From: Andrew Cooper <andrew.cooper3@citrix.com>
To: Razvan Cojocaru <rcojocaru@bitdefender.com>, xen-devel@lists.xen.org
Cc: tim@xen.org
Subject: Re: [PATCH RFC 2/9] xen: Optimize introspection access to guest state
Date: Wed, 2 Jul 2014 16:31:00 +0100 [thread overview]
Message-ID: <53B425B4.3010808@citrix.com> (raw)
In-Reply-To: <1404308041-15461-2-git-send-email-rcojocaru@bitdefender.com>
On 02/07/14 14:33, Razvan Cojocaru wrote:
> Speed optimization for introspection purposes: a handful of registers
> are sent along with each mem_event. This requires enlargement of the
> mem_event_request / mem_event_response stuctures, and additional code
> to fill in relevant values.
>
> Signed-off-by: Razvan Cojocaru <rcojocaru@bitdefender.com>
The public API already has struct hvm_hw_cpu in
xen/include/public/arch-x86/hvm/save.h
It might be better to reuse that rather than defining a new structure to
contain a subset of the information.
> ---
> xen/arch/x86/hvm/hvm.c | 33 +++++++++++++++++
> xen/arch/x86/hvm/vmx/vmx.c | 1 +
> xen/arch/x86/mm/p2m.c | 61 ++++++++++++++++++++++++++++++++
> xen/include/public/arch-x86/hvm/save.h | 4 +++
> xen/include/public/mem_event.h | 36 +++++++++++++++++++
> 5 files changed, 135 insertions(+)
>
> diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
> index 17ff011..f65a5f5 100644
> --- a/xen/arch/x86/hvm/hvm.c
> +++ b/xen/arch/x86/hvm/hvm.c
> @@ -6016,6 +6016,38 @@ int hvm_debug_op(struct vcpu *v, int32_t op)
> return rc;
> }
>
> +static inline void hvm_mem_event_fill_regs(mem_event_request_t *req)
> +{
> + struct cpu_user_regs *regs = guest_cpu_user_regs();
> + struct vcpu *v = current;
> +
> + req->regs.rax = regs->eax;
> + req->regs.rcx = regs->ecx;
> + req->regs.rdx = regs->edx;
> + req->regs.rbx = regs->ebx;
> + req->regs.rsp = regs->esp;
> + req->regs.rbp = regs->ebp;
> + req->regs.rsi = regs->esi;
> + req->regs.rdi = regs->edi;
> +
> + req->regs.r8 = regs->r8;
> + req->regs.r9 = regs->r9;
> + req->regs.r10 = regs->r10;
> + req->regs.r11 = regs->r11;
> + req->regs.r12 = regs->r12;
> + req->regs.r13 = regs->r13;
> + req->regs.r14 = regs->r14;
> + req->regs.r15 = regs->r15;
> +
> + req->regs.rflags = regs->eflags;
> + req->regs.rip = regs->eip;
> +
> + req->regs.msr_efer = v->arch.hvm_vcpu.guest_efer;
> + req->regs.cr0 = v->arch.hvm_vcpu.guest_cr[0];
> + req->regs.cr3 = v->arch.hvm_vcpu.guest_cr[3];
> + req->regs.cr4 = v->arch.hvm_vcpu.guest_cr[4];
> +}
> +
> static int hvm_memory_event_traps(long p, uint32_t reason,
> unsigned long value, unsigned long old,
> bool_t gla_valid, unsigned long gla)
> @@ -6060,6 +6092,7 @@ static int hvm_memory_event_traps(long p, uint32_t reason,
> req.gla = old;
> }
>
> + hvm_mem_event_fill_regs(&req);
> mem_event_put_request(d, &d->mem_event->access, &req);
>
> return 1;
> diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
> index 2caa04a..fed21b6 100644
> --- a/xen/arch/x86/hvm/vmx/vmx.c
> +++ b/xen/arch/x86/hvm/vmx/vmx.c
> @@ -425,6 +425,7 @@ static void vmx_vmcs_save(struct vcpu *v, struct hvm_hw_cpu *c)
> c->cr4 = v->arch.hvm_vcpu.guest_cr[4];
>
> c->msr_efer = v->arch.hvm_vcpu.guest_efer;
> + c->guest_x86_mode = vmx_guest_x86_mode(v);
guest_x86_mode is a linear function of cr0, eflags and efer. It can be
calculated by userspace doesn't need to transmitted individually.
