xen-devel.lists.xenproject.org archive mirror
 help / color / mirror / Atom feed
From: Corneliu ZUZU <czuzu@bitdefender.com>
To: xen-devel@lists.xen.org
Cc: Kevin Tian <kevin.tian@intel.com>,
	Tamas K Lengyel <tamas@tklengyel.com>,
	Jan Beulich <jbeulich@suse.com>,
	Razvan Cojocaru <rcojocaru@bitdefender.com>,
	Andrew Cooper <andrew.cooper3@citrix.com>,
	Jun Nakajima <jun.nakajima@intel.com>
Subject: Re: [PATCH 05/16] x86/monitor: relocate code more appropriately
Date: Mon, 11 Jul 2016 09:19:28 +0300	[thread overview]
Message-ID: <a993806d-fd0b-5605-ad27-11bf5e16f7fb@bitdefender.com> (raw)
In-Reply-To: <1468037707-6663-1-git-send-email-czuzu@bitdefender.com>

This was 2/8 in the old v3-series "x86/vm-event: Adjustments & fixes".
Transferring code-review from "Tian, Kevin <kevin.tian@intel.com>" (July 
11, 2016) and responding...


On 7/9/2016 7:15 AM, Corneliu ZUZU wrote:
> For readability:
>
> * Add function monitor_ctrlreg_write_data() (in x86/monitor.c) and separate
> handling of monitor_write_data there (previously done directly in
> hvm_do_resume()).
>
> * Separate enabling/disabling of CPU_BASED_CR3_LOAD_EXITING for CR3 monitor
> vm-events from CR0 node @ vmx_update_guest_cr(v, 0) into newly added function
> vmx_vm_event_update_cr3_traps(d); also, for a better interface, create generic
> functions monitor_ctrlreg_adjust_traps() and monitor_ctrlreg_disable_traps() (in
> x86/monitor.c) which deal with setting/unsetting any traps for cr-write monitor
> vm-events.
>
> Signed-off-by: Corneliu ZUZU <czuzu@bitdefender.com>
> ---
>   xen/arch/x86/hvm/hvm.c            | 46 ++++++++++----------------
>   xen/arch/x86/hvm/vmx/vmx.c        | 59 ++++++++++++++++++++++++++++++---
>   xen/arch/x86/monitor.c            | 68 +++++++++++++++++++++++++++++++++++----
>   xen/include/asm-x86/hvm/vmx/vmx.h |  1 +
>   xen/include/asm-x86/monitor.h     |  2 ++
>   5 files changed, 135 insertions(+), 41 deletions(-)
>
> diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
> index 7f99087..79ba185 100644
> --- a/xen/arch/x86/hvm/hvm.c
> +++ b/xen/arch/x86/hvm/hvm.c
> @@ -473,8 +473,6 @@ void hvm_do_resume(struct vcpu *v)
>   
>       if ( unlikely(v->arch.vm_event) )
>       {
> -        struct monitor_write_data *w = &v->arch.vm_event->write_data;
> -
>           if ( unlikely(v->arch.vm_event->emulate_flags) )
>           {
>               enum emul_kind kind = EMUL_KIND_NORMAL;
> @@ -492,29 +490,7 @@ void hvm_do_resume(struct vcpu *v)
>               v->arch.vm_event->emulate_flags = 0;
>           }
>   
> -        if ( w->do_write.msr )
> -        {
> -            hvm_msr_write_intercept(w->msr, w->value, 0);
> -            w->do_write.msr = 0;
> -        }
> -
> -        if ( w->do_write.cr0 )
> -        {
> -            hvm_set_cr0(w->cr0, 0);
> -            w->do_write.cr0 = 0;
> -        }
> -
> -        if ( w->do_write.cr4 )
> -        {
> -            hvm_set_cr4(w->cr4, 0);
> -            w->do_write.cr4 = 0;
> -        }
> -
> -        if ( w->do_write.cr3 )
> -        {
> -            hvm_set_cr3(w->cr3, 0);
> -            w->do_write.cr3 = 0;
> -        }
> +        monitor_ctrlreg_write_data(v);
>       }
>   
>       /* Inject pending hw/sw trap */
> @@ -2204,7 +2180,10 @@ int hvm_set_cr0(unsigned long value, bool_t may_defer)
>   
>           if ( hvm_monitor_crX(CR0, value, old_value) )
>           {
> -            /* The actual write will occur in hvm_do_resume(), if permitted. */
> +            /*
