Date: Tue, 25 Feb 2003 13:31:20 -0800 From: Marcel Moolenaar <marcel@xcllnt.net> To: Arun Sharma <arun.sharma@intel.com> Cc: freebsd-ia64@FreeBSD.ORG Subject: Re: Slow loading of the kernel from efifs ? Message-ID: <20030225213120.GA49775@ns1.xcllnt.net> In-Reply-To: <200302251946.h1PJktr26424@unix-os.sc.intel.com> References: <200302251946.h1PJktr26424@unix-os.sc.intel.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Feb 25, 2003 at 11:46:55AM -0800, Arun Sharma wrote:
>
> I've noticed that it takes a long time for the boot loader to load the
> kernel from the efifs (especially for the GENERIC kernel which is 9MB
> or so in size). I also noticed that the efifs read function is using a
> 8K block size to read data. Has anyone tried bigger block sizes ?
>
> Also, Why not read it using just one big EFI_FILE.Read call ?
The EFI loader hooks into an existing infrastructure that has to
work with anything from floppy disks, CD to BOOTP/DHCP. Clearly,
a non-optimized hook-up like we have now may not be what we want,
but it works. I don't think we'll ever have a fully EFI optimized
loader, because that conflicts with a platform independent frame-
work. In time when more important work has been done (or we have
more people to do the work) we may want to improve the EFI loader
further, possibly tweaking the infrastructure to the benefit of all.
There may be some low-hanging fruit or even some genuine bugs that
may improve this without much work, but I haven't looked at it.
BTW: It would be nice if we could fix the loader to not have fs0
hardcoded, but instead use device paths to figure out which
mapping to use (maybe even create our own mapping). Would
you be willing/able to give that a shot?
--
Marcel Moolenaar USPA: A-39004 marcel@xcllnt.net
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ia64" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030225213120.GA49775>