Smalltalk › Cincom › VisualWorks

Unable to Allocate Memory

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

5 messages Options

Travis Griggs

Unable to Allocate Memory

Even after porting to VW74, we still see this error. It seems to
happen "after a while." Sounds like a memory leak, huh? Originally it
was. And by switching to Announcements we got rid of that. And now we
still get it after a while. This time I'm sure there's not a memory
leak. Top, printing all memory, etc. There's lots of memory left. The
common theme I've noticed now seem to be allocating _fixed_ memory
though.

I'm off to do dig through the VM source. I wonder if others when
they've seen these problems have been using Fixed memory. Does anyone
else even really used Fixed memory. Our app uses it heavily. In large
4MB chunks.

--
Travis Griggs
Objologist
"Dying men never wish they'd spent more time at the office"

-----------------------------------------
DISCLAIMER: This email is bound by the terms and conditions
described at
http://www.key.net/disclaimer.htm

Joerg Beekmann, DeepCove Labs (YVR)

RE: Unable to Allocate Memory

When is fixed space compacted? I don't think that ever happens while the
image is running. If this is true, smaller allocations may be
fragmenting the space preventing the allocation of a contiguous piece of
4MB.

Joerg

-----Original Message-----
From: Travis Griggs [mailto:[hidden email]]
Sent: Thursday, February 16, 2006 8:31 AM
To: VW NC
Subject: Unable to Allocate Memory

Even after porting to VW74, we still see this error. It seems to
happen "after a while." Sounds like a memory leak, huh? Originally it
was. And by switching to Announcements we got rid of that. And now we
still get it after a while. This time I'm sure there's not a memory
leak. Top, printing all memory, etc. There's lots of memory left. The
common theme I've noticed now seem to be allocating _fixed_ memory
though.

I'm off to do dig through the VM source. I wonder if others when
they've seen these problems have been using Fixed memory. Does anyone
else even really used Fixed memory. Our app uses it heavily. In large
4MB chunks.

--
Travis Griggs
Objologist
"Dying men never wish they'd spent more time at the office"

-----------------------------------------
DISCLAIMER: This email is bound by the terms and conditions
described at
http://www.key.net/disclaimer.htm

Travis Griggs

Re: Unable to Allocate Memory

On Feb 16, 2006, at 8:39, Joerg Beekmann wrote:

> When is fixed space compacted? I don't think that ever happens
> while the
> image is running. If this is true, smaller allocations may be
> fragmenting the space preventing the allocation of a contiguous
> piece of
> 4MB.

I guess I've always taken it for granted. Not that I'm naive enough I
don't understand fragmentation and the effects thereof. But after
reading enough GC literature, I guess I've bought into the
"fragmentation worries are usually overrated" sentiment. I will look
into this some more because I'm curious what really is going on. It
would be handy of the VM had some mechanism of showing fixed space
fragmentation...

That aside. It still doesn't make sense why it fails. OK, so I'm
making poor use of my fixed space (possibly). But I've got lots more
of it. Why is it deciding prematurely to cease letting me fragment
even more fixed space. Unless there's a limit hidden somewhere that I
don't see. I've looked at MemoryPolicy and don't see anything that
looks like it does that. Though I'm not sure I understand completely
the context of "threaded data" in said classes comment.

I've put together a little "showFixedSpaceStats" on ObjectMemory to
put in my end application and watch what it reports over time. The
interesting thing is to run it in my dev image. Where there are no
fixed space objects, or are there? Because it shows that I have:
204800 fixed space bytes total
0 used
1 fixed object

That doesn't make sense. Maybe there's an object out there with a
fixed space body of 0 size? So I do an:

ObjectMemory allObjectsDo: [:each | each occupiesFixedSpace ifTrue:
[each halt]]

Run that, and none comes up.
(occupiesFixedSpace is a variant of isFixedArgument, but it does not
include immediates). So that's just weird.

I hope Eliot is watching this thread. Would love to have his thoughts
on some of this.

