[PATCH 8/18 V2]: PVH xen: domain creation code changes

[PATCH 8/18 V2]: PVH xen: domain creation code changes

[Mukesh Rathor]

This patch contains changes to arch/x86/domain.c to allow for a PVH
domain. Also, since PVH uses lot of HVM data structs and code paths, in
hvm_vcpu {} a sub struct to store PVH specific info is created.  Right
now it only has one field, but it can grow over time.

Changes in V2:
  - changes to read_segment_register() moved to this patch.
  - The other comment was to create NULL functions for pvh_set_vcpu_info
    and pvh_read_descriptor which are implemented in later patch, but since
    I disable PVH creation until all patches are checked in, it is not needed.
    But it helps breaking down of patches.

Signed-off-by: Mukesh Rathor <mukesh.rathor@oracle.com>
---
 xen/arch/x86/domain.c          |   99 +++++++++++++++++++++++++++-------------
 xen/include/asm-x86/hvm/hvm.h  |   18 +++++++
 xen/include/asm-x86/hvm/vcpu.h |    9 ++++
 xen/include/asm-x86/system.h   |    8 +++-
 4 files changed, 101 insertions(+), 33 deletions(-)

diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index 768c19d..5b5444f 100644
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -387,8 +387,11 @@ int vcpu_initialise(struct vcpu *v)
 
     vmce_init_vcpu(v);
 
-    if ( is_hvm_domain(d) )
+    if ( is_hvm_or_pvh_domain(d) )
     {
+        if ( is_pvh_domain(d) )
+            v->arch.hvm_vcpu.hvm_pvh.vcpu_info_mfn = INVALID_MFN;
+
         rc = hvm_vcpu_initialise(v);
         goto done;
     }
@@ -455,7 +458,7 @@ void vcpu_destroy(struct vcpu *v)
 
     vcpu_destroy_fpu(v);
 
-    if ( is_hvm_vcpu(v) )
+    if ( is_hvm_or_pvh_vcpu(v) )
         hvm_vcpu_destroy(v);
     else
         xfree(v->arch.pv_vcpu.trap_ctxt);
@@ -467,7 +470,7 @@ int arch_domain_create(struct domain *d, unsigned int domcr_flags)
     int rc = -ENOMEM;
 
     d->arch.hvm_domain.hap_enabled =
-        is_hvm_domain(d) &&
+        is_hvm_or_pvh_domain(d) &&
         hvm_funcs.hap_supported &&
         (domcr_flags & DOMCRF_hap);
     d->arch.hvm_domain.mem_sharing_enabled = 0;
@@ -515,7 +518,7 @@ int arch_domain_create(struct domain *d, unsigned int domcr_flags)
     mapcache_domain_init(d);
 
     HYPERVISOR_COMPAT_VIRT_START(d) =
-        is_hvm_domain(d) ? ~0u : __HYPERVISOR_COMPAT_VIRT_START;
+        is_hvm_or_pvh_domain(d) ? ~0u : __HYPERVISOR_COMPAT_VIRT_START;
 
     if ( (rc = paging_domain_init(d, domcr_flags)) != 0 )
         goto fail;
@@ -557,7 +560,7 @@ int arch_domain_create(struct domain *d, unsigned int domcr_flags)
             goto fail;
     }
 
-    if ( is_hvm_domain(d) )
+    if ( is_hvm_or_pvh_domain(d) )
     {
         if ( (rc = hvm_domain_initialise(d)) != 0 )
         {
@@ -569,9 +572,9 @@ int arch_domain_create(struct domain *d, unsigned int domcr_flags)
     {
         /* 64-bit PV guest by default. */
         d->arch.is_32bit_pv = d->arch.has_32bit_shinfo = 0;
-
-        spin_lock_init(&d->arch.pv_domain.e820_lock);
     }
+    if ( !is_hvm_domain(d) )
+        spin_lock_init(&d->arch.pv_domain.e820_lock);
 
