@kas1e

Waited on google for 30 minutes circa then tried to type an address, after pressing "enter" the loading gauge at the bottom start to be corrupted (the text i mean) and then OWB freeze as always.

For an auto-refresh site goto --> www.repubblica.it

Edit:

It crash alone after a few time because when it's true that it doesn't crash early then it crash aniway after a long/short period of time, for example if you load google and type somethings in search it works without problems, but if you load the same page but waiting on the main page then it crash as explained above.

So i presume that even when you don't touch OWB at all it continue to works in someway in background.

BTW: on OWB about page this problem doesn't exists

Last edited by samo79 on 25-Feb-2011 at 12:43 PM.
Last edited by samo79 on 25-Feb-2011 at 12:38 PM.

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@saimo79

So, its crashes when downloading is starts exactly, not on empty place. Exactly when OWB dl something. I notice now that all crashes happens when downloading in progress , and then crash.

@all
Have 3st crash log, but now it looks strange: OWB crash info just skips, and all the crash sounds like "digiclock" are crashes (but its not, its just OWB info skips by some reassons). But all the other info are with stack trace, maybe can point on some ideas .. Can be of course just fully not related , but well more info the better and all those words about "locale.library" sounds a bit strange ..

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e
Can't think of others right now, but it should work fine without Flash (it's just for the ads). Have used it in Reaction OWB many times.

_________________

@kas1e

Don't know for sure, maybe ...

For restrict the area of the possible responsable you may try to comment step by step all the code related to:

locale.library
codeset.library
cURL (maybe trying to report it)

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@samo79

For first need to understand 100% when exactly crashes happens, in which condition, and what can cause it. Timers, locale or when you type something or not type. When you wait or not wait. I.e. something which can be still reproducable somehow.

Last edited by kas1e on 25-Feb-2011 at 12:53 PM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

For my experience just "the time" is responsable of the lookups, not a particular site or a particular user interaction/site visited ...

BTW: the internal about page it's the only page without problems, you can leave it open for ours without any lookups

Last edited by samo79 on 25-Feb-2011 at 01:00 PM.

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@samo79

Its never crashes "fast" ? I.e. only after a while ? I for now will try to remove from wbstartup digiclock, in hope that somehow , on one of crashes i again can catch stack trace.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@samo79
Btw, did you use digiclock ? (just to avoid any possible "maybe its related to digiclock")

EDIT: nope, not related, already remove it, and have 4st crash, which looks heavy enough now (but visually , all the same, the same "stop on a second", and then lockup).

4st crash

Now its again skip owb info crash, and show only "ramlib" crash

Last edited by kas1e on 25-Feb-2011 at 01:09 PM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@all
And crash #5: here

For now i browse the net for 10-15 mins, then on some site press "download", download window spawns and then all stops on 1 second, and then lockup.

As i say before, in the download manager names of files are trashed, what already point on the memory corruption imho , and very possible its related.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

Quote:

Its never crashes "fast" ? I.e. only after a while ? I for now will try to remove from wbstartup digiclock, in hope that somehow , on one of crashes i again can catch stack trace.

Ya only after a while ...

Quote:

Btw, did you use digiclock ? (just to avoid any possible "maybe its related to digiclock")

Yes i use it but i'm not entire sure that problems is caused by it, i try to remove it aniway.

BTW: Roman i think that all that experiments are quite useless, better if you try commentings some code instead, locale.library support for example, start from it

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@samo79

Quote:

BTW: Roman i think that all that experiments are quite useless, better if you try commentings some code instead, locale.library support for example, start from it

Yep, i will start from charsets.library first (that one which we use for 2 small functions), and its easy to remove.

Then next will be locale stuff (harder).

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

Its not codesets.library realated. The same crash without (6st crash )

Last edited by kas1e on 25-Feb-2011 at 03:23 PM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

Yeah it's sad and annoying i know, but i think it would be the unique mothod to understand where problem is

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@samo79

With help of Fab already fixed now "garbage in filenames", but very possible its non-related.

The good news, that after crash i can still ping os4, what mean that tcp/ip stack working, and that mean that there is possibility to use some remote stuff, and try to dump some memory content there and there.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

Well atleast a couple of good news

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

Hello kas1e.

first congratulations for your hard and fast progressing!

Probably you already know but i would try to delete every OWB-MUI mui prefs in Envarc:MUI/ after every crash you get. A corrupted mui prefs, could land you to some un-predictable/random issues.

Good luck

_________________

@kas1e

Deniil made good point elsewhere that r12 often (always?) contains 0xDEADF00D. Maybe it oculd be good idea disassemble whole OWB binary and check if there are many references to r12. It is not necessarily used very much because there are plenty of other spare registers used first.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@kas1e

Quote:

kas1e wrote: @COBRA Yes, always in background.

Does the same crash happen when you're not using memguard?

@kas1e Quote:

The nature of crash the same as before: all works works, then at some point small stop on a second, then all continue to works for 2 seconds more, and then window totally lockup. On this stage OWB crashes, GR/digiclock and workbnench itself crashes.

Just a wild guess (and maybe stating the obvious) : The brief pause could be it following some garbage/dangling pointer in a list (or similar), it then spends a long time following garbage nodes (effectively busy looping hence the pause), until it happens to reach a NIL memory location which causes it to terminate (as it interprets it as the end of the list). But that leaves OWB in a corrupt state (and/or corrupts memory during list traversal) so that some moments later it runs into the results of that corruption (doing different operation) and fully crashes.

But even if that were true, not sure how that helps you track down the exact location of the crash...

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@kas1e Quote:

After i leave google.com for 40 minuts , then i type in google "blalba", do search. Then go to arosworld.org - same crash. If i reboot, and go to arosworld.org - no crash.

I guess that most likely, after a reboot, memory is mostly 0x00000000, and so dangling/garbage pointer is most likely to point to such locations, causing list traversal (or similar) to terminate without bad effect. But after using OWB or OS4 for a while, memory contains garbage from being used, and so dangling/garbage pointer more likely to not point to memory containing 0x00000000, and so crashes more likely.

Maybe if you filled all memory with non-0x00000000 before running OWB, then the crash would happen sooner & more reliably?

Last edited by ChrisH on 27-Feb-2011 at 11:07 AM.
Last edited by ChrisH on 27-Feb-2011 at 11:06 AM.

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...