@lfb6 - I could try the other 'Good bios', but it's a full 16MB bios. Can I use ME Tools to flash just the bios region, or would we need to flash using the CH341?

All P50 laptops had discrete graphics. I think there were three different Nvidia models available, but it wasn't available without discrete graphics, so we should be fine.

I'm pretty confident nothing got damaged while I was working on the eeprom. Discrete graphics is just a chip soldered to the motherboard. You can't even see it unless you remove the cooling system/fans. I didn't disturb any of this to access the eeprom.

@Lost_N_BIOS - I've tried booting with the external display connected, but it makes no difference.

Before all this happened, my laptop was plugged into a docking station, with the two external monitors connected to the dock via Display Port (DP) cables. I could hot dock/undock, and the screens would be detected straight away. I'm only using an HDMI cable directly into the back of the laptop for testing, just to rule out any potential dock problems. I never tried with HDMI before, but when plugging in via DP, I used to get a beep that it had detected new hardware. I get nothing with HDMI. I don't have a mini DP cable to plug directly into the laptop though. It's all full size DP on the dock.

Yes, tried new Nvidia driver. Completely removed old, and installed new. Also tried older versions to see if that helped, but no change. All the settings in the bios look as I remember.

Regardless of the Nvidia discrete graphics, I would expect the Intel HD 530 to be working, as this is all built into the main i7 processor. It obviously works well enough to drive the laptop screen, but doesn't seem to be able to route it to external devices. That, or the devices are just not being recognised, so the processor doesn't know it needs to re-route the video.

Thanks,
Ian.

@lfb6 - Yeah, I'd already read those. Mine would have been bricked again if I didn't have a bios backup and a programmer.

Thanks,
Ian.

@IanP50 It's not very probable that you can flash the bios region with fpt, I assume it'll be locked, but it's worth a try. I think flashing the bios region only from a full image is possible, but I never tried- I flash always from a bios region. Anyway- you can open the 'Good bios' in UefitoolNE and right click on the bios region- 'extract as' is should give you the bios region, 'fpt???.exe -bios -f Good_BIOS_region_own_EVSA_PAD.rgn' would start flashing it.

Attached the "Good Bios region" with your own EVSA store. You could combine the original first 7 MB of your old bios (FD, GbE, ME- not cleaned from #11) to create a complete image. Or try to flash the region only, as described. In case this works you could simply use it since you'd have MAC from your own GbE and own Windows license/ confguration bytes...

@lfb6 - Thanks. I'm off to bed now. will try in the morning.

@IanP50 Seems that on many (?) systems the external display ports are routed/ powered through the discrete GPU. This would at least make your symptoms fit together: Nvidia graphics gone/ not available. If this is software/ bios or a hardware fault we'll have to find out...

https://www.reddit.com/r/pop_os/comments...tors_on_laptop/, https://forum.thinkpads.com/viewtopic.php?t=129853 (older)

You didn't change Thunderbolt settings in bios or used Thunderbolt lately?

@lfb6 - Looks like using the Lenovo bios update utility has closed down some of the security.

I'm getting this error when I try to flash with fpt:

Error 368: Failed to disable write protection for the BIOS space.
FPT Operation Failed.

That's a shame, as I really didn't want to strip the laptop down again.

Interesting about external ports being routed through the discrete GPU. As you say, that would tie in with what I'm seeing. I don't think the discrete GPU is completely dead, as it does show up if I scan with HWiNFO64. Interestingly, this shows the GPU also has some sort of bios. Not sure we can do much with it though.

@IanP50 Same security was there before bios update. Disabling bios safety would require testing/ flashing different versions, too.

Interesting that you can see the Nvidia device in HwInfo but not in device manager! These devices have their own OROM (bios) for legacy boot and a GOP driver/ GOP VBT for UEFI boot built into system bios. But the video bios information in you screenshot is 'pretty much Intel', that's the Intel legacy video bios/OROM (in 6A5F1EF9-C478-42BA-A4DB-C7846CC7DB57). The video bios for GM 107 (B17608F9-E955-4DA8-8762-03AE634A52EB) has version 82.07.9D.00.14 - but device- ID and Nvidia driver seem to be correct.