> +             * The actual write will occur in monitor_ctrlreg_write_data(), if
> +             * permitted.
> +             */
>               v->arch.vm_event->write_data.do_write.cr0 = 1;
>               v->arch.vm_event->write_data.cr0 = value;
>   
> @@ -2306,7 +2285,10 @@ int hvm_set_cr3(unsigned long value, bool_t may_defer)
>   
>           if ( hvm_monitor_crX(CR3, value, old) )
>           {
> -            /* The actual write will occur in hvm_do_resume(), if permitted. */
> +            /*
> +             * The actual write will occur in monitor_ctrlreg_write_data(), if
> +             * permitted.
> +             */
>               v->arch.vm_event->write_data.do_write.cr3 = 1;
>               v->arch.vm_event->write_data.cr3 = value;
>   
> @@ -2386,7 +2368,10 @@ int hvm_set_cr4(unsigned long value, bool_t may_defer)
>   
>           if ( hvm_monitor_crX(CR4, value, old_cr) )
>           {
> -            /* The actual write will occur in hvm_do_resume(), if permitted. */
> +            /*
> +             * The actual write will occur in monitor_ctrlreg_write_data(), if
> +             * permitted.
> +             */
>               v->arch.vm_event->write_data.do_write.cr4 = 1;
>               v->arch.vm_event->write_data.cr4 = value;
>   
> @@ -3765,7 +3750,10 @@ int hvm_msr_write_intercept(unsigned int msr, uint64_t msr_content,
>       {
>           ASSERT(v->arch.vm_event);
>   
> -        /* The actual write will occur in hvm_do_resume() (if permitted). */
> +        /*
> +         * The actual write will occur in monitor_ctrlreg_write_data(), if
> +         * permitted.
> +         */
>           v->arch.vm_event->write_data.do_write.msr = 1;
>           v->arch.vm_event->write_data.msr = msr;
>           v->arch.vm_event->write_data.value = msr_content;
> diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
> index 0798245..7a31582 100644
> --- a/xen/arch/x86/hvm/vmx/vmx.c
> +++ b/xen/arch/x86/hvm/vmx/vmx.c
> @@ -1441,11 +1441,6 @@ static void vmx_update_guest_cr(struct vcpu *v, unsigned int cr)
>               if ( !hvm_paging_enabled(v) && !vmx_unrestricted_guest(v) )
>                   v->arch.hvm_vmx.exec_control |= cr3_ctls;
>   
> -            /* Trap CR3 updates if CR3 memory events are enabled. */
> -            if ( v->domain->arch.monitor.write_ctrlreg_enabled &
> -                 monitor_ctrlreg_bitmask(VM_EVENT_X86_CR3) )
> -                v->arch.hvm_vmx.exec_control |= CPU_BASED_CR3_LOAD_EXITING;
> -
>               if ( old_ctls != v->arch.hvm_vmx.exec_control )
>                   vmx_update_cpu_exec_control(v);
>           }
> @@ -3898,6 +3893,60 @@ void vmx_vmenter_helper(const struct cpu_user_regs *regs)
>   }
>   
>   /*
> + * Adjusts domain CR3 load-exiting execution control for CR3-write monitor
> + * vm-event.
> + */
> +void vmx_vm_event_update_cr3_traps(const struct domain *d)
> +{
> +    struct vcpu *v;
> +    struct arch_vmx_struct *avmx;
     [Kevin wrote]:

	better move to inner later loop.

---


Ack.


> +    unsigned int cr3_bitmask;
> +    bool_t cr3_vmevent, cr3_ldexit;
> +
> +    /* Domain must be paused. */
> +    ASSERT(atomic_read(&d->pause_count));

     [Kevin wrote]:

	The comment doesn't add more info than the code itself, unless you want to
	explain the reason why domain must be paused here.
---


Ack, will remove the comment (or explain why the domain is paused: probably because we're writing exec_control of all its vcpus).