     /* initialize default tsc behavior in case tools don't */
     tsc_set_info(d, TSC_MODE_DEFAULT, 0UL, 0, 0);
@@ -593,9 +596,10 @@ int arch_domain_create(struct domain *d, unsigned int domcr_flags)
 
 void arch_domain_destroy(struct domain *d)
 {
-    if ( is_hvm_domain(d) )
+    if ( is_hvm_or_pvh_domain(d) )
         hvm_domain_destroy(d);
-    else
+
+    if ( !is_hvm_domain(d) )
         xfree(d->arch.pv_domain.e820);
 
     free_domain_pirqs(d);
@@ -663,7 +667,7 @@ int arch_set_info_guest(
 #define c(fld) (compat ? (c.cmp->fld) : (c.nat->fld))
     flags = c(flags);
 
-    if ( !is_hvm_vcpu(v) )
+    if ( !is_hvm_or_pvh_vcpu(v) )
     {
         if ( !compat )
         {
@@ -716,7 +720,7 @@ int arch_set_info_guest(
     v->fpu_initialised = !!(flags & VGCF_I387_VALID);
 
     v->arch.flags &= ~TF_kernel_mode;
-    if ( (flags & VGCF_in_kernel) || is_hvm_vcpu(v)/*???*/ )
+    if ( (flags & VGCF_in_kernel) || is_hvm_or_pvh_vcpu(v)/*???*/ )
         v->arch.flags |= TF_kernel_mode;
 
     v->arch.vgc_flags = flags;
@@ -727,7 +731,7 @@ int arch_set_info_guest(
     if ( !compat )
     {
         memcpy(&v->arch.user_regs, &c.nat->user_regs, sizeof(c.nat->user_regs));
-        if ( !is_hvm_vcpu(v) )
+        if ( !is_hvm_or_pvh_vcpu(v) )
             memcpy(v->arch.pv_vcpu.trap_ctxt, c.nat->trap_ctxt,
                    sizeof(c.nat->trap_ctxt));
     }
@@ -743,10 +747,13 @@ int arch_set_info_guest(
 
     v->arch.user_regs.eflags |= 2;
 
-    if ( is_hvm_vcpu(v) )
+    if ( is_hvm_or_pvh_vcpu(v) )
     {
         hvm_set_info_guest(v);
-        goto out;
+        if ( is_hvm_vcpu(v) || v->is_initialised )
+            goto out;
+        else 
+            goto pvh_skip_pv_stuff;
     }
 
     init_int80_direct_trap(v);
@@ -755,7 +762,8 @@ int arch_set_info_guest(
     v->arch.pv_vcpu.iopl = (v->arch.user_regs.eflags >> 12) & 3;
     v->arch.user_regs.eflags &= ~X86_EFLAGS_IOPL;
 
-    /* Ensure real hardware interrupts are enabled. */
+    /* Ensure real hardware interrupts are enabled. Note: PVH may not have 
+     * IDT set on all vcpus so don't enable IF for it yet. */
     v->arch.user_regs.eflags |= X86_EFLAGS_IF;
 
     if ( !v->is_initialised )
@@ -852,6 +860,7 @@ int arch_set_info_guest(
     if ( rc != 0 )
         return rc;
 
+pvh_skip_pv_stuff:
     if ( !compat )
     {
         cr3_gfn = xen_cr3_to_pfn(c.nat->ctrlreg[3]);
@@ -859,19 +868,26 @@ int arch_set_info_guest(
 
         if ( !cr3_page )
         {
-            destroy_gdt(v);
+            if ( !is_pvh_vcpu(v) )
+                destroy_gdt(v);
             return -EINVAL;
         }
         if ( !paging_mode_refcounts(d)
              && !get_page_type(cr3_page, PGT_base_page_table) )
         {
             put_page(cr3_page);
-            destroy_gdt(v);
+            if ( !is_pvh_vcpu(v) )
+                destroy_gdt(v);
             return -EINVAL;
         }
 
