[linux-lvm] q: BLOCK_SIZE

[linux-lvm] q: BLOCK_SIZE

[Michael Tokarev]

This is a repost, basically.  When I posted this question
for a first time (8 Dec 2000), noone answered, but there
was a way low traffic on this list that time...

Little question that I already posted here a long time ago.
What drawbacks can suffer if one changes BLOCK_SIZE in
include/linux/lvm.h from 1024 (or 4096 on S/390) to 512?
I can't understand from code what't a real purpose of
this constant, -- lvm seemed to work well with BLOCK_SIZE=512,
no troubles so far, -- but i'm still not shure if it is
"legal" to change that.  At least there is some purpose on
having different values for this constant on S/390 and other
platforms, -- again, can't see that purposes...

This is about using lvm with rawio -- on 2.2, rawio wan't
work with i/o size be not a multiple of blocksize, while
on 2.4 it works.  Or at least this was on Dec-2000.

Oh yes, this is about 2.2 kernel and lvm-0.9.

Thank you.

Regards,
 Michael.

Re: [linux-lvm] q: BLOCK_SIZE

[Andreas Dilger]

Michael writes:
> Little question that I already posted here a long time ago.
> What drawbacks can suffer if one changes BLOCK_SIZE in
> include/linux/lvm.h from 1024 (or 4096 on S/390) to 512?

On S/390 it would appear that the DASD block devices have a 4096
byte sector size.  However, on most other systems there is only
a 512 byte sector size, so I don't see any reason there would
be problems with BLOCK_SIZE = 512.

I think Andrea Arcangeli was complaining about this as well.

> This is about using lvm with rawio -- on 2.2, rawio wan't
> work with i/o size be not a multiple of blocksize, while
> on 2.4 it works.  Or at least this was on Dec-2000.
> Oh yes, this is about 2.2 kernel and lvm-0.9.

If it works for you at 512 bytes, leave it then.  Maybe Heinz
will let us know why it is set to 1024 bytes.

Cheers, Andreas
-- 
Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
                 \  would they cancel out, leaving him still hungry?"
http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert

Re: [linux-lvm] q: BLOCK_SIZE

[Joe Thornber]

On Tue, Feb 27, 2001 at 01:25:42PM -0700, Andreas Dilger wrote:
> Michael writes:
> > Little question that I already posted here a long time ago.
> > What drawbacks can suffer if one changes BLOCK_SIZE in
> > include/linux/lvm.h from 1024 (or 4096 on S/390) to 512?
> 
> On S/390 it would appear that the DASD block devices have a 4096
> byte sector size.  However, on most other systems there is only
> a 512 byte sector size, so I don't see any reason there would
> be problems with BLOCK_SIZE = 512.
> 
> I think Andrea Arcangeli was complaining about this as well.

I checked Andrea's patch into cvs, try and see.

- Joe

Re: [linux-lvm] q: BLOCK_SIZE

[Heinz J. Mauelshagen]

On Tue, Feb 27, 2001 at 01:25:42PM -0700, Andreas Dilger wrote:
> Michael writes:
> > Little question that I already posted here a long time ago.
> > What drawbacks can suffer if one changes BLOCK_SIZE in
> > include/linux/lvm.h from 1024 (or 4096 on S/390) to 512?
> 
> On S/390 it would appear that the DASD block devices have a 4096
> byte sector size.  However, on most other systems there is only
> a 512 byte sector size, so I don't see any reason there would
> be problems with BLOCK_SIZE = 512.
> 
> I think Andrea Arcangeli was complaining about this as well.
> 
> > This is about using lvm with rawio -- on 2.2, rawio wan't
> > work with i/o size be not a multiple of blocksize, while
> > on 2.4 it works.  Or at least this was on Dec-2000.
> > Oh yes, this is about 2.2 kernel and lvm-0.9.
> 
> If it works for you at 512 bytes, leave it then.

Yep.

> Maybe Heinz
> will let us know why it is set to 1024 bytes.

It is basically an approach to support larger LV sizes than 2T :)

Because the Linux kernel supports 2^32 blocks per block minor up to version 2.4,
this is one approach (with tradeoffs for sure) to raise that limit.

With the 4k block size on S/390 DASD devices this means 16T per LV!

Sure: we won't have a satisfying solution before Linux 2.5.

> 
> Cheers, Andreas
> -- 
> Andreas Dilger  \ "If a man ate a pound of pasta and a pound of antipasto,
>                  \  would they cancel out, leaving him still hungry?"
> http://www-mddsp.enel.ucalgary.ca/People/adilger/               -- Dogbert
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm

-- 

Regards,
Heinz    -- The LVM Guy --

*** Software bugs are stupid.
    Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen                                 Sistina Software Inc.
Senior Consultant/Developer                       Am Sonnenhang 11
                                                  56242 Marienrachdorf
                                                  Germany
Mauelshagen@Sistina.com                           +49 2626 141200
                                                       FAX 924446
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-