* What is ANAND header?
@ 2007-01-13 6:36 Seongsu Lee
2007-01-14 17:07 ` Thomas Gleixner
0 siblings, 1 reply; 12+ messages in thread
From: Seongsu Lee @ 2007-01-13 6:36 UTC (permalink / raw)
To: linux-mtd
Hello,
I've a embedded board with 1GB NAND flash memory from Samsung.
(The model number of the flash memory is
SAMSUNG 640, K9K8G08U0A)
I want to use NFTL to mount the flash memory by FAT filesystem.
But NFTL does not work for me. I found that MTD failed to
find boot record in nftlmount.c source.
It try to find out ANAND header from the MTD device. But it fails
to find out.
if (retlen < 6 || memcmp(buf, "ANAND", 6)) {
continue;
}
What is ANAND header?
Is it possible to use NFTL for mounting FAT on MTD device in my
environment?
Thank you in advance!
--
Seongsu's blog - http://www.senux.com/
Some people have parts that are so private they
themselves have no knowledge of them.
^ permalink raw reply [flat|nested] 12+ messages in thread* Re: What is ANAND header?
2007-01-13 6:36 What is ANAND header? Seongsu Lee
@ 2007-01-14 17:07 ` Thomas Gleixner
2007-01-24 5:47 ` Seongsu Lee
0 siblings, 1 reply; 12+ messages in thread
From: Thomas Gleixner @ 2007-01-14 17:07 UTC (permalink / raw)
To: Seongsu Lee; +Cc: linux-mtd
On Sat, 2007-01-13 at 15:36 +0900, Seongsu Lee wrote:
> I want to use NFTL to mount the flash memory by FAT filesystem.
>
> But NFTL does not work for me. I found that MTD failed to
> find boot record in nftlmount.c source.
NFTL is only for M-Systems DiskOnChip devices. It can not be used on
bare NAND chips.
tglx
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: What is ANAND header?
2007-01-14 17:07 ` Thomas Gleixner
@ 2007-01-24 5:47 ` Seongsu Lee
2007-01-24 12:14 ` Ricard Wanderlöf
0 siblings, 1 reply; 12+ messages in thread
From: Seongsu Lee @ 2007-01-24 5:47 UTC (permalink / raw)
To: linux-mtd
On Sun, Jan 14, 2007 at 06:07:00PM +0100, Thomas Gleixner wrote:
> On Sat, 2007-01-13 at 15:36 +0900, Seongsu Lee wrote:
> > I want to use NFTL to mount the flash memory by FAT filesystem.
> >
> > But NFTL does not work for me. I found that MTD failed to
> > find boot record in nftlmount.c source.
>
> NFTL is only for M-Systems DiskOnChip devices. It can not be used on
> bare NAND chips.
Ok, I see.
Then, is there any sub module, like NFTL, that serves as
translation layer on NAND flash memory?
How can I mount 'conventional filesystem' such as FAT, EXT{2,3}
on NAND flash memory instead of mouting on mtdblock because of
its poor functionality. (no wear-leveling)
What I want to do in a final is to mount FAT filesystem on bare
NAND flash memories without performance degradation.
Thank you very much in advance.
--
Seongsu's blog - http://www.senux.com/
LILO, you've got me on my knees! (from David Black,
dblack@pilot.njin.net, with apologies to Derek and the
Dominos, and Werner Almsberger)
^ permalink raw reply [flat|nested] 12+ messages in thread* RE: What is ANAND header?
2007-01-24 5:47 ` Seongsu Lee
@ 2007-01-24 12:14 ` Ricard Wanderlöf
2007-01-25 6:49 ` Seongsu Lee
0 siblings, 1 reply; 12+ messages in thread
From: Ricard Wanderlöf @ 2007-01-24 12:14 UTC (permalink / raw)
To: Seongsu Lee, linux-mtd
> How can I mount 'conventional filesystem' such as FAT, EXT{2,3}
> on NAND flash memory instead of mouting on mtdblock because of
> its poor functionality. (no wear-leveling)
>
> What I want to do in a final is to mount FAT filesystem on bare
> NAND flash memories without performance degradation.
Use UBI, then mount the filesystem on an UBI volume? (Check the mtd list archives or web pages for more info on UBI)
/Ricard
^ permalink raw reply [flat|nested] 12+ messages in thread* Re: What is ANAND header?
