All of lore.kernel.org
 help / color / mirror / Atom feed
From: Isaila Alexandru <aisaila@bitdefender.com>
To: Paul Durrant <Paul.Durrant@citrix.com>,
	"xen-devel@lists.xen.org" <xen-devel@lists.xen.org>
Cc: Ian Jackson <Ian.Jackson@citrix.com>,
	Wei Liu <wei.liu2@citrix.com>,
	"jbeulich@suse.com" <jbeulich@suse.com>,
	Andrew Cooper <Andrew.Cooper3@citrix.com>
Subject: Re: [PATCH v12 03/11] x86/hvm: Introduce hvm_save_cpu_ctxt_one func
Date: Tue, 17 Jul 2018 15:25:32 +0300	[thread overview]
Message-ID: <1531830332.10865.4.camel@bitdefender.com> (raw)
In-Reply-To: <492015aaf7164a48bc2a2e635ef0277d@AMSPEX02CL02.citrite.net>

On Lu, 2018-07-16 at 15:29 +0000, Paul Durrant wrote:
> > 
> > -----Original Message-----
> > From: Alexandru Isaila [mailto:aisaila@bitdefender.com]
> > Sent: 16 July 2018 15:55
> > To: xen-devel@lists.xen.org
> > Cc: Ian Jackson <Ian.Jackson@citrix.com>; Wei Liu <wei.liu2@citrix.
> > com>;
> > jbeulich@suse.com; Andrew Cooper <Andrew.Cooper3@citrix.com>; Paul
> > Durrant <Paul.Durrant@citrix.com>; Alexandru Isaila
> > <aisaila@bitdefender.com>
> > Subject: [PATCH v12 03/11] x86/hvm: Introduce hvm_save_cpu_ctxt_one
> > func
> > 
> > This is used to save data from a single instance.
> > 
> > Signed-off-by: Alexandru Isaila <aisaila@bitdefender.com>
> > 
> > ---
> > Changes since V11:
> > 	- hvm_save_cpu_ctxt() now returns err from
> > 	  hvm_save_cpu_ctxt_one().
> > ---
> >  xen/arch/x86/hvm/hvm.c | 216 ++++++++++++++++++++++++++-----------
> > -
> > -----------
> >  1 file changed, 113 insertions(+), 103 deletions(-)
> > 
> > diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c
> > index dd88751..e20a25c 100644
> > --- a/xen/arch/x86/hvm/hvm.c
> > +++ b/xen/arch/x86/hvm/hvm.c
> > @@ -787,119 +787,129 @@ static int hvm_load_tsc_adjust(struct
> > domain *d,
> > hvm_domain_context_t *h)
> >  HVM_REGISTER_SAVE_RESTORE(TSC_ADJUST, hvm_save_tsc_adjust,
> >                            hvm_load_tsc_adjust, 1, HVMSR_PER_VCPU);
> > 
> > +static int hvm_save_cpu_ctxt_one(struct vcpu *v,
> > hvm_domain_context_t
> > *h)
> > +{
> > +    struct segment_register seg;
> > +    struct hvm_hw_cpu ctxt;
> > +
> > +    memset(&ctxt, 0, sizeof(ctxt));
> Why not use an = {} initializer instead of the memset here like
> elsewhere?
> 
>   Paul

I wanted to make less change as possible and I only added a initializer
where there was none. 

