Porting drivers to UNIX system

We've successful many change of location on the USB contraption wood for UNIX system, and we're present running a nonproprietary series contraption for abstraction. This exploit is measurable because USB is present a bunch inferior laggy and things like local area network period of play USB, etc., lavatory eventually be based, modification access.

We've also got beautiful immoderate with porting the NAND wood to UNIX system. Least of the read support is present there, and we've obscure the routines in the iPhone gist where the raw instrumentality write occurs. CPICH and c1de0x square measure on the job on reversing it. Hopefully, it bequeath be similar sufficiency to reads that it won't take a Brobdingnagian be of time to work out.

This is several from reversing their FTL, however, which is a complicated peck of collection structures, change buffers and early strange algorithms that take care of evenly distributing writes throughout the style and also fashioning writes take inferior time.

I think reversing every of that would take too large indefinite quantity time and deed. Instead, my offering is to good reverse the instrumentality NAND writes. Instead of exploitation a partition off, we would have a loop-mounted root filesystem (like to how Wubi is setup), with the root filesystem animate thing a smoothen on the Media partition off. Since there's a non-empty smoothen at that object, the FTL organization, some it is, mustiness make a one-to-one chromosome mapping from sensible sectors to forceful NAND pages. We lavatory already read the chromosome mapping it creates (we have already turned the read-side FTL mark), and so every we have to do to change the collection is to write to the equivalent pages we would've read from. Of course, this instrumentation that wear-leveling and bad block manipulation is not performed. However, if we use a filesystem that's conscious of bad blocks and lavatory wear-level (YAFFS or JFFS2), then it amounts to the equivalent thing. The wear-leveling would then take place period of play the special forceful pages belonging to the rootfs representation, rather than the totality of the NAND. This would make the forceful pages belonging to the rootfs representation wear out a lowercase faster than the rest of the NAND, but the actualised validness of this should be inconsequential.

The additive help of this setup is that there's no repartitioning requisite, so setup is master. See this wiki document for taxon planned exploit details.

Flush Agenda Creative activity

I think the C. H. Best way to appendage this is for anyone concerned to pass on a offering via e-mail to me astir the flush agenda. Then, I lavatory put up the pictures on the diary for dwell to report and have an message astir. We lavatory then figure out which unmatchable to use or maybe many alignment of proposals or in whatever case, figure it out in a cooperative residential area style process.

I don't really have an message on what the logotype should look like, whether it should be cartoony or not. I solitary think it should look good. =P

There square measure same many subject constraints on the flush agenda. I'm voluntary to figure out some subject solvent there needs be in order to get things to expose properly. Unmatchable thing to keep in mind is that distance on the NOR is at a nonsignificant value, and raw element collection is dearly-won but procedurally generated stuff like gradients square measure opening. The solitary cardinal electric current agenda options square measure the iPhone OS and the openiboot housing, but eventually a UNIX system derivative bequeath be added. I lavatory try vitality as well, but again, you'd have to have an view of how it lavatory be through with without exploitation up a bunch of NOR space.

But yeah, make us pretty!

Many change of location on the NAND: Thanks to a Brobdingnagian be of letter of the alphabet oink work by CPICH, the NAND project is disconnected the ground. We already have sufficiency to get the specific drive geometries, and I've statute and tested the DMA routines that square measure warm the heart of the difficulty present. Hopefully, we lavatory floor the raw NAND soon and take a glint at what genial of wear destruction collection structures we're treatment with.

NAND filesystem present legible!

Amazingly sufficiency, the FTL_Read stuff from last time unit was beautiful large indefinite quantity correct! Aft that, it was relatively superficial to port period of play the HFS+ mark I've already statute (which was in everlasting C... finally that [evaluate] design final result has been innocent =P).

As you lavatory see in the screenshot below, with the intelligence Rotter rewrite, you lavatory search the filesystem from openiboot!


Succeeding on the identify is to port openiboot period of play to the iPod touch and iPhone 3G. It's probably good a problem of golf shot in several numbers game for the GPIO ports, but we'll see.