+        if ( is_pvh_vcpu(v) ) {
+            v->arch.cr3 = page_to_mfn(cr3_page);
+            v->arch.hvm_vcpu.guest_cr[3] = c.nat->ctrlreg[3];
+        }
+
         v->arch.guest_table = pagetable_from_page(cr3_page);
-        if ( c.nat->ctrlreg[1] )
+        if ( c.nat->ctrlreg[1] && !is_pvh_vcpu(v) )
         {
             cr3_gfn = xen_cr3_to_pfn(c.nat->ctrlreg[1]);
             cr3_page = get_page_from_gfn(d, cr3_gfn, NULL, P2M_ALLOC);
@@ -896,7 +912,8 @@ int arch_set_info_guest(
         }
         else if ( !(flags & VGCF_in_kernel) )
         {
-            destroy_gdt(v);
+            if ( !is_pvh_vcpu(v) )
+                destroy_gdt(v);
             return -EINVAL;
         }
     }
@@ -938,6 +955,13 @@ int arch_set_info_guest(
 
     update_cr3(v);
 
+    if ( is_pvh_vcpu(v) )
+    {
+        /* guest is bringing up non-boot SMP vcpu */
+        if ( (rc=hvm_pvh_set_vcpu_info(v, c.nat)) != 0 )
+            return rc;
+    }
+
  out:
     if ( flags & VGCF_online )
         clear_bit(_VPF_down, &v->pause_flags);
@@ -968,16 +992,21 @@ void arch_vcpu_reset(struct vcpu *v)
 static void
 unmap_vcpu_info(struct vcpu *v)
 {
-    unsigned long mfn;
+    unsigned long mfn, *mfnp;
+
+    if ( is_pvh_vcpu(v) )
+        mfnp = &v->arch.hvm_vcpu.hvm_pvh.vcpu_info_mfn;
+    else
+        mfnp = &v->arch.pv_vcpu.vcpu_info_mfn;
 
-    if ( v->arch.pv_vcpu.vcpu_info_mfn == INVALID_MFN )
+    mfn = *mfnp;
+    if ( mfn == INVALID_MFN )
         return;
 
-    mfn = v->arch.pv_vcpu.vcpu_info_mfn;
     unmap_domain_page_global(v->vcpu_info);
 
     v->vcpu_info = &dummy_vcpu_info;
-    v->arch.pv_vcpu.vcpu_info_mfn = INVALID_MFN;
+    *mfnp = INVALID_MFN;
 
     put_page_and_type(mfn_to_page(mfn));
 }
@@ -996,11 +1025,17 @@ map_vcpu_info(struct vcpu *v, unsigned long gfn, unsigned offset)
     vcpu_info_t *new_info;
     struct page_info *page;
     int i;
+    unsigned long *mfnp;
+
+    if ( is_pvh_vcpu(v) )
+        mfnp = &v->arch.hvm_vcpu.hvm_pvh.vcpu_info_mfn;
+    else
+        mfnp = &v->arch.pv_vcpu.vcpu_info_mfn;
 
     if ( offset > (PAGE_SIZE - sizeof(vcpu_info_t)) )
         return -EINVAL;
 
-    if ( v->arch.pv_vcpu.vcpu_info_mfn != INVALID_MFN )
+    if ( *mfnp != INVALID_MFN )
         return -EINVAL;
 
     /* Run this command on yourself or on other offline VCPUS. */
@@ -1037,7 +1072,7 @@ map_vcpu_info(struct vcpu *v, unsigned long gfn, unsigned offset)
     }
 
     v->vcpu_info = new_info;
-    v->arch.pv_vcpu.vcpu_info_mfn = page_to_mfn(page);
+    *mfnp = page_to_mfn(page);
 
