my system makes heavy use of CAMD, all seems to be working fine)

So, I guess I'm the only one with this problem. I haven't seen any reports like this one so far. If I can help to track this problem down I will, but on a separate hard drive. Otherwise I'll stay at final release until the next big release happens.

Keep up the progress, and if I can help, just ask.
LyleHaze

@LyleHaze

That is indeed the first time I hear about this sort of issue. Do you have a serial line attached? It might be interesting to see if there is anything on the serial output prior to lockup.

@Rogue

At the moment my serial port is connected to an X-10 interface to water my plants. I thought that it MIGHT be getting debug output there, so I turned it off for the last three lockups. No change.

I'm just starting a 3 day weekend, so I should have time to try this:
I'll pull my current hard drive for safe keeping. I find that I get very irritable when my Amiga is down.

I'll add a clean drive and install the latest full release, test for a while, then add the July Update and test some more. I'll try to keep a record of all changes/additions that I make (classes for SimpleMail, CAMD stuff, etc.)

It shouldn't take too long to find it. it seems to lock up pretty fast. When I have more to report, I'll feed it back into this forum, unless you'd like me to write you directly.

Now where did I see a spare hard drive around here??

LyleHaze

@LyleHaze

ISTR you have to change something in the U-Boot boot args to redirect debug output back to serial. You can also do it after OS4 is up and running with:
kdebug "console serial"
(I think)

I've been getting a few hard freezes here, previously this did happen but only due to a highlighting bug in NotePad, now they are more frequent and seem to be random.

Chris

@LyleHaze

One thing to watch Lyle, although maybe not related to your issue.
If everything worked fine before, check your s-s and paste any missing entries from your working setup back into the new s-s.

#6

@Chris

Quote:

It's actually "kdebug {command}", which is a combination of level and redirection based on the level of debug required.
"0" was suggested for most people, and that's what I use unless told otherwise.

@lylehaze

See docs/kernel/kdebug.readme

My s-s reads:

; $VER: Startup-Sequence 52.5 (03.07.2007)

kdebug "debuglevel 0" >NIL:
kdebug "console memory" >NIL:

etc.

Yes, serial already goes to console by default now, but that would only read things like "dropped irq" when using "dumpdebugbuffer > {filename]. You need the above, in addition, for "kernel".

#6

@Rogue

OK, some more information.
I think we can rule out any add-on software from me.
Here's the current scoop:

I removed my 2 precious HD's and added a Samsung 120 that had an older version installed.
Side note: had 4 partitions, 2 are now bad, so I just removed their "automount" ticks. DH0 and DH1 partitions still working.
I quick formatted DH0: , and installed SLB_V2 from the december 2006 release.
I did the full usual install from Dec 2006.
Did a network wizard thing to find my LAN, OK.
Added screenmode 1280x768@60 to my radeon tooltypes, rebooted, then selected same screenmode from prefs.
download the July2007 update, unarc and install.
type "reboot" into an open shell.

2 minutes after the reboot, machine is locked up. Input device does not see CTRL-ALT-ALT

Now I'll go find a serial cable (I may have to make one)
I'll update more as I proceed.

Maybe a complete re-install was the long way around, but it certainly eliminates a lot of little possibilities.

LyleHaze

@LyleHaze
Probably just needs a new battery.

@ssolie

Battery was replaced with the December update, and since the A1 is powered up 24/7, there is almost no drain on it.
Besides, the problem appeared exactly when I added the July update, and disappeared again when I got rid of the July update (twice each so far).

I mean, I'm open for suggestions, really!
But I am confident that this is not battery related.

About the only thing I can think of right now is UBoot settings, since that is all that can be "carried over" from the previous installation.
I was beginning to wonder if I was getting the wrong kick modules from a different partition than I was booting from, but that has now been eliminated too.

Anyway, I found the null modem cable, and stretched it over to the win98 machine. Now I'm off to figure out the baud settings for the debug stream..
Will report back when I have more to tell.

LyleHaze

@all

Just a repost of the link to the article by wegster:
"Serial debug output on A1s for AOS4"
http://amigaworld.net/modules/feature ... ?op=r&cat_id=4&rev_id=125

#6

@LyleHaze
Quote:
I bet you the $2.50 CAD it costs for the new battery you are wrong.

@ssolie

OK, first the status update:
I am unable to get any debug output from OS4.

After reviewing the guide by #6, I can only guess that it's because I don't have a debug boot option, which he states must be selected.