There might be the possibility, that Windows got 'confused' as well. What information about display adapters is in device manager now? Please check events, too, sometimes Windows recognizes devices wrongly as different device 'known from before'.



EDIT Regarding "external display ports are routed/ powered through the discrete GPU": This wouldn't correspond to the setups referred to when explaining Optimus, muxed and muxless.

@lfb6 - Here's what I see in Device Manager.

@IanP50 - Try remove the device in device manager (Delete driver is asked).
I would actually uninstall the Nvidia driver and clean with driver cleaner first, then boot to windows, and uninstall device from device manager (then also delete driver if asked), then reboot and re-scan for changes if it does not auto-reinstall.
Then install nvidia driver again and see if fixed.

Did you maybe bump card while inside, little loose maybe, or was it part of the PCB itself? I think it's a MXM card, so should be removable, but you'd have to pull off heatsink to reseat the card.
Did you ever dump other chips? It's possible the EC FW was messed up during the initial brick, it would be on a similar chip as the BIOS possibly (unless in QFP Package), 128K-1MB usually.

vBIOS itself (For several models) is within the main BIOS region, inside the main DXE volume, so it should not be messed up

We can get past error 368 possibly, but it may leave error 167 or 28 and still require flash programmer, hard to know without trying.
Also, you may be able to flash certain volumes in the BIOS region, without flashing it all at once, and not have any error (Please show me image result >> FPTw.exe -i)
Keep Boot Guard is enabled in mind here, you cannot edit/change any colored area of the BIOS as seen in UEFITool NE or = brick

To try and bypass 368 without flash programmer, you can either do yourself with RU method, or you can dump vars via H2OUVE and send to me and I will edit, then you program back, both methods info below
Or, it may be possibly without either of the below, if your BIOS has S3 sleep bug >> Put system to Sleep (S3, not hibernate) for one minute, then wake it up and try again, if you still get error 368 move onto next, your BIOS does not have the bug

RU method to unlock BIOS Lock is here, Go to section 2.2 and make bootable USB with RU program, then read 2.3-2.5
http://forum.notebookreview.com/threads/...-issues.812372/

You will change the following, and you will NOT be making the change in "Setup" as discussed in the guide
BIOS Lock, VarStoreInfo (VarOffset/VarName): 0x14
Location you need to make edit in >> "PCHSetup" >> GUID >> 4570B7F1-ADE8-4943-8DC3-406472842384

Change offset 0x14 (Line 10, out to column 4 = 0x14), change 01 to 00 (If already 00, then nothing to do here, programmer required, or pinmod may work as well to unlock FD, if lock in FD is only one preventing BIOS region flash)

Dump vars method >> download the following package, and run the command below from each versions folder that directly contains it's exe. Before doing this, make sure you have secure boot disabled in BIOS, any BIOS password removed, and TPM/Encryption disabled.
Once done, copy the entire folder somewhere, delete everything but any created vars.txt and then repackage this and send to me, this way all created vars.txt remain in place in the folders of the version that created them.

https://www.sendspace.com/file/8v98ya

H2OUVE.exe -gv vars.txt

@IanP50 You normally don't have to scroll/ search through the event viewer. Device manager, open device, tab 'events', klick on 'view all events' should give you the list to all events related to that device.



There should at least be some warnings. Check both Intel and Nvidia. I'd like to see the last 3 to 4 entries on the events tab for both devices, if possible. Code 43 means 'driver reported problem, that could mean a lot of things...

I'd still give the 'good bios' with own parts a try- but I understand that disassembling isn't a very attractive option right now.


@Lost_N_BIOS Nvidia chip is on mainboard, nothing to change/ remove/ reseat.

EC firmware is first MB of Bios region (in paddng), do you think there is another/ second location? TPM has it's own firmware, and Thunderbolt chip has own firmware, this chip is known to brick if filled with garbage, but was never in use here?

@Lost_N_BIOS - I've tried removing/deleting the driver multiple times, but the outcome is the same.

There's no way to even see the Nvidia GPU without removing the fan assembly. Even then it's soldered directly onto the motherboard, so I can't really do anything with the hardware.

Thanks for all the detail about flashing the bios. I think it's probably easier for me to just take it all apart again. I'm getting pretty good at it now.

