UEFI PXE on Dell Optiplex 7010 hangs

[BardWood]

@sebastian-roth 04_ipxe.efi:

[Sebastian Roth]

@BardWood Ok, here is ready to go 05_ipxe.efi (DEBUG=init,efi_utils,efi_driver,efi_pci). This time I guess you need to take a video of the messages passing by on screen. Make sure to rest the camera/smartphone e.g. on a pile of books so we get a decent steady picture. Would be great if you could even take a video with 60 fps which most smartphones are capable of these days!

But if there’s too much output on screen and you can’t take a proper video you might test 06_ipxe.efi (DEBUG=init,efi_utils,efi_pci)…

@george1421 Re-reading all the stuff we tried back in 2015 I get the impression that the issue with those 7010s is that for some reason the native Intel driver is not being used. Then it tries SNP and for some reason that fails. More or less a guess though.

[BardWood]

@sebastian-roth

RE: 05 = Wowsa! Yea, much output. My phone won’t let me video yet because the battery is too low (???). I’ll capture a video later but I cant even read it in person it scrolls by really fast. I’ll do what I can.

RE: 06 =

[Sebastian Roth]

@BardWood Ok, seems like it actually does see it as being a proper Intel NIC. I think we are on the right track. Though it definitely needs more work than I had expected. Keep it up and hopefully you can get a good video (60 fps!!) of it. This is where digital devices have a great advantage over our eyes.

[george1421]

@sebastian-roth Well let me say I’ve very disappointed in this latest iphone upgrade. Its really driving me crazy at the moment.

I finally have the download figured out. I’ll get you the link to the 05 = Wowsa! video in a moment, or when ever my iphone decides to email me the video
Intersting messages in video

00:16.0 (8086:1e3d class 0708000) has no driver
00:16.1 cannot read PCI configuration: No such device http://ipxe.org/0x2c044087
00:16.3 (8086:1e3d class 0708002) has no driver
00:19.0 (8086:1502 class 020000) has driver "82579lm"
has driver "PCI"
disconnecting existing drivers

Then loops forever

Has driver "MII"
DRIVER_START
Refusing to start during disconnection
has driver "SNP"
DRIVER_START
refusing to start during disconnection

[BardWood]

@sebastian-roth Unfortunately I can’t capture 60 FPS video on a Motorola Droid Turbo. I thought it was an artificial limitation since the camera is capable of it spec-wise but when I tried in Open Camera, I got the same message. I’ll see if I can get our photographer buf to do it on her iPhone 6.

[BardWood]

@sebastian-roth OK have a ~1Gb video file I’m trying to upload to flickr of 05 but a quick glance tells me it’s scrolling the same info in a loop. Here is a screenshot:

[BardWood]

@sebastian-roth here is the video of 05 on flickr https://www.flickr.com/gp/138098812@N03/D0569a

[george1421]

@bardwood FWIW: Your video is repeating in the same spot as mine.

It would be interesting to know what is causing the “Disconnecting existing drivers” line to post. I think that may be the root of the issue.

[Sebastian Roth]

@BardWood Wow! You’ve done an amazing job getting a stable picture with the camera and I am sure we got all the information from that debug build in the picture you posted! I am still watching the video just to make absolutely sure we don’t miss anything else. Thanks heaps for that.

I will dig through the code and hopefully will find what’s up here. Though we might need to do more of these rounds I’m afraid.

[Sebastian Roth]

@BardWood @george1421 Hmmm, digging through the code I get the impression that we are pretty close already. Running that same binary on a different machine not having this issue I see:

...
EFIPCI ... has driver "82579lm"
EFIDRV ... has driver "PCI"
EFIDRV ... disconnecting existing drivers
EFIDRV ... connecting new drivers
EFIPCI ... has driver "82579lm"
EFIDRV ... has driver "PCI"
EFIDRV ... DRIVER_START
EFIPCI ... has driver "82579lm"
EFIPCI ... using driver "82579lm"
EFIDRV ... using driver "PCI"
...

As you see there is kind of a handshake going on, first disconnecting existing drivers which I suppose might be drivers loaded by UEFI firmware beforehand. Then connecting new drivers (the iPXE driver I suppose) and finally using driver.

On the Dell 7010 though we don’t see the connecting new drivers message. The “funny” thing is that in the code between “disconnecting existing drivers” and “connecting new drivers” there are only a dozen lines of code - all in the very same function. My guess is that something goes wrong when iPXE tries to disconnect the drivers and is in a state of “trying to disconnect” from then on as we see from the messages saying “refusing to start during disconnection”.

So I compiled the next binary 07_ipxe.efi for you which should dump a lot of things like the actual EFI protocol calls. I also added some sleeps to the code so it hopefully doesn’t scroll past too fast. Although I can’t add sleeps in the middle of EFI protocol calls so it will be a challenge to get a clear picture/video again.

Edit: Updated the iPXE binary number as I somehow mixed those up with a different debug… 07 is definitely the newest one here.

[BardWood]

@sebastian-roth I’m not sure a video is necessary since the message loops and the full message is captured in this image but am happy to oblige if it helps.

07_ipxe =

[Sebastian Roth]

@BardWood I am sorry for loosing track of this. As mentioned in the private chat there is an interesting answer on the iPXE forum: http://forum.ipxe.org/showthread.php?tid=7877

Would you be able to test the stuff described there. Do you need assistance with that? This seems to be very specific issue and we need you to try things out on the hardware to get an idea.

[BardWood]

@sebastian-roth Greetings. I had to get the first rev of W10 images out so the 7010 image is legacy this time around. I should have time this week or next. I did look at the link and didn’t look terribly difficult. I’ll give it a crack.

[Sebastian Roth]

@BardWood Bump… Any news on this?