kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Glauber Costa <glommer@parallels.com>
To: Marcelo Tosatti <mtosatti@redhat.com>
Cc: <kvm@vger.kernel.org>, <johnstul@us.ibm.com>, <jeremy@goop.org>,
	<zamsden@gmail.com>, <gleb@redhat.com>, <avi@redhat.com>,
	<pbonzini@redhat.com>
Subject: Re: [patch 07/16] x86: pvclock: generic pvclock vsyscall initialization
Date: Thu, 1 Nov 2012 18:19:46 +0400	[thread overview]
Message-ID: <50928502.6000309@parallels.com> (raw)
In-Reply-To: <20121031224824.149072621@redhat.com>

On 11/01/2012 02:47 AM, Marcelo Tosatti wrote:
> Originally from Jeremy Fitzhardinge.
> 
> Introduce generic, non hypervisor specific, pvclock initialization 
> routines.
> 
> Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
> 
> Index: vsyscall/arch/x86/kernel/pvclock.c
> ===================================================================
> --- vsyscall.orig/arch/x86/kernel/pvclock.c
> +++ vsyscall/arch/x86/kernel/pvclock.c
> @@ -17,6 +17,10 @@
>  
>  #include <linux/kernel.h>
>  #include <linux/percpu.h>
> +#include <linux/notifier.h>
> +#include <linux/sched.h>
> +#include <linux/gfp.h>
> +#include <linux/bootmem.h>
>  #include <asm/pvclock.h>
>  
>  static u8 valid_flags __read_mostly = 0;
> @@ -122,3 +126,67 @@ void pvclock_read_wallclock(struct pvclo
>  
>  	set_normalized_timespec(ts, now.tv_sec, now.tv_nsec);
>  }
> +
> +static aligned_pvti_t *pvclock_vdso_info;
is there a non-aligned pvti ?

I would also add that "pvti" is not exactly the most descriptive name
I've seen.

> +
> +static struct pvclock_vsyscall_time_info *
> +pvclock_get_vsyscall_user_time_info(int cpu)
> +{
> +	if (pvclock_vdso_info == NULL) {
> +		BUG();
> +		return NULL;
> +	}
> +
> +	return &pvclock_vdso_info[cpu].info;
> +}
> +
if (!pvclock_vdso_info)
    BUG();
return &pvclock...

> +struct pvclock_vcpu_time_info *pvclock_get_vsyscall_time_info(int cpu)
> +{
> +	return &pvclock_get_vsyscall_user_time_info(cpu)->pvti;
> +}
> +
> +int pvclock_task_migrate(struct notifier_block *nb, unsigned long l, void *v)
> +{
> +	struct task_migration_notifier *mn = v;
> +	struct pvclock_vsyscall_time_info *pvti;
> +
> +	pvti = pvclock_get_vsyscall_user_time_info(mn->from_cpu);
> +
> +	if (pvti == NULL)
> +		return NOTIFY_DONE;
> +

When is it NULL? IIUC, this is when the vsyscall is disabled, right?
Would you mind adding comments for it?

Also, this is supposed to be an unlikely branch, right?

> +	pvti->migrate_count++;
> +
> +	return NOTIFY_DONE;
> +}
> +
> +static struct notifier_block pvclock_migrate = {
> +	.notifier_call = pvclock_task_migrate,
> +};
> +
> +/*
> + * Initialize the generic pvclock vsyscall state.  This will allocate
> + * a/some page(s) for the per-vcpu pvclock information, set up a
> + * fixmap mapping for the page(s)
> + */
> +int __init pvclock_init_vsyscall(void)
> +{
> +	int idx;
> +	unsigned int size = PVCLOCK_VSYSCALL_NR_PAGES*PAGE_SIZE;
> +
> +	pvclock_vdso_info = __alloc_bootmem(size, PAGE_SIZE, 0);

This will be the most critical path for reading pvclock's time data. So
why can't we:
1) Make it per-node.
2) Of course, as a consequence, make sure all info structures in the
same page are in the same node ?



