Subject: Re: [PATCH v4 2/3] x86/xen/time: setup vcpu 0 time info page
To: Joao Martins <joao.m.martins@oracle.com>
References: <20170927134623.3147-1-joao.m.martins@oracle.com>
 <20170927134623.3147-3-joao.m.martins@oracle.com>
 <b38ff455-d225-67d5-cf20-047ea54abce4@oracle.com>
 <a1fb7112-f8fd-7973-a48a-e1c6ef46b1c8@oracle.com>
Cc: linux-kernel@vger.kernel.org, xen-devel@lists.xenproject.org,
        Thomas Gleixner <tglx@linutronix.de>, Ingo Molnar <mingo@redhat.com>,
        "H. Peter Anvin" <hpa@zytor.com>, x86@kernel.org,
        Juergen Gross <jgross@suse.com>, Andy Lutomirski <luto@amacapital.net>
From: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Message-ID: <0f58594c-ee42-7813-a0c2-1cbbc1e2a576@oracle.com>
Date: Wed, 27 Sep 2017 16:22:25 -0400
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101
 Thunderbird/45.8.0
MIME-Version: 1.0
In-Reply-To: <a1fb7112-f8fd-7973-a48a-e1c6ef46b1c8@oracle.com>
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2635
Lines: 76

On 09/27/2017 11:26 AM, Joao Martins wrote:
> On 09/27/2017 03:40 PM, Boris Ostrovsky wrote:
>>> +static void xen_setup_vsyscall_time_info(void)
>>> +{
>>> +	struct vcpu_register_time_memory_area t;
>>> +	struct pvclock_vsyscall_time_info *ti;
>>> +	struct pvclock_vcpu_time_info *pvti;
>>> +	int ret;
>>> +
>>> +	pvti = &__this_cpu_read(xen_vcpu)->time;
>>> +
>>> +	/*
>>> +	 * We check ahead on the primary time info if this
>>> +	 * bit is supported hence speeding up Xen clocksource.
>>> +	 */
>>> +	if (!(pvti->flags & PVCLOCK_TSC_STABLE_BIT))
>>> +		return;
>>> +
>>> +	pvclock_set_flags(PVCLOCK_TSC_STABLE_BIT);
>> Is it OK to have this flag set if anything below fails?
>>
> Yes - if anything below fails it will only affect userspace mapped page.

Then should it be set somewhere else, like in xen_time_init()?

-boris

>  What I
> do above is just allowing xen clocksource to use/check that bit (consequently
> speeding up sched_clock) given the necessary support is there in the master
> copy. The secondary copy (i.e. what's being set up below, mapped/used in vdso)
> has the same data from the master copy, just separate memory regions. The checks
> below are just for the unlikely cases of failing to register the secondary copy
> or if its content were to differ from master copy in future releases - and
> therefore we handle those more gracefully.
>
>> (I can see in the changelog that apparently at some point I've asked
>> about this at v1 but I can't remember/find what exactly it was)
>>
>>> +
>>> +	ti = (struct pvclock_vsyscall_time_info *)get_zeroed_page(GFP_KERNEL);
>>> +	if (!ti)
>>> +		return;
>>> +
>>> +	t.addr.v = &ti->pvti;
>>> +
>>> +	ret = HYPERVISOR_vcpu_op(VCPUOP_register_vcpu_time_memory_area, 0, &t);
>>> +	if (ret) {
>>> +		pr_notice("xen: VCLOCK_PVCLOCK not supported (err %d)\n", ret);
>>> +		free_page((unsigned long)ti);
>>> +		return;
>>> +	}
>>> +
>>> +	/*
>>> +	 * If the check above succedded this one should too since it's the
>>> +	 * same data on both primary and secondary time infos just different
>>> +	 * memory regions. But we still check it in case hypervisor is buggy.
>>> +	 */
>>> +	pvti = &ti->pvti;
>>> +	if (!(pvti->flags & PVCLOCK_TSC_STABLE_BIT)) {
>>> +		t.addr.v = NULL;
>>> +		ret = HYPERVISOR_vcpu_op(VCPUOP_register_vcpu_time_memory_area,
>>> +					 0, &t);
>>> +		if (!ret)
>>> +			free_page((unsigned long)ti);
>>> +
>>> +		pr_notice("xen: VCLOCK_PVCLOCK not supported (tsc unstable)\n");
>>> +		return;
>>> +	}
>>> +
>>> +	xen_clock = ti;
>>> +	pvclock_set_pvti_cpu0_va(xen_clock);
>>> +
>>> +	xen_clocksource.archdata.vclock_mode = VCLOCK_PVCLOCK;
>>> +}
>>> +