Re: [Xen-devel] nvmx deadlock with MSR bitmaps

2020-03-12 Thread Jürgen Groß
On 12.03.20 16:11, Jan Beulich wrote: On 12.03.2020 14:44, Roger Pau Monné wrote: On Thu, Mar 12, 2020 at 12:12:12PM +0100, Jürgen Groß wrote: On 12.03.20 11:56, Roger Pau Monné wrote: On Thu, Mar 12, 2020 at 09:59:48AM +0100, Jan Beulich wrote: On 11.03.2020 19:04, Andrew Cooper wrote: Spec

Re: [Xen-devel] nvmx deadlock with MSR bitmaps

2020-03-12 Thread Jan Beulich
On 12.03.2020 14:44, Roger Pau Monné wrote: > On Thu, Mar 12, 2020 at 12:12:12PM +0100, Jürgen Groß wrote: >> On 12.03.20 11:56, Roger Pau Monné wrote: >>> On Thu, Mar 12, 2020 at 09:59:48AM +0100, Jan Beulich wrote: On 11.03.2020 19:04, Andrew Cooper wrote: > Specifically, this is a switc

Re: [Xen-devel] nvmx deadlock with MSR bitmaps

2020-03-12 Thread Roger Pau Monné
On Thu, Mar 12, 2020 at 12:12:12PM +0100, Jürgen Groß wrote: > On 12.03.20 11:56, Roger Pau Monné wrote: > > On Thu, Mar 12, 2020 at 09:59:48AM +0100, Jan Beulich wrote: > > > On 11.03.2020 19:04, Andrew Cooper wrote: > > > > Specifically, this is a switch from an HVM vcpu, to a PV vcpu, where the

Re: [Xen-devel] nvmx deadlock with MSR bitmaps

2020-03-12 Thread Jürgen Groß
On 12.03.20 11:56, Roger Pau Monné wrote: On Thu, Mar 12, 2020 at 09:59:48AM +0100, Jan Beulich wrote: On 11.03.2020 19:04, Andrew Cooper wrote: Specifically, this is a switch from an HVM vcpu, to a PV vcpu, where the mapcache code tries to access the per-domain mappings on the HVM monitor tabl

Re: [Xen-devel] nvmx deadlock with MSR bitmaps

2020-03-12 Thread Jan Beulich
On 12.03.2020 11:56, Roger Pau Monné wrote: > On Thu, Mar 12, 2020 at 09:59:48AM +0100, Jan Beulich wrote: >> On 11.03.2020 19:04, Andrew Cooper wrote: >>> Specifically, this is a switch from an HVM vcpu, to a PV vcpu, where the >>> mapcache code tries to access the per-domain mappings on the HVM m

Re: [Xen-devel] nvmx deadlock with MSR bitmaps

2020-03-12 Thread Roger Pau Monné
On Thu, Mar 12, 2020 at 09:59:48AM +0100, Jan Beulich wrote: > On 11.03.2020 19:04, Andrew Cooper wrote: > > Specifically, this is a switch from an HVM vcpu, to a PV vcpu, where the > > mapcache code tries to access the per-domain mappings on the HVM monitor > > table.  It ends up trying to recursi

Re: [Xen-devel] nvmx deadlock with MSR bitmaps

2020-03-12 Thread Jan Beulich
On 11.03.2020 19:04, Andrew Cooper wrote: > Specifically, this is a switch from an HVM vcpu, to a PV vcpu, where the > mapcache code tries to access the per-domain mappings on the HVM monitor > table.  It ends up trying to recursively acquire the mapcache lock while > trying to walk %cr2 to identif

[Xen-devel] nvmx deadlock with MSR bitmaps

2020-03-11 Thread Andrew Cooper
Hello, Testing has encountered this deadlock: (XEN) Watchdog timer detects that CPU0 is stuck! (XEN) [ Xen-4.14.0-9.0.4-d  x86_64  debug=y   Not tainted ] (XEN) CPU:    0 (XEN) RIP:    e008:[] _spin_lock+0x34/0x5e (XEN) RFLAGS: 0002   CONTEXT: hypervisor (d0v0) (XEN) rax: