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

Re: [Xen-devel] Debian Wheezy installer crashes guest when EPT is enabled



>>> On 18.04.14 at 16:08, <wei.liu2@xxxxxxxxxx> wrote:
> (d2) Booting from DVD/CD...
> (d2) Booting from 0000:7c00
> (XEN) stdvga.c:151:d2v0 leaving stdvga
> [  261.416031] xenbr0: port 3(vif2.0-emu) entered forwarding state
> (XEN) hap.c:273: d2 failed to allocate from HAP pool<G><0>vmx.c:3063:d2v0 Bad 
> vmexit (reason 0x31)
> (XEN) domain_crash called from vmx.c:3064
> (XEN) Domain 2 (vcpu#0) crashed on cpu#7:
> (XEN) ----[ Xen-4.5-unstable  x86_64  debug=y  Tainted:    C ]----
> (XEN) CPU:    7
> (XEN) RIP:    0010:[<ffffffff811b2ab2>]
> (XEN) RFLAGS: 0000000000010016   CONTEXT: hvm guest
> (XEN) rax: 0000000000000000   rbx: ffffffff81601d30   rcx: 000000000000003f
> (XEN) rdx: 8000000000000163   rsi: ffffffff81732ba0   rdi: ffffffffff4b8000
> (XEN) rbp: 00000000efffc000   rsp: ffffffff81601cd0   r8:  8000000000000163
> (XEN) r9:  0000000000000000   r10: 0000000000000000   r11: 0000000000000000
> (XEN) r12: ffffffffff4b8000   r13: 0000000148000000   r14: 0000000000000000
> (XEN) r15: 0000000148000000   cr0: 0000000080050033   cr4: 00000000000000b0
> (XEN) cr3: 0000000001605000   cr2: 0000000000000000
> (XEN) ds: 0000   es: 0000   fs: 0000   gs: 0000   ss: 0000   cs: 0010

Are you (or the tools you use) perhaps setting a particularly small
shadow allocation size (which is what gets interpreted as the HAP pool
one)? When I run a 2-vCPU guest (SLES, not Wheezy), I see
- for memory=2048 0x46 P2M pages out of a HAP pool of 0x1200
- for memory=4096 0x4d P2M pages out of a HAP pool of 0x2200
- for memory=4864 0x76 P2M pages out of a HAP pool of 0x2800
And I have no reason to believe that the numbers would dramatically
change if I bumped the domain size to 5G (the system has got only 6G,
and I don't want to trouble Dom0 too much). The almost doubling of
the P2M page count is presumably attributed to the fact that there are
not enough contiguous 2Mb pages left anymore, i.e. 4k pages (and
hence more page tables) are being used.

If that's not it, or if you're not certain, would you mind giving the
attached debugging patch a try?

Jan

Attachment: hap.diff
Description: Text document

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

 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.