@ChrisH

This is pretty interesting. You don't have to rely on the Workbench titlebar to report free memory. You can use the avail command in a shell window.

_________________
X5000 / microA1(OS4.1 FE U2) / CodeBench / Imagine / Blender
Lightwave 2019 / Microsoft Visual C++

@Deniil715 Quote:

I don't see this problem on my A1-XE

The memory loss problem appears limited to OS4 on the Sam440, going by Xenic's post. You seem to confirm this...

Quote:

My guess is that some log buffer is getting larger and larger over time. Check T:

I already did, without finding anything. :(

I do NOT have "kdebug" in my startup-sequence.

@gerograph
Quote:

My SAM runs since 6 hours, I started of with 410 MB memory at the moment I am at 341MB. I did nothing than moving an clicking mouse once an hour. I did not even open a directory. BUT "memory loss" seems to become less with the time It lost 50 MB the first hour (you could watch the countdown nearly lineraly), the rest was only a loss of 17 MB.

Your experience matches what you would expect with a Slab allocator: Initially it creates a lot of Slabs for a lot of different allocation sizes, which appears as wasted memory (but isn't), but over time Slabs for all the sizes are created, so eventually no more Slabs need creating (so no more memory appears wasted).

For those with a technical interest: Slabs are NOT really wasted space, because the OS will free them if memory runs out. Also, the Slabs allocator includes an Object Cache, which will also appear to waste memory - again unless memory runs out. Thus incredibly naive comparisons of MOS memory usage vs OS4.1 memory usage will simply give totally wrong results for OS4.

Quote:

I have NOT installed the QuickFix !!!

VERY INTERESTING! It seems the memory loss bug may have been introduced by the QuickFix...

Last edited by ChrisH on 17-Sep-2009 at 05:38 PM.

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

@ChrisH

Quote:

VERY INTERESTING! It seems the memory loss bug may have been introduced by the QuickFix...

This is possible, considering the fact that the 53.8 Quickfix installs a kernel that has already been proved to be flawed (with many users reverting to the debug version).

_________________
The Rear Window blog

AmigaOne X5000/020 @ 2GHz / 4GB RAM / Radeon RX 560 / ESI Juli@ / AmigaOS 4.1 Final Edition
SAM440ep-flex @ 667MHz / 1GB RAM / Radeon 9250 / AmigaOS 4.1 Final Edition

@rigo

Moving the mouse on SAM440ep and SAM440flex appears to use up a lot of memory. It doesn't happen on an AmigaOne.

> In light of this, I shall reitterate what has been said many times before,
> free memory figures on OS4.1 are meaningless, so don't believe what you read
> in the WB titlebar.

Yet the OS4 team puts the free memory counter in the titlebar by default. Why?

Kind regards,

Dennis

@ChrisH

Since I don't have a SAM, I can't verify this situation. However, it *might* be related to USB.

In the AmigaOne version OS4.1 Prefs/USB the default settings are to log errors and log warnings, but not log information. When I enable "log information", the system logs insertion and removal of my usb devices in T:USB.log. Any other unusual activity is also logged.

Since I don't have a SAM, I don't know if the default settings are the same. If the preferences are set to log information, any unusual activity on the USB interfaces would show up in the USB.log.

---
redfox

@A3000T

Quote:

Moving the mouse on SAM440ep and SAM440flex appears to use up a lot of memory. It doesn't happen on an AmigaOne.

Then that rules out the Kernel, as the memory handling routines are not machine specific. I have my suspicions on what this may be, and if my suspicions are correct, the bug has already been fixed.

This is currently being investigated.

Simon

_________________
Simon

Comments made by me on any public fora are not representative of, or on behalf of, any company I may have, or assumed by the reader to have, any association with.

Any comments are a personal opinion, and should be accepted as such.

Quote:

VERY INTERESTING! It seems the memory loss bug may have been introduced by the QuickFix...

Yep, thats what I thought right away as I read this thread, so that's actually why I posted + to make people aware that there is a quickfix, and that a "quicked fixed SAM" behaves different" than an "out of the box" one. Because sometimes I have the impression, that people in our and threads forget that there are to OS4.1 for SAM (one normal and one quickfixed).

So, yes, If I have time, I will recheck with the Quickfix installed on my other harddisk (but this once again might take a couple of weeks... I am bussy right now..)

BTW I left the computer running another 4 hours and memory changed only about
-2 MB, then I opend up a few drawers...closed them and got +3MB back. O.k. lets forget about 1MB (not using an A500 here !!!). => NO MORE MEMORY LOSS AFTER A COUPLE OF HOURS

just grabbed a calculator, to find out how many times I have to open a few drawers to overrule 512MB every 1MB plus can thene be soled to A500 users !!!!

_________________
Geomarketing at
www.geobiz.de
www.gebietsplanung.net
www.geomarketing-consultant.de

@thread

OK, my machine has been idle for 2.5 hours here, and the memory has stayed stable. Reported free memory is the same (give or take a few K) as it was when the machine was booted.

I think it's fair to say this problem has been cured in the latest updates.

Simon

_________________
Simon

Comments made by me on any public fora are not representative of, or on behalf of, any company I may have, or assumed by the reader to have, any association with.

Any comments are a personal opinion, and should be accepted as such.

@rigo

so it was a Kernel Problem ??

_________________
GDG-Entertainment
OS4Games
Iksnet
Bitplane

@nubechecorre

No, it can't be. The memory management code in the kernel is the same on all platforms.

I think I know what it was, but it doesn't matter now, as it is fixed. Thats why the AmigaOS4.1 release for the Sam was labelled as "beta".

Simon

_________________
Simon

Comments made by me on any public fora are not representative of, or on behalf of, any company I may have, or assumed by the reader to have, any association with.

Any comments are a personal opinion, and should be accepted as such.

@rigo

I'm drooling for an update...is it soon for the rest of us?
Or should i just revert to orginal version. That graphic glitch hopefully was nailed down too...

I've seen memory going down...probably why the machine crashes after a day or so.
So I'd like to either revert back if it isn't soon or if it is I can wait...


Why does virtual show 1/2 GB7, when I think I had the swap as 1GB.


7.Workbench:> avail virtual
56553472
Workbench:> avail total
56545280

I just contribute to this one (even though should get solved in upcoming OS4.1 update for SAM):

Did a fresh Install - no memory loss at all
put Quickfix on it - no memory loss at all
plugged my USB (hama) Hub including SanDisk in memory loss began (over night it lost about all, had 5 MB left in the morning)

_________________
Geomarketing at
www.geobiz.de
www.gebietsplanung.net
www.geomarketing-consultant.de

I wish hyperion would comment on this issue... I really hope it is fixed in the final version or next update.

@rigo

Quote:

rigo wrote: @A3000T Quote: Moving the mouse on SAM440ep and SAM440flex appears to use up a lot of memory. It doesn't happen on an AmigaOne. Then that rules out the Kernel, as the memory handling routines are not machine specific. I have my suspicions on what this may be, and if my suspicions are correct, the bug has already been fixed. This is currently being investigated. Simon

Could still be the kernel.. Could be something else that is leaking memory like for example USB like some people mentioned.

@Tomas

It is the OHCI driver that leaks a little memory every time it had to handle an IORequest inside the HCD driver.

The question is now if the UHCI driver has the same problem as it uses the same memory handling code, people say no I'm not fully convinced as I have not tested it... my A1 has a dead HardDisk so I can't test it at the moment.

RWO

_________________
Debugging is a state of mind