1) Wondering what the best way to report possible bugs?

The couple of things I saw so far is (OS4 Classic):

2) version c:fs_#? brings up the grim reaper (I get a 8000 0003 guru).

"Version" does not like one or more of these:
fs_plugin_cache fs_plugin_encrypt fs_set_flush_strategy

[EDIT] @Framiga:
5.AOS4:C> version fs_plugin_cache
fs_plugin_cache 52.1

is ok, while "fs_plugin_encrypt" and "fs_set_flush_strategy" makes the command "version" crash


3) MediaToolbox doesn't seem to recognize ID15 for the cyppc.device (i.e. setting HDD jumper to unit 15)...need to try this one more time to confirm but I think this is true. [EDIT: Yes, I am still seeing this]

Some minor things:
4) Page 28 in Quickstart guide mentions the SDK but I didn't find it on the CD...
Found it here -- for the curious like me:
ftp://ftp.hyperion-entertainment.biz/AmigaOS4_SDK
5) Think there should be some mention that the user needs to install the network device (i.e. like for me it was the Ariadne-II device driver).

[EDIT]

6) Memory not getting free -- just noticed this in the "Sound" prefs program:
In the "System startup sound", when I select a new sample and then repeatedly
test the sound by clicking on the speaker, more and more of my chip mem
is allocated, until I get a AmigaDOS requester that says it no longer can
load/play the sound. Quiting the program and then using "avail flush" does not free memory...right now I have 30,896 bytes of chip mem.

Last edited by my_pc_is_amiga on 01-Jan-2008 at 05:55 PM.
Last edited by my_pc_is_amiga on 01-Jan-2008 at 05:53 PM.

There should be a sticky thread about problems with OS4.

For intance:

On my Highway USB card the system is prone to lock-ups under heavy disk-access (possibly only write). This has happened on different devices, too - A4000 IDE and CSPPC SCSI.

Also memory reporting is a little knackered. Try filling your RAM disk with a 30MB file. Then delete the file. Watch how your free memory doesn't go back up. It's freeing the memory ok, it's just not noticing it.

There should be somewhere for this sort of problem, I think.. as long as people don't hijack the thread. It'd perhaps have to be heavily moderated to stay on topic.

@IanO

Has anyone noticed that the Acube systems web site has disappeared!!! Looks like they forgot to renew their subs for 2008. The Compatibility list has disappeared as well

http://www.acube-systems.com/compatibility/

_________________
Peter J Hutchison
http://www.pjhutchison.org/

5.AOS4:C> version fs_plugin_cache
fs_plugin_cache 52.1

is ok, while "fs_plugin_encrypt" and "fs_set_flush_strategy" makes the command "version" crash

Last edited by Framiga on 01-Jan-2008 at 03:33 PM.
Last edited by Framiga on 01-Jan-2008 at 03:27 PM.

_________________

@pjhutch

Quote:

pjhutch wrote: @IanO Has anyone noticed that the Acube systems web site has disappeared!!! Looks like they forgot to renew their subs for 2008. The Compatibility list has disappeared as well http://www.acube-systems.com/compatibility/

http://www.acube-systems.biz/compatibility/

they are switching server ... don't worry thay are alive and kicking!

Official source

Last edited by Framiga on 01-Jan-2008 at 11:51 AM.

_________________

@Spirantho

Quote:

Spirantho wrote: Also memory reporting is a little knackered. Try filling your RAM disk with a 30MB file. Then delete the file. Watch how your free memory doesn't go back up. It's freeing the memory ok, it's just not noticing it.

This is expected and documented behaviour. OS4 caches free memory.
See http://os4.hyperion-entertainment.biz/

@Framiga

Quote:

Framiga wrote: 5.AOS4:C> version fs_plugin_cache fs_plugin_cache 52.1 is ok, while "fs_plugin_encrypt" and "fs_set_flush_strategy" makes the command "version" crash

Tried to reproduce this on A1 but no luck. Version command 52.2, the utilities have
version 52.1 or no version string here.