     /* Set new vcpu_info pointer /before/ setting pending flags. */
     wmb();
@@ -1443,7 +1478,7 @@ static void update_runstate_area(struct vcpu *v)
 
 static inline int need_full_gdt(struct vcpu *v)
 {
-    return (!is_hvm_vcpu(v) && !is_idle_vcpu(v));
+    return (!is_hvm_or_pvh_vcpu(v) && !is_idle_vcpu(v));
 }
 
 static void __context_switch(void)
@@ -1571,7 +1606,7 @@ void context_switch(struct vcpu *prev, struct vcpu *next)
         /* Re-enable interrupts before restoring state which may fault. */
         local_irq_enable();
 
-        if ( !is_hvm_vcpu(next) )
+        if ( !is_hvm_or_pvh_vcpu(next) )
         {
             load_LDT(next);
             load_segments(next);
@@ -1690,12 +1725,12 @@ unsigned long hypercall_create_continuation(
         regs->eax  = op;
 
         /* Ensure the hypercall trap instruction is re-executed. */
-        if ( !is_hvm_vcpu(current) )
+        if ( !is_hvm_or_pvh_vcpu(current) )
             regs->eip -= 2;  /* re-execute 'syscall' / 'int $xx' */
         else
             current->arch.hvm_vcpu.hcall_preempted = 1;
 
-        if ( !is_hvm_vcpu(current) ?
+        if ( !is_hvm_or_pvh_vcpu(current) ?
              !is_pv_32on64_vcpu(current) :
              (hvm_guest_x86_mode(current) == 8) )
         {
@@ -2011,7 +2046,7 @@ int domain_relinquish_resources(struct domain *d)
         for_each_vcpu ( d, v )
             vcpu_destroy_pagetables(v);
 
-        if ( !is_hvm_domain(d) )
+        if ( !is_hvm_or_pvh_domain(d) )
         {
             for_each_vcpu ( d, v )
             {
@@ -2086,7 +2121,7 @@ int domain_relinquish_resources(struct domain *d)
         BUG();
     }
 
-    if ( is_hvm_domain(d) )
+    if ( is_hvm_or_pvh_domain(d) )
         hvm_domain_relinquish_resources(d);
 
     return 0;
@@ -2167,7 +2202,7 @@ void vcpu_mark_events_pending(struct vcpu *v)
     if ( already_pending )
         return;
 
-    if ( is_hvm_vcpu(v) )
+    if ( is_hvm_or_pvh_vcpu(v) )
         hvm_assert_evtchn_irq(v);
     else
         vcpu_kick(v);
diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
index 2fa2ea5..31aa04f 100644
--- a/xen/include/asm-x86/hvm/hvm.h
+++ b/xen/include/asm-x86/hvm/hvm.h
@@ -190,6 +190,11 @@ struct hvm_function_table {
                                 paddr_t *L1_gpa, unsigned int *page_order,
                                 uint8_t *p2m_acc, bool_t access_r,
                                 bool_t access_w, bool_t access_x);
+    /* PVH functions */
+    int (*pvh_set_vcpu_info)(struct vcpu *v, struct vcpu_guest_context *ctxtp);
+    int (*pvh_read_descriptor)(unsigned int sel, const struct vcpu *v,
+                         const struct cpu_user_regs *regs, unsigned long *base,
+                         unsigned long *limit, unsigned int *ar);
 };
 
 extern struct hvm_function_table hvm_funcs;
@@ -323,6 +328,19 @@ static inline unsigned long hvm_get_shadow_gs_base(struct vcpu *v)
     return hvm_funcs.get_shadow_gs_base(v);
 }
 