> +
> +    /* Non-hap domains trap CR3 writes unconditionally. */
> +    if ( !paging_mode_hap(d) )
> +    {
> +#ifndef NDEBUG
> +        for_each_vcpu ( d, v )
> +            ASSERT(v->arch.hvm_vmx.exec_control & CPU_BASED_CR3_LOAD_EXITING);
> +#endif
> +        return;
> +    }
> +
> +    cr3_bitmask = monitor_ctrlreg_bitmask(VM_EVENT_X86_CR3);
> +    cr3_vmevent = !!(d->arch.monitor.write_ctrlreg_enabled & cr3_bitmask);
> +
> +    for_each_vcpu ( d, v )
> +    {
> +        avmx = &v->arch.hvm_vmx;
> +        cr3_ldexit = !!(avmx->exec_control & CPU_BASED_CR3_LOAD_EXITING);
> +
> +        if ( cr3_vmevent == cr3_ldexit )
> +            continue;
> +
> +        /*
> +         * If domain paging is disabled (CR0.PG=0) and
> +         * the domain is not in real-mode, then CR3 load-exiting
> +         * must remain enabled (see vmx_update_guest_cr(v, cr) when cr = 0).
> +         */

     [Kevin wrote]:

	Please give the real informative message here instead of asking people to
	look at other place
---


I only now realized I might have mistakenly assumed something when writing up this function.
I'll amend that in a v2 and consider making this comment clearer (will ask for feedback from you again here before sending v2).


> +        if ( cr3_ldexit && !hvm_paging_enabled(v) && !vmx_unrestricted_guest(v) )
> +            continue;
     [Kevin wrote]:

	if !cr3_ldexit is not expected to occur in such scenario, is it more strict as below?

	if ( !hvm_paging_enabled(v) && !vmx_unrestricted_guest(v) )
	{
		ASSERT(cr3_ldexit);
		continue;
	}
---

That actually seems to be correct, I thought it was possible to have CR3 
load-exiting disabled with the above condition true (by looking @ old 
the code). I must attentively re-review this entire function so that it 
preserves the old behavior.

> +
> +        if ( cr3_vmevent )
> +            avmx->exec_control |= CPU_BASED_CR3_LOAD_EXITING;
> +        else
> +            avmx->exec_control &= ~CPU_BASED_CR3_LOAD_EXITING;
> +
> +        vmx_vmcs_enter(v);
> +        vmx_update_cpu_exec_control(v);
> +        vmx_vmcs_exit(v);
> +    }
> +}
> +
> +/*
>    * Local variables:
>    * mode: C
>    * c-file-style: "BSD"
> diff --git a/xen/arch/x86/monitor.c b/xen/arch/x86/monitor.c
> index 06d21b1..29188e5 100644
> --- a/xen/arch/x86/monitor.c
> +++ b/xen/arch/x86/monitor.c
> @@ -19,10 +19,39 @@
>    * License along with this program; If not, see <http://www.gnu.org/licenses/>.
>    */
>   
> +#include <asm/hvm/vmx/vmx.h>
>   #include <asm/monitor.h>
>   #include <asm/vm_event.h>
>   #include <public/vm_event.h>
>   
> +static inline
> +void monitor_ctrlreg_adjust_traps(struct domain *d, unsigned int index)
> +{
> +    /* For now, VMX only. */
> +    ASSERT(cpu_has_vmx);
> +
> +    /* Other CRs than CR3 are always trapped. */
> +    if ( VM_EVENT_X86_CR3 == index )
> +        vmx_vm_event_update_cr3_traps(d);
     [Kevin wrote]:

	Please add above into a hvm function instead of directly calling
	vmx in common file. (other arch can have it unimplemented).
	Then you don't need change this common code again later when
	other archs are added

---


This is not common code, it's in arch/x86/monitor.c (?) and currently, 
as the above ASSERT indicates, only VMX is supported. If in the future 
support for SVM for example will be added, then the hvm move you suggest 
must be done (Jan also suggested this).
Or, I only now realize, if you guys prefer doing this now I could add a 
vm_event_update_cr3_traps field in hvm_function_table, but BUG() in the 
SVM one..

