Meltdown and Spectre Patched BIOS for X58 Motherboards

[LDeacon]

Hey @Regeneration .
Would it be possible if you can mod the Gigabyte G1 Sniper F2 (final) bios please?
The one you have on here is the F4E which is beta, and after doing numerous tests the F2 bios has much better performance.
The official link is here:

https://download.gigabyte.com/FileList/BIOS/mb_bios_g1.sniper_f2.exe

Thank you it would be much appreciated.

 

[Regeneration]

Some ASUS motherboards have a buggy implementation of Intel VT-d (Virtualization Technology for Directed I/O) in the BIOS. If enabled, some operating systems like Windows 10 will not boot. The workaround is to disable the "USB 2.0 Controller" from "USB Configuration" in the BIOS. On the P6T series, it just disables some ACPI tables and there is no negative impact in USB 2.0 bandwidth or performance whatsoever. If you can't wake up from sleep afterwards, set "Suspend Mode" to "S1 (POS) only" in the power menu.

 

[Dhiru]

Some ASUS motherboards have a buggy implementation of Intel VT-d (Virtualization Technology for Directed I/O) in the BIOS. If enabled, some operating systems like Windows 10 will not boot. The workaround is to disable the "USB 2.0 Controller" from "USB Configuration" in the BIOS. On the P6T series, it just disables some ACPI tables and there is no negative impact in USB 2.0 bandwidth or performance whatsoever.

This works on my Asus P6X58D-E. I am able to boot to Windows 10 with VT-d enabled. Thanks for the workaround!

Is there any way the BIOS can be patched to fix the ACPI tables?

 

[Regeneration]

This works on my Asus P6X58D-E. I am able to boot to Windows 10 with VT-d enabled. Thanks for the workaround!

Is there any way the BIOS can be patched to fix the ACPI tables?

Might break other things and corrupt profiles.

 

[b3bis]

Some ASUS motherboards have a buggy implementation of Intel VT-d (Virtualization Technology for Directed I/O) in the BIOS. If enabled, some operating systems like Windows 10 will not boot. The workaround is to disable the "USB 2.0 Controller" from "USB Configuration" in the BIOS. On the P6T series, it just disables some ACPI tables and there is no negative impact in USB 2.0 bandwidth or performance whatsoever.

Gonna try this on my P6T WS Pro and see if it works. It would be really interesting. One hell of a platform if it wasn't for these small problems.

 

[PurSpyk!!]

I have a Asus P6T Deluxe, with VT-D enabled and USB 2.0 disabled, Windows 10 does not wake up from sleep. Computer switches on, but keyboard and monitor remain off.

 

[Regeneration]

I have a Asus P6T Deluxe, with VT-D enabled and USB 2.0 disabled, Windows 10 does not wake up from sleep. Computer switches on, but keyboard and monitor remain off.

It's better to use S1 sleep mode on ASUS X58 motherboads since memory timings may change when waking from S3.

 

[PurSpyk!!]

It's better to use S1 sleep mode on ASUS X58 motherboads since memory timings may change when waking from S3.

Hi

How do I enable this?

 

[Regeneration]

Hi

How do I enable this?

Power > Suspend mode > S1

 

[Edwired]

Some ASUS motherboards have a buggy implementation of Intel VT-d (Virtualization Technology for Directed I/O) in the BIOS. If enabled, some operating systems like Windows 10 will not boot. The workaround is to disable the "USB 2.0 Controller" from "USB Configuration" in the BIOS. On the P6T series, it just disables some ACPI tables and there is no negative impact in USB 2.0 bandwidth or performance whatsoever. If you can't wake up from sleep afterwards, set "Suspend Mode" to "S1 (POS) only" in the power menu.

Sorry to bump the thread. Got a question about this method does it disable the usb 2.0 controller altogether after the vt-d is enabled. Or is there any benefit using vt-d?

 

[Regeneration]

Sorry to bump the thread. Got a question about this method does it disable the usb 2.0 controller altogether after the vt-d is enabled. Or is there any benefit using vt-d?

Just on some ports and devices. VT-d is useful for virtualization (VMware, VirtualBox and such).

If you don't use these apps it won't be a big deal to turn it off.

 

[Edwired]

Just on some ports and devices. VT-d is useful for virtualization (VMware, VirtualBox and such).

If you don't use these apps it won't be a big deal to turn it off.