+static inline int hvm_pvh_set_vcpu_info(struct vcpu *v, 
+                                        struct vcpu_guest_context *ctxtp)
+{
+    return hvm_funcs.pvh_set_vcpu_info(v, ctxtp);
+}
+
+static inline int hvm_pvh_read_descriptor(unsigned int sel, 
+               const struct vcpu *v, const struct cpu_user_regs *regs, 
+               unsigned long *base, unsigned long *limit, unsigned int *ar)
+{
+    return hvm_funcs.pvh_read_descriptor(sel, v, regs, base, limit, ar);
+}
+
 #define is_viridian_domain(_d)                                             \
  (is_hvm_domain(_d) && ((_d)->arch.hvm_domain.params[HVM_PARAM_VIRIDIAN]))
 
diff --git a/xen/include/asm-x86/hvm/vcpu.h b/xen/include/asm-x86/hvm/vcpu.h
index e8b8cd7..2725a62 100644
--- a/xen/include/asm-x86/hvm/vcpu.h
+++ b/xen/include/asm-x86/hvm/vcpu.h
@@ -104,6 +104,13 @@ struct nestedvcpu {
 
 #define vcpu_nestedhvm(v) ((v)->arch.hvm_vcpu.nvcpu)
 
+/* add any PVH specific fields here */
+struct pvh_hvm_vcpu_ext
+{
+    /* Guest-specified relocation of vcpu_info. */
+    unsigned long vcpu_info_mfn;
+};
+
 struct hvm_vcpu {
     /* Guest control-register and EFER values, just as the guest sees them. */
     unsigned long       guest_cr[5];
@@ -170,6 +177,8 @@ struct hvm_vcpu {
     struct hvm_trap     inject_trap;
 
     struct viridian_vcpu viridian;
+
+    struct pvh_hvm_vcpu_ext hvm_pvh;
 };
 
 #endif /* __ASM_X86_HVM_VCPU_H__ */
diff --git a/xen/include/asm-x86/system.h b/xen/include/asm-x86/system.h
index d8dc6f2..5681806 100644
--- a/xen/include/asm-x86/system.h
+++ b/xen/include/asm-x86/system.h
@@ -4,9 +4,15 @@
 #include <xen/lib.h>
 #include <asm/bitops.h>
 
+/* We need vcpu because during context switch, going from pure PV to PVH,
+ * in save_segments(), current has been updated to next, and no longer pointing
+ * to the pure PV. Note: for PVH, we update regs->selectors on each vmexit */
 #define read_segment_register(vcpu, regs, name)                 \
 ({  u16 __sel;                                                  \
-    asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
+    if (is_pvh_vcpu(vcpu))                                         \
+        __sel = regs->name;                                     \
+    else                                                         \
+        asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
     __sel;                                                      \
 })
 
-- 
1.7.2.3

Re: [PATCH 8/18 V2]: PVH xen: domain creation code changes

[Jan Beulich]

>>> On 16.03.13 at 01:36, Mukesh Rathor <mukesh.rathor@oracle.com> wrote:
> --- a/xen/include/asm-x86/system.h
> +++ b/xen/include/asm-x86/system.h
> @@ -4,9 +4,15 @@
>  #include <xen/lib.h>
>  #include <asm/bitops.h>
>  
> +/* We need vcpu because during context switch, going from pure PV to PVH,
> + * in save_segments(), current has been updated to next, and no longer pointing
> + * to the pure PV. Note: for PVH, we update regs->selectors on each vmexit */
>  #define read_segment_register(vcpu, regs, name)                 \
>  ({  u16 __sel;                                                  \
> -    asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
> +    if (is_pvh_vcpu(vcpu))                                         \
> +        __sel = regs->name;                                     \
> +    else                                                         \
> +        asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
>      __sel;                                                      \
>  })

In a generic macro like this, please make sure you evaluate each
argument exactly once, and you properly parenthesize all uses of
macro arguments.