> +}
> +
> +static inline void monitor_ctrlreg_disable_traps(struct domain *d)
> +{
> +    unsigned int old = d->arch.monitor.write_ctrlreg_enabled;
> +
> +    d->arch.monitor.write_ctrlreg_enabled = 0;
> +
> +    if ( old )
> +    {
> +        /* For now, VMX only. */
> +        ASSERT(cpu_has_vmx);
> +
> +        /* Was CR3 load-exiting enabled due to monitoring? */
> +        if ( old & monitor_ctrlreg_bitmask(VM_EVENT_X86_CR3) )
> +            vmx_vm_event_update_cr3_traps(d);
> +    }
> +}
> +
>   int monitor_init_domain(struct domain *d)
>   {
>       if ( !d->arch.monitor.msr_bitmap )
> @@ -38,6 +67,8 @@ void monitor_cleanup_domain(struct domain *d)
>   {
>       xfree(d->arch.monitor.msr_bitmap);
>   
> +    monitor_ctrlreg_disable_traps(d);
> +
>       memset(&d->arch.monitor, 0, sizeof(d->arch.monitor));
>       memset(&d->monitor, 0, sizeof(d->monitor));
>   }
> @@ -79,6 +110,35 @@ void monitor_ctrlreg_write_resume(struct vcpu *v, vm_event_response_t *rsp)
>       }
>   }
>   
> +void monitor_ctrlreg_write_data(struct vcpu *v)
> +{
> +    struct monitor_write_data *w = &v->arch.vm_event->write_data;
> +
> +    if ( w->do_write.msr )
> +    {
> +        hvm_msr_write_intercept(w->msr, w->value, 0);
> +        w->do_write.msr = 0;
> +    }
> +
> +    if ( w->do_write.cr0 )
> +    {
> +        hvm_set_cr0(w->cr0, 0);
> +        w->do_write.cr0 = 0;
> +    }
> +
> +    if ( w->do_write.cr4 )
> +    {
> +        hvm_set_cr4(w->cr4, 0);
> +        w->do_write.cr4 = 0;
> +    }
> +
> +    if ( w->do_write.cr3 )
> +    {
> +        hvm_set_cr3(w->cr3, 0);
> +        w->do_write.cr3 = 0;
> +    }
> +}
> +
>   static unsigned long *monitor_bitmap_for_msr(const struct domain *d, u32 *msr)
>   {
>       ASSERT(d->arch.monitor.msr_bitmap && msr);
> @@ -197,13 +257,7 @@ int arch_monitor_domctl_event(struct domain *d,
>           else
>               ad->monitor.write_ctrlreg_enabled &= ~ctrlreg_bitmask;
>   
> -        if ( VM_EVENT_X86_CR3 == mop->u.mov_to_cr.index )
> -        {
> -            struct vcpu *v;
> -            /* Latches new CR3 mask through CR0 code. */
> -            for_each_vcpu ( d, v )
> -                hvm_update_guest_cr(v, 0);
> -        }
> +        monitor_ctrlreg_adjust_traps(d, mop->u.mov_to_cr.index);
>   
>           domain_unpause(d);
>   
> diff --git a/xen/include/asm-x86/hvm/vmx/vmx.h b/xen/include/asm-x86/hvm/vmx/vmx.h
> index 359b2a9..301df56 100644
> --- a/xen/include/asm-x86/hvm/vmx/vmx.h
> +++ b/xen/include/asm-x86/hvm/vmx/vmx.h
> @@ -101,6 +101,7 @@ void vmx_update_debug_state(struct vcpu *v);
>   void vmx_update_exception_bitmap(struct vcpu *v);
>   void vmx_update_cpu_exec_control(struct vcpu *v);
>   void vmx_update_secondary_exec_control(struct vcpu *v);
> +void vmx_vm_event_update_cr3_traps(const struct domain *d);
>   
>   #define POSTED_INTR_ON  0
>   #define POSTED_INTR_SN  1
> diff --git a/xen/include/asm-x86/monitor.h b/xen/include/asm-x86/monitor.h
> index adca378..3ae24dd 100644
> --- a/xen/include/asm-x86/monitor.h
> +++ b/xen/include/asm-x86/monitor.h
> @@ -93,6 +93,8 @@ void monitor_cleanup_domain(struct domain *d);
>   
>   void monitor_ctrlreg_write_resume(struct vcpu *v, vm_event_response_t *rsp);
>   
> +void monitor_ctrlreg_write_data(struct vcpu *v);
> +
>   bool_t monitored_msr(const struct domain *d, u32 msr);
>   
>   #endif /* __ASM_X86_MONITOR_H__ */

Thanks,
Zuzu C.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel

  reply	other threads:[~2016-07-11  6:19 UTC|newest]