Is it the i/o usb ports on the back of the motherboard that get disabled or the ones on the bottom of the board as it hard to know which one as I almost use all of the back usb ports for connecting devices except the bottom ports of the board which isn't connected to anything. I don't use vmware or virtualbox. If I wanted to use other operating systems I just use the boot selection via f8 on the asus p6x58d-e to boot standalone operating system like example I use windows xp for getting videos of my old handheld camcorder which runs on it own hard drive and so on

 

[Regeneration]

Is it the i/o usb ports on the back of the motherboard that get disabled or the ones on the bottom of the board as it hard to know which one as I almost use all of the back usb ports for connecting devices except the bottom ports of the board which isn't connected to anything. I don't use vmware or virtualbox. If I wanted to use other operating systems I just use the boot selection via f8 on the asus p6x58d-e to boot standalone operating system like example I use windows xp for getting videos of my old handheld camcorder which runs on it own hard drive and so on

All ports should work fine but few USB2/3 devices may work slowly or not at all. Keyboards, mice, flash drives should work fine.

 

[X0r]

Hi, @Regeneration

I'm new to the forum and came across this thread after painful searching for Foxconn bios files.

It seems that Foxconn ended channel support for consumer products sometime in 2019 and shutdown two websites containing official drivers, bios, manuals for all Foxconn motherboards.

http://www.foxconnchannel.com

http://www.foxconnsupport.com

Most of these files were not archived.

I was curious if you have the original bios for the Foxconn boards you have listed in your main post.

Additional questions:

* Checksum
(crc32, md5, sha256, etc) or similar gpg signatures are not included in the main post, on your website, or on the google drive backup link. Do you have these available for verification? (would be good to offer)

* CPUID
do you have a list of processors supported by the mod? There was a comment on page 4 mentioning "extended CPU support for less known brands." This was a few days before your last update for all ROM (2018-08-11). I'm trying to confirm that the Xeon X5675 is supported by Foxconn bios. It's part of the Westmere-EP family (206c2), but X5675 was released in 2011 after Foxconn had stopped releasing updates for Bloodrage and Renaissance products in April 2010. I suspect that X5670 (Mar 2010) is supported but X5675 (Feb 2011) is not.

* ECC support
do the original bios or modded bios offer ECC support ? (Registered, or Unbuffered)

* Verification
lastly, as someone who is hesitant to install firmware by strangers due to potential ring0 security vulnerabilities is there any chance you might consider sharing information on what tools you used to mod the firmware, what was modded, and how someone might retrace your steps. This is a big ask I know, but it also enables people to learn and provides transparency. I don't have experience in this area. I tried looking for linux bios tools for AMI bios and didn't have much luck at all. A hex editor probably won't cut it.

To close, thank you for creating this thread. I read all twelve pages.

 

[Regeneration]

Hi, @Regeneration

I'm new to the forum and came across this thread after painful searching for Foxconn bios files.

It seems that Foxconn ended channel support for consumer products sometime in 2019 and shutdown two websites containing official drivers, bios, manuals for all Foxconn motherboards.

http://www.foxconnchannel.com

http://www.foxconnsupport.com

Most of these files were not archived.

I was curious if you have the original bios for the Foxconn boards you have listed in your main post.

Additional questions:

* Checksum
(crc32, md5, sha256, etc) or similar gpg signatures are not included in the main post, on your website, or on the google drive backup link. Do you have these available for verification? (would be good to offer)

* CPUID
do you have a list of processors supported by the mod? There was a comment on page 4 mentioning "extended CPU support for less known brands." This was a few days before your last update for all ROM (2018-08-11). I'm trying to confirm that the Xeon X5675 is supported by Foxconn bios. It's part of the Westmere-EP family (206c2), but X5675 was released in 2011 after Foxconn had stopped releasing updates for Bloodrage and Renaissance products in April 2010. I suspect that X5670 (Mar 2010) is supported but X5675 (Feb 2011) is not.

* ECC support
do the original bios or modded bios offer ECC support ? (Registered, or Unbuffered)

* Verification
lastly, as someone who is hesitant to install firmware by strangers due to potential ring0 security vulnerabilities is there any chance you might consider sharing information on what tools you used to mod the firmware, what was modded, and how someone might retrace your steps. This is a big ask I know, but it also enables people to learn and provides transparency. I don't have experience in this area. I tried looking for linux bios tools for AMI bios and didn't have much luck at all. A hex editor probably won't cut it.

To close, thank you for creating this thread. I read all twelve pages.

Yes. The Foxconn files came from the official site before it went down.

Most BIOSes have internal checksum verification to prevent flashing of a corrupted file.

All X58 six core CPUs have CPUID of 206C2 and are supported by the BIOSes here. However, some motherboards require DDR ratio of 1:10 with FSB to work with Xeons. Some motherboards have CPUID lock in the circuitry so even modded BIOS won't unlock support.

ECC support is enabled on ECC supported motherboards.

This project is more than 3 years ago and I don't remember what and how at this time. I have done many other projects ever since. X58 uses primitive non-UEFI BIOS that is almost invulnerable to ring0 attacks. Let alone that most chips and files are just 1MB and 90%+ already in-use.

 

[X0r]

Yes. The Foxconn files came from the official site before it went down.

Most BIOSes have internal checksum verification to prevent flashing of a corrupted file.

All X58 six core CPUs have CPUID of 206C2 and are supported by the BIOSes here. However, some motherboards require DDR ratio of 1:10 with FSB to work with Xeons. Some motherboards have CPUID lock in the circuitry so even modded BIOS won't unlock support.

ECC support is enabled on ECC supported motherboards.

This project is more than 3 years ago and I don't remember what and how at this time. I have done many other projects ever since. X58 uses primitive non-UEFI BIOS that is almost invulnerable to ring0 attacks. Let alone that most chips and files are just 1MB and 90%+ already in-use.

Hi @Regeneration

There is a misunderstanding with my first comment. I apologize.

I was asking

* Do you still have the original untouched Foxconn Bios for the motherboards you listed and would you be willing to share those zip files.

* Is the Xeon X5670 and X5675 supported by Foxconn motherboards you listed and can this be verified from the microcode? if so how?

* Is ecc supported on the Foxconn motherboards you listed?

* Would you be willing to outline how you modded the files and what tools you used?


Since my first comment I've discovered the windows AMI tool suite (not for linux =/), but am not sure which utilities to use and how to use them. It does appear though that AMIBCP.EXE v3.x.x series applies to these Foxconn bios.

I was trying to learn while being able to verify the steps you went through to mod.

It's mostly moot if you can't share the original bios.

* checksum

I was not referring to checksum verification during a bios flash procedure.

I was suggesting checksum (crc32, md5, sha256sum) or gpg verification being made available for files you have hosted by Mediafire and Google Drive. These are bios files and you don't control the host. It's good security practice to include checksum on a separate host enabling users to verify integrity of the files. This safeguards in the event that if Google or Mediafire were compromised, someone who downloads might notice that your bios files had been tampered with by a third party.

 

[SMITHBEATZ]

I just did a budget build for a customer on X58 with a new X5570 and Chinese motherboard. I always remember Brian from Tech Yes City preaching the importance of InSpectre on older Xeon platforms. Is it still necessary to do? Thanks

 

[Zyll Goliat]

I just did a budget build for a customer on X58 with a new X5570 and Chinese motherboard. I always remember Brian from Tech Yes City preaching the importance of InSpectre on older Xeon platforms. Is it still necessary to do? Thanks

As far as I remember it depend of the system(version).....I believe that on my X58 with W10 installed didn't have almost any impact when I turned ON/OFF inSpectre/Meltdown.......Sometimes maybe there was some gains here and there when it was turned off but all in all it was mostly negligible......

 

[Edwired]

Yeah I am aware about the dram changes when using the S3 mode after putting the system to sleep and wake it up the next day it changes the first dram setting from 8 to 9 in windows 10 using hwinfo64 but the interesting thing to note cpu-z reports a different valve to 7 on the first value of dram, as Aida64 does the same thing changes the first value either 8 or 9 of dram (can't remember the exact info of the top of my mind). Just almost thinks it like there cap leakage on the board somewhere once it goes to sleep this just changes something once leakage stop and it sticks until you restart the computer and it will revert back to normal. Only thing that annoys me abit is the
dram round trip latency on CHA Auto
dram round trip latency on CHB Auto
dram round trip latency on CHC Auto
Above will fluctuates to different numbers every time from cold boot/restart
Like example from cold boot it be like this
2N-52-54-56 = Cold boot
Then exit from bios let the system restart then enter back into bios to view the changes
2N-54-54-56 = Restarted
Then Restarted again this happens it changes two values
2N-53-55-56 = Second restart from bios
2N-54-56-56 = Third restart from bios
2N-54-56-57 = Fourth restart from bios changes the last value
What I don't understand it why is it constantly changing on restart. It shouldn't have to keep doing this but only for other changes in dram setting it should do change once or twice at most all the other settings stay the same value till you change value on each one

 

[Regeneration]

Yeah I am aware about the dram changes when using the S3 mode after putting the system to sleep and wake it up the next day it changes the first dram setting from 8 to 9 in windows 10 using hwinfo64 but the interesting thing to note cpu-z reports a different valve to 7 on the first value of dram, as Aida64 does the same thing changes the first value either 8 or 9 of dram (can't remember the exact info of the top of my mind). Just almost thinks it like there cap leakage on the board somewhere once it goes to sleep this just changes something once leakage stop and it sticks until you restart the computer and it will revert back to normal. Only thing that annoys me abit is the
dram round trip latency on CHA Auto
dram round trip latency on CHB Auto
dram round trip latency on CHC Auto
Above will fluctuates to different numbers every time from cold boot/restart
Like example from cold boot it be like this
2N-52-54-56 = Cold boot
Then exit from bios let the system restart then enter back into bios to view the changes
2N-54-54-56 = Restarted
Then Restarted again this happens it changes two values
2N-53-55-56 = Second restart from bios
2N-54-56-56 = Third restart from bios
2N-54-56-57 = Fourth restart from bios changes the last value
What I don't understand it why is it constantly changing on restart. It shouldn't have to keep doing this but only for other changes in dram setting it should do change once or twice at most all the other settings stay the same value till you change value on each one

