Hi everyone, I am posting a help request here as I seem to be at the final stages of a seemingly impossible challenge and wanted to just check what I'm missing to get this complete.

I have a Dell XPS 9360 (ME 11.8.x) which by design has the OPI Link Speed locked to 2x PCIe lanes in its ME configuration. The quasi-identical boards in the XPS 9370 and XPS 9380 have unlocked the PCIe lanes to x4, and I've found the offending parameter in the ME configuration, using FIT. Using the guide outlined in the first post, I can successfully dump the ME (by unlocking the ME region through a UEFI variable), clean it and build a new image with FIT.exe. I can also successfully flash the new outimage.bin back to the ME and the laptop boots fine. It's got BootGuard and VerifiedBoot enabled but allows changes to the ME partition. In fact I have upgraded the ME version several times using this method.

The problem is - I cannot seem to make any changes to the ME configuration. If I set the specific parameters to change PCIe speed from x2 to x4 in FIT.exe (Internal PCH Buses -> OPI Configuration -> OPI Link Speed -> 4GT/s) and build the image, the settings do not persist in the binary. This is not something related to the Flash Descriptor as it's the new outimage.bin that's missing the changes, so I just wanted to check what I'm doing wrong here.

I've linked my full chip dump and ME dumps to this fileshare here: https://mega.nz/#F!I0ZFECwK!FJvqWpiB0NJPMysuPGdOnA

Happy to share it on another fileshare if it helps.

Thanks in advance

@jkbuha - As I mentioned to you in PM, QPI link speed is not same/has nothing to do with PCIE lane width, two VERY different things there.

Yes thanks - I do understand the difference between the two but the entry clearly says OPI not QPI on my Flash Image Tool (11.8 r28). Isn't that the correct parameter to modify?

Also - it still doesn't explain why no settings are preserved when I build the new image. What am I doing wrong here?

OPI??? As I've said all along, show me an image of exactly what you are looking at, so I can look at same in FITc myself.
OPI (DMI link speed) is also not PCIE Lane width (That, PCIE Lane width, is only that, which I showed you image of in FITc)
But, I do see this, so you may be onto something - https://egpu.io/forums/pc-setup/skykaby-...andwidth-issue/
However, changing that, even if it actually changed immediately, would still leave your PCIE lanes set exactly as I showed you in ME FW and exactly how they are set in BIOS too (They wouldn't automatically change)

I will look at this again now, sorry, this is first time I noticed you said OPI and not QPI

Please do keep in mind, your original MW FW is corrupted, and may still be same if you rebuilt/updated using your own ME/BIOS dump as base, you better check it with ME Analyzer and see if you get 100 errors like the original dump you sent me.

* Edit @jkbuha - OK, I checked, and your error is do to the broken ME FW you have as base.
I downloaded another dump BIOS from this same system with older ME FW, and without as many ME FW errors (3 vs 30+ in yours), and I can change that setting and rebuild without issue and the setting is retained in outimage rebuild
If you use that ME FW, I would follow the cleanup guide, update it to latest version, and then at last step before build image, make your setting change to OPI

Yes that's correct thanks - now we're getting somewhere :)

I've managed to flash the ME image you found but for some reason I can't change any settings in the outimage.bin. It must be something I'm doing wrong, because when I load the outimage.bin the settings I changed are not there.

Just to be clear, this is what I'm doing:

1) Dumping the existing ME (or with the file you showed me, loading that ME instead)
2) Save the config.xml
3) Replace file with new ME Region.bin
4) Opening FIT again and choosing config.xml
5) Making proposed settings changes (eg: OPI -> 4x)
6) Build Image (successful)

When I reload the resulting outimage.bin back into FIT.exe, it does not have the new settings I applied in 5). What am I doing wrong?

@jkbuha - You're welcome It must be your process? Drop the BIOS I linked you to (in-full) on FITc, run through the cleanup guide and update the ME FW (this not required, but suggested), then right before you do your build image, change that setting.
In outimage, replace BIOS region with yours instead using UEFITool 25. Unlock FD via straight hex edit, then program BIOS back into board. If you need me to do, let me know, I can do all that, since I have your programmer backup.

Or, maybe you simply need to delete the config.ini file before you re-open your outimage, because when "Config.ini" is in folder that is auto-loaded upon opening of FITc
In that case, you'd see settings shown of whatever is in that config, so if you didn't change setting before you made config.ini, then the setting change wouldn't be there

Ok thanks, so I've managed to successfully change the ME settings if I use the FULL dump. But unfortunately I haven't found a way to unlock the FD on the 9360 board and the bios chip is WSON not SOIC (so the CH341 method doesn't work). So until I find a way to flash the full chip without using a programmer I cannot use this method.

The only way I can flash the ME back is if I directly write to the ME Region using FPT. And using this method I've discovered an issue - when I only edit the ME.BIN, any changes aren't saved back. I maybe asking something really stupid here - but do any new changes in ME get written back to the ME Region or the FD? If it's the FD region then I'm stuck as I cannot write to the FD :(

@jkbuha - Well you don't have to unlock the FD, I only mentioned that as a good thing to do since you were programming BIOS back (I assumed, forgot about WSON)
You'll have to either get adapter and remove WSON, or replace it with SOIC8, or do pinmod to unlock FD and then dump it with FPT, edit, flash it back.

If you can write ME region with FPT, then what you write should be placed into the ME FW region. I don't understand what you mean >> "when I only edit the ME.BIN, any changes aren't saved back"
Are you saying, that you write ME FW with FPT, and then dump it again, and your written file is not what is dumped out? If yes, is the new version dumped out and just not the changed setting, or not even the same version you wrote?
Are you sure FPT is writing ME FW region at all? Usually this requires a jumper, or unlocked FD to allow ME FW region to even be written, unless you have a BIOS option that you enabled to allow it.
If you are sure it's writing, then you can do this, and then there should be no way old ME FW remains in place (unless replaced by some backup ME FW and some BIOS auto-recovery function) >> FPTw.exe -rewrite -me -f me.bin
That will erase the entire ME FW region before it writes anything, so be 100% everything is OK in what you write in, and be sure you do the command exactly and when doing that I would also only use ME FW region extracted from the final mod BIOS with UEFITool as the source file.

I can correctly flash ME FW properly with FPT and confirm that the region is flashed properly (it happens only when I unlock ME region for flashing with a specific UEFI variable).

The problem is generating the outimage.bin itself. If I edit the FULL BIOS dump (32MB) (from FPT -D ALL.BIN) and make changes (eg: OPI Link Speed -> GT4) the outimage.bin is successfully changed (confirmed because I reload the outimage.bin and the OPI are changes saved). However, when I only edit the ME partition dump (8MB) (from FPT -ME -D ME.BIN) and make the same changes and build the image, there is no change in the outimage.bin.

@jkbuha - Don't edit your ME FW from FPT -D ALL.BIN dumped off your board, your ME FW is corrupted as I've mentioned all along (or at least has 30+ errors the other one doesn't) Edit the ME FW in the BIOS I linked you to, update if you want or not, then make your changes.
Then with that BIOS, replace the "BIOS Region" with BIOS region from your original FPT Dumped BIOS region and you then have whole good BIOS specific to your board again, then reflash ME FW.
Or, simply extract the ME FW from the final edited BIOS (not yours, the one I linked), then flash that with FPT. Don't edit only the ME FW itself, edit the BIOS as a whole, then extract the ME FW and FPT flash it.

Unfortunately it didn't work. I extracted the ME FW from the modded BIOS and flashed it (FPT -ME -F ME_NEW.bin was successful and new ME partition re-initialised successfully), but the old settings persisted. This leads me to conclude that the OPI settings must be stored in the FD region, for which I don't have write access yet.

There's a way to unlock the FD from the system board using a jumper short, but I haven't managed to get it working yet. Will resort to try using a modified SOIC clip to better fit WSON chips again...

