<html><head><title>nvme-device hangs after power-on - after soft-reboot all is okay</title> <META http-equiv=Content-Type content="text/html; charset=iso-8859-1"> </head> <body> Hi, I have a problem with a NVMe-Device Samsung SM951 in a virtualized Illumos-environment. My default setup is a HP G8 Server (I have tried others) as ESXi6.02-Host and a newest OmniOS-bloody-VM. I´ve learned to setup the nvme from here: <a style=" font-family:'Courier New'; font-size: 9pt;" href="https://www.mail-archive.com/smartos-discuss@lists.smartos.org/msg02873.html">https://www.mail-archive.com/smartos-discuss@lists.smartos.org/msg02873.html.</a> Also I have read this: <a style=" font-family:'Courier New'; font-size: 9pt;" href="https://www.illumos.org/issues/6908">https://www.illumos.org/issues/6908</a> and other informations on web. The progress of my installation you can read here (if you speak german) - please follow all posts of effemmess: <a style=" font-family:'Courier New'; font-size: 9pt;" href="http://www.hardwareluxx.de/community/f101/zfs-stammtisch-570052-272.html#post24644356">http://www.hardwareluxx.de/community/f101/zfs-stammtisch-570052-272.html#post24644356</a> The problem in <a style=" font-family:'Courier New'; font-size: 9pt;" href="http://www.hardwareluxx.de/community/f101/zfs-stammtisch-570052-272.html#post24660135">http://www.hardwareluxx.de/community/f101/zfs-stammtisch-570052-272.html#post24660135</a> is solved -it was a old updated vm, now I use an fresh installed vm... But previosly I thought, there is something with the powerstates of the nvme. Because of that I made these additions to /etc/vmware/passthru.map (info is from here: <a style=" font-family:'Courier New'; font-size: 9pt;" href="http://www.vmware.com/pdf/vsp_4_vmdirectpath_host.pdf">http://www.vmware.com/pdf/vsp_4_vmdirectpath_host.pdf</a>: # Samsung NVMe 144d a801 d3d0 false 144d a802 d3d0 false These additions are active till now... The actual problem is, that after every cold-boot of the vm the nvme hangs. I test this by running the command "format". I cant see in any logs something interesting. Now I reset the vm via second ssh and the command "reboot -p". Now works the nvme very well for hours. As long as I only reboot the vm all is okay, but if I cold-boot the vm, the nvme hangs again. The whole process is reproducible. Gea says in hardwareluxx-forum, he has also problems with "format" with intel-NVMe. Btw, in all Windows-VMs the NVMe runs promptly after every cold-vm-start. Now I testet it in a freshly installed Openindiana-vm - full updated. Here I have a critical fault after every cold-boot and soft-boot. Also a "fmadm repaired dev:////pci@0,0/pci15ad,7a0@15/pci144d,a801@0" doesn't help. In summary, the NVMe runs always very well on bar-metal and on Windows-vm's and after soft-boot also very well in OmniOS. Can anybody help? Best regards, Frank </body></html>