This issue appears mostly on ASUS motherboards and the solution is to use S1 or to enter timings and sub-timings manually.

You should stress test with -1 or -2 when its like 1T-52-53-54 (the lowest set by the BIOS plus -1 on the first) so if that ever changes the system still remains stable.

Different timings may require more voltage for DRAM and VTT/QPI. Even the slight change may cause instabilities (BSODs, data corruption).

 

[Edwired]

Voltages are manually set only some of the dram timings is in manual as matching according to the spd of the ram. It pretty good with some minor hanging from time to time as a reboot would fix it
Still I do the memtest that loaded from usb it seems to drop down to a snail speed in one of the memory tests don't know if that related to some of the memory timing in auto

 

[Regeneration]

Voltages are manually set only some of the dram timings is in manual as matching according to the spd of the ram. It pretty good with some minor hanging from time to time as a reboot would fix it
Still I do the memtest that loaded from usb it seems to drop down to a snail speed in one of the memory tests don't know if that related to some of the memory timing in auto

All of the automatic DDR values remain persistent except of tRTL values (round trip latency) that change on every reboot (including S3 wakeup).

 

[Edwired]

Yeah it a strange behavior but as before with the timings used before I was getting 24000mb, read 19000mb write and 27000mb copy but as soon I set them all to manual except one cas back to back delay in auto this setting has me confused if I set it to 10 the read, write and copy decrease massively at around same number 14000mb to 15000mb give or take in aida64 memory benchtest but dropping back to back to 8 the above numbers goes up the same way. But the the one thing that I can't get my head around this is it better to have even numbers in memory benchtest or uneven as in auto

 

[chromecut]

Im curious do you know of anybody or a friend of a friend that has ever been affected by spectre or meltdown?
or have they been affected by a malicious bios attack? Its all nonsense and is a lame excuse to "protect" us by locking down the bios and these sham microcode updates further the "planned obsolesce" by the greedy oems
its tricky to show folks on newer hardware that most microcode updates unless specifically for new cpu support or enhanced features actually LOWER the performance of the cpu by around 30%
you need to have full read/write of your bios to confirm but that was my result when I checked what puters I had full bios access to-lenovo thinkcentre Mp3-I hacked my bios to support 4790k and using the supplied versions I had to go back 10 bios versions to get 31% speed back when downgrading-I wish I could thank the person personally for putting that list together-same thing for my lenovo notepad and alienware aurora R7 & R8 which are the exact same computer in every way except for a simple bios mod inserting 9th gen cpu support-dell and lenovo are the worst violators-Ive cracked open more bios than I care to think about with almost all having more anti modding bios updates than feature or stability upgrades-infecting a computer via bios is crazy difficult and pointless when there are easier ways-If anything bios hackers labelled "criminal" are using their skills to combat oem lockdown-so check it out for yourself

 

[MaksDampf]

I just tried the x58USB3 BIOS and sadly i can't get it to work.

My board was BIOS version F2 and I wasn't able to install any newer Bios than F3 because those newer BIOSes are 2MB and the flash is formatted to 1MB only.

Found the solution by using a deprecated version of Gigabyte @bios, see here for instructions: https://www.tweaktownforum.com/foru...abyte/45693-q-flash-error-incorrect-file-size
@bios seems to ignore if the filesize or checksum does not match.

I can flash the original F5C version from Gigabyte and it works sort of. The System does not POST anymore when its hot (reset, restart, ctrl+alt+delete), but still manages to post every time when it is cold booted. Strange behaviour, but at least the original F5C BIOS sort of works.
I always enabled CMOS Clear and delete DMI pool Data when flashing any new BIOS.

But the NVME version simply does not POST. The only way to flash it was @bios again. Neither a DOS Booitstick works nor the integrated Q-flash. Is there some kind of magic involved?

If i flash it with the included FLASHSPI.EXE, the exe says " !!! BAD BIOS FILE !!" and the Qflash utility says the BIOS file is corrupted and will not allow it. I re-downloaded it 3 times and also tried a different USB Stick and made a New DOS bootstick with rufus twice.

Any ideas guys?