Alex 
> > 
> > +
> > +    /* Architecture-specific vmcs/vmcb bits */
> > +    hvm_funcs.save_cpu_ctxt(v, &ctxt);
> > +
> > +    ctxt.tsc = hvm_get_guest_tsc_fixed(v, v->domain-
> > > 
> > > arch.hvm_domain.sync_tsc);
> > +
> > +    ctxt.msr_tsc_aux = hvm_msr_tsc_aux(v);
> > +
> > +    hvm_get_segment_register(v, x86_seg_idtr, &seg);
> > +    ctxt.idtr_limit = seg.limit;
> > +    ctxt.idtr_base = seg.base;
> > +
> > +    hvm_get_segment_register(v, x86_seg_gdtr, &seg);
> > +    ctxt.gdtr_limit = seg.limit;
> > +    ctxt.gdtr_base = seg.base;
> > +
> > +    hvm_get_segment_register(v, x86_seg_cs, &seg);
> > +    ctxt.cs_sel = seg.sel;
> > +    ctxt.cs_limit = seg.limit;
> > +    ctxt.cs_base = seg.base;
> > +    ctxt.cs_arbytes = seg.attr;
> > +
> > +    hvm_get_segment_register(v, x86_seg_ds, &seg);
> > +    ctxt.ds_sel = seg.sel;
> > +    ctxt.ds_limit = seg.limit;
> > +    ctxt.ds_base = seg.base;
> > +    ctxt.ds_arbytes = seg.attr;
> > +
> > +    hvm_get_segment_register(v, x86_seg_es, &seg);
> > +    ctxt.es_sel = seg.sel;
> > +    ctxt.es_limit = seg.limit;
> > +    ctxt.es_base = seg.base;
> > +    ctxt.es_arbytes = seg.attr;
> > +
> > +    hvm_get_segment_register(v, x86_seg_ss, &seg);
> > +    ctxt.ss_sel = seg.sel;
> > +    ctxt.ss_limit = seg.limit;
> > +    ctxt.ss_base = seg.base;
> > +    ctxt.ss_arbytes = seg.attr;
> > +
> > +    hvm_get_segment_register(v, x86_seg_fs, &seg);
> > +    ctxt.fs_sel = seg.sel;
> > +    ctxt.fs_limit = seg.limit;
> > +    ctxt.fs_base = seg.base;
> > +    ctxt.fs_arbytes = seg.attr;
> > +
> > +    hvm_get_segment_register(v, x86_seg_gs, &seg);
> > +    ctxt.gs_sel = seg.sel;
> > +    ctxt.gs_limit = seg.limit;
> > +    ctxt.gs_base = seg.base;
> > +    ctxt.gs_arbytes = seg.attr;
> > +
> > +    hvm_get_segment_register(v, x86_seg_tr, &seg);
> > +    ctxt.tr_sel = seg.sel;
> > +    ctxt.tr_limit = seg.limit;
> > +    ctxt.tr_base = seg.base;
> > +    ctxt.tr_arbytes = seg.attr;
> > +
> > +    hvm_get_segment_register(v, x86_seg_ldtr, &seg);
> > +    ctxt.ldtr_sel = seg.sel;
> > +    ctxt.ldtr_limit = seg.limit;
> > +    ctxt.ldtr_base = seg.base;
> > +    ctxt.ldtr_arbytes = seg.attr;
> > +
> > +    if ( v->fpu_initialised )
> > +    {
> > +        memcpy(ctxt.fpu_regs, v->arch.fpu_ctxt,
> > sizeof(ctxt.fpu_regs));
> > +        ctxt.flags = XEN_X86_FPU_INITIALISED;
> > +    }
> > +
> > +    ctxt.rax = v->arch.user_regs.rax;
> > +    ctxt.rbx = v->arch.user_regs.rbx;
> > +    ctxt.rcx = v->arch.user_regs.rcx;
> > +    ctxt.rdx = v->arch.user_regs.rdx;
> > +    ctxt.rbp = v->arch.user_regs.rbp;
> > +    ctxt.rsi = v->arch.user_regs.rsi;
> > +    ctxt.rdi = v->arch.user_regs.rdi;
> > +    ctxt.rsp = v->arch.user_regs.rsp;
> > +    ctxt.rip = v->arch.user_regs.rip;
> > +    ctxt.rflags = v->arch.user_regs.rflags;
> > +    ctxt.r8  = v->arch.user_regs.r8;
> > +    ctxt.r9  = v->arch.user_regs.r9;
> > +    ctxt.r10 = v->arch.user_regs.r10;
> > +    ctxt.r11 = v->arch.user_regs.r11;
> > +    ctxt.r12 = v->arch.user_regs.r12;
> > +    ctxt.r13 = v->arch.user_regs.r13;
> > +    ctxt.r14 = v->arch.user_regs.r14;
> > +    ctxt.r15 = v->arch.user_regs.r15;
> > +    ctxt.dr0 = v->arch.debugreg[0];
> > +    ctxt.dr1 = v->arch.debugreg[1];
> > +    ctxt.dr2 = v->arch.debugreg[2];
> > +    ctxt.dr3 = v->arch.debugreg[3];
> > +    ctxt.dr6 = v->arch.debugreg[6];
> > +    ctxt.dr7 = v->arch.debugreg[7];
> > +
> > +    return hvm_save_entry(CPU, v->vcpu_id, h, &ctxt);
> > +}
> > +
> >  static int hvm_save_cpu_ctxt(struct domain *d,
> > hvm_domain_context_t *h)
> >  {
> >      struct vcpu *v;
> > -    struct hvm_hw_cpu ctxt;
> > -    struct segment_register seg;
> > +    int err = 0;
> > 
> >      for_each_vcpu ( d, v )
> >      {
> > -        /* We don't need to save state for a vcpu that is down;
> > the restore
> > -         * code will leave it down if there is nothing saved. */
> > +        /*
> > +         * We don't need to save state for a vcpu that is down;
> > the restore
> > +         * code will leave it down if there is nothing saved.
