4100 - LBA48 support

[wallonthefly]

Let me know if I can provide anything. It looks like we have some momentum. Very cool!

[blue68f100]

Good picture, The flash is were the OS is stored. The v1 2000 had 2 of them, held the complete OS & Bios functions. I think the 4100 only have the basic BIOS & OS, NO help files. See that you have the Promise controller. Need to do a search to see if its the same ones that have the problem. It's highly posiable that the chip was fixed, and need to just turn on LBA48bit on some models.

[wallonthefly]

Cool, so does someone need me to send them a BIOS chip?

[Phoenix32]

David,

The 4100 and 4000 do not keep the Help files and such in the Flashram. If you remove the drives and replace them (all at once), you have to reinstall the OS to get the Help files back...


Ya know, taking into consideration my "this is no easy task" message since it still applies, the BIOS update may only require a change of the controller (promise) driver and maybe turning the option on.

[blue68f100]

Originally Posted by wallonthefly

Cool, so does someone need me to send them a BIOS chip?

I have the programmer to read and burn it. But we will need someone that has the knowledge of BIOS and IntelHex, which is way the OS is loaded. Out of my capability. But the first step would be reading whats there.

I bought the programmer because I was working on old pc's in which had crashed during bios updates. Gave me away to revive pc's people were throwing away.

Phoenix, Promise released an update to correct the bad controller chips.

[wallonthefly]

Ok, where is the guru, we have the chips.

[blue68f100]

With the 4100 only have 1 falsh ram, I wonder how much of the chip is used for BIOS and for OS.

How good are you at un-soldering surface mounted devices. The tsop-40 package only has ~0.020" spacing between leads. Not much room for error.

Did you loacte the clock chip? Anap use to store unit drive config, network settings and admin password in the clock chip.

I wonder if the pic chip has the signature ID for the snap.

[wgong]

Originally Posted by Phoenix32

The 4100 and 4000 do not keep the Help files and such in the Flashram. If you remove the drives and replace them (all at once), you have to reinstall the OS to get the Help files back...

With the 705N there are two SUP files - one is OS+Help files and the other is just the Help files. When I got my unit it only had one drive and it was dead so I put a spare 6GB Quantum Fireball CX drive in just to test it. I only had to load the help files and not the full OS. I am going to buy some 120 or 160 GB drives this weekend.

[wallonthefly]

I can either ship someone that is going to do the work the unit/parts, or attempt it myself if someone tells me what needs to be done, but I'd likely prefer the former as I don't have equipment to read PROM's, etc.

[Phoenix32]

Originally Posted by wgong

With the 705N there are two SUP files - one is OS+Help files and the other is just the Help files. When I got my unit it only had one drive and it was dead so I put a spare 6GB Quantum Fireball CX drive in just to test it. I only had to load the help files and not the full OS. I am going to buy some 120 or 160 GB drives this weekend.

Same with SNAP. I was just being general. I should have said, a SUP file has to be reloaded.

[Phoenix32]

Originally Posted by blue68f100

With the 4100 only have 1 falsh ram, I wonder how much of the chip is used for BIOS and for OS.

How good are you at un-soldering surface mounted devices. The tsop-40 package only has ~0.020" spacing between leads. Not much room for error.

Did you loacte the clock chip? Anap use to store unit drive config, network settings and admin password in the clock chip.

I wonder if the pic chip has the signature ID for the snap.

This has been my concern throughout this David. I keep feeling like we got part of the OS or drivers or BIOS or.... Loaded in some hidden place we have not thought of, on on a flash chip we did not identify.


Okay, so we have a flash chip (that we are hoping is good and can be unsoldered) and we have an EPROM reader/programmer available to read this chip. So as someone else asked, do we have a programmer to read and edit the code once we have it? I would havr to go back through the messages, but it seems like we did have somebody say they were able to do this type stuff.

Does this experimental 4100 board work? Are we going to be able to test the new code on that board or are we going to need a test subject?

[wallonthefly]

Please let's not let this fade away. Do we have a programmer that can look into this?

[blue68f100]

1. Is this a working 4100?
2. Do you have the skills required to unsolder the Flashram with out damaging it?
I can read the chip and send the code to anyone willing to attempt this.

3. The most important. Do we have a person that can program the BIOS or patch the OS?

If I recall, someone has allready read the PIC chip, I have the code somewhere.

[re3dyb0y]

Yes

Ive got 2 code snippets somewhere aswell from the same guy i think

But i think they were from something unrelevant

[wallonthefly]

Yes, they are working 4100's, and yes I believe I'm capable of unsoldering it.

[wgong]

Originally Posted by blue68f100

Did you loacte the clock chip? Anap use to store unit drive config, network settings and admin password in the clock chip.