Jan

Re: [PATCH 8/18 V2]: PVH xen: domain creation code changes

[Konrad Rzeszutek Wilk]

> diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm-x86/hvm/hvm.h
> index 2fa2ea5..31aa04f 100644
> --- a/xen/include/asm-x86/hvm/hvm.h
> +++ b/xen/include/asm-x86/hvm/hvm.h
> @@ -190,6 +190,11 @@ struct hvm_function_table {
>                                  paddr_t *L1_gpa, unsigned int *page_order,
>                                  uint8_t *p2m_acc, bool_t access_r,
>                                  bool_t access_w, bool_t access_x);
> +    /* PVH functions */
> +    int (*pvh_set_vcpu_info)(struct vcpu *v, struct vcpu_guest_context *ctxtp);
> +    int (*pvh_read_descriptor)(unsigned int sel, const struct vcpu *v,
> +                         const struct cpu_user_regs *regs, unsigned long *base,
> +                         unsigned long *limit, unsigned int *ar);

Ewww.. Please remove the 'pvh_' part and have a comment saying:
	/* These two functions are used only in PVH mode. */

>  };
>  
>  extern struct hvm_function_table hvm_funcs;
> @@ -323,6 +328,19 @@ static inline unsigned long hvm_get_shadow_gs_base(struct vcpu *v)
>      return hvm_funcs.get_shadow_gs_base(v);
>  }
>  
> +static inline int hvm_pvh_set_vcpu_info(struct vcpu *v, 
> +                                        struct vcpu_guest_context *ctxtp)
> +{
> +    return hvm_funcs.pvh_set_vcpu_info(v, ctxtp);
> +}
> +
> +static inline int hvm_pvh_read_descriptor(unsigned int sel, 
> +               const struct vcpu *v, const struct cpu_user_regs *regs, 
> +               unsigned long *base, unsigned long *limit, unsigned int *ar)
> +{
> +    return hvm_funcs.pvh_read_descriptor(sel, v, regs, base, limit, ar);
> +}
> +
>  #define is_viridian_domain(_d)                                             \
>   (is_hvm_domain(_d) && ((_d)->arch.hvm_domain.params[HVM_PARAM_VIRIDIAN]))
>  
> diff --git a/xen/include/asm-x86/hvm/vcpu.h b/xen/include/asm-x86/hvm/vcpu.h
> index e8b8cd7..2725a62 100644
> --- a/xen/include/asm-x86/hvm/vcpu.h
> +++ b/xen/include/asm-x86/hvm/vcpu.h
> @@ -104,6 +104,13 @@ struct nestedvcpu {
>  
>  #define vcpu_nestedhvm(v) ((v)->arch.hvm_vcpu.nvcpu)
>  
> +/* add any PVH specific fields here */
> +struct pvh_hvm_vcpu_ext
> +{
> +    /* Guest-specified relocation of vcpu_info. */
> +    unsigned long vcpu_info_mfn;
> +};
> +
>  struct hvm_vcpu {
>      /* Guest control-register and EFER values, just as the guest sees them. */
>      unsigned long       guest_cr[5];
> @@ -170,6 +177,8 @@ struct hvm_vcpu {
>      struct hvm_trap     inject_trap;
>  
>      struct viridian_vcpu viridian;
> +
> +    struct pvh_hvm_vcpu_ext hvm_pvh;

Can you remove the two 'hvm' parts? So it is
	struct pvh_vcpu_ext pvh;

?
>  };
>  
>  #endif /* __ASM_X86_HVM_VCPU_H__ */
> diff --git a/xen/include/asm-x86/system.h b/xen/include/asm-x86/system.h
> index d8dc6f2..5681806 100644
> --- a/xen/include/asm-x86/system.h
> +++ b/xen/include/asm-x86/system.h
> @@ -4,9 +4,15 @@
>  #include <xen/lib.h>
>  #include <asm/bitops.h>
>  
> +/* We need vcpu because during context switch, going from pure PV to PVH,
> + * in save_segments(), current has been updated to next, and no longer pointing
> + * to the pure PV. Note: for PVH, we update regs->selectors on each vmexit */
>  #define read_segment_register(vcpu, regs, name)                 \
>  ({  u16 __sel;                                                  \
> -    asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
> +    if (is_pvh_vcpu(vcpu))                                         \
> +        __sel = regs->name;                                     \
> +    else                                                         \
> +        asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
>      __sel;                                                      \
>  })
>  
> -- 
> 1.7.2.3
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
> 

