lkml.org 
[lkml]   [1999]   [Nov]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: FIBMAP ioctl restrictions: why?
Hi,

On Sun, 7 Nov 1999 22:44:43 +0100, Martin Mares <mj@ucw.cz> said:

> Hello, world!\n
> I'm currently writting my own database library (a much faster alternative
> to GDBM) and I want to sort file accesses before feeding them to the kernel,
> so that they will be executed in linear sector order and seek time will be
> minimized. Unfortunately, this plan breaks on the FIBMAP ioctl requiring
> CAP_SYS_RAWIO capability. To be honest, I don't see why -- can the kernel
> gurus advise me what deep reasons are behind this restriction?

If we're going to modify the kernel for this, far better to go in the
direction of kernel async IO. That way you can just submit all the IOs
to the kernel at once and let it sort the IOs itself. The kernel may
have a much better idea about the ideal IO ordering than you do,
especially if there is software raid or LVM involved; and the kernel
can also delegate the ordering to the disks by using scsi tagged command
queueing where appropriate.

--Stephen

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:54    [W:0.039 / U:0.244 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site