RE: arm64: virtio broken in upstream kernel

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 




> -----Original Message-----
> From: Richard W.M. Jones [mailto:rjones@xxxxxxxxxx]
> Sent: Tuesday, December 02, 2014 5:54 PM
> To: Wang, Yalin
> Cc: kvmarm@xxxxxxxxxxxxxxxxxxxxx; Wei Huang
> Subject: Re: arm64: virtio broken in upstream kernel
> 
> On Tue, Dec 02, 2014 at 05:51:48PM +0800, Wang, Yalin wrote:
> > > -----Original Message-----
> > > From: Richard W.M. Jones [mailto:rjones@xxxxxxxxxx]
> > > Sent: Tuesday, December 02, 2014 5:47 PM
> > > To: Wang, Yalin
> > > Cc: kvmarm@xxxxxxxxxxxxxxxxxxxxx; Wei Huang
> > > Subject: Re: arm64: virtio broken in upstream kernel
> > >
> > > On Tue, Dec 02, 2014 at 05:45:08PM +0800, Wang, Yalin wrote:
> > > > > -----Original Message-----
> > > > > From: Richard W.M. Jones [mailto:rjones@xxxxxxxxxx]
> > > > > Sent: Monday, December 01, 2014 10:02 PM
> > > > > To: kvmarm@xxxxxxxxxxxxxxxxxxxxx
> > > > > Cc: Wang, Yalin; Wei Huang
> > > > > Subject: Re: arm64: virtio broken in upstream kernel
> > > > >
> > > > > OK I fixed it ...
> > > > >
> > > > > This patch to qemu fixes the problem for me.  I will post it to
> > > > > qemu-devel shortly.
> > > > >
> > > > > Rich.
> > > > Oh , I see the BUG ,
> > > > It load dtb in the same page with initrd, This patch can fix, I
> > > > think more better solution is that You can get page size from
> > > > kernel by
> > > > sysinfo() system call, Then align most data into PAGE_SIZE, This
> > > > is good for both 32 / 64 arch .
> > >
> > > I guess the problem is that the host might be using 4K pages and the
> > > guest might be using 64K pages.  (Is that possible?)
> > >
> > No , I don't think so,
> > I think you just need align address to the page size of guest kernel,
> > Then it will work.
> >
> > So this is a problem, sysinfo() can't work, How to get
> > CONFIG_PAGE_SIZE from guest kernel image before Boot it?
> 
> Sounds horrible :-(
> 
> Did you see also my kernel patch?
> 
> https://lists.cs.columbia.edu/pipermail/kvmarm/2014-December/012567.html
> 
> This is what I ended up using.
> 
Oh, this patch just revert my origin kernel patch,
This solution will also fix qemu problem.
But It will not free the head or tail page of inidrd
if the address is not aligned if you do like this.

This means some page will lost and can't be used by kernel.

	





_______________________________________________
kvmarm mailing list
kvmarm@xxxxxxxxxxxxxxxxxxxxx
https://lists.cs.columbia.edu/mailman/listinfo/kvmarm




[Index of Archives]     [Linux KVM]     [Spice Development]     [Libvirt]     [Libvirt Users]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux