Re: Forcing device loss for debugging
Under the sysfs directory for each PCI device you should have a virtual file called "remove". So if you just do an "echo 1 > /sys/devices/pci/remove" you basically simulate hot unplugging the specific PCI device IIRC the PCI bridge where the device is connected to should then have a
Forcing device loss for debugging
Is there any way to force a Vulkan device to report itself lost, in order to exercise one's application's recovery paths? I looked through the Mesa sources but didn't find anything that seemed suitable.