> +	if (!pvclock_vdso_info)
> +		return -ENOMEM;
> +
> +	memset(pvclock_vdso_info, 0, size);
> +
> +	for (idx = 0; idx <= (PVCLOCK_FIXMAP_END-PVCLOCK_FIXMAP_BEGIN); idx++) {
> +		__set_fixmap(PVCLOCK_FIXMAP_BEGIN + idx,
> +			     __pa_symbol(pvclock_vdso_info) + (idx*PAGE_SIZE),
> +			     PAGE_KERNEL_VVAR);
> +	}
> +
> +	register_task_migration_notifier(&pvclock_migrate);
> +
> +	return 0;
> +}
> Index: vsyscall/arch/x86/include/asm/fixmap.h
> ===================================================================
> --- vsyscall.orig/arch/x86/include/asm/fixmap.h
> +++ vsyscall/arch/x86/include/asm/fixmap.h
> @@ -19,6 +19,7 @@
>  #include <asm/acpi.h>
>  #include <asm/apicdef.h>
>  #include <asm/page.h>
> +#include <asm/pvclock.h>
>  #ifdef CONFIG_X86_32
>  #include <linux/threads.h>
>  #include <asm/kmap_types.h>
> @@ -81,6 +82,10 @@ enum fixed_addresses {
>  	VVAR_PAGE,
>  	VSYSCALL_HPET,
>  #endif
> +#ifdef CONFIG_PARAVIRT_CLOCK
> +	PVCLOCK_FIXMAP_BEGIN,
> +	PVCLOCK_FIXMAP_END = PVCLOCK_FIXMAP_BEGIN+PVCLOCK_VSYSCALL_NR_PAGES-1,
> +#endif
>  	FIX_DBGP_BASE,
>  	FIX_EARLYCON_MEM_BASE,
>  #ifdef CONFIG_PROVIDE_OHCI1394_DMA_INIT
> Index: vsyscall/arch/x86/include/asm/pvclock.h
> ===================================================================
> --- vsyscall.orig/arch/x86/include/asm/pvclock.h
> +++ vsyscall/arch/x86/include/asm/pvclock.h
> @@ -13,6 +13,8 @@ void pvclock_read_wallclock(struct pvclo
>  			    struct pvclock_vcpu_time_info *vcpu,
>  			    struct timespec *ts);
>  void pvclock_resume(void);
> +int __init pvclock_init_vsyscall(void);
> +struct pvclock_vcpu_time_info *pvclock_get_vsyscall_time_info(int cpu);
>  
>  /*
>   * Scale a 64-bit delta by scaling and multiplying by a 32-bit fraction,
> @@ -85,4 +87,17 @@ unsigned __pvclock_read_cycles(const str
>  	return version;
>  }
>  
> +struct pvclock_vsyscall_time_info {
> +	struct pvclock_vcpu_time_info pvti;
> +	u32 migrate_count;
> +};
> +
> +typedef union {
> +	struct pvclock_vsyscall_time_info info;
> +	char pad[SMP_CACHE_BYTES];
> +} aligned_pvti_t ____cacheline_aligned;

Please, just align pvclock_vsyscall_time_info. It is 10 thousand times
more descriptive.


  reply	other threads:[~2012-11-01 14:19 UTC|newest]

