Mailinglist Archive: opensuse (3666 mails)
| < Previous | Next > |
Re: [SLE] New File System On a Flash Disk?
- From: Randall R Schulz <rschulz@xxxxxxxxx>
- Date: Tue, 15 Mar 2005 08:09:00 -0800
- Message-id: <200503150809.00713.rschulz@xxxxxxxxx>
Danny,
On Tuesday 15 March 2005 06:45, Danny Sauer wrote:
> On Sunday 13 March 2005 11:25 am, Randall R Schulz wrote:
> > Hi,
> >
> > On Sunday 13 March 2005 08:51, Greg Freemyer wrote:
> > > If all you need is Linux support, I guess ext2 would be the best
> > > choice. Journaling would not make since, at least not with SUSE
> > > 9.2 because it mounts with the SYNC option that defeats the whole
> > > purpose of a jorunal.
> >
> > As I said, the mounting and unmounting are handled by the system
> > (some function of subfs, automount and possibly SuSEPlugger,
> > too--as I said, I don't really understand this stuff at all).
>
> I'd suggest against a journaled system, largely because of the extra
> disk writes. Flash can be read a lot, but can't be written to
> indefinitely, and adding journaling just increases the writes.
I was thinking plain old extfs2, but this cramfs thing sounds
interesting. What are the implications / requirements for getting it
working on an otherwise stock SuSE 9.1 (with all YOU patches, including
new kernels, installed)?
> As an aside, I'm amazed at how well the newest kernels handle USB
> devices.
Hmmm... When I bought this 1GB flash-RAM "disk," I removed the old 128M
version and plugged in the new one. Much confusion ensued, culminating
in all processes that attempt to access the device hanging in an
unkillable state. After a reboot, things were back to normal and the
device can be accessed normally.
Of course, I just upgraded my BIOS (subsequent to changing the flash
memory stick) and one of the changes incorporated since the previous
BIOS version was "Ported latest AMI USB support and also fixed USB Card
reader issue." (vagueness only an engineer or a politician could
create). So perhaps Linux or SuSE's packaging of it or embellishments
to it are not to blame.
I'm not enthusiastic about repeating this exercise, so I'll just ignore
it for now.
> --Danny, running cramfs on his compact-flash cards (the CPU time to
> decompress is more than offset by the bandwidth savings to the card),
> but they're used for a different purpose (booting a car mp3 player,
> etc)
That must be from Rio / Diamond Networks (or one of the firms it
absorbed), right? I have a Rio Karma, and it's an oustanding player.
Randall Schulz
On Tuesday 15 March 2005 06:45, Danny Sauer wrote:
> On Sunday 13 March 2005 11:25 am, Randall R Schulz wrote:
> > Hi,
> >
> > On Sunday 13 March 2005 08:51, Greg Freemyer wrote:
> > > If all you need is Linux support, I guess ext2 would be the best
> > > choice. Journaling would not make since, at least not with SUSE
> > > 9.2 because it mounts with the SYNC option that defeats the whole
> > > purpose of a jorunal.
> >
> > As I said, the mounting and unmounting are handled by the system
> > (some function of subfs, automount and possibly SuSEPlugger,
> > too--as I said, I don't really understand this stuff at all).
>
> I'd suggest against a journaled system, largely because of the extra
> disk writes. Flash can be read a lot, but can't be written to
> indefinitely, and adding journaling just increases the writes.
I was thinking plain old extfs2, but this cramfs thing sounds
interesting. What are the implications / requirements for getting it
working on an otherwise stock SuSE 9.1 (with all YOU patches, including
new kernels, installed)?
> As an aside, I'm amazed at how well the newest kernels handle USB
> devices.
Hmmm... When I bought this 1GB flash-RAM "disk," I removed the old 128M
version and plugged in the new one. Much confusion ensued, culminating
in all processes that attempt to access the device hanging in an
unkillable state. After a reboot, things were back to normal and the
device can be accessed normally.
Of course, I just upgraded my BIOS (subsequent to changing the flash
memory stick) and one of the changes incorporated since the previous
BIOS version was "Ported latest AMI USB support and also fixed USB Card
reader issue." (vagueness only an engineer or a politician could
create). So perhaps Linux or SuSE's packaging of it or embellishments
to it are not to blame.
I'm not enthusiastic about repeating this exercise, so I'll just ignore
it for now.
> --Danny, running cramfs on his compact-flash cards (the CPU time to
> decompress is more than offset by the bandwidth savings to the card),
> but they're used for a different purpose (booting a car mp3 player,
> etc)
That must be from Rio / Diamond Networks (or one of the firms it
absorbed), right? I have a Rio Karma, and it's an oustanding player.
Randall Schulz
| < Previous | Next > |