2007-01-24 12:14 ` Ricard Wanderlöf
@ 2007-01-25 6:49 ` Seongsu Lee
2007-01-25 9:12 ` Artem Bityutskiy
0 siblings, 1 reply; 12+ messages in thread
From: Seongsu Lee @ 2007-01-25 6:49 UTC (permalink / raw)
To: linux-mtd
On Wed, Jan 24, 2007 at 01:14:55PM +0100, Ricard Wanderlöf wrote:
> > How can I mount 'conventional filesystem' such as FAT, EXT{2,3}
> > on NAND flash memory instead of mouting on mtdblock because of
> > its poor functionality. (no wear-leveling)
> >
> > What I want to do in a final is to mount FAT filesystem on bare
> > NAND flash memories without performance degradation.
>
> Use UBI, then mount the filesystem on an UBI volume? (Check the mtd list archives or web pages for more info on UBI)
Thank you for the reply.
Then, is there any technical reason that translation layer
for bare NAND flash memory is *not* exist?
(NFTL is not for base flash memory but for DiskOnChip)
(I prefer to use translation layer in MTD because UBI is still
experimental and is not included stable kernel tree.)
If something wrong, let me know. Thank you in advance.
--
Seongsu's blog - http://www.senux.com/
The human race never solves any of its problems. It
merely outlives them. -- David Gerrold
^ permalink raw reply [flat|nested] 12+ messages in thread* Re: What is ANAND header?
2007-01-25 6:49 ` Seongsu Lee
@ 2007-01-25 9:12 ` Artem Bityutskiy
2007-01-25 15:30 ` David Woodhouse
0 siblings, 1 reply; 12+ messages in thread
From: Artem Bityutskiy @ 2007-01-25 9:12 UTC (permalink / raw)
To: Seongsu Lee; +Cc: linux-mtd
On Thu, 2007-01-25 at 15:49 +0900, Seongsu Lee wrote:
> Then, is there any technical reason that translation layer
> for bare NAND flash memory is *not* exist?
Just because it is a damn difficult task.
http://www.linux-mtd.infradead.org/faq/general.html#L_ext2_mtd
--
Best regards,
Artem Bityutskiy (Битюцкий Артём)
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: What is ANAND header?
2007-01-25 9:12 ` Artem Bityutskiy
@ 2007-01-25 15:30 ` David Woodhouse
2007-01-25 15:57 ` Artem Bityutskiy
0 siblings, 1 reply; 12+ messages in thread
From: David Woodhouse @ 2007-01-25 15:30 UTC (permalink / raw)
To: dedekind; +Cc: linux-mtd, Seongsu Lee
On Thu, 2007-01-25 at 20:12 +1100, Artem Bityutskiy wrote:
> On Thu, 2007-01-25 at 15:49 +0900, Seongsu Lee wrote:
> > Then, is there any technical reason that translation layer
> > for bare NAND flash memory is *not* exist?
>
> Just because it is a damn difficult task.
No it isn't. It's just not an _interesting_ task. A 'normal' file system
on top of a translation layer is always going to be suboptimal when
compared to a file system designed to run directly on flash.
--
dwmw2
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: What is ANAND header?
2007-01-25 15:30 ` David Woodhouse
@ 2007-01-25 15:57 ` Artem Bityutskiy
2007-01-25 16:17 ` David Woodhouse
0 siblings, 1 reply; 12+ messages in thread
From: Artem Bityutskiy @ 2007-01-25 15:57 UTC (permalink / raw)
To: David Woodhouse; +Cc: linux-mtd, Seongsu Lee
On Thu, 2007-01-25 at 23:30 +0800, David Woodhouse wrote:
> On Thu, 2007-01-25 at 20:12 +1100, Artem Bityutskiy wrote:
> > On Thu, 2007-01-25 at 15:49 +0900, Seongsu Lee wrote:
> > > Then, is there any technical reason that translation layer
> > > for bare NAND flash memory is *not* exist?
> >
> > Just because it is a damn difficult task.
>
> No it isn't. It's just not an _interesting_ task. A 'normal' file system
> on top of a translation layer is always going to be suboptimal when
> compared to a file system designed to run directly on flash.
Well, FTL is somewhat similar to flash FS in terms of complexity. A
reasonably good FTL is probably not a damn difficult task. But a
scalable FTL where you do not need to scan your device when initializing
it is really difficult.
Of course, FTL will always be slower then flash-oriented FS. But if you
have a very quick HW (like Cafe?) you may end-up with an acceptable
speed - this depends on requirements.
Also, if you teach the FS to inform FTL about blocks that are free -
it'll be much faster.
--
Best regards,
Artem Bityutskiy (Битюцкий Артём)
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: What is ANAND header?
