Re: [PATCH v7 09/13] KVM: x86: Implement Intel Processor Trace context switch

From: Paolo Bonzini <pbonzini@redhat.com>
To: Alexander Shishkin <alexander.shishkin@linux.intel.com>,
	Luwei Kang <luwei.kang@intel.com>
Cc: kvm@vger.kernel.org, tglx@linutronix.de, mingo@redhat.com,
	hpa@zytor.com, x86@kernel.org, rkrcmar@redhat.com,
	linux-kernel@vger.kernel.org, joro@8bytes.org,
	peterz@infradead.org, chao.p.peng@linux.intel.com
Subject: Re: [PATCH v7 09/13] KVM: x86: Implement Intel Processor Trace context switch
Date: Fri, 4 May 2018 23:49:44 +0200	[thread overview]
Message-ID: <c339bc07-c852-cb76-6c65-393488bdfe7e@redhat.com> (raw)
In-Reply-To: <20180504102958.u6c3mqackl4kzd54@um.fi.intel.com>

On 04/05/2018 12:29, Alexander Shishkin wrote:
> On Thu, May 03, 2018 at 08:08:39PM +0800, Luwei Kang wrote:
>> +static void pt_guest_enter(struct vcpu_vmx *vmx)
>> +{
>> +	if (pt_mode == PT_MODE_HOST || pt_mode == PT_MODE_HOST_GUEST)
>> +		rdmsrl(MSR_IA32_RTIT_CTL, vmx->pt_desc.host.ctl);
>> +
>> +	if (pt_mode == PT_MODE_HOST_GUEST &&
>> +		vmx->pt_desc.guest.ctl & RTIT_CTL_TRACEEN) {
>> +		wrmsrl(MSR_IA32_RTIT_CTL, 0);
>> +		pt_save_msr(&vmx->pt_desc.host, vmx->pt_desc.range_cnt);
>> +		pt_load_msr(&vmx->pt_desc.guest, vmx->pt_desc.range_cnt);
>> +	}
>> +}
>> +
>> +static void pt_guest_exit(struct vcpu_vmx *vmx)
>> +{
>> +	if (pt_mode == PT_MODE_HOST_GUEST &&
>> +		vmx->pt_desc.guest.ctl & RTIT_CTL_TRACEEN) {
>> +		pt_save_msr(&vmx->pt_desc.guest, vmx->pt_desc.range_cnt);
>> +		pt_load_msr(&vmx->pt_desc.host, vmx->pt_desc.range_cnt);
>> +	}
>> +
>> +	if (pt_mode == PT_MODE_HOST || pt_mode == PT_MODE_HOST_GUEST)
>> +		wrmsrl(MSR_IA32_RTIT_CTL, vmx->pt_desc.host.ctl);
>> +}
> 
> This means that a host PT event won't get the guest traces and won't get
> any notification as to what'd happened or why. At the minimum we need to
> send a PARTIAL AUX record at the pt_guest_enter(), when we turn the host
> tracing off.

How do you do that? :)

Paolo