The clock chip is part of U9 (the 82371EB (PIIX4E) south bridge just to the right of the Intel Flash chip) as can be seen by its proximity to Y1 (the 32768Hz watch crystal).

According to the data sheet for the 82371AB (an earlier version than the 82371EB found on my PCB) the RTC has a 256 byte battery-backed CMOS SRAM with two 8 byte lockout ranges.

http://www.intel.com/design/intarch/...s/29056201.pdf

[Phoenix32]

While we know what ASSUME means, I would have to assume the only thing stored in that CMOS SRAM is settings, not OS.



EDIT:

After a little more clear thinking, it is possible for some of the CMOS settings to affect our little problem here. There may be a bit set in there turning LBA48bit off. Hell, it might even be as simple as turning a bit on somewhere turning it on and no driver change or BIOS change needed. Doubtful, but possible.

[wallonthefly]

Tivo's, Xboxes and various other hardware appliances are always hacked to support bigger HD's. It's hard for me to believe that Snap has spent more resources trying to protect drive upgrades then the former.

Can we not just try baby steps and modify the BIOS, or are we sure we need to modify all these subsystems, and continue to make assumptions of what may or may not be?

[blue68f100]

The clock storage is where the info that is cleared with reset cmds, found this out on my 2000. The older clock chip were large in size and had a battery backup build into the chip, good for 7 yrs. The 2000 v1 did not have a battery backing up cmos, like the newer boards.

I think there is proably a flag set some where that disables the LBA support on the 4100's. But if they got defective chips in them (atleast early models) turning it on WILL NOT be a good idea. If it has an error in calculating the parity bit all data will be lost. Now if the chips have been updated with a newer or different supplier it should be a go.

Who is a programmer that can find this flag or code?

The easier way to locate it may be to monitor the data flow (port capture) to see what part of the mem address is being changed. Since the code is loaded by address into the flashram (Intel Hex code). This is the same technique used by hackers, follow the data flow.

[wallonthefly]

I'm willing to take risks and loose data or even cook one of these things. Let me know, thanks.

[wallonthefly]

And into the depths of forum archives we go...

[blue68f100]

The 1st step would be to read the chips, so extract them and send them to me. Then I will send the code to anyone wanting it. This should be Intel Hex Code.

This will not go any further unless some one can re-program the chip. But it may be better to setup a switch or hub to capture traffic to and from the 4100 while you change the 48 bit flag. This will give some idea of where to look.

I think we have a means to write the code back to the 4100. Using the std sup file.

Another option would be to get a std bios setup in it and run FreeNAS (Freebsd 6.1).

[Harbinger]

Originally Posted by wallonthefly

Yes, they are working 4100's, and yes I believe I'm capable of unsoldering it.

Best/easiest/safest way to unsolder a surface mount chip is to get a ChipQuick kit. It includes flux and a special low-temp solder that melts ~135 degrees or so (I think). Just squirt some flux around the pins, move your iron around one side and heat the pins while melting the low-temp solder, then move to the other. You should then have enough time to either lift the chip after quickly heating the pins with your iron, or just use a heat gun (on low temp) or a hair dryer to keep the joints hot while you pick up the chip with a set of tweezers.

After picking up the chip and holding it firmly with the tweezers, apply some heat with the hot air gun and tap the tweezers on the edge of a table. The excess blobbed solder will fall right off. You can then heat the board again and clean off the blobbed solder from it.

It's a bit tricky to resolder the chip afterward just because the pins are so small/narrow, but it can be done with a steady hand and care not to heat a joint too much. Make sure the chip gets soldered back in in the same direction you pulled it out!

[krzee]

blue: I have a 4100 I would be happy to ship to you for testing, let me know if this will help the quest for lba48... It was given to me and is quite useless to me with the current size limitations (4x500 gig drives would be very cool tho). If it works send it back, if you break it in the process, se la vi.

[Davesworld]

Did we get off on a tangent here somehow? The issue has always been the simple fact that the os contains no updated promise driver that will support lba48 and Snap never was able to spend the man hours necessary to make it possible as they were up to their armpits developing the Guardian OS based units. The reason why lba48 works in the 4000 is because it uses the native Intel controllers in the chipset which supports lba48 in this version of BSD already although the throughput is much slower than the promise controllers. If it was just setting a lba48 switch with the current promise driver module, we'd be able to do it in much the same way as a 4000, through the debug menu. I would rather see a way to put FreeNas on it myself since it is a 16MB flash based NAS os and we can tell Adaptec to stuff it and quit trying to bilk us for money on upgrades that don't help us get the most out of the hardware anyway.