The A1 is still locking up frequently.

If I set debuglevel and console by kdebug from a shell, I still get no output. After lunch, I'm going to write a script to test the cable connection, just to be sure.

Just to be a friendly type, I'm going to go out to lunch and buy a new battery for my AOne.

I mean, the bet is a big one!

Be back in about 90 minutes.
(God I hate this Windows machine!)

LyleHaze

@LyleHaze

A-ha! Just crashed on me, but I don't get any serial output either, except:

[sii3112ide/irq_wait] timed out
[sii3112ide/exec_pci_ata_cmd] <- here
[sii3112ide/ata_rw_blocks] unit 0 lba28 write error 255, retries left : 4
[sii3112ide/irq_wait] timed out
[sii3112ide/exec_pci_ata_cmd] <- here
[sii3112ide/ata_rw_blocks] unit 0 lba28 write error 255, retries left : 3
[sii3112ide/irq_wait] timed out
[sii3112ide/exec_pci_ata_cmd] <- here
[sii3112ide/ata_rw_blocks] unit 0 lba28 write error 255, retries left : 2
[sii3112ide/irq_wait] timed out
[sii3112ide/exec_pci_ata_cmd] <- here
[sii3112ide/ata_rw_blocks] unit 0 lba28 write error 255, retries left : 1

Sound is loopy, display is still updating, input is gone.

I probably do need to change my battery, but this seems like odd symptoms for that.

Hmm, it's reading 2.86V which doesn't seem bad for a 3V battery, it's the original one too. I've put it back in and had to reset my U-Boot prefs (the "load config" thingy in U-Boot Prefs doesn't appear to work btw). There were loads of settings in my U-Boot config that aren't there now, so maybe that's done something.

Chris

@Chris

You need a new battery! A lithium cell has the characteristics of holding a constant voltage output over its lifetime. A1s have shown boot problems (black screen) when the cell voltage drops to 2.95V or less. A drop in voltage below 3V indicates end of life is in progress. Not saying that will solve what you're seeing, but you're primed to have a system that won't boot at all & it's very sudden & scary.

Uboot config loading works, you need to hit the save button in the lower left to actually save it back. It will pop up a window saying what variables its going to change/save & an option to chicken out.

@LyleHaze

Quote:

Be carefull!

If the plants get all that horrific debug data, they might die :)

@Thread

OK, this is getting less funny by the minute.
I have installed a new battery.

On boot, the kickstart modules are loaded, it then goes to the "Boing Ball" screen that usually means "insert workbench". It refuses to boot to either the IDE HD or the IDE CDROM.
I seem to recall reading something about this problem back in January, but it didn't happen to me then.

How is it that simply getting serial debug information should be such a pain in the ass?

I'll review the installation instructions from the CD-ROM, and try to figure out which UBOOT parameter is wrong.

Any hints will be gratefully accepted.

Lyle

@Thread

OK, I recall something about the Pioneer DVR-107 being the cause, something had to be set in UBoot, then later reset in the startup sequence to prevent this problem.

I'm still searching, its been a LONG time.

Lyle

From IntuitionBase: (Thanks Ryu)
Notes: While the drive's firmware messes up the interrupts during detection, you should make a workaround to make it work with UDMA on OS4: turn it's IRQ off (Ubootprefs) and put this into your startup sequence: idetool -i a1ide.device 1 1 &

OK, so setenv a1ide_irq 1101
should have worked, but didn't.
setenv a1ide_irq 0000
does work, I'm booting again.

I love the details!
Somebody write that down for me.

Off to figure out the serial debug setup..

Lyle

OK, problem solved.

It looks like I owe SSolie a toonie.

Replacing the battery has fixed the problem.

Big thanks to everyone for the help.

Maybe I'll design a battery replacement that never goes below 3Volts, based on a bigger, higher voltage battery.
Then include an audible alarm when it gets low.

Time to re-install my regular HDs again.

LyleHaze

Back to my usual HD's, and no lockups unless I open IBrowse.
That knocks it out in about two seconds.

I wonder if it's related to the update, or all the mucking around I did in UBOOT?
I wish there was a UBOOT backup/restore utility.

Oh Well, at least the plants are getting watered again.

Lyle

@LyleHaze

Quote:

You still have the option of joining us in irc, where we often discuss this very topic.

#6

@LyleHaze

Quote:

Prefs/UBoot has just that. Just check the menu items....