Thread overview: 94+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-10-24 13:13 [patch 00/18] pvclock vsyscall support + KVM hypervisor support (v2) Marcelo Tosatti
2012-10-24 13:13 ` [patch 01/18] KVM: x86: retain pvclock guest stopped bit in guest memory Marcelo Tosatti
2012-10-24 13:13 ` [patch 02/18] x86: pvclock: make sure rdtsc doesnt speculate out of region Marcelo Tosatti
2012-10-24 13:13 ` [patch 03/18] x86: pvclock: remove pvclock_shadow_time Marcelo Tosatti
2012-10-30  9:23   ` Avi Kivity
2012-10-30  9:24     ` Avi Kivity
2012-10-24 13:13 ` [patch 04/18] x86: pvclock: create helper for pvclock data retrieval Marcelo Tosatti
2012-10-24 13:13 ` [patch 05/18] x86: pvclock: fix flags usage race Marcelo Tosatti
2012-10-24 13:13 ` [patch 06/18] x86: pvclock: introduce helper to read flags Marcelo Tosatti
2012-10-24 13:13 ` [patch 07/18] sched: add notifier for cross-cpu migrations Marcelo Tosatti
2012-10-24 13:13 ` [patch 08/18] x86: pvclock: generic pvclock vsyscall initialization Marcelo Tosatti
2012-10-29 14:18   ` Glauber Costa
2012-10-29 14:54     ` Marcelo Tosatti
2012-10-29 17:46       ` Jeremy Fitzhardinge
2012-10-29 14:39   ` Glauber Costa
2012-10-24 13:13 ` [patch 09/18] KVM: x86: introduce facility to support vsyscall pvclock, via MSR Marcelo Tosatti
2012-10-29 14:45   ` Glauber Costa
2012-10-29 17:44     ` Jeremy Fitzhardinge
2012-10-29 18:40       ` Marcelo Tosatti
2012-10-30  7:41         ` Glauber Costa
2012-10-30  9:39         ` Avi Kivity
2012-10-31  3:12           ` Marcelo Tosatti
2012-11-02 10:21             ` Glauber Costa
2012-10-30  7:38       ` Glauber Costa
2012-10-24 13:13 ` [patch 10/18] x86: kvm guest: pvclock vsyscall support Marcelo Tosatti
2012-10-24 13:13 ` [patch 11/18] x86: vsyscall: pass mode to gettime backend Marcelo Tosatti
2012-10-29 14:47   ` Glauber Costa
2012-10-29 18:41     ` Marcelo Tosatti
2012-10-30  7:42       ` Glauber Costa
2012-10-24 13:13 ` [patch 12/18] x86: vdso: pvclock gettime support Marcelo Tosatti
2012-10-29 14:59   ` Glauber Costa
2012-10-29 18:42     ` Marcelo Tosatti
2012-10-30  7:49       ` Glauber Costa
2012-10-31  3:16         ` Marcelo Tosatti
2012-10-24 13:13 ` [patch 13/18] KVM: x86: pass host_tsc to read_l1_tsc Marcelo Tosatti
2012-10-29 15:04   ` Glauber Costa
2012-10-29 18:45     ` Marcelo Tosatti
2012-10-30  7:55       ` Glauber Costa
2012-10-24 13:13 ` [patch 14/18] time: export time information for KVM pvclock Marcelo Tosatti
2012-11-10  1:02   ` John Stultz
2012-11-13 21:07     ` Marcelo Tosatti
2012-10-24 13:13 ` [patch 15/18] KVM: x86: implement PVCLOCK_TSC_STABLE_BIT pvclock flag Marcelo Tosatti
2012-10-30  8:34   ` Glauber Costa
2012-10-31  3:19     ` [patch 15/18] KVM: x86: implement PVCLOCK_TSC_STABLE_BIT pvclock flag\ Marcelo Tosatti
2012-10-24 13:13 ` [patch 16/18] KVM: x86: notifier for clocksource changes Marcelo Tosatti
2012-10-24 13:13 ` [patch 17/18] KVM: x86: add kvm_arch_vcpu_postcreate callback, move TSC initialization Marcelo Tosatti
2012-10-24 13:13 ` [patch 18/18] KVM: x86: require matched TSC offsets for master clock Marcelo Tosatti
2012-10-31 22:46 ` [patch 00/16] pvclock vsyscall support + KVM hypervisor support (v3) Marcelo Tosatti
