Re: Agenda for ext3 developer interlock call (June 21th, 2006, 8:00PCT)

Hi,

On Tue, 2006-06-27 at 16:10 -0700, Andrew Morton wrote:

> > A similar split of the fs/jbd will be made in order to
> > support 64-bit jbd, which will be used by fs/ext4 and future versions
> > of ocfs2.
> 
> That's a bit sad.  I'd have thought that the JBD codebase could remain
> unforked - use a typedef and some kbuild trickery.
> 
> Or is something more that 64-bit-conversion being proposed?

Not at first, but there are other things we may want to do which would
be much more intrusive to the jbd logic.  In particular, there are 3
performance changes I'd like to try, one of which is on-disk
incompatible with current jbd, the other two of which only change the
logic internally:

      * Checksummed commit blocks (from Iron FS, allows us to avoid the
        barrier before commit write);

      * More extensive use of bio barriers rather than waiting for IOs;

      * Pessimistic copying of journaled metadata for IO, to prevent
        other transactions from stalling when they try to modify a
        metadata block that is already in-flight to the journal.

Given that 64-bits jbd is format-incompatible anyway, this seems like a
sensible place to fork the code if we're doing that for ext3 anyway.

--Stephen


Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642