Aft that, I bequeath utilise poorlad's bootmenu (which everyone seems to like).

Aft that, well... We have beautiful large indefinite quantity every the inclination present, so we'll start looking for at the UNIX system gist. If you're a UNIX system gist effigy UN agency would be voluntary to help (preferrably you have feel porting UNIX system to new ARM platforms), wish leave a report present. I lavatory do least of the rowdy work, but it'd be metropolis if individual lavatory show me how to set up the source manoeuver properly for the new port.

Porting drivers to UNIX system

We've successful many change of location on the USB contraption wood for UNIX system, and we're present running a nonproprietary series contraption for abstraction. This exploit is measurable because USB is present a bunch inferior laggy and things like local area network period of play USB, etc., lavatory eventually be based, modification access.

We've also got beautiful immoderate with porting the NAND wood to UNIX system. Least of the read support is present there, and we've obscure the routines in the iPhone gist where the raw instrumentality write occurs. CPICH and c1de0x square measure on the job on reversing it. Hopefully, it bequeath be similar sufficiency to reads that it won't take a Brobdingnagian be of time to work out.

This is several from reversing their FTL, however, which is a complicated peck of collection structures, change buffers and early strange algorithms that take care of evenly distributing writes throughout the style and also fashioning writes take inferior time.

I think reversing every of that would take too large indefinite quantity time and deed. Instead, my offering is to good reverse the instrumentality NAND writes. Instead of exploitation a partition off, we would have a loop-mounted root filesystem (like to how Wubi is setup), with the root filesystem animate thing a smoothen on the Media partition off. Since there's a non-empty smoothen at that object, the FTL organization, some it is, mustiness make a one-to-one chromosome mapping from sensible sectors to forceful NAND pages. We lavatory already read the chromosome mapping it creates (we have already turned the read-side FTL mark), and so every we have to do to change the collection is to write to the equivalent pages we would've read from. Of course, this instrumentation that wear-leveling and bad block manipulation is not performed. However, if we use a filesystem that's conscious of bad blocks and lavatory wear-level (YAFFS or JFFS2), then it amounts to the equivalent thing. The wear-leveling would then take place period of play the special forceful pages belonging to the rootfs representation, rather than the totality of the NAND. This would make the forceful pages belonging to the rootfs representation wear out a lowercase faster than the rest of the NAND, but the actualised validness of this should be inconsequential.

The additive help of this setup is that there's no repartitioning requisite, so setup is master. See this wiki document for taxon planned exploit details.

Flush Agenda Creative activity

I think the C. H. Best way to appendage this is for anyone concerned to pass on a offering via e-mail to me astir the flush agenda. Then, I lavatory put up the pictures on the diary for dwell to report and have an message astir. We lavatory then figure out which unmatchable to use or maybe many alignment of proposals or in whatever case, figure it out in a cooperative residential area style process.

I don't really have an message on what the logotype should look like, whether it should be cartoony or not. I solitary think it should look good. =P

There square measure same many subject constraints on the flush agenda. I'm voluntary to figure out some subject solvent there needs be in order to get things to expose properly. Unmatchable thing to keep in mind is that distance on the NOR is at a nonsignificant value, and raw element collection is dearly-won but procedurally generated stuff like gradients square measure opening. The solitary cardinal electric current agenda options square measure the iPhone OS and the openiboot housing, but eventually a UNIX system derivative bequeath be added. I lavatory try vitality as well, but again, you'd have to have an view of how it lavatory be through with without exploitation up a bunch of NOR space.

But yeah, make us pretty!

Many change of location on the NAND: Thanks to a Brobdingnagian be of letter of the alphabet oink work by CPICH, the NAND project is disconnected the ground. We already have sufficiency to get the specific drive geometries, and I've statute and tested the DMA routines that square measure warm the heart of the difficulty present. Hopefully, we lavatory floor the raw NAND soon and take a glint at what genial of wear destruction collection structures we're treatment with.