>
> __vmread(GUEST_SYSENTER_CS, &c->sysenter_cs);
> __vmread(GUEST_SYSENTER_ESP, &c->sysenter_esp);
> diff --git a/xen/arch/x86/mm/p2m.c b/xen/arch/x86/mm/p2m.c
> index 642ec28..93252d9 100644
> --- a/xen/arch/x86/mm/p2m.c
> +++ b/xen/arch/x86/mm/p2m.c
> @@ -1314,6 +1314,64 @@ void p2m_mem_paging_resume(struct domain *d)
> }
> }
>
> +static inline void p2m_mem_event_fill_regs(mem_event_request_t *req)
> +{
> + struct cpu_user_regs *regs = guest_cpu_user_regs();
> + struct segment_register seg;
> + struct hvm_hw_cpu ctxt;
> + struct vcpu *v = current;
> +
> + memset(&ctxt, 0, sizeof(struct hvm_hw_cpu));
> +
> + /* Architecture-specific vmcs/vmcb bits */
> + hvm_funcs.save_cpu_ctxt(v, &ctxt);
> +
> + req->regs.rax = regs->eax;
> + req->regs.rcx = regs->ecx;
> + req->regs.rdx = regs->edx;
> + req->regs.rbx = regs->ebx;
> + req->regs.rsp = regs->esp;
> + req->regs.rbp = regs->ebp;
> + req->regs.rsi = regs->esi;
> + req->regs.rdi = regs->edi;
> +
> +#ifdef __x86_64__
There is no need to code for __i386__ inside xen/arch/x86
~Andrew
next prev parent reply other threads:[~2014-07-02 15:31 UTC|newest]
Thread overview: 64+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-07-02 13:33 [PATCH RFC 1/9] xen: Emulate with no writes; compute current instruction length Razvan Cojocaru
2014-07-02 13:33 ` [PATCH RFC 2/9] xen: Optimize introspection access to guest state Razvan Cojocaru
2014-07-02 15:31 ` Andrew Cooper [this message]
2014-07-07 14:50 ` Razvan Cojocaru
2014-07-10 8:05 ` Razvan Cojocaru
2014-07-10 8:17 ` Andrew Cooper
2014-07-10 8:23 ` Razvan Cojocaru
2014-07-10 11:57 ` Razvan Cojocaru
2014-07-10 12:16 ` Razvan Cojocaru
2014-07-10 13:01 ` Andrew Cooper
2014-07-02 15:37 ` Jan Beulich
2014-07-03 8:12 ` Razvan Cojocaru
2014-07-03 8:54 ` Jan Beulich
2014-07-02 13:33 ` [PATCH RFC 3/9] xen: Force-enable relevant MSR events; optimize the number of sent MSR events Razvan Cojocaru
2014-07-02 15:35 ` Andrew Cooper
2014-07-02 15:43 ` Jan Beulich
2014-07-09 8:02 ` Razvan Cojocaru
2014-07-23 7:56 ` Jan Beulich
2014-07-23 8:03 ` Razvan Cojocaru
2014-07-02 13:33 ` [PATCH RFC 4/9] xenctrl: Make the headers C++ friendly Razvan Cojocaru
2014-07-02 15:37 ` Andrew Cooper
2014-07-02 13:33 ` [PATCH RFC 5/9] xen: Support for VMCALL mem_events Razvan Cojocaru
2014-07-02 15:47 ` Jan Beulich
2014-07-02 15:54 ` Razvan Cojocaru
2014-07-02 16:11 ` Jan Beulich
2014-07-02 16:23 ` Razvan Cojocaru
2014-07-03 6:28 ` Jan Beulich
2014-07-03 7:29 ` Razvan Cojocaru
2014-07-02 15:54 ` Andrew Cooper
2014-07-02 15:59 ` Razvan Cojocaru
2014-07-02 13:33 ` [PATCH RFC 6/9] xen, libxc: Request page fault injection via libxc Razvan Cojocaru
2014-07-02 15:51 ` Jan Beulich
2014-07-02 16:00 ` Andrew Cooper
2014-07-02 16:58 ` Mihai Donțu
2014-07-02 17:07 ` Andrew Cooper
2014-07-03 8:23 ` Mihai Donțu
2014-07-03 9:32 ` Andrew Cooper
2014-07-03 9:40 ` Razvan Cojocaru
2014-07-02 16:06 ` Razvan Cojocaru
2014-07-02 16:13 ` Jan Beulich
2014-07-02 13:33 ` [PATCH RFC 7/9] xen: Handle resumed instruction based on previous mem_event reply Razvan Cojocaru
2014-07-02 15:56 ` Jan Beulich
2014-07-03 8:55 ` Razvan Cojocaru
2014-07-03 9:02 ` Jan Beulich
2014-07-03 9:12 ` Razvan Cojocaru
2014-07-03 9:18 ` Andrew Cooper
2014-07-03 9:22 ` Jan Beulich
2014-07-03 9:34 ` Razvan Cojocaru
2014-07-03 10:14 ` Jan Beulich
2014-07-02 13:34 ` [PATCH RFC 8/9] xen: Generic instruction re-execution mechanism for execute faults Razvan Cojocaru
2014-07-02 16:04 ` Andrew Cooper
2014-07-02 13:34 ` [PATCH RFC 9/9] mm: mark pages that have their permissions controlled by a domain Razvan Cojocaru
2014-07-03 10:19 ` Jan Beulich
2014-07-03 11:27 ` Razvan Cojocaru
2014-07-03 12:15 ` Jan Beulich
2014-07-02 15:20 ` [PATCH RFC 1/9] xen: Emulate with no writes; compute current instruction length Andrew Cooper
2014-07-03 7:42 ` Razvan Cojocaru
2014-07-02 15:21 ` Jan Beulich
2014-07-02 15:43 ` Razvan Cojocaru
2014-07-02 16:08 ` Jan Beulich
2014-07-02 16:18 ` Razvan Cojocaru
2014-07-03 6:24 ` Jan Beulich
2014-07-03 7:38 ` Razvan Cojocaru
2014-07-03 8:05 ` Jan Beulich
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=53B425B4.3010808@citrix.com \
--to=andrew.cooper3@citrix.com \
--cc=rcojocaru@bitdefender.com \
--cc=tim@xen.org \
--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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.