J3455B-ITX I/O errors with VT-d and ESXi 6.5

Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic

   I have an LSI 9211-8i HBA that I am passing through over VT-d to my ESXi 6.5 install. I'm running OpenMediaVault, but get the same issues on FreeNAS.
After a few seconds of dd'ing 10 GB of zeros (dd if=/dev/zero of=tempfile bs=1MB count=10000) to my ZFS pool, I get a bunch of error messages in dmesg. Eventually all drives come up with errors and the VM hangs.

mpt2sas0: log_info(0x31120322): originator(PL), code(0x12), sub_code(0x0322)blk_update_request: I/0 error, dev sdc, sector ...

I mapped the log_info using this header file and came up with:

MPI_IOCLOGINFO_TYPE_SAS | IOC_LOGINFO_ORIGINATOR_PL | PL_LOGINFO_CODE_ABORT | PL_LOGINFO_SUB_CODE_WRONG_REL_OFF_OR_FRAME_LENGTH 

I originally thought that it was a bad PCIe port, so I RMA'd the board. The replacement board has the same problem. I then eliminated ESXi 6.5 and just installed OpenMediaVault on bare metal. It works!

The problem is definitely ESXi 6.5 on this board. I have the proper mpt2sas and lsi drivers installed for ESXi, but note that the issue only occurs when copying large files. I  have tried P16-P20 firmwares on the HBA as well. It is in IT mode.

Could this be the result of an improper VT-d implementation on this board? I have a feeling there's some sort of overflow happening.

Author	Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic
siddhartha77 Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 23 Jun 2017 Status: Offline Points: 4	Post Options Post Reply Quote siddhartha77 Report Post Thanks(0) Quote Reply Topic: J3455B-ITX I/O errors with VT-d and ESXi 6.5 Posted: 23 Jun 2017 at 4:00am
	I have an LSI 9211-8i HBA that I am passing through over VT-d to my ESXi 6.5 install. I'm running OpenMediaVault, but get the same issues on FreeNAS. After a few seconds of dd'ing 10 GB of zeros (dd if=/dev/zero of=tempfile bs=1MB count=10000) to my ZFS pool, I get a bunch of error messages in dmesg. Eventually all drives come up with errors and the VM hangs. mpt2sas0: log_info(0x31120322): originator(PL), code(0x12), sub_code(0x0322) blk_update_request: I/0 error, dev sdc, sector ... I mapped the log_info using this header file and came up with: MPI_IOCLOGINFO_TYPE_SAS \| IOC_LOGINFO_ORIGINATOR_PL \| PL_LOGINFO_CODE_ABORT \| PL_LOGINFO_SUB_CODE_WRONG_REL_OFF_OR_FRAME_LENGTH I originally thought that it was a bad PCIe port, so I RMA'd the board. The replacement board has the same problem. I then eliminated ESXi 6.5 and just installed OpenMediaVault on bare metal. It works! The problem is definitely ESXi 6.5 on this board. I have the proper mpt2sas and lsi drivers installed for ESXi, but note that the issue only occurs when copying large files. I have tried P16-P20 firmwares on the HBA as well. It is in IT mode. Could this be the result of an improper VT-d implementation on this board? I have a feeling there's some sort of overflow happening.

siddhartha77 Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 23 Jun 2017 Status: Offline Points: 4	Post Options Post Reply Quote siddhartha77 Report Post Thanks(0) Quote Reply Posted: 27 Jun 2017 at 11:16pm
	Does anyone have any similar errors?