xen-devel.lists.xenproject.org archive mirror
 help / color / mirror / Atom feed
From: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
To: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: jacob.shin@amd.com, haitao.shan@intel.com,
	jun.nakajima@intel.com, dietmar.hahn@ts.fujitsu.com,
	xen-devel@lists.xen.org
Subject: Re: [PATCH 3/8] x86/AMD: Read VPMU MSRs from context when it is not loaded into HW
Date: Thu, 11 Apr 2013 13:26:19 -0500	[thread overview]
Message-ID: <5167004B.2000905@amd.com> (raw)
In-Reply-To: <1365528379-2516-4-git-send-email-boris.ostrovsky@oracle.com>

Boris,

I tried booting the guest HVM after the patch, I still see PERF only 
working in Software mode only.  I'll look more into this.

Suravee
On 4/9/2013 12:26 PM, Boris Ostrovsky wrote:
> Mark context as LOADED on any MSR write (and on vpmu_restore()). This
> will allow us to know whether to read an MSR from HW or from context:
> guest may write into an MSR without enabling it (thus not marking the
> context as RUNNING) and then be migrated. Reading this MSR again will
> not match the pervious write since registers will not be loaded into HW.
>
> In addition, we should be saving the context when it is LOADED, not
> RUNNING --- otherwise we need to save it any time it becomes non-RUNNING,
> which may be a frequent occurrence.
>
> Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
> ---
>   xen/arch/x86/hvm/svm/vpmu.c | 48 +++++++++++++++++++++++++++++++++++++++++++--
>   1 file changed, 46 insertions(+), 2 deletions(-)
>
> diff --git a/xen/arch/x86/hvm/svm/vpmu.c b/xen/arch/x86/hvm/svm/vpmu.c
> index f194975..3115923 100644
> --- a/xen/arch/x86/hvm/svm/vpmu.c
> +++ b/xen/arch/x86/hvm/svm/vpmu.c
> @@ -225,6 +225,8 @@ static void amd_vpmu_restore(struct vcpu *v)
>   
>       context_restore(v);
>       apic_write(APIC_LVTPC, ctxt->hw_lapic_lvtpc);
> +
> +    vpmu_set(vpmu, VPMU_CONTEXT_LOADED);
>   }
>   
>   static inline void context_save(struct vcpu *v)
> @@ -246,7 +248,7 @@ static void amd_vpmu_save(struct vcpu *v)
>       struct amd_vpmu_context *ctx = vpmu->context;
>   
>       if ( !(vpmu_is_set(vpmu, VPMU_CONTEXT_ALLOCATED) &&
> -           vpmu_is_set(vpmu, VPMU_RUNNING)) )
> +           vpmu_is_set(vpmu, VPMU_CONTEXT_LOADED)) )
>           return;
>   
>       context_save(v);
> @@ -256,6 +258,35 @@ static void amd_vpmu_save(struct vcpu *v)
>   
>       ctx->hw_lapic_lvtpc = apic_read(APIC_LVTPC);
>       apic_write(APIC_LVTPC,  ctx->hw_lapic_lvtpc | APIC_LVT_MASKED);
> +    vpmu_reset(vpmu, VPMU_CONTEXT_LOADED);
> +}
> +
> +static void context_read(unsigned int msr, u64 *msr_content)
> +{
> +    unsigned int i;
> +    struct vcpu *v = current;
> +    struct vpmu_struct *vpmu = vcpu_vpmu(v);
> +    struct amd_vpmu_context *ctxt = vpmu->context;
> +
> +    if ( k7_counters_mirrored &&
> +        ((msr >= MSR_K7_EVNTSEL0) && (msr <= MSR_K7_PERFCTR3)) )
> +    {
> +        msr = get_fam15h_addr(msr);
> +    }
> +
> +    for ( i = 0; i < num_counters; i++ )
> +    {
> +       if ( msr == ctrls[i] )
> +       {
> +           *msr_content = ctxt->ctrls[i];
> +           return;
> +       }
> +       else if (msr == counters[i] )
> +       {
> +           *msr_content = ctxt->counters[i];
> +           return;
> +       }
> +    }
>   }
>   
>   static void context_update(unsigned int msr, u64 msr_content)
> @@ -318,6 +349,12 @@ static int amd_vpmu_do_wrmsr(unsigned int msr, uint64_t msr_content)
>           release_pmu_ownship(PMU_OWNER_HVM);
>       }
>   
> +    if ( !vpmu_is_set(vpmu, VPMU_CONTEXT_LOADED) )
> +    {
> +        context_restore(v);
> +        vpmu_set(vpmu, VPMU_CONTEXT_LOADED);
> +    }
> +
>       /* Update vpmu context immediately */
>       context_update(msr, msr_content);
>   
> @@ -328,7 +365,14 @@ static int amd_vpmu_do_wrmsr(unsigned int msr, uint64_t msr_content)
>   
>   static int amd_vpmu_do_rdmsr(unsigned int msr, uint64_t *msr_content)
>   {
> -    rdmsrl(msr, *msr_content);
> +    struct vcpu *v = current;
> +    struct vpmu_struct *vpmu = vcpu_vpmu(v);
> +
> +    if ( !vpmu_is_set(vpmu, VPMU_CONTEXT_LOADED) )
> +        context_read(msr, msr_content);
> +    else
> +        rdmsrl(msr, *msr_content);
> +
>       return 1;
>   }
>   

  reply	other threads:[~2013-04-11 18:26 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-04-09 17:26 [PATCH 0/8] Various VPMU patches Boris Ostrovsky
2013-04-09 17:26 ` [PATCH 1/8] x86/AMD: Allow more fine-grained control of VMCB MSR Permission Map Boris Ostrovsky
2013-04-09 17:26 ` [PATCH 2/8] x86/AMD: Do not intercept access to performance counters MSRs Boris Ostrovsky
2013-04-10 13:25   ` Jan Beulich
2013-04-09 17:26 ` [PATCH 3/8] x86/AMD: Read VPMU MSRs from context when it is not loaded into HW Boris Ostrovsky
2013-04-11 18:26   ` Suravee Suthikulpanit [this message]
2013-04-11 18:34     ` Boris Ostrovsky
2013-04-11 19:30       ` Suravee Suthikulpanit
2013-04-16 15:41       ` Konrad Rzeszutek Wilk
2013-04-16 17:12         ` Jacob Shin
2013-04-16 18:36           ` Konrad Rzeszutek Wilk
2013-06-19 22:56             ` Suravee Suthikulanit
2013-06-19 23:32               ` Boris Ostrovsky
2013-06-19 23:53                 ` Boris Ostrovsky
2013-04-09 17:26 ` [PATCH 4/8] x86/AMD: Stop counters on VPMU save Boris Ostrovsky
2013-04-09 17:26 ` [PATCH 5/8] x86/VPMU: Add Haswell support Boris Ostrovsky
2013-04-09 17:26 ` [PATCH 6/8] x86/VPMU: Factor out VPMU common code Boris Ostrovsky
2013-04-10 16:03   ` Nakajima, Jun
2013-04-09 17:26 ` [PATCH 7/8] x86/VPMU: Save/restore VPMU only when necessary Boris Ostrovsky
2013-04-10  8:57   ` Dietmar Hahn
2013-04-10 12:53     ` Boris Ostrovsky
2013-04-09 17:26 ` [PATCH 8/8] x86/AMD: Clean up context_update() in AMD VPMU code Boris Ostrovsky
2013-04-11 19:48   ` Suravee Suthikulpanit
2013-04-11 20:42     ` Boris Ostrovsky
2013-04-10  8:57 ` [PATCH 0/8] Various VPMU patches Dietmar Hahn
2013-04-10 18:49 ` Suravee Suthikulanit
2013-04-10 19:10   ` Boris Ostrovsky

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=5167004B.2000905@amd.com \
    --to=suravee.suthikulpanit@amd.com \
    --cc=boris.ostrovsky@oracle.com \
    --cc=dietmar.hahn@ts.fujitsu.com \
    --cc=haitao.shan@intel.com \
    --cc=jacob.shin@amd.com \
    --cc=jun.nakajima@intel.com \
    --cc=xen-devel@lists.xen.org \
    /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).