@TetiSoft

5.AOS4:C> version C:version full
Version 52.2 ( 9-Dic-2006)


the 2 fs_ componets that crashes the version command are:

version fs_plugin_encrypt

version fs_set_flush_strategy

from shell with default stack 36.856 byte and even rised to 100.000

elf.library.kmod involved too

Crashlog here

_________________

@Framiga

Well, it seems the two binaries on the OS4Classic install CD are broken.
Trying to run them (or using objdump) causes a crash too.
The crashlog looks like a stack overflow caused by endless recursion.
If you have access to an A1 with OS4, copy the binaries from there,
otherwise delete them. Sorry.

@TetiSoft

no problems, i don't need it

_________________

@TetiSoft

Quote:

TetiSoft wrote: @Spirantho Quote: Spirantho wrote: Also memory reporting is a little knackered. Try filling your RAM disk with a 30MB file. Then delete the file. Watch how your free memory doesn't go back up. It's freeing the memory ok, it's just not noticing it. This is expected and documented behaviour. OS4 caches free memory. See http://os4.hyperion-entertainment.biz/

I don't mean to sound argumentative here, but no, it's not. :)

It doesn't matter if OS4 caches free memory or not, the fact is that the memory is available to be used and therefore should be reported as such.

For instance:

You have 64 MB free RAM.
You allocate 30MB by copying a 30MB file into the RAM disk.
You then delete the file. This will free 30MB won't it? It must do - before the delete, you could safely allocate 34MB. After the delete you can safely allocate 64MB.
Therefore, the memory system should be reporting that 64MB is free. It doesn't matter what's going on under the hood, if 64MB is available to be used, then it should say 64MB is free, not 34MB, otherwise what's the point in it?

If you want proof that the free memory reporting is broken I'm happy to send you a screenshot of it saying that I have a negative amount of memory left! Just for a laugh last time it went negative (-400KB I think) I wanted to see how far it could go by opening WB windows - I got it to -1.7MB free before the system hung. Maybe something to do with Chip RAM?

I know you've been involved in OS 4 and I know you stick up for it whenever possible but the fact is there are bugs in OS 4 and denying it is just shooting yourself in the foot. For instance if you want proof that the HighwayUSB driver causes quite severe instability at least in my system then I'm more than willing to show you, and given the fact that nobody else has said "works fine here" I'm not sure how well it's been tested. I may well drop Rogue a line and see if I can't get the Highway driver source and see what's happening. I wonder if it's related to interrupt-handling, given the behaviour is precisely what I observed with the Bt8x8 driver when that fired off too many interrupts (that was on my AmigaIOne of course).

What we need is not an excuse for odd behaviour in the OS, we need a place where we as users can report problems as specifically as possible to held the people behind the OS fix the bugs. OS 4 is a superb OS - it's an amazing job. But it's not perfect and unless we the users are listened to it'll never mature!

a little warn for Classic users...

avoid to use CPU-Watcher on Classic. The cpu temp rised about 4 degrees C

Fortunately my cooling system alarm, bleeped for me!

No such a problem with the docky tools

_________________

@Spirantho

"Just for a laugh last time it went negative (-400KB I think) I wanted to see how far it could go by opening WB windows - I got it to -1.7MB free before the system hung."

I also have seen negative memory for "other mem" in the Workbench title bar in the same neighborhood...(about -1MB)

"we need a place where we as users can report problems as specifically as possible to held the people behind the OS fix the bugs."

I wouldn't mind seeing this also...

Last edited by my_pc_is_amiga on 02-Jan-2008 at 04:15 PM.

@everybody
Happy new year

@TetiSoft
I guess I`ll have to use a separate ACTIVE terminator, just to be sure (I only have to get one of these things to my hands).

@Framiga
Is it safe to set the flag with this utility?
I`ve used once the CSPPC-SCSI disk to recover-set the rdb iirc, when I tried to install sfs with NetBSD & FFS partitions... (I`l have to check my posts)

_________________
Ordell Robbie: Is she dead, yes or no?
Louis: Pretty much.

@Cass

yes absolutely safe.

_________________

@TetiSoft
In case this is at all useful to compare with the AmigaOne version,
here's the checksum for the versions on the classic release.

4.AmigaOS4:> md5sum c:fs_plugin_cache c:fs_plugin_encrypt c:fs_set_flush_strategy
7ca370f396b5190fd13f97f978ec3727 c:fs_plugin_cache
20a1ad9b3905e722d231dc9668a87c2c c:fs_plugin_encrypt
99897cab84f4183a8c1349dcf670cc30 c:fs_set_flush_strategy

@my_pc_is_amiga

It should be possible for everybody with the public OS4SDK to recompile the fs_#? files. The source is in the SDK:Examples/FastFileSystem dir.

@my_pc_is_amiga

Your md5sums disagree with those on my A1. The md5sum for c:fs_plugin_cache is correct, the other two are not.

@Spirantho:
The main problem with the memory reporting is that there is no longer any easy and accurate way for the Exec to count or estimate the amount of "unused, available" memory. Old functions and applications that "grew up" depending on values returned by AvailMem are still supported (in a way, ie they don't crash), but the approach of programmers has to change with the OS.

In the old days of devices like the C64, it was possible to calculate exactly how much memory was free because the Exec knew exactly where the "free" part was and how big it is. However, as the OS has grown up, memory is now reserved for so many reasons by so many applications that it becomes impractical to go through and add them all up just so that someone can see a number on his WB title bar. Hence the number that you get now is at best a guess at the amount of free memory.

The only real test of whether you have enough free memory to do such-and-such is to try allocating it - if it worked, you had enough, if it didn't work, then the system tried really hard to get enough memory for you, but couldn't scrape up enough for your request. You can imagine the Exec saying "Sorry. By the way, here is a guess of how much there is spare. That guess was taken at 11:44:58 and may have changed since then."

_________________
cheers
tony

Hyperion Support Forum: http://forum.hyperion-entertainment.biz/index.php

@tonyw

Thanks for the response - it makes sense. However, I think if that's the case then the "free memory" amount should be removed completely. I'd rather not have a readout at all than one which at best was inaccurate and at worst completely meaningless (e.g. negative memory amounts).

Given that the OS is in control of all memory allocation and de-allocation, though, would it not be possible just to keep a tally of how much is allocated? You know exactly how many bytes you start out with, and you know exactly how many bytes are allocated with each call to alloc() (or the equivalent). Therefore at any given time the OS should know from its own internal counter how much memory is available to the system. Even with a garbage collection system the internal calls should still use free() so the memory readout could be accurate.
What this wouldn't do, of course, is guarantee that an allocation of a certain size would work, because of the nature of fragmented memory - although this should be somewhat lessened if not eliminated with the new memory system.

When we have working virtual memory, though, we can probably do without the free memory reporting altogether.

@ZeroG

Based on threads, there seems to be something wrong with those 2 fs_plugn#? files...but the one obvious thing that I haven't heard is why "version" should
crash...instead of crashing it should give an error instead...i.e. the old OS3.9 version command works fine...no guru!

Note:
Workbench:c/version --> OS3.9
c/version --> OS4

---------------------------------------------------------
6.AmigaOS4:> workbench:c/Version workbench:c/Version full
Version 44.4 (09/30/1999)
6.AmigaOS4:> workbench:c/Version c/Version full
Version 52.2 (12/09/2006)
6.AmigaOS4:> workbench:c/Version c/fs_plugin_cache full
fs_plugin_cache 52.1 (12/09/2006)
6.AmigaOS4:> workbench:c/Version c/fs_plugin_encrypt full
fs_plugin_encrypt 52.1 (12/09/2006)
6.AmigaOS4:> workbench:c/Version c/fs_set_flush_strategy full
Could not find version information for 'c/fs_set_flush_strategy'