On Fri, 28 Jul 2017 08:22:45 +0300 Dmitry Fleytman <dmi...@daynix.com> wrote:
> > On 28 Jul 2017, at 07:51 AM, Zhang, Xiong Y <xiong.y.zh...@intel.com> wrote: > > > >>> On 26 Jul 2017, at 08:22 AM, Zhang, Xiong Y <xiong.y.zh...@intel.com> > >> wrote: > >>> > >>> Sorry, we indeed found Intel windows guest graphic driver couldn't be > >>> bind > >> when GMS memory size is zero. And we have fixed it and the next intel > >> windows driver release will contain this fix. > >>> So currently please use x-igd-gms in legacy mode to work around this. > >>> > >>> BTW, how did you know window driver allocate extra ~4G memory when > >> GMS size set to 0 ? > >> > >> We noticed that with new IGD driver memory usage on VMs raised by around > >> 4G. > >> > >> Then we examined memory allocations with poolmon > >> (https://docs.microsoft.com/en-us/windows-hardware/drivers/devtest/poolm > >> on) > >> and saw that around 4G of memory is allocated with IGD driver pool tag. > >> Additionally we noticed that on VMs with less than 5G of memory IGD driver > >> does not load and system does fallback to standard VGA driver. > >> > > [Zhang, Xiong Y] I tried our internal driver and didn't found the above > > two issues, please wait for the next intel graphic driver release. > > Thanks! So to close on this, we should just be recommending that users encountering this problem add the x-igd-gms=1 option to the device until the next Intel graphics release. I won't plan to push this change. Thanks all, Alex