How to make sense out of soft resets?

[vovka1965]

Palm OS seems to love soft resets. Anything that it is unhappy about and boom! No error codes, nothing..
Is there an app there that can log error codes with each soft reset and hopefully let the user make sense out of them?

I have heard of Crash Pro.. Is that basically the only application that can do this??

[stevec]

Crash Pro is probably not the only one but it's the only one I'm aware of - it has a solid reputation and I've used it for years. What is it you want to do that's not covered by Crash Pro?

[BaDZeD]

Hmm, my TX runs without resets wor days, sometimes weeks. The only thing that has consistently caused a reset for me is browsing thew internet while mundu IM is on.

[cyberdude]

Crash Pro is excellent and I've used it for years. But... CP doesn't log errors for most soft resets. I'm not sure why some get logged and some don't. If you check the box in CP to log soft resets soft resets that don't have an error to go along with them are logged as soft reset - not much help. But when there is an error, you get a detailed messsage of what app caused it (Not always correct) the time/date and the type and line number.

Another app is Reset Doctor, but it gives next to zero info. What it does do is log resets as either fatal or non-fatal (it sounds worse than it is).

Both apps together can be more helpful in that some resets logged as a soft reset in CP get logged as a fatal one in RD - go figure!

[FerdFerd]

I finally bought Crash Pro, hoping to have more information about why my TX crashes and resets, almost always when I'm testing something new that is still in development. I've had it set not to log soft resets. Most of the crashes I've had since I bought it have not produced any logging by Crash Pro at all. A few have been logged.

I've just set it to log soft resets too, and I'll see what happens next time I get a crash and reset. I expect that the ones not now being logged will just get recorded as a soft reset, thereby providing no useful information at all.

If I had it to do again, I probably wouldn't buy Crash Pro.

[vovka1965]

Sounds like Crash Pro is the one, then..
Basically, I am finding out that a number of built-in Palm Apps or official releases cause crashes and as I am sorting through all these crashes, it is good to have extra info

Example of apps that came with Palm that crash every so often?

1. VersaMail-- this is a real piece of crap! I am trying Snappermail now. And I am not willing to fork over even $10 for VersaMail update unless I can test-ride that update..
2. CradlePatch fix-- some fix! (got rid of that one!)
3. Blazer-- this is just a joke of an application...
4. DTG, the free version, crashes after syncing.. Out it goes!

[vovka1965]

I finally bought Crash Pro, hoping to have more information about why my TX crashes and resets, almost always when I'm testing something new that is still in development. I've had it set not to log soft resets. Most of the crashes I've had since I bought it have not produced any logging by Crash Pro at all. A few have been logged.

I've just set it to log soft resets too, and I'll see what happens next time I get a crash and reset. I expect that the ones not now being logged will just get recorded as a soft reset, thereby providing no useful information at all.

If I had it to do again, I probably wouldn't buy Crash Pro.

Not Crash Pro? Then there is nothing at all there?

[cyberdude]

Sounds like Crash Pro is the one, then..
Basically, I am finding out that a number of built-in Palm Apps or official releases cause crashes and as I am sorting through all these crashes, it is good to have extra info

Example of apps that came with Palm that crash every so often?

1. VersaMail-- this is a real piece of crap! I am trying Snappermail now. And I am not willing to fork over even $10 for VersaMail update unless I can test-ride that update..
I rarely get a crash with VM but when I do CP doesn't log anything.
2. CradlePatch fix-- some fix! (got rid of that one!)
This was discussed a while back and yes it is a piece of crap. It CAUSES resets.
3. Blazer-- this is just a joke of an application...
It works better than most browsers although I think that Universe (still in beta) will become the default - universewpa.com I think.
4. DTG, the free version, crashes after syncing.. Out it goes!
I use v 10.x and its a dream.

As to whether to get Crash Pro, I would. The information from the crashes that are not logged as soft resets is extensive enough for developers to figure out what's happening - for you it will only point to what might be the offending app, although this may sometimes point to your launcher :(.

I have both Crash Pro and Reset Doctor and find that together they provide a decent amount of info (CP more than RD). Neither cost very much RD barely anything, IIRC. I don't know how much CP costs since I won that app on a forum contest. :)

[FerdFerd]

Crash Pro costs $9.99.

I agree that when it catches and logs a crash, it does provide information that is useful to pass on to the developer. It just doesn't do that very often in my hands.

[phreakonaleash]

It works better than most browsers although I think that Universe (still in beta) will become the default - universewpa.com I think.

wapuniverse.com ;-)

CrashPro is definately your best option, i;m afraid. I will get in contact with there developers and try to get some things sorted out soon.

