HP 840 EFI ipx boot failure about ipxe HOT 18 CLOSED

Once #250 is closed I can re-run tests on the Z840 to confirm.

from ipxe.

This issue is also happening on the most recent build on HP Z6's. Going to a commit previous to the one linked above resolves it for me as well. It was indeed with ipxe.efi, I did not try snponly.efi.

from ipxe.

After further investigation. The file were it was erroring on was was src/interface/efi/efi_pci.c (reading error message). After a couple of permutations the commit the build fails on is 9b25f6e

Any recommendations for getting around the error?

from ipxe.

After further investigation. The file were it was erroring on was was src/interface/efi/efi_pci.c (reading error message). After a couple of permutations the commit the build fails on is 9b25f6e

Are you saying that it works before commit 9b25f6e but fails on that commit or later? That would be very odd, since that particular commit is itself restoring an earlier behaviour.

from ipxe.

Yes, the ipxe build before commit 9b25f6e succeeds. but any commit later it is failing on.

from ipxe.

May plausibly be the same issue as #250

from ipxe.

We're seeing this issue with several HP models when in UEFI mode. No issues with legacy mode. We use Quest KACE SDA which utilises iPXE. They recently incorporated the current release and it's led to this for us.

from ipxe.

Could we verify if this is still a current issue?

I assume this was with ipxe.efi ?, Did you ever try with snponly.efi?

from ipxe.

I was able to reproduce the error on a hp Z8 machine. snponly.efi succeeds to boot but then both nics are unseen in ipxe, so I am unable to chain https boot after. I booted the intel.ipxe and it succeeds but ipxe is still not seeing one of the onboard nics. On the z8 it would appear that ipxe doesnt like the intel x722 nic. This is still an issue.

from ipxe.

This error is still persistent in the latest version of ipxe

from ipxe.

This error is still persistent in the latest version of ipxe

and

appear that ipxe doesnt like the intel x722 nic

Using that x722 for further steps:

stappers@myhost:~/src/mailinglists/ipxe
$ git grep x722
src/drivers/infiniband/linda.c: PCI_ROM ( 0x1077, 0x7220, "iba7220", "QLE7240/7280 HCA driver", 0 ),
src/drivers/net/intelxl.c:      PCI_ROM ( 0x8086, 0x37ce, "x722-kx", "X722 10GbE backplane", 0 ),
src/drivers/net/intelxl.c:      PCI_ROM ( 0x8086, 0x37cf, "x722-qsfp", "X722 10GbE QSFP+", 0 ),
src/drivers/net/intelxl.c:      PCI_ROM ( 0x8086, 0x37d0, "x722-sfp", "X722 10GbE SFP+", 0 ),
src/drivers/net/intelxl.c:      PCI_ROM ( 0x8086, 0x37d1, "x722-1gt", "X722 1GBASE-T", 0 ),
src/drivers/net/intelxl.c:      PCI_ROM ( 0x8086, 0x37d2, "x722-10gt", "X722 10GBASE-T", 0 ),
src/drivers/net/intelxl.c:      PCI_ROM ( 0x8086, 0x37d3, "x722-sfp-i", "X722 10GbE SFP+", 0 ),
src/drivers/net/intelxlvf.c:    PCI_ROM ( 0x8086, 0x37cd, "x722-vf", "X722 VF", 0 ),
src/drivers/net/intelxlvf.c:    PCI_ROM ( 0x8086, 0x37d9, "x722-vf-hv", "X722 VF (Hyper-V)", 0 ),
stappers@myhost:~/src/mailinglists/ipxe
$ 

The basenames without the .c are intelxl and intelxlvf.

Use those basenames in:

make [platform]/[driver].[extension] DEBUG=intelxl,intelxlvf

[platform], [driver] and [extension] inspired by https://ipxe.org/appnote/buildtargets
If only [driver].[extension] is good enough, say so. Express explicite which command is used for building the iPXE executable.

from ipxe.

Ok, This is using our standard Ipxe build bin-x86_64-efi/ipxe.usb DEBUG=intelxl,intelxlvf,pcimsix

I added the pcimsix debug after going to https://ipxe.org/2dcc408e and seeing it reference pcimsix.c

from ipxe.

OCR, Optical Character Recognitation. of the above photo:

iPXE initialising devices...INTELXL 0x654b40f0 PF 7 using port 3 queues [0x07ff-0x07ff]
MSI-X 0x654b4130 has 129 vectors for 0000:04:00.2
MSI-X 0x654b4130 table could not find BAR3
INTELXL 0x654b40f0 could not enable MSI-X: No such file or directory (https://ipxe.org/2dcc408e)

from ipxe.

To me it looks that the original issue is solved. At least there is a complete different error.

This snippet is from #731 (comment)

INTELXL 0x9ea1b910 PF 7 using port 3 queues [0x07ff-0x07ff]
MSI-X 0x9ea1b950 has 129 vectors for 0000:44:00.3
MSI-X 0x9ea1b950 table could not find BAR3
INTELXL 0x9ea1b910 could not enable MSI-X: No such file or directory (https://ipxe.org/2dcc408e)

Yes, that is the same https://ipxe.org/2dcc408e

from ipxe.

Hmm.. Ok if its a different issue i would have to go back to the commit before a202de3 and if my builds work then I would agree that I'm running into the same error as #731
I'll try and confirm that today. Thank you.

from ipxe.

I'll try and confirm that today.

How did that go? (BTW: What is the ipx in the current title of this issue?)

from ipxe.

This issue started as a good issue. It was about a single aspect of iPXE. Something worth to be an iPXE issue.

The

Yes, the ipxe build before commit 9b25f6e succeeds. but any commit later it is failing on.

was also a good thing.

Follow up question Could we verify if this is still a current issue? got after about ten weeks a blunt This error is still persistent in the latest version of ipxe

Then there was the original error is gone and this issue became iPXE does NOT work for me, you must fix it.

This comment is for asking to close this issue. At least to prevent another vague comment like

On the z8 it would appear that ipxe doesnt like the intel x722 nic. This is still an issue.

expressing you must fix my problem

from ipxe.

Merge request #801 closed #731. Those who can close this issue, please do.

from ipxe.