Re: [RFC PATCH 13/21] i386/xen: implement HYPERVISOR_hvm_op

From: David Woodhouse <dwmw2@infradead.org>
To: "Philippe Mathieu-Daudé" <philmd@linaro.org>, qemu-devel@nongnu.org
Cc: Paolo Bonzini <pbonzini@redhat.com>, Paul Durrant <paul@xen.org>,
	Joao Martins <joao.m.martins@oracle.com>,
	Ankur Arora <ankur.a.arora@oracle.com>
Subject: Re: [RFC PATCH 13/21] i386/xen: implement HYPERVISOR_hvm_op
Date: Tue, 06 Dec 2022 01:18:14 +0000	[thread overview]
Message-ID: <ddda9a672fb8beba420f988f513420ad66e1c717.camel@infradead.org> (raw)
In-Reply-To: <4d6b112e-ab74-5ae3-7206-9b1aafb25034@linaro.org>

[-- Attachment #1: Type: text/plain, Size: 1523 bytes --]

On Mon, 2022-12-05 at 23:13 +0100, Philippe Mathieu-Daudé wrote:
> > +static int kvm_xen_hcall_hvm_op(struct kvm_xen_exit *exit,
> > +                                int cmd, uint64_t arg)
> > +{
> > +    switch (cmd) {
> > +    case HVMOP_pagetable_dying: {
> > +            exit->u.hcall.result = -ENOSYS;
> > +            return 0;
> > +        }
> > +    }
> 
> Could it be helpful to have a trace event here, or log a GUEST_ERROR?
> 
> > +    exit->u.hcall.result = -ENOSYS;
> > +    return HCALL_ERR;

We already have a trace event for hypercalls. So the unimplemented ones
look something like this (e.g. failing to set FIFO event channels, then
failing to get the CONSOLE_EVTCHN HVM param)...

[    0.151084] NR_IRQS: 524544, nr_irqs: 256, preallocated irqs: 16
kvm_xen_hypercall xen_hypercall: cpu 0 cpl 0 input 32 a0 0xb a1 0xffffffffbda03da0 a2 0xffff93cc3ec2e9a0 ret 0xffffffffffffffda
[    0.152018] xen:events: Using 2-level ABI
kvm_xen_hypercall xen_hypercall: cpu 0 cpl 0 input 34 a0 0x0 a1 0xffffffffbda03dd8 a2 0x7ff0 ret 0x0
[    0.152731] xen:events: Xen HVM callback vector for event delivery is enabled
[    0.154158] rcu: srcu_init: Setting srcu_struct sizes based on contention.
[    0.170239] Console: colour VGA+ 80x25
kvm_xen_hypercall xen_hypercall: cpu 0 cpl 0 input 34 a0 0x1 a1 0xffffffffbda03e60 a2 0x7ff0 ret 0xffffffffffffffda
[    0.170966] Cannot get hvm parameter CONSOLE_EVTCHN (18): -38!

(I just fixed a PRIu64 to PRIx64 in the trace event definition)

[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 5965 bytes --]