Freeing boot memory

Freeing boot memory

[Leandro Motta Barros]

Hello,

I and a colleague are studying the VM subsystem (actually this is the first 
time we are examining the Linux source code more closely) and have a question 
or two.

Well, the questions concern the boot memory allocator. To be more precise, 
We're interested in the memory deallocation routines. We have seen that it is 
only possible to free full pages. So, theoretically, if we make several 
allocations smaller than one page, we will not be able to actually free this 
memory. I just don't know of this kind of situation happens in real life. Do 
we currently have some pages of memory "wasted" because the boot memory 
allocator was not able to free small allocations? Is there any estimate (or 
benchmark or whatever) on the number of pages that could be freed but are 
not?

We have interest in hacking a little bit in the VM, and we thought that trying 
to find out ways to avoid this problem (if this is really a problem) could be 
nice. Do you have any thoughts about this?

Thanks a lot,

LMB
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: Freeing boot memory

[Martin J. Bligh]

> I and a colleague are studying the VM subsystem (actually this is the first 
> time we are examining the Linux source code more closely) and have a question 
> or two.
> 
> Well, the questions concern the boot memory allocator. To be more precise, 
> We're interested in the memory deallocation routines. We have seen that it is 
> only possible to free full pages. So, theoretically, if we make several 
> allocations smaller than one page, we will not be able to actually free this 
> memory. I just don't know of this kind of situation happens in real life. Do 
> we currently have some pages of memory "wasted" because the boot memory 
> allocator was not able to free small allocations? Is there any estimate (or 
> benchmark or whatever) on the number of pages that could be freed but are 
> not?
> 
> We have interest in hacking a little bit in the VM, and we thought that trying 
> to find out ways to avoid this problem (if this is really a problem) could be 
> nice. Do you have any thoughts about this?

What would you *do* with this half a page? There's no main memory allocator
to stick it in, as far as I can see.

M.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: Freeing boot memory

[William Lee Irwin III]

On Tue, Sep 16, 2003 at 06:17:37PM -0300, Leandro Motta Barros wrote:
> Well, the questions concern the boot memory allocator. To be more precise, 
> We're interested in the memory deallocation routines. We have seen that it is 
> only possible to free full pages. So, theoretically, if we make several 
> allocations smaller than one page, we will not be able to actually free this 
> memory. I just don't know of this kind of situation happens in real life. Do 
> we currently have some pages of memory "wasted" because the boot memory 
> allocator was not able to free small allocations? Is there any estimate (or 
> benchmark or whatever) on the number of pages that could be freed but are 
> not?

When I rewrote this, I got crapped on. I highly doubt anyone will
listen this time, either.


-- wli
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>