Date: Sat, 18 Sep 1999 22:12:17 -0700 (PDT) From: Matthew Dillon <dillon@apollo.backplane.com> To: Greg Lehey <grog@lemis.com> Cc: cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG Subject: Re: User block device access (was: cvs commit: src/sys/miscfs/specfs spec_vnops.c src/sys/sys vnode.h src/sys/kern vfs_subr.c) Message-ID: <199909190512.WAA68295@apollo.backplane.com> References: <199909170610.XAA56084@freefall.freebsd.org> <19990919105045.T55065@freebie.lemis.com>
next in thread | previous in thread | raw e-mail | index | archive | help
:On Thursday, 16 September 1999 at 23:10:28 -0700, Matt Dillon wrote:
:> dillon 1999/09/16 23:10:28 PDT
:>
:> Modified files:
:> sys/miscfs/specfs spec_vnops.c
:> sys/sys vnode.h
:> sys/kern vfs_subr.c
:> Log:
:> Add vfs.enable_userblk_io sysctl to control whether user reads and writes
:> to buffered block devices are allowed. The default is to be backwards
:> compatible, i.e. reads and writes are allowed.
:>
:> The idea is for a larger crowd to start running with this disabled and
:> see what problems, if any, crop up, and then to change the default to
:> off and see if any problems crop up in the next 6 months prior to
:> potentially removing support entirely. There are still a few people,
:> Julian and myself included, who believe the buffered block device
:> access from usermode to be useful.
:
:Add me to that list. What reasons have been given for disabling it?
:
:Greg
This is one of the major issues that led to the problems between Poul
and I which spilled over into -committers two weeks ago. A number of
people, including me, are not willing to allow block device access
from usermode to be removed, at least not in the manner it was (prior
to being backed-out). Others make the point that nobody seems to be
using buffered block devices anymore and that their original creation
was a fluke anyway, so they should be removed. You can revisit that
thread to get a recap, I'd rather not reopen that can of worms here.
The pseudo-compromise is to make it a sysctl and procedure that will
determine how many people actually use buffered block devices vs raw
devices (raw devices are obviously used and not going away). I
personally believe that block devices are useful. I am willing to go
through a process to decide the point once and for all.
A number of problems have already been found and fixed. For example:
problems with savecore. On the otherhand, there are still some problems
which have not been fixed, such as fsck assuming a 512 byte sector size.
fsck cannot be run on filesystems with a larger physical sector size
except via a buffered block device. And nobody really knows how many
people outside the core developers use buffered block devices for other
purposes.
-Matt
Matthew Dillon
<dillon@backplane.com>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199909190512.WAA68295>