Thread overview: 54+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-07-09  4:11 [PATCH 00/16] x86/vm-event: numerous adjustments & fixes Corneliu ZUZU
2016-07-09  4:12 ` [PATCH 01/16] x86/vmx_update_guest_cr: minor optimization Corneliu ZUZU
2016-07-11  6:24   ` Corneliu ZUZU
2016-07-09  4:12 ` [PATCH 02/16] x86: fix: make atomic_read() param const Corneliu ZUZU
2016-07-11 15:18   ` Andrew Cooper
2016-07-12  5:11     ` Corneliu ZUZU
2016-07-12  9:42       ` Andrew Cooper
2016-07-12 10:11         ` Corneliu ZUZU
2016-07-12 10:22           ` Andrew Cooper
2016-07-12 10:35             ` Corneliu ZUZU
2016-07-12 10:38             ` Corneliu ZUZU
2016-07-12 12:49               ` Andrew Cooper
2016-07-12 13:45                 ` Corneliu ZUZU
2016-07-09  4:13 ` [PATCH 03/16] x86/monitor: mechanical renames Corneliu ZUZU
2016-07-09 18:10   ` Tamas K Lengyel
2016-07-09 18:46     ` Corneliu ZUZU
2016-07-11 16:43       ` Tamas K Lengyel
2016-07-12  6:10         ` Corneliu ZUZU
2016-07-15  7:18           ` Corneliu ZUZU
2016-07-18 18:07             ` Andrew Cooper
2016-07-19  9:36               ` Corneliu ZUZU
2016-07-09  4:14 ` [PATCH 04/16] x86/monitor: relocate vm_event_register_write_resume() function to monitor code Corneliu ZUZU
2016-07-09 18:14   ` Tamas K Lengyel
2016-07-09 18:47     ` Corneliu ZUZU
2016-07-09  4:15 ` [PATCH 05/16] x86/monitor: relocate code more appropriately Corneliu ZUZU
2016-07-11  6:19   ` Corneliu ZUZU [this message]
2016-07-12  7:45     ` Tian, Kevin
2016-07-12  8:07       ` Corneliu ZUZU
2016-07-15 11:41         ` Corneliu ZUZU
2016-07-09  4:15 ` [PATCH 06/16] x86/monitor: fix: set msr_bitmap to NULL after xfree Corneliu ZUZU
2016-07-09  4:16 ` [PATCH 07/16] x86/vm-event: fix: call cleanup when init fails, to free partial allocs Corneliu ZUZU
2016-07-09  4:17 ` [PATCH 08/16] x86/vm-event: call monitor init & cleanup funcs from respective vm_event funcs Corneliu ZUZU
2016-07-09  4:18 ` [PATCH 09/16] arm/monitor: move d->monitor cleanup to monitor_cleanup_domain() Corneliu ZUZU
2016-07-09  4:19 ` [PATCH 10/16] x86/vm-event: centralize vcpu-destroy cleanup in vm-events code Corneliu ZUZU
2016-07-09  4:20 ` [PATCH 11/16] x86/monitor: fix: treat -monitor- properly, as a subsys of the vm-event subsys Corneliu ZUZU
2016-07-09 17:34   ` Tamas K Lengyel
2016-07-09 17:46     ` Corneliu ZUZU
2016-07-11 16:38       ` Tamas K Lengyel
2016-07-11 20:20         ` Corneliu ZUZU
2016-07-11 21:27           ` Tamas K Lengyel
2016-07-11 21:47             ` Corneliu ZUZU
2016-07-09  4:20 ` [PATCH 12/16] x86/vm-event: fix: move cleanup of mem_access_emulate_each_rep to monitor stub Corneliu ZUZU
2016-07-09  4:21 ` [PATCH 13/16] x86/monitor: introduce writes_pending field in monitor_write_data Corneliu ZUZU
2016-07-09  4:22 ` [PATCH 14/16] x86/monitor: clarify separation between monitor subsys and vm-event as a whole Corneliu ZUZU
2016-07-09 18:26   ` Tamas K Lengyel
2016-07-09 18:57     ` Corneliu ZUZU
2016-07-13  4:26       ` Corneliu ZUZU
2016-07-13 18:56         ` Tamas K Lengyel
2016-07-09  4:23 ` [PATCH 15/16] x86/monitor: fix: don't compromise a monitor_write_data with pending CR writes Corneliu ZUZU
2016-07-09  4:23 ` [PATCH 16/16] x86/monitor: fix: xc_monitor _write_ctrlreg w/o previous _enable must fail Corneliu ZUZU
2016-07-09  4:34   ` Corneliu ZUZU
2016-07-11  2:54 ` [PATCH 00/16] x86/vm-event: numerous adjustments & fixes Tian, Kevin
2016-07-11  5:32   ` Corneliu ZUZU
2016-07-12  7:42     ` Tian, Kevin

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=a993806d-fd0b-5605-ad27-11bf5e16f7fb@bitdefender.com \
    --to=czuzu@bitdefender.com \
    --cc=andrew.cooper3@citrix.com \
    --cc=jbeulich@suse.com \
    --cc=jun.nakajima@intel.com \
    --cc=kevin.tian@intel.com \
    --cc=rcojocaru@bitdefender.com \
    --cc=tamas@tklengyel.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).