Disable IOMMU in unRAID6 Beta 21

[Trunkz]

Hi,

 

I'm wondering if IOMMU can be disabled in unRAID? I've got a Marvel 9230 SATA card which has issues in unRAID 6. (Due to IOMMU support) Now I dont run KVM etc.. only Docker. In prior occasions, I would recall seeing DMA errors and segfaults in the logs. My parity and 3 data drives currently live off the onboard SATA connections thus are not affected by this however my two cache drives and unused 4TB 2.5" data drive (no data currently housed on this drive due to the aforementioned issues - however it is part of the array). I'll get some diagnostic logs of the faults I seem to experience tonight; but just thought I'd put the question out there.

 

Thanks!

[itimpi]

Hi,

 

I'm wondering if IOMMU can be disabled in unRAID? I've got a Marvel 9230 SATA card which has issues in unRAID 6. (Due to IOMMU support) Now I dont run KVM etc.. only Docker. In prior occasions, I would recall seeing DMA errors and segfaults in the logs. My parity and 3 data drives currently live off the onboard SATA connections thus are not affected by this however my two cache drives and unused 4TB 2.5" data drive (no data currently housed on this drive due to the aforementioned issues - however it is part of the array). I'll get some diagnostic logs of the faults I seem to experience tonight; but just thought I'd put the question out there.

i would expect there to be a BIOS option for this.  Not sure exactly what it will be called as it seems to vary a lot between systems.  I suspect if you mentioned exactly which motherboard you have it is highly likely someone can help.

[Trunkz]

Its an HP Microserver N54L. I've flashed TheBay's BIOS ROM to enable AHCI and full-speed SATA. I've made sure SVM is disabled in BIOS. Checking the logs last night; I saw alot of segfaults pertaining to nzbget. (This is running in a docker container)

[SpaceInvaderOne]

Navigate through each BIOS screen  and find and disable the "IOMMU," "I/O Memory Management Unit," or "Intel  VT-d"setting (usually located under the "Advanced" or "Chipset/Northbridge/Tylersburg IOH/Intel VT for Directed I/O Configuration"settings menu). I think yours is amd chip in the n54L

 

[Trunkz]

I've gone through each option and I've made sure to disable any and all references to virtualization. The only one I could find was SVM which I've disabled. Syslog:

 

Apr 14 21:23:01 Tower kernel: nzbget[4324]: segfault at 0 ip 000000000042f3b9 sp 00002b07e7f47550 error 4 in nzbget[400000+92000]

Apr 14 21:23:03 Tower kernel: nzbget[4481]: segfault at 0 ip 000000000042f3b9 sp 00002b2a90e0f550 error 4 in nzbget[400000+92000]

Apr 14 21:24:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:24:04 Tower kernel: nzbget[4495]: segfault at 0 ip 000000000042f3b9 sp 00002b09ea6da550 error 4 in nzbget[400000+92000]

Apr 14 21:25:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:25:05 Tower kernel: nzbget[5290]: segfault at 0 ip 000000000042f3b9 sp 00002b17582f1550 error 4 in nzbget[400000+92000]

Apr 14 21:26:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:26:06 Tower kernel: nzbget[5949]: segfault at 0 ip 000000000042f3b9 sp 00002b687142d550 error 4 in nzbget[400000+92000]

Apr 14 21:27:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:27:07 Tower kernel: nzbget[6539]: segfault at 0 ip 000000000042f3b9 sp 00002b735a444550 error 4 in nzbget[400000+92000]

Apr 14 21:28:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:28:08 Tower kernel: nzbget[7284]: segfault at 0 ip 000000000042f3b9 sp 00002b97147f5550 error 4 in nzbget[400000+92000]

Apr 14 21:29:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:29:09 Tower kernel: nzbget[7947]: segfault at 0 ip 000000000042f3b9 sp 00002b26f929b550 error 4 in nzbget[400000+92000]

Apr 14 21:30:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:30:10 Tower kernel: nzbget[8652]: segfault at 0 ip 000000000042f3b9 sp 00002ba3d7c30550 error 4 in nzbget[400000+92000]

Apr 14 21:31:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:31:11 Tower kernel: nzbget[9270]: segfault at 0 ip 000000000042f3b9 sp 00002b12c59b0550 error 4 in nzbget[400000+92000]

Apr 14 21:32:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:32:12 Tower kernel: nzbget[9935]: segfault at 0 ip 000000000042f3b9 sp 00002ada8a497550 error 4 in nzbget[400000+92000]

Apr 14 21:33:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:33:12 Tower kernel: nzbget[10761]: segfault at 0 ip 000000000042f3b9 sp 00002b1a9846f550 error 4 in nzbget[400000+92000]

Apr 14 21:34:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:34:13 Tower kernel: nzbget[11635]: segfault at 0 ip 000000000042f3b9 sp 00002b8bb8239550 error 4 in nzbget[400000+92000]

Apr 14 21:35:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:35:14 Tower kernel: nzbget[12487]: segfault at 0 ip 000000000042f3b9 sp 00002b58c7242550 error 4 in nzbget[400000+92000]

Apr 14 21:36:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:36:15 Tower kernel: nzbget[13276]: segfault at 0 ip 000000000042f3b9 sp 00002acb81a0f550 error 4 in nzbget[400000+92000]

Apr 14 21:37:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:37:16 Tower kernel: nzbget[13947]: segfault at 0 ip 000000000042f3b9 sp 00002b94ae02b550 error 4 in nzbget[400000+92000]

Apr 14 21:38:01 Tower crond[1514]: exit status 127 from user root /usr/local/sbin/monitor &> /dev/null

Apr 14 21:38:17 Tower kernel: nzbget[14674]: segfault at 0 ip 000000000042f3b9 sp 00002b236f013550 error 4 in nzbget[400000+92000]

Apr 14 21:38:32 Tower kernel: nzbget[15276]: segfault at 0 ip 000000000042f3b9 sp 00002b0cec6de550 error 4 in nzbget[400000+92000]

Apr 14 21:38:36 Tower kernel: nzbget[15470]: segfault at 0 ip 000000000042f3b9 sp 00002adf20b07550 error 4 in nzbget[400000+92000]

[RobJ]

I've gone through each option and I've made sure to disable any and all references to virtualization. The only one I could find was SVM which I've disabled.

 

Segfaults tend to be either RAM issues, conflicts in dependencies, or just plain program bugs.  I'm not aware of a direct association with virtualization.  I would look for a different version of nzbget.