2012-10-31 22:46   ` [patch 01/16] KVM: x86: retain pvclock guest stopped bit in guest memory Marcelo Tosatti
2012-11-01 10:39     ` Gleb Natapov
2012-11-01 20:51       ` Marcelo Tosatti
2012-11-01 13:44     ` Glauber Costa
2012-10-31 22:46   ` [patch 02/16] x86: pvclock: make sure rdtsc doesnt speculate out of region Marcelo Tosatti
2012-11-01 11:48     ` Gleb Natapov
2012-11-01 13:49       ` Glauber Costa
2012-11-01 13:51         ` Gleb Natapov
2012-11-01 20:56         ` Marcelo Tosatti
2012-11-01 22:13           ` Gleb Natapov
2012-11-01 22:21             ` Marcelo Tosatti
2012-11-02  6:02               ` Gleb Natapov
2012-10-31 22:46   ` [patch 03/16] x86: pvclock: remove pvclock_shadow_time Marcelo Tosatti
2012-11-01 13:52     ` Glauber Costa
2012-10-31 22:47   ` [patch 04/16] x86: pvclock: create helper for pvclock data retrieval Marcelo Tosatti
2012-11-01 14:04     ` Glauber Costa
2012-11-01 20:57       ` Marcelo Tosatti
2012-10-31 22:47   ` [patch 05/16] x86: pvclock: introduce helper to read flags Marcelo Tosatti
2012-11-01 14:07     ` Glauber Costa
2012-11-01 21:08       ` Marcelo Tosatti
2012-10-31 22:47   ` [patch 06/16] sched: add notifier for cross-cpu migrations Marcelo Tosatti
2012-11-01 14:08     ` Glauber Costa
2012-10-31 22:47   ` [patch 07/16] x86: pvclock: generic pvclock vsyscall initialization Marcelo Tosatti
2012-11-01 14:19     ` Glauber Costa [this message]
2012-10-31 22:47   ` [patch 08/16] KVM: x86: introduce facility to support vsyscall pvclock, via MSR Marcelo Tosatti
2012-11-01 14:28     ` Glauber Costa
2012-11-01 21:39       ` Marcelo Tosatti
2012-11-02 10:23         ` Glauber Costa
2012-11-02 13:00           ` Marcelo Tosatti
2012-11-05  8:03             ` Glauber Costa
2012-10-31 22:47   ` [patch 09/16] x86: kvm guest: pvclock vsyscall support Marcelo Tosatti
2012-11-02  9:42     ` Glauber Costa
2012-11-05  8:35       ` Marcelo Tosatti
2012-10-31 22:47   ` [patch 10/16] x86: vdso: pvclock gettime support Marcelo Tosatti
2012-11-01 14:41     ` Glauber Costa
2012-11-01 21:42       ` Marcelo Tosatti
2012-11-02  0:33         ` Marcelo Tosatti
2012-11-02 10:25           ` Glauber Costa
2012-11-14 10:42     ` Gleb Natapov
2012-11-14 22:42       ` Marcelo Tosatti
2012-10-31 22:47   ` [patch 11/16] KVM: x86: pass host_tsc to read_l1_tsc Marcelo Tosatti
2012-10-31 22:47   ` [patch 12/16] time: export time information for KVM pvclock Marcelo Tosatti
2012-10-31 22:47   ` [patch 13/16] KVM: x86: implement PVCLOCK_TSC_STABLE_BIT pvclock flag Marcelo Tosatti
2012-10-31 22:47   ` [patch 14/16] KVM: x86: notifier for clocksource changes Marcelo Tosatti
2012-10-31 22:47   ` [patch 15/16] KVM: x86: add kvm_arch_vcpu_postcreate callback, move TSC initialization Marcelo Tosatti
2012-10-31 22:47   ` [patch 16/16] KVM: x86: require matched TSC offsets for master clock Marcelo Tosatti

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=50928502.6000309@parallels.com \
    --to=glommer@parallels.com \
    --cc=avi@redhat.com \
    --cc=gleb@redhat.com \
    --cc=jeremy@goop.org \
    --cc=johnstul@us.ibm.com \
    --cc=kvm@vger.kernel.org \
    --cc=mtosatti@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=zamsden@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).