--
Travis Griggs
Objologist
"Conference calls: the appearance of doing work without the substance
thereof" - Bruce Boyer

-----------------------------------------
DISCLAIMER: This email is bound by the terms and conditions
described at
http://www.key.net/disclaimer.htm

Boris Popov, DeepCove Labs (SNN)

RE: Unable to Allocate Memory

In reply to this post by Travis Griggs

From page 8 of doc\TechNotes\vwMemoryMgmt.pdf,

<quote>
Since the contents of Fixed Space can't move, it cannot be compacted,
and so quickly becomes fragmented. Fixed Space is coallesced at image
start-up, so can be compacted by saving the image, quitting, and then
restarting.
</quote>

Hope this helps, it's a somewhat comprehensive document depending on how
much detail you may be looking for,

Cheers!

-Boris

--
+1.604.689.0322
DeepCove Labs Ltd.
4th floor 595 Howe Street
Vancouver, Canada V6C 2T5

[hidden email]

CONFIDENTIALITY NOTICE

This email is intended only for the persons named in the message
header. Unless otherwise indicated, it contains information that is
private and confidential. If you have received it in error, please
notify the sender and delete the entire message including any
attachments.

Thank you.

> -----Original Message-----
> From: Travis Griggs [mailto:[hidden email]]
> Sent: Thursday, February 16, 2006 9:37 AM
> To: VW NC
> Subject: Re: Unable to Allocate Memory
>
>
> On Feb 16, 2006, at 8:39, Joerg Beekmann wrote:
>
> > When is fixed space compacted? I don't think that ever
> happens while
> > the image is running. If this is true, smaller allocations may be
> > fragmenting the space preventing the allocation of a
> contiguous piece
> > of 4MB.
>
> I guess I've always taken it for granted. Not that I'm naive
> enough I don't understand fragmentation and the effects
> thereof. But after reading enough GC literature, I guess I've
> bought into the "fragmentation worries are usually overrated"
> sentiment. I will look into this some more because I'm
> curious what really is going on. It would be handy of the VM
> had some mechanism of showing fixed space fragmentation...
>
> That aside. It still doesn't make sense why it fails. OK, so
> I'm making poor use of my fixed space (possibly). But I've
> got lots more of it. Why is it deciding prematurely to cease
> letting me fragment even more fixed space. Unless there's a
> limit hidden somewhere that I don't see. I've looked at
> MemoryPolicy and don't see anything that looks like it does
> that. Though I'm not sure I understand completely the context
> of "threaded data" in said classes comment.
>
> I've put together a little "showFixedSpaceStats" on
> ObjectMemory to put in my end application and watch what it
> reports over time. The interesting thing is to run it in my
> dev image. Where there are no fixed space objects, or are
> there? Because it shows that I have:
> 204800 fixed space bytes total
> 0 used
> 1 fixed object
>
> That doesn't make sense. Maybe there's an object out there
> with a fixed space body of 0 size? So I do an:
>
> ObjectMemory allObjectsDo: [:each | each occupiesFixedSpace ifTrue:
> [each halt]]
>
> Run that, and none comes up.
> (occupiesFixedSpace is a variant of isFixedArgument, but it
> does not include immediates). So that's just weird.
>
> I hope Eliot is watching this thread. Would love to have his
> thoughts on some of this.
>
> --
> Travis Griggs
> Objologist
> "Conference calls: the appearance of doing work without the
> substance thereof" - Bruce Boyer
>
>
>
> -----------------------------------------
> DISCLAIMER: This email is bound by the terms and conditions
> described at http://www.key.net/disclaimer.htm
>
>

smime.p7s (4K) Download Attachment

Terry Raymond

RE: Unable to Allocate Memory

This is a bit extreme, but you should be able to "compact"
fixed space by emptying it, doing a gc, then refilling it.
Of course this may not be practical.

Terry

===========================================================
Terry Raymond Smalltalk Professional Debug Package
Crafted Smalltalk
80 Lazywood Ln.
Tiverton, RI 02878
(401) 624-4517 [hidden email]
<http://www.craftedsmalltalk.com>
===========================================================

> -----Original Message-----
> From: Boris Popov [mailto:[hidden email]]
> Sent: Thursday, February 16, 2006 12:51 PM
> To: Travis Griggs; VW NC
> Subject: RE: Unable to Allocate Memory
>
> From page 8 of doc\TechNotes\vwMemoryMgmt.pdf,
>
> <quote>
> Since the contents of Fixed Space can't move, it cannot be compacted,
> and so quickly becomes fragmented. Fixed Space is coallesced at image
> start-up, so can be compacted by saving the image, quitting, and then
> restarting.
> </quote>
>
> Hope this helps, it's a somewhat comprehensive document depending on how
> much detail you may be looking for,
>
> Cheers!
>
> -Boris
>
> --
> +1.604.689.0322
> DeepCove Labs Ltd.
> 4th floor 595 Howe Street
> Vancouver, Canada V6C 2T5
>
> [hidden email]
>
> CONFIDENTIALITY NOTICE
>
> This email is intended only for the persons named in the message
> header. Unless otherwise indicated, it contains information that is
> private and confidential. If you have received it in error, please
> notify the sender and delete the entire message including any
> attachments.
>
> Thank you.
>
>
> > -----Original Message-----
> > From: Travis Griggs [mailto:[hidden email]]
> > Sent: Thursday, February 16, 2006 9:37 AM
> > To: VW NC
> > Subject: Re: Unable to Allocate Memory
> >
> >
> > On Feb 16, 2006, at 8:39, Joerg Beekmann wrote:
> >
> > > When is fixed space compacted? I don't think that ever
> > happens while
> > > the image is running. If this is true, smaller allocations may be
> > > fragmenting the space preventing the allocation of a
> > contiguous piece
> > > of 4MB.
> >
> > I guess I've always taken it for granted. Not that I'm naive
> > enough I don't understand fragmentation and the effects
> > thereof. But after reading enough GC literature, I guess I've
> > bought into the "fragmentation worries are usually overrated"
> > sentiment. I will look into this some more because I'm
> > curious what really is going on. It would be handy of the VM
> > had some mechanism of showing fixed space fragmentation...
> >
> > That aside. It still doesn't make sense why it fails. OK, so
> > I'm making poor use of my fixed space (possibly). But I've
> > got lots more of it. Why is it deciding prematurely to cease
> > letting me fragment even more fixed space. Unless there's a
> > limit hidden somewhere that I don't see. I've looked at
> > MemoryPolicy and don't see anything that looks like it does
> > that. Though I'm not sure I understand completely the context
> > of "threaded data" in said classes comment.
> >
> > I've put together a little "showFixedSpaceStats" on
> > ObjectMemory to put in my end application and watch what it
> > reports over time. The interesting thing is to run it in my
> > dev image. Where there are no fixed space objects, or are
> > there? Because it shows that I have:
> > 204800 fixed space bytes total
> > 0 used
> > 1 fixed object
> >
> > That doesn't make sense. Maybe there's an object out there
> > with a fixed space body of 0 size? So I do an:
> >
> > ObjectMemory allObjectsDo: [:each | each occupiesFixedSpace ifTrue:
> > [each halt]]
> >
> > Run that, and none comes up.
> > (occupiesFixedSpace is a variant of isFixedArgument, but it
> > does not include immediates). So that's just weird.
> >
> > I hope Eliot is watching this thread. Would love to have his
> > thoughts on some of this.
> >
> > --
> > Travis Griggs
> > Objologist
> > "Conference calls: the appearance of doing work without the
> > substance thereof" - Bruce Boyer
> >
> >
> >
> > -----------------------------------------
> > DISCLAIMER: This email is bound by the terms and conditions
> > described at http://www.key.net/disclaimer.htm
> >
> >