@lfb6 - Wow, you do indeed learn something new every day. I didn't know you could show Event Viewer entries like that.

Just noticed that both the Intel and Nvidia are showing the same error.

Intel - Device PCI\VEN_8086&DEV_191B&SUBSYS_222E17AA&REV_06\3&e89b380&0&10 requires further installation.
Nvidia - Device PCI\VEN_10DE&DEV_13B0&SUBSYS_222E17AA&REV_A2\4&374f1360&0&0008 requires further installation.

Event Viewer doesn't show any errors or warning for the Intel, but this is what I get for the Nvidia.

@IanP50 Thanks.

I still think it's 'bios'/ bios settings. The 'Good bios' has one entry pchsetup in NVRAM, your latest dump has 5 entries that contain pchsetup in the header, but all are 'invalid'. The PCH is quite a central element, I'd think it might have some tasks in connecting graphics, too. Maybe it's just not initialised properly?

That's a thing I disagree with your brother. Hardware initialisation is done by bios, distributing PCIe lines for example and such things can't be undone/ changed by OS afaik.

'Require further installation'- that's always the first entry when Windows finds out that the devuce needs a driver. In device manager you have to scroll down to see latest entry. One of the first entries almost always will be "requires further installation".

Pictures: The easiest way is to have them in the forum, means to attach them. This way the posts remain consistent/ complete even if you some time begin to clean up your image 'cloud'. On the other hand there are now some pictures and bios dumps here which have for example you serial and embedded Windows license. (But there's no limit for time to edit posts.)
=> Attach file, Browse, Upload, Insert

@lfb6 - EC FW in BIOS may be copy, it's also in stock BIOS package separately, that's why I asked if he saw or dumped any other similar to BIOS SOIC8 chips.
He should look on board if he did not dump or check before, hopefully if this is part of the issue, the EC FW is in SOIC package not QFP. Yes, often EC FW/KBC FW is in it's own chip, and only in BIOS as copy (like some BIOS have ME FW backup file too)
If GPU is part of the main PCB then yeah, nothing to worry about there! I was not sure due to vBIOS modules in BIOS all named MXMxxxx

What is current final state in BIOS now of bytes following "LNVBBSEC" and "INVALIDINVALID"?
If they do not match original BIOS content, these should be put back first, this may be the issue
Byte following "LNVBBSEC" >> FB
Bytes following "INVALIDINVALID" >> 15 DF 01 49 81 B0 29 8D 69

* Edit - Hmm, what you found above about PCHSetup in NVRAM is not good, at least one should be valid or yes as you suspect that may be an issue


@IanP50 - Thanks for confirmation about your testing removal of drivers and such, does that include actual "Removal" of device via device manager and rebooting too, or just driver removals?
This is two very different things, I was suggesting removing/cleaning driver and also remove device in device manager (and driver if asked)
I think the issue may be above, since I think I remember you guys wanting to leave that as is from the stock BIOS recovery stuff, but I'm not 100% sure what's in BIOS now.

To embed image, you need to use the attach function located in advanced reply >> Button on far right at each post, in there you will find attach button far right as well
Guide/info here - [Guide] How to insert pictures or attach files to a post

@Lost_N_BIOS Last state was 'own padding restored'. But maybe it was an error to begin with a stock bios without this information... But a valid pchsetup should be in NVRAM in any case.

Byte following "LNVBBSEC" >> FB
Bytes following "INVALIDINVALID" >> 15 DF 01 49 81 B0 29 8D 69

@lfb6 - I'll tear the machine down this evening and flash the other 'good' bios'. Hopefully this will improve things. I think I also disagree with my brother. If you look at the details on the Lenovo bios download page, it does list bios updates for Linux systems.

@Lost_N_BIOS - Sorry, forgot to mention that I didn't dump any other chips from the motherboard. Most of the board is covered in a black protective cover which I didn't remove. Not sure if there are other SOIC type chips fitted. I'll have a look later.

PCH sounds promising.

I completely removed the Nvidia driver. I completely removed it in device manager, and then used a program called DDU to completely remove everything. System was rebooted, and windows installed new driver.

Thanks,
Ian.