Yes, it could be that, or it could be that BIOS settings also need changed too as I mentioned. I'd try once more with this command (don't always use this either) >> FPTw.exe -rewrite -me -f MENew.bin << Remove that underscore
Then issue FPTw.exe -greset
If it does not auto-reboot at -greset command, then shut down, remove all power (cables and batteries etc), then short power on pins or button for 10-15 seconds, then let sit for 1+ minute without power
Then start it up again and load optimal default in BIOS, then save and reboot back to BIOS and make any changes you need, and carry on with testing.

If there is FD/ME jumper on board, put jumper on there, reboot to windows twice, then do your FD dump/edit/flash back. You could easily see if FD is holding some change, by comparing FD extracted from source before/after ME FITc edit.

There is also pinmod method to unlock FD, see E.1 on page one of this thread.

I will look at BIOS and see if I can find anything OPI related * Edit = Nothing found Do you know what is this an abbreviation for? Maybe in BIOS it's using full name instead of OPI

FPT -rewrite doesn't change anything unfortunately. Same results. The pinmod method worked on my old XPS 15 (HDAOUT), but hasn't worked so far on the XPS 13. Will try again when I dismantle the laptop...

OPI stands for "On Package (DMI interconnect) Interface".
More info here: https://egpu.io/forums/pc-setup/skykaby-...andwidth-issue/

Did you do the -greset or power drain timeout too? Pinmod only needed if you can't write to ME FW area, or want to unlock FD etc. But yes, sorry, I forgot you suspected maybe something changes at FD to, so that may be needed after all.
Did you compare FD of input BIOS and outimage file, and see changes at FD? If not, then there is your answer.

Thank you for the link and extended name, I will look through BIOS again and see if I notice anything related. Funny, that page you linked is the one I had open the other day when I first looked up OPI

*Edit - In BIOS @ Chipset >> DMI Config >> DMI Max Link Speed (This set Auto, may need set Gen3)
If that does not help alone, there may be some other BIOS module that needs updated by Dell (And Adandtech mentions DMI Gen3 too, so auto may be setting Gen2 on this BIOS)

I have Z170N used with G3258 Haswell and I3-9100F Coffee CPU and I checked my BIOS/ME, and it's set to OPI 2GT/s too, but that article says only Sky/Kaby ULV CPU's/system affected. Does this mean only those need the 4GT's setting due to some bug, and others are OK on 2GT/s?
My speeds are correct for my NVME specs and do not seem limited in any way like shown in the article. I checked this, because I was going to test outcome on my end, but I guess I can't and may not be affected by this bug due to not using a ULV CPU?

Zitat von Lost_N_BIOS im Beitrag #239

---

Did you do the -greset or power drain timeout too? Pinmod only needed if you can't write to ME FW area, or want to unlock FD etc. But yes, sorry, I forgot you suspected maybe something changes at FD to, so that may be needed after all.
Did you compare FD of input BIOS and outimage file, and see changes at FD? If not, then there is your answer.

---

Zitat von Lost_N_BIOS im Beitrag #239

---

*Edit - In BIOS @ Chipset >> DMI Config >> DMI Max Link Speed (This set Auto, may need set Gen3)
If that does not help alone, there may be some other BIOS module that needs updated by Dell (And Adandtech mentions DMI Gen3 too, so auto may be setting Gen2 on this BIOS)

---

Zitat von Lost_N_BIOS im Beitrag #239

---

I have Z170N used with G3258 Haswell and I3-9100F Coffee CPU and I checked my BIOS/ME, and it's set to OPI 2GT/s too, but that article says only Sky/Kaby ULV CPU's/system affected. Does this mean only those need the 4GT's setting due to some bug, and others are OK on 2GT/s?
My speeds are correct for my NVME specs and do not seem limited in any way like shown in the article. I checked this, because I was going to test outcome on my end, but I guess I can't and may not be affected by this bug due to not using a ULV CPU?

---