> > +         */
> >          if ( v->pause_flags & VPF_down )
> >              continue;
> > 
> > -        memset(&ctxt, 0, sizeof(ctxt));
> > -
> > -        /* Architecture-specific vmcs/vmcb bits */
> > -        hvm_funcs.save_cpu_ctxt(v, &ctxt);
> > -
> > -        ctxt.tsc = hvm_get_guest_tsc_fixed(v, d-
> > >arch.hvm_domain.sync_tsc);
> > -
> > -        ctxt.msr_tsc_aux = hvm_msr_tsc_aux(v);
> > -
> > -        hvm_get_segment_register(v, x86_seg_idtr, &seg);
> > -        ctxt.idtr_limit = seg.limit;
> > -        ctxt.idtr_base = seg.base;
> > -
> > -        hvm_get_segment_register(v, x86_seg_gdtr, &seg);
> > -        ctxt.gdtr_limit = seg.limit;
> > -        ctxt.gdtr_base = seg.base;
> > -
> > -        hvm_get_segment_register(v, x86_seg_cs, &seg);
> > -        ctxt.cs_sel = seg.sel;
> > -        ctxt.cs_limit = seg.limit;
> > -        ctxt.cs_base = seg.base;
> > -        ctxt.cs_arbytes = seg.attr;
> > -
> > -        hvm_get_segment_register(v, x86_seg_ds, &seg);
> > -        ctxt.ds_sel = seg.sel;
> > -        ctxt.ds_limit = seg.limit;
> > -        ctxt.ds_base = seg.base;
> > -        ctxt.ds_arbytes = seg.attr;
> > -
> > -        hvm_get_segment_register(v, x86_seg_es, &seg);
> > -        ctxt.es_sel = seg.sel;
> > -        ctxt.es_limit = seg.limit;
> > -        ctxt.es_base = seg.base;
> > -        ctxt.es_arbytes = seg.attr;
> > -
> > -        hvm_get_segment_register(v, x86_seg_ss, &seg);
> > -        ctxt.ss_sel = seg.sel;
> > -        ctxt.ss_limit = seg.limit;
> > -        ctxt.ss_base = seg.base;
> > -        ctxt.ss_arbytes = seg.attr;
> > -
> > -        hvm_get_segment_register(v, x86_seg_fs, &seg);
> > -        ctxt.fs_sel = seg.sel;
> > -        ctxt.fs_limit = seg.limit;
> > -        ctxt.fs_base = seg.base;
> > -        ctxt.fs_arbytes = seg.attr;
> > -
> > -        hvm_get_segment_register(v, x86_seg_gs, &seg);
> > -        ctxt.gs_sel = seg.sel;
> > -        ctxt.gs_limit = seg.limit;
> > -        ctxt.gs_base = seg.base;
> > -        ctxt.gs_arbytes = seg.attr;
> > -
> > -        hvm_get_segment_register(v, x86_seg_tr, &seg);
> > -        ctxt.tr_sel = seg.sel;
> > -        ctxt.tr_limit = seg.limit;
> > -        ctxt.tr_base = seg.base;
> > -        ctxt.tr_arbytes = seg.attr;
> > -
> > -        hvm_get_segment_register(v, x86_seg_ldtr, &seg);
> > -        ctxt.ldtr_sel = seg.sel;
> > -        ctxt.ldtr_limit = seg.limit;
> > -        ctxt.ldtr_base = seg.base;
> > -        ctxt.ldtr_arbytes = seg.attr;
> > -
> > -        if ( v->fpu_initialised )
> > -        {
> > -            memcpy(ctxt.fpu_regs, v->arch.fpu_ctxt,
> > sizeof(ctxt.fpu_regs));
> > -            ctxt.flags = XEN_X86_FPU_INITIALISED;
> > -        }
> > -
> > -        ctxt.rax = v->arch.user_regs.rax;
> > -        ctxt.rbx = v->arch.user_regs.rbx;
> > -        ctxt.rcx = v->arch.user_regs.rcx;
> > -        ctxt.rdx = v->arch.user_regs.rdx;
> > -        ctxt.rbp = v->arch.user_regs.rbp;
> > -        ctxt.rsi = v->arch.user_regs.rsi;
> > -        ctxt.rdi = v->arch.user_regs.rdi;
> > -        ctxt.rsp = v->arch.user_regs.rsp;