Re: [PATCH 8/18 V2]: PVH xen: domain creation code changes

[Tim Deegan]

At 11:57 +0000 on 18 Mar (1363607863), Jan Beulich wrote:
> >>> On 16.03.13 at 01:36, Mukesh Rathor <mukesh.rathor@oracle.com> wrote:
> > --- a/xen/include/asm-x86/system.h
> > +++ b/xen/include/asm-x86/system.h
> > @@ -4,9 +4,15 @@
> >  #include <xen/lib.h>
> >  #include <asm/bitops.h>
> >  
> > +/* We need vcpu because during context switch, going from pure PV to PVH,
> > + * in save_segments(), current has been updated to next, and no longer pointing
> > + * to the pure PV. Note: for PVH, we update regs->selectors on each vmexit */
> >  #define read_segment_register(vcpu, regs, name)                 \
> >  ({  u16 __sel;                                                  \
> > -    asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
> > +    if (is_pvh_vcpu(vcpu))                                         \
> > +        __sel = regs->name;                                     \
> > +    else                                                         \
> > +        asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
> >      __sel;                                                      \
> >  })
> 
> In a generic macro like this, please make sure you evaluate each
> argument exactly once, and you properly parenthesize all uses of
> macro arguments.

And please don't disturb the neat row of backslashes. :)

Tim.

Re: [PATCH 8/18 V2]: PVH xen: domain creation code changes

[Mukesh Rathor]

On Mon, 18 Mar 2013 11:57:43 +0000
"Jan Beulich" <JBeulich@suse.com> wrote:

> >>> On 16.03.13 at 01:36, Mukesh Rathor <mukesh.rathor@oracle.com>
> >>> wrote:
> > --- a/xen/include/asm-x86/system.h
> > +++ b/xen/include/asm-x86/system.h
> > @@ -4,9 +4,15 @@
> >  #include <xen/lib.h>
> >  #include <asm/bitops.h>
> >  
> > +/* We need vcpu because during context switch, going from pure PV
> > to PVH,
> > + * in save_segments(), current has been updated to next, and no
> > longer pointing
> > + * to the pure PV. Note: for PVH, we update regs->selectors on
> > each vmexit */ #define read_segment_register(vcpu, regs,
> > name)                 \ ({  u16
> > __sel;                                                  \
> > -    asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
> > +    if (is_pvh_vcpu(vcpu))
> > \
> > +        __sel = regs->name;                                     \
> > +    else                                                         \
> > +        asm volatile ( "movw %%" STR(name) ",%0" :
> > "=r" (__sel) );  \
> > __sel;                                                      \ })
> 
> In a generic macro like this, please make sure you evaluate each
> argument exactly once, and you properly parenthesize all uses of
> macro arguments.

Hmm... at a loss. The only think I'm able to come up with here is
paranthesis around regs, and spaces in the if statement. Both vcpu and
regs are used only once.

Re: [PATCH 8/18 V2]: PVH xen: domain creation code changes

[Jan Beulich]