[maceyr]

You may also want to check the TX for possible corruptions and problems that may cause them.

Check out my posting on resets (http://palmdiscovery.netfirms.com/nfblog/2005/11/28/palm-problems-pt-1-reset-loops-crashes/) for some suggestions of software that can help.

[vovka1965]

You may also want to check the TX for possible corruptions and problems that may cause them.

Check out my posting on resets (http://palmdiscovery.netfirms.com/nfblog/2005/11/28/palm-problems-pt-1-reset-loops-crashes/) for some suggestions of software that can help.


That was a useful writeup. But the link to "stability test" is dead.. :(

[FerdFerd]

But the link to "stability test" is dead.. :(I wouldn't worry about it much. I used to use Stability Test, and it never once showed my TX to be unstable, even when I knew it was. It's gone from my device now.

[cyberdude]

I still have Stability Test if you really want it. But FerdFerd is correct, all the times I've used only a couple of times did it show my device was unstable (it passed all the tests but reset after exit which it warns you means the device is unstable). I still have it because I use it every so often (actually infrequently now, I used to run after I installed ANYTHING) and I tend to be a hoarder in general, so I have a lot of stuff on my device that really doesn't need to be there.

[BrentDC]

In my mind the best way to test the stability of your device is to: Go to the default palm launcher (if your device crashes there is probable a non .prc/.pdb file in the card launcher directory), flush db cache (if your device crashes you probable have a non NVFS aware app installed), backup with RE, (seems to crash if there is anything wrong), or do a manual soft reset (if your device goes in a reset loop you got a problem).

[BaDZeD]

4. DTG, the free version, crashes after syncing.. Out it goes!
Actually that one is quite easily fixed. You can download the DataViz Tech tool here: http://download.dataviz.com/support/patches/documentstogo/dataviztech.prc

go to "advanced options"-->delete preferences, delete database, delete temp database


That solves the reset after hotsync issue in a jiffy in most cases

[FerdFerd]

flush db cache (if your device crashes you probable have a non NVFS aware app installed), backup with RE, (seems to crash if there is anything wrong)Flushing cache when you suspect an unstable unit or corrupt file can be dangerous. See this (http://www.1src.com/forums/showpost.php?p=953277&postcount=27) post and the rest of the thread it came from before you begin using cache flushing as an intentional stabilty assay.

[BrentDC]

This problem has never happened to me, but I'm sure it's happened to others. I always use RE's flush option and not any other, but I believe that was also implemented by Pruss. I mentioned this because it frequently finds problems I did not know were there, but another way to test this is by launching a large application that needs a lot of cache space (aka Blazer), but I don't know if this is any safer. I've sometimes found just launching RLock will cause a crash if there is a non NVFS aware app installed. You would think this might be a safer way to test, although that is just a guess, and might be 100% wrong.

[maceyr]

StabilityTest (http://mytreo.net/downloads/details-522.html) at mytreo.net

I'm updating my link in the posting right now.

EDIT. I've also added a link to PalmInternals as well since that is a much more exhaustive program.

[_Em]

This problem has never happened to me, but I'm sure it's happened to others. I always use RE's flush option and not any other, but I believe that was also implemented by Pruss. I mentioned this because it frequently finds problems I did not know were there, but another way to test this is by launching a large application that needs a lot of cache space (aka Blazer), but I don't know if this is any safer. I've sometimes found just launching RLock will cause a crash if there is a non NVFS aware app installed. You would think this might be a safer way to test, although that is just a guess, and might be 100% wrong.

The RE flush option is based on Pruss' newer flushing code, which is fairly safe (as I mentioned in that other thread). Personally, I think that you have the same chance of bricking your device using RE or OffFlush (latest) as you do by forcing an OS flush as you've described. None of the methods are 100% safe, but you're also not all that likely to run into the memory manager bug.

I've bricked 2 TXes: one was using Pruss' very early flushing routine, as implemented in NVBackup (note: it's not there now), the other was done by having UDMH installed, filling up the internal NVFS partition, filling up my DB Cache, and then trying to run a memory intensive program, even after receiving an error that there was a critical memory problem.

Also: I think there's a thread on here somewhere where we're trying to build a custom live Linux CD designed specifically to "un-brick" NVFS devices hosed in the above manner. Nothing's completed yet (for one thing, we're still trying to figure out how to reliably enter debug mode over USB), but this is definitely something to help out with / look for if your NVFS device is out of warranty.