2007-01-25 15:57 ` Artem Bityutskiy
@ 2007-01-25 16:17 ` David Woodhouse
2007-01-25 16:44 ` Artem Bityutskiy
0 siblings, 1 reply; 12+ messages in thread
From: David Woodhouse @ 2007-01-25 16:17 UTC (permalink / raw)
To: dedekind; +Cc: linux-mtd, Seongsu Lee
On Thu, 2007-01-25 at 17:57 +0200, Artem Bityutskiy wrote:
> Well, FTL is somewhat similar to flash FS in terms of complexity. A
> reasonably good FTL is probably not a damn difficult task. But a
> scalable FTL where you do not need to scan your device when
> initializing it is really difficult.
No more difficult than a file system, which will give better results.
> Of course, FTL will always be slower then flash-oriented FS. But if
> you have a very quick HW (like Cafe?) you may end-up with an
> acceptable speed - this depends on requirements.
>
> Also, if you teach the FS to inform FTL about blocks that are free -
> it'll be much faster.
Well, yes. The more you teach the FS about the actual requirements of
flash hardware, the more efficient it'll be. Hence the observation that
FTL isn't an interesting problem; filesystems are.
--
dwmw2
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: What is ANAND header?
2007-01-25 16:17 ` David Woodhouse
@ 2007-01-25 16:44 ` Artem Bityutskiy
2007-01-25 16:55 ` David Woodhouse
0 siblings, 1 reply; 12+ messages in thread
From: Artem Bityutskiy @ 2007-01-25 16:44 UTC (permalink / raw)
To: David Woodhouse; +Cc: linux-mtd, Seongsu Lee
On Fri, 2007-01-26 at 00:17 +0800, David Woodhouse wrote:
> No more difficult than a file system, which will give better results.
Agree, an FS is a much more interesting problem. Although a good FTL is
is something which would be nice to have.
--
Best regards,
Artem Bityutskiy (Битюцкий Артём)
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: What is ANAND header?
2007-01-25 16:44 ` Artem Bityutskiy
@ 2007-01-25 16:55 ` David Woodhouse
2007-01-25 17:31 ` Artem Bityutskiy
0 siblings, 1 reply; 12+ messages in thread
From: David Woodhouse @ 2007-01-25 16:55 UTC (permalink / raw)
To: dedekind; +Cc: linux-mtd, Seongsu Lee
On Thu, 2007-01-25 at 18:44 +0200, Artem Bityutskiy wrote:
>
> Agree, an FS is a much more interesting problem. Although a good FTL is
> is something which would be nice to have.
Not really. If we had a well-maintained FTL then people would only use
it. :)
NFTL/INFTL/SAFTL are actually quite good designs as FTLs go; it's just
that we have the potential patent issues with them and nobody really
cares enough about them to make sure _our_ implementation is as good as
it should be, or even that we tie in block-deletion events from the file
system.
--
dwmw2
^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: What is ANAND header?
2007-01-25 16:55 ` David Woodhouse
@ 2007-01-25 17:31 ` Artem Bityutskiy
0 siblings, 0 replies; 12+ messages in thread
From: Artem Bityutskiy @ 2007-01-25 17:31 UTC (permalink / raw)
To: David Woodhouse; +Cc: linux-mtd, Seongsu Lee
On Fri, 2007-01-26 at 00:55 +0800, David Woodhouse wrote:
> > Agree, an FS is a much more interesting problem. Although a good FTL is
> > is something which would be nice to have.
>
> Not really. If we had a well-maintained FTL then people would only use
> it. :)
Heh, users are morons and they would use inefficient FTL instead of
JFFS2, so having an FTL is bad - good point :-)
--
Best regards,
Artem Bityutskiy (Битюцкий Артём)
^ permalink raw reply [flat|nested] 12+ messages in thread
end of thread, other threads:[~2007-01-25 17:31 UTC | newest]
Thread overview: 12+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-01-13 6:36 What is ANAND header? Seongsu Lee
2007-01-14 17:07 ` Thomas Gleixner
2007-01-24 5:47 ` Seongsu Lee
2007-01-24 12:14 ` Ricard Wanderlöf
2007-01-25 6:49 ` Seongsu Lee
2007-01-25 9:12 ` Artem Bityutskiy
2007-01-25 15:30 ` David Woodhouse
2007-01-25 15:57 ` Artem Bityutskiy
2007-01-25 16:17 ` David Woodhouse
2007-01-25 16:44 ` Artem Bityutskiy
2007-01-25 16:55 ` David Woodhouse
2007-01-25 17:31 ` Artem Bityutskiy
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox