I got a purple screen when I start up a vm with a pci device
With Esxi5 was working fine.
I tested it with 2 vm on different hosts ( but same hardware)
i7 2600 32gb
What can I try
Other than performance and convenience the biggest reason is because they only emulate a few devices if you do not pass through, and for everything else if you can't pass through the USB controller, you can't use it, which is a serious problem.
HI guys ,
Just to report
I can passtrough a ATI 6850, and a LSI 2008 raid
But cannot pass my usb 2.0 and usb 3.0 Witch are Inter Corp Panther Point enhanced host onboard of my Asrock Z77 Pro 3
Lets just hope its fix soon.
Update to the brand new controller : got the same problem with an Adaptec 29320LP and reopened the case at VMware.
A quality testing methodology would have prevented this.
Have this release been through UAT?
Maybe it would be helpful to given them a link to this thread so they can see that this isn't an isolated problem solely with a few unusual devices.
I am having similar problems trying to pass through two devices:
Intel 82579LM NIC
I can pass both through both on 5.0 and on 5.0 U1.
On 5.1, as soon (20%) as I start the VM I get #PF Exception 14...Code start: 0x418036200000...PCI_getExtCapIdx@vmkernel...
I would be nice to have a solution to this problem that seems relatively common.
@srwsol : As I already said VMware is aware of the pass-through problem (see post above) and a patch is being released in the next two or three months after an express patch will be released for other more important issues with ESXi 5.1.
@knoxel : I had the same problem with a LSI U320 adapter - which I unsuccessfully replaced by an Adaptec 29320LPE U320 adapter - seems like the problem (at least on my side) is that our HP DL360G6 server lets ESXi include the underlying PCIe to PCI-X bridge which is in my opinion the core problem. As the VMware support pointed out the problem is that PCIe commands are issued when starting up ESXi which should list PCIe capabilities - returning values on a native PCIe devices, but return nothing on bridged PCI-X (non-native PCIe) devices and therefore letting the system crash. So the new ESXi release seems to not have been tested thoroughly enough for all possibilities. Additional to this the VMware supporter pointed out that they have now included automatic testing procedures for these devices.
Btw. I have worked around this issue using the VMware unsupported directly attached SCSI devices by creating two new SCSI devices in the VM's configuration. But since my device is a multiple LUN aware HP Autoloader 1/8 I needed to issue this command...
esxcli storage nmp satp rule add --satp VMW_SATP_LOCAL --driver="aic79xx" --description="Multiple LUN support for Adaptec Card"
...on the ESXi's shell and reboot in order to let both devices (tape and changer) appear. In my case I used "LSI Logic Parallel" as both SCSI controllers for both devices. But like USB arbitration the speed is not nearly the same as with pass-through (around 30% to 50%).
Just as an update : the performance of the new configuration does not look as bad as I wrote above, it seems to be more or less comparable.
As there is an update already planned to solve the pass-through issue I have closed my case at VMware and will continue to use this configuration for my backup device.
To find out which driver you need use:-
Once I worked out we didn't use an adaptec it worked fine.
I am having similar problem on an Asus P8B WS (C206 chipset). I used to pass thought one of the USB 2.0 controllers and the USB 3.0 controller to virtual machines. After upgrading to ESXi 5.1, the USB 3.0 shows up as ASMedia device, not as unknown, which is positive. But when I select it for passthrough and restart the box, it resets to non-passthrough device. So basically it's impossible to add. Also, the two USB 2.0 controllers disappeared from the list, I cannot even select and add them.
Going back to 5.0u1 as soon as I can afford a restart
Has anyone tried patch ESXi510-201210001? Does it resolve the issue?
iv upgrade my esxi with ESXi510-201210001 and its still not resove
still open bug
I'm new to ESXi and think I am having a similar issue. i am trying to pass through a Mellanox MHEA28-XT but ESXI keeps crashing to the purple screen. What is the recommeneded fix for this? I am using ESXi 5.1 and have enabled passthrough on my motherboard.
I'm afraid the only fix available now is downgrade to the latest esxi 5.0u1 release.
edit: changed version 4 in 5.0, sorry for the mistake.
But why would an older version of ESXi work that seems a bit counter-intuitive!?
I will try with ESXi 4.1 Installable Update 3
I agree, I think like many others that the release of version 5 isn't well tested, cause as so many peaple encounter problems with VMDirectpath/PCI-Passthrough it is a fault/mistake that should have been come up in the testing procedure.
The only thing we can do now is wait till they release a fix for version 5.
Whoops little edit: you can go to version 5.0 update 1, the problem occurs in 5.1, sorry for the mistake.
Thanks I'll give that a try. Do i need to downgrade my Vsphere Client as well?
VSphere should work fine with older esxi installations.
Unfortunately it is still giving the same error
My specs are:
Xeon 1230 V2
Gigabyte G1 Sniper 3
ESXI version 5.0 update 1
I am trying to pass through a Mellanox MHEA28-XT card in one of the PCie slots.