Re: [Xen-devel] XEN[ARM] Master not working on Allwinner A20

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Mon, 2013-12-16 at 12:04 +0100, peter wrote:
> But I have also found another bug between 720f45ad..d9639236 in the DT
> bootargs parsing.
> When my DTS is:
> xen,dom0-bootargs = "console=hvc0,115200n8 debug ignore_loglevel rw
> rootwait earlyprintk=xen clk_ignore_unused ";
> It successfully boots (but I can't mount my rootfs :( ), but when
> dom0-bootargs is:
> xen,dom0-bootargs = "console=hvc0,115200n8 debug ignore_loglevel rw
> rootwait earlyprintk=xen clk_ignore_unused root=/dev/nfs
> nfsroot=10.31.5.161:/export/cubie/rootfs,v3,tcp
> ip=10.31.5.165:10.31.5.161:10.31.4.1:255.255.252.0:cubie:eth0:off";

"console=hvc0" is sufficient, the ",115200n8" is not needed. But that
isn't your problem.

I can't see anything about the second command line which would cause it
to be ignored -- which is what seems to be happening because the next
line in the log should be dom0 console stuff.

You first log (which was for a different case) showed:
        MODULE[1]: 00000000402ef700 - 00000000402f1c57 
        MODULE[2]: 0000000050000000 - 0000000050f00000 
        MODULE[3]: 0000000060000000 - 0000000066000000 

but I would expect MODULE[2] (the kernel) to include a command line
after its memory range. Is that the case here too?

I'm not really sure what is going on here. A full log and visibility
into your DTB might help. It might also be worth adding some extra
printk logging to xen/arch/arm/domain_build.c:write_properties() where
the xen,dom0-bootargs are processed and perhaps to
xen/common/device_tree.c where the host DTB is parsed for options (use
dt_printk here).

> 
> 720f45ad Is working for me I can create multiple vm's but I get some
> kernel messages when creating them:
> [  278.215653] Failed to unmap pfn:5e035 rc:-2
> [  278.219786] Failed to unmap pfn:5e035 rc:-2
> [  278.233617] Failed to unmap pfn:5f033 rc:-2
> [  278.237857] Failed to unmap pfn:5e035 rc:-2
> [  278.248553] Failed to unmap pfn:5e033 rc:-2
> [  278.256166] Failed to unmap pfn:5e032 rc:-2

That's a known issue, it is a dom0 memory leak so it's mostly benign
unless you are creating/destroying lots of domains. There is a fix in
the pipeline.

> 
> And when I destroy a VM I get this:
> (XEN) mm.c:1250:d0 gnttab_mark_dirty not implemented yet
> 
> The rootfs of dom0 is Ubuntu 13.04 and is mounted using NFS.

Ah, NFS doesn't need DMA, which is how it managed to work without the
1:1 workaround I think. I'm using SATA which does need DMA. Not sure
about MMC.

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel