Removing AHCI option rom?

[elrf]

Even with the latest EX58-UD3R bios from 2010, I notice that it includes an ancient, slow and buggy AHCI option rom from 2008 (version 1.20E).

It appears that only Gigabyte AWARD BIOS includes this particular option rom - other X58 boards don't appear to have it. Does anyone know what purpose it serves?

If it's absolutely necessary, is there a reason that Gigabyte can't include the version from last year ("Serial ATA AHCI BIOS, Version iSrc 1.20_E.0024 12212009").

The 1.20E version of this code appears to have a bug where it will crash (error code 23) based on values on the hard drive. Specifically, if the ICH SATA is in AHCI mode, and the option rom executes, it will hang if a SATA disk contains small values in the first partition table entry's Ending CHS values.

[Lsdmeasap]

Re: Removing AHCI option rom?

What? The 1.20E is the fastest AHCI ROM ever, especially for Gigabyte anyway.

Many other boards use this, and we fought for a long time to get them in Gigabyte BIOSes.

Yes, all other boards also use an AHCI ROM, and it is invoked anytime AHCI or RAID is set in the BIOS.

As for your bug, this is the first I've heard of it, and honestly I don't understand the cause of the issue but that's not saying you are wrong at all.

Just saying I've never seen the error, nor seen anyone report any issues caused by it or issues with 1.20E either.

Do you have that file you mention, if so post it here so I can compare with the 1.20E used currently.

I find it hard to believe that Intel would keep the 1.20E name/version from 2008 to a new version in 2009, are you sure?

[kopierkatze]

Re: Removing AHCI option rom?

Just saying I've never seen the error, nor seen anyone report any issues caused by it or issues with 1.20E either.

There are MANY reports about trouble with this option ROM!

Just google for "ahci isrc error 23" and you will find some. I am currently having the same issue right now!

I find it hard to believe that Intel would keep the 1.20E name/version from 2008 to a new version in 2009

And yet it's true. There are at least three different option ROMs labeled "1.20E" out there:

iSrc 1.20_E.0024 12212009
iSrc 1.20_E.0012 11252008
iSrc 1.20E

They are not only using different version strings, but they also differ in size:

iSrc 1.20_E.0024 12212009 = 17,408 Bytes
iSrc 1.20_E.0012 11252008 = 16,896 Bytes
iSrc 1.20E = 16,384 Bytes

So each new version increased by one 512 Byte page in size. So it's not just a new name, they really added some functionality while keeping the name "1.20E".

I extracted these option ROMs from different BIOSes. Other vendors than Gigabyte DO update their BIOSes (e.g. HP does)! If you want, I can send you these ROMs so you can compare them yourself!

[Lsdmeasap]

Re: Removing AHCI option rom?

Ahh ok, thanks! I just had never personally seen any, nor seen any reports of it here or any other forums I help in.

Very interesting info! Can you give me the newer ROM, or link me to a BIOS you know that has it?

Yes, please send me the ROM's, then I can update Gigabyte BIOSes with them when asked.

Have you tried using the newer ones yet? If you need me to insert it for you I can, let me know

[sstahlman]

Re: Removing AHCI option rom?

I wrote a post on my blog about this issue.

Short way of breaking Serial ATA AHCI BIOS, Version iSrc 1.20E:

WARNING: THIS WILL BREAK STUFF. DON'T DO THIS UNLESS YOU UNDERSTAND WHAT YOU'RE DOING AND WHY.

• Use dskprobe or hxd to edit your partition table.
  
• Write down bytes 0x1C3 and 0x1C4 (they'll be 0xFE and 0xFF with one partition, something smaller on a dual partition setup; doesn't matter, they're not used for anything important other than crashing your BIOS)
  
• Change these to something small: 0x06 and 0x05 will work nicely (giving you 7, 6, 5)
• Save, reboot in AHCI mode, and enjoy the boot hang!

Fixing: Disconnect the drive, enter BIOS, switch back to IDE mode and reconnect (or swap to a non-ICH SATA port ) Put proper values back in 0x1C3 and 0x1C4.

All the 1.20E-XX option ROMs I've seen have come from Insyde EFI BIOS, so I don't know if they'll be compatible with AWARD.

Oh, and the v1.20E-24 could be even faster if it didn't read values from the hard drive, discard the values it just got done reading, and go off and calculate C/H/S anyway.

[Lsdmeasap]

Re: Removing AHCI option rom?

The ones he's talking about are from HP BIOS, so I think they'll be ok.

I plan to test, and can recover from a bad flash easily, so I'll be ok.

[sstahlman]

Re: Removing AHCI option rom?

The ones he's talking about are from HP BIOS, so I think they'll be ok.

The HPs with 1.20E-24 do fix the behavior I've seen; I'm interested in hearing if my instructions on how to break the previous 1.20Es happen to work on your setup.

[kopierkatze]

Re: Removing AHCI option rom?

• Write down bytes 0x1C3 and 0x1C4 (they'll be 0xFE and 0xFF with one partition, something smaller on a dual partition setup; doesn't matter, they're not used for anything important other than crashing your BIOS)

So you say these bytes can be manually changed to FE FF without messing up the partition table?
This would be a nice workaround for the "code 23 error".

In a few hours I will arive home, and I will report if option ROM v1.20E_0024 (extracted from HP Insyde BIOS) will work on my Gigabyte motherboard.

Very interesting article on your blog, by the way!

[Lsdmeasap]

Re: Removing AHCI option rom?

Ohh sorry sstahlman, I didn't get ya until now :)

Also, sorry I thought the insyde BIOSes were all Intel boards only, didn't know others used them too.

Guess I'll find out how it goes, I'll try on P55 and X58 :)

[sstahlman]

Re: Removing AHCI option rom?

So you say these bytes can be manually changed to FE FF without messing up the partition table?

Yes, because you're simply putting nonsense values for the ending head and sector value, which nothing really cares about any more, with modern hdds. Well, except certain AHCI ROMs, of course. ;-)

A helpful link: NTFS.com Partition Table. Hard Drive Partition.

Lsdmeasap: I've seen Insydes predominately on laptops, but the important bit is any given BIOS with the particular Intel AHCI ROM... so, you're right, this would only affect Intel boards. :-)

A very good question might be just why anything stock in the BIOS thinks it ought to be reading values from a partition on the hard drive in the first place; especially since there's fallback code to simply ask the hard drive controller itself what its legacy geometry values are; code that, due to the buggy 'fix' for the previous 1.20E bugs, in 1.20E-24, is executed every time anyway!

[Lsdmeasap]

Re: Removing AHCI option rom?

Haha, well it doesn't work in P55

And yes, I can now answer your original question. You can remove the ROM if you want!

Flash went fine, I'm in windows now from the modified BIOS, but the AHCI ROM fails to load. I use RAID mode with a single SSD, so the AHCI ROM screen was skipped and just the RAID screen showed, then windows.

I did try AHCI mode, and it doesn't work. Well at least disks are not visible in the BIOS disk priority screen and since no AHCI ROM Screen loaded I assumed it would be a waste of time trying to load windows that way.

But yes, if you want to remove the ROM you can! Just use single non-member RAID mode for AHCI instead, that's how I always run anyway.

You only save 2-3 seconds off the boot though, unless you have disks that are really slow at detecting?

Should I try X58 also, I don't think the AHCI rom matter between chipsets, but I suppose it probably does since the orig P55 is named PCH and X58 is named ICH :)

What chipset were those BIOSes originally intended for anyway?

[Lsdmeasap]

Re: Removing AHCI option rom?

So should I test on X58 or not?

I waited since you guys disappeared on me

[sstahlman]

Re: Removing AHCI option rom?

I'm not sure of all the chipsets; my Acer laptop with Insyde BIOS originally had 1.20E-19, and that has an Intel GS45 Express Chipset.

The chipset doesn't matter though; what's happening is that code in the AHCI ROM is causing an exception when it reads unexpected values in the hard drive's partition table. 1.20e-24 seems to always calculate values, even though it does try to read from the partition table. To see buggy behavior, you need to use a version of the AHCI ROM prior to -24.

Kopierkatze: Did you have luck with your error code 23 by placing 0xFE 0xFF in 0x1C3 and 0x1C4, respectively?

[Lsdmeasap]

Re: Removing AHCI option rom?

Well I mean what chipset was the BIOSes the OP originally intended for, just asking so I could know if it should work or not in P55 or X58 - Aside from failing due to being an Insyde file because the contents look the same as normal 1.20E files, so I think it should work on the right chipset.

[elrf]

Re: Removing AHCI option rom?

Yeah, I'm still on the X58. If it'd work to use the 1.20E-24, that'd be useful.
Knowing that nothing will go horrifically wrong from having the wrong option rom is comforting - I wonder if Win7 would still scan the AHCI even if there's no option rom to fill the ACPI tables.
Using the RAID mode is also a nice workaround for the #23 problem.. I wasn't sure if that still resulted in AHCI performance once it got to the OS?