Subject: Re: RAW access to files
To: Jonathan Stone <jonathan@DSG.Stanford.EDU>
From: Thor Lancelot Simon <email@example.com>
Date: 12/13/2001 01:41:54
On Wed, Dec 12, 2001 at 12:12:53PM -0800, Jonathan Stone wrote:
> I didnt dare try an ffs with 1meg blocks and 128k fragments and
> 2-block readahead: would that have worked?
No, because you couldn't have mounted such an FFS without serious surgery
on various device drivers, and, at that time, the buffer cache. The
FFS block size is constrained by MAXBSIZE and by MAXPHYS; on the i386
both are 64K. The MAXBSIZE constraint is (or should be!) now gone, though
I think you'd have trouble because I think the "old" buffer cache is still
used for access to directories, and so you'd be unable to read in a 1MB
directory block... ugh. The MAXPHYS constraint remains, which kinda sucks.
I had a plan for replacing MAXPHYS with a device property inherited from
devices' parents but I got busy and lost access to the big storage array
I was planning to use this to do RAID experments anyway. Sigh.
Sometimes, you really want to issue a single SCSI command for N megabytes
of data. We just can't. It's a shame. But the real loser here is
RAIDframe, which ends up doing zillions of tiny I/Os out to the disks because
the I/Os the filesystem hands it can't ever get big enough for it not to.
Thor Lancelot Simon firstname.lastname@example.org
But as he knew no bad language, he had called him all the names of common
objects that he could think of, and had screamed: "You lamp! You towel! You
plate!" and so on. --Sigmund Freud