>>> On 26.03.13 at 02:29, Mukesh Rathor <mukesh.rathor@oracle.com> wrote:
> On Mon, 18 Mar 2013 11:57:43 +0000
> "Jan Beulich" <JBeulich@suse.com> wrote:
> 
>> >>> On 16.03.13 at 01:36, Mukesh Rathor <mukesh.rathor@oracle.com>
>> >>> wrote:
>> > --- a/xen/include/asm-x86/system.h
>> > +++ b/xen/include/asm-x86/system.h
>> > @@ -4,9 +4,15 @@
>> >  #include <xen/lib.h>
>> >  #include <asm/bitops.h>
>> >  
>> > +/* We need vcpu because during context switch, going from pure PV
>> > to PVH,
>> > + * in save_segments(), current has been updated to next, and no
>> > longer pointing
>> > + * to the pure PV. Note: for PVH, we update regs->selectors on
>> > each vmexit */ #define read_segment_register(vcpu, regs,
>> > name)                 \ ({  u16
>> > __sel;                                                  \
>> > -    asm volatile ( "movw %%" STR(name) ",%0" : "=r" (__sel) );  \
>> > +    if (is_pvh_vcpu(vcpu))
>> > \
>> > +        __sel = regs->name;                                     \
>> > +    else                                                         \
>> > +        asm volatile ( "movw %%" STR(name) ",%0" :
>> > "=r" (__sel) );  \
>> > __sel;                                                      \ })
>> 
>> In a generic macro like this, please make sure you evaluate each
>> argument exactly once, and you properly parenthesize all uses of
>> macro arguments.
> 
> Hmm... at a loss. The only think I'm able to come up with here is
> paranthesis around regs, and spaces in the if statement. Both vcpu and
> regs are used only once.

Not really - vcpu is, but regs has one path where it gets evaluated,
and one path where it doesn't get used.

Btw, no matter whether there are other precedents, I do think that
the use of STR() here is misguided too - #name seems like the way
to go to me. STR() really is needed when you want the argument to
be further macro expanded before getting converted to a string,
but here you want the exact opposite - the guarantee that no
macro expansion happens (or else the "regs->name" use would
break).

Jan

Re: [PATCH 8/18 V2]: PVH xen: domain creation code changes

[Mukesh Rathor]

On Tue, 26 Mar 2013 07:39:51 +0000
"Jan Beulich" <JBeulich@suse.com> wrote:

> >>> On 26.03.13 at 02:29, Mukesh Rathor <mukesh.rathor@oracle.com>
> >>> wrote:
> > On Mon, 18 Mar 2013 11:57:43 +0000
> > "Jan Beulich" <JBeulich@suse.com> wrote:
> > 
> >> >>> On 16.03.13 at 01:36, Mukesh Rathor <mukesh.rathor@oracle.com>
> >> >>> wrote:
> > Hmm... at a loss. The only think I'm able to come up with here is
> > paranthesis around regs, and spaces in the if statement. Both vcpu
> > and regs are used only once.
> 
> Not really - vcpu is, but regs has one path where it gets evaluated,
> and one path where it doesn't get used.
> 
> Btw, no matter whether there are other precedents, I do think that
> the use of STR() here is misguided too - #name seems like the way
> to go to me. STR() really is needed when you want the argument to
> be further macro expanded before getting converted to a string,
> but here you want the exact opposite - the guarantee that no
> macro expansion happens (or else the "regs->name" use would
> break).

Got it, thanks a lot. 

#define read_segment_register(vcpu, regs, name)                   \
({  u16 __sel;                                                    \
    struct cpu_user_regs *_regs = (regs);                         \
                                                                  \
    if ( is_pvh_vcpu(vcpu) )                                      \
        __sel = _regs->name;                                      \
    else                                                          \
        asm volatile ( "movw %%" #name ",%0" : "=r" (__sel) );    \
    __sel;                                                        \
})

Re: [PATCH 8/18 V2]: PVH xen: domain creation code changes

[Jan Beulich]

>>> On 26.03.13 at 22:04, Mukesh Rathor <mukesh.rathor@oracle.com> wrote:
> #define read_segment_register(vcpu, regs, name)                   \
> ({  u16 __sel;                                                    \
>     struct cpu_user_regs *_regs = (regs);                         \
>                                                                   \
>     if ( is_pvh_vcpu(vcpu) )                                      \
>         __sel = _regs->name;                                      \
>     else                                                          \
>         asm volatile ( "movw %%" #name ",%0" : "=r" (__sel) );    \
>     __sel;                                                        \
> })

Almost: If you really want to keep the _regs variable, then it
should be "const struct cpu_user_regs *". But is there a reason
not to simply have the declaration of __sel have an initializer?

Jan