> > -        ctxt.rip = v->arch.user_regs.rip;
> > -        ctxt.rflags = v->arch.user_regs.rflags;
> > -        ctxt.r8  = v->arch.user_regs.r8;
> > -        ctxt.r9  = v->arch.user_regs.r9;
> > -        ctxt.r10 = v->arch.user_regs.r10;
> > -        ctxt.r11 = v->arch.user_regs.r11;
> > -        ctxt.r12 = v->arch.user_regs.r12;
> > -        ctxt.r13 = v->arch.user_regs.r13;
> > -        ctxt.r14 = v->arch.user_regs.r14;
> > -        ctxt.r15 = v->arch.user_regs.r15;
> > -        ctxt.dr0 = v->arch.debugreg[0];
> > -        ctxt.dr1 = v->arch.debugreg[1];
> > -        ctxt.dr2 = v->arch.debugreg[2];
> > -        ctxt.dr3 = v->arch.debugreg[3];
> > -        ctxt.dr6 = v->arch.debugreg[6];
> > -        ctxt.dr7 = v->arch.debugreg[7];
> > -
> > -        if ( hvm_save_entry(CPU, v->vcpu_id, h, &ctxt) != 0 )
> > -            return 1;
> > +        err = hvm_save_cpu_ctxt_one(v, h);
> > +        if ( err )
> > +            break;
> >      }
> > -    return 0;
> > +    return err;
> >  }
> > 
> >  /* Return a string indicating the error, or NULL for valid. */
> > --
> > 2.7.4

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

  reply	other threads:[~2018-07-17 12:25 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-07-16 14:55 [PATCH v12 00/11] x86/domctl: Save info for one vcpu instance Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 01/11] x86/cpu: Introduce vmce_save_vcpu_ctxt_one() func Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 02/11] x86/hvm: Introduce hvm_save_tsc_adjust_one() func Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 03/11] x86/hvm: Introduce hvm_save_cpu_ctxt_one func Alexandru Isaila
2018-07-16 15:29   ` Paul Durrant
2018-07-17 12:25     ` Isaila Alexandru [this message]
2018-07-17 14:03       ` Jan Beulich
2018-07-17 14:57         ` Isaila Alexandru
2018-07-16 14:55 ` [PATCH v12 04/11] x86/hvm: Introduce hvm_save_cpu_xsave_states_one Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 05/11] x86/hvm: Introduce hvm_save_cpu_msrs_one func Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 06/11] x86/hvm: Introduce hvm_save_mtrr_msr_one func Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 07/11] x86/hvm: Introduce viridian_save_vcpu_ctxt_one() func Alexandru Isaila
2018-07-16 15:34   ` Paul Durrant
2018-07-16 14:55 ` [PATCH v12 08/11] x86/hvm: Add handler for save_one funcs Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 09/11] x86/domctl: Don't pause the whole domain if only getting vcpu state Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 10/11] x86/hvm: Remove redundant save functions Alexandru Isaila
2018-07-16 14:55 ` [PATCH v12 11/11] x86/hvm: Remove save_one handler Alexandru Isaila

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=1531830332.10865.4.camel@bitdefender.com \
    --to=aisaila@bitdefender.com \
    --cc=Andrew.Cooper3@citrix.com \
    --cc=Ian.Jackson@citrix.com \
    --cc=Paul.Durrant@citrix.com \
    --cc=jbeulich@suse.com \
    --cc=wei.liu2@citrix.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 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.