Rockbox.org home
release
dev builds
extras
themes manual
wiki
device status forums
mailing lists
IRC bugs
patches
dev guide



Rockbox mail archive

Subject: Re: slow/delaying usb connection to archos/rockbox
From: Paul Suade (paul.suade_at_laposte.net)
Date: 2002-09-21


Sorry but you are half wrong, Björn. The multisector feature you're speaking
about is not concerned because it always switched on in ide-disk.c if this
feature is present on the harddisk (which is now 100% of harddisks). VFAT or
FAT drivers doesn't know about multisector feature or DMA feature, and still
if you ask for transfering for large contiguous sectors in a FAT partition,
those features would be used. What is slow indeed must be the sector caching
for FAT table (as you said) or thing alike the operating system needs to do
to speedup transfer at the end. It perhaps comes from the fact that linux
FAT doesn't handle removable harddisk as expected or because it thinks the
jukebox harddisk as not a removable harddisk. Windows is more respectful in
regard with the removability of a harddisk.

Another important point : the ISD200 or ISD300 might not handle multisector
transfers. If that is so, it means we are unable to use the multisector
feature and of course I forgot it - you must know it better than I : linux
doesn't use disk IDE drivers to access jukebox harddisk but a disk SCSI
driver. So to my mind, there is no possibility to use any feature on IDE
harddisk which cannot be handled first by ISD200 and ISD300 and secondly by
disk SCSI driver. If modificiations should occur, it would be in SCSI driver
code not in FAT code.

----- Original Message -----
From: "Björn Stenberg" <bjorn_at_haxx.se>
To: <rockbox_at_cool.haxx.se>
Sent: Saturday, September 21, 2002 11:55 AM
Subject: Re: slow/delaying usb connection to archos/rockbox

> Christian Folini wrote:
> > in a mail to linux-usb-devel
> > -> http://www.geocrawler.com/archives/3/2571/2001/5/0/5860868/
> > Björn Stenberg mentions a dead slow connection when writing to
> > his Archos player.
>
> That's only for the first copy operation. The second will be much faster
(~700 KB/s).
>
> The reason is, as I speculated in that old mail, that the FAT is very big
and it needs to be read before the vfat driver can write to the disk.
> Adding to the slowness is the fact that the vfat driver reads the FAT
sector-by-sector instead of in larger and faster multisector chunks.
>
> There is, I'm afraid, not much you can do about this without writing code.
Adding multisector fat reading to the vfat driver would speed things up, but
is non-trivial.
>
> --
> Björn



Page was last modified "Jan 10 2012" The Rockbox Crew
aaa