|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Xen-changelog] [xen-unstable] x86: Remove global percpu_mm_info structure, to make dataflow through
# HG changeset patch
# User Keir Fraser <keir.fraser@xxxxxxxxxx>
# Date 1250759818 -3600
# Node ID 974260bfd3d8f250bbc94758538a027a8f547b51
# Parent e83bb28d48feee796837261247e0918255b42f85
x86: Remove global percpu_mm_info structure, to make dataflow through
mm code clearer.
The FOREIGNDOM method was just confusing and pointless. The deferred
TLB flushing is of questionable value now that much automatic flushing has to be
synchronous to avoid guest SMP races.
Signed-off-by: Keir Fraser <keir.fraser@xxxxxxxxxx>
---
xen/arch/x86/mm.c | 235 +++++++++++++++++++++---------------------------------
1 files changed, 92 insertions(+), 143 deletions(-)
diff -r e83bb28d48fe -r 974260bfd3d8 xen/arch/x86/mm.c
--- a/xen/arch/x86/mm.c Thu Aug 20 08:26:51 2009 +0100
+++ b/xen/arch/x86/mm.c Thu Aug 20 10:16:58 2009 +0100
@@ -134,23 +134,6 @@ l1_pgentry_t __attribute__ ((__section__
#define PTE_UPDATE_WITH_CMPXCHG
#endif
-/* Used to defer flushing of memory structures. */
-struct percpu_mm_info {
-#define DOP_FLUSH_TLB (1<<0) /* Flush the local TLB. */
-#define DOP_FLUSH_ALL_TLBS (1<<1) /* Flush TLBs of all VCPUs of current dom. */
-#define DOP_RELOAD_LDT (1<<2) /* Reload the LDT shadow mapping. */
- unsigned int deferred_ops;
- /* If non-NULL, specifies a foreign subject domain for some operations. */
- struct domain *foreign;
-};
-static DEFINE_PER_CPU(struct percpu_mm_info, percpu_mm_info);
-
-/*
- * Returns the current foreign domain; defaults to the currently-executing
- * domain if a foreign override hasn't been specified.
- */
-#define FOREIGNDOM (this_cpu(percpu_mm_info).foreign ?: current->domain)
-
/* Private domain structs for DOMID_XEN and DOMID_IO. */
struct domain *dom_xen, *dom_io;
@@ -1679,10 +1662,10 @@ static inline int update_intpte(intpte_t
/* Update the L1 entry at pl1e to new value nl1e. */
static int mod_l1_entry(l1_pgentry_t *pl1e, l1_pgentry_t nl1e,
unsigned long gl1mfn, int preserve_ad,
- struct vcpu *vcpu)
+ struct vcpu *pt_vcpu, struct domain *pg_dom)
{
l1_pgentry_t ol1e;
- struct domain *d = vcpu->domain;
+ struct domain *pt_dom = pt_vcpu->domain;
unsigned long mfn;
p2m_type_t p2mt;
int rc = 1;
@@ -1690,55 +1673,55 @@ static int mod_l1_entry(l1_pgentry_t *pl
if ( unlikely(__copy_from_user(&ol1e, pl1e, sizeof(ol1e)) != 0) )
return 0;
- if ( unlikely(paging_mode_refcounts(d)) )
- {
- rc = UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, vcpu, preserve_ad);
+ if ( unlikely(paging_mode_refcounts(pt_dom)) )
+ {
+ rc = UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, pt_vcpu, preserve_ad);
return rc;
}
if ( l1e_get_flags(nl1e) & _PAGE_PRESENT )
{
/* Translate foreign guest addresses. */
- mfn = mfn_x(gfn_to_mfn(FOREIGNDOM, l1e_get_pfn(nl1e), &p2mt));
+ mfn = mfn_x(gfn_to_mfn(pg_dom, l1e_get_pfn(nl1e), &p2mt));
if ( !p2m_is_ram(p2mt) || unlikely(mfn == INVALID_MFN) )
return 0;
ASSERT((mfn & ~(PADDR_MASK >> PAGE_SHIFT)) == 0);
nl1e = l1e_from_pfn(mfn, l1e_get_flags(nl1e));
- if ( unlikely(l1e_get_flags(nl1e) & l1_disallow_mask(d)) )
+ if ( unlikely(l1e_get_flags(nl1e) & l1_disallow_mask(pt_dom)) )
{
MEM_LOG("Bad L1 flags %x",
- l1e_get_flags(nl1e) & l1_disallow_mask(d));
+ l1e_get_flags(nl1e) & l1_disallow_mask(pt_dom));
return 0;
}
/* Fast path for identical mapping, r/w and presence. */
if ( !l1e_has_changed(ol1e, nl1e, _PAGE_RW | _PAGE_PRESENT) )
{
- adjust_guest_l1e(nl1e, d);
- rc = UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, vcpu,
+ adjust_guest_l1e(nl1e, pt_dom);
+ rc = UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, pt_vcpu,
preserve_ad);
return rc;
}
- if ( unlikely(!get_page_from_l1e(nl1e, d, FOREIGNDOM)) )
+ if ( unlikely(!get_page_from_l1e(nl1e, pt_dom, pg_dom)) )
return 0;
- adjust_guest_l1e(nl1e, d);
- if ( unlikely(!UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, vcpu,
+ adjust_guest_l1e(nl1e, pt_dom);
+ if ( unlikely(!UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, pt_vcpu,
preserve_ad)) )
{
ol1e = nl1e;
rc = 0;
}
}
- else if ( unlikely(!UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, vcpu,
+ else if ( unlikely(!UPDATE_ENTRY(l1, pl1e, ol1e, nl1e, gl1mfn, pt_vcpu,
preserve_ad)) )
{
return 0;
}
- put_page_from_l1e(ol1e, d);
+ put_page_from_l1e(ol1e, pt_dom);
return rc;
}
@@ -2482,91 +2465,63 @@ int new_guest_cr3(unsigned long mfn)
return 1;
}
-static void process_deferred_ops(void)
-{
- unsigned int deferred_ops;
- struct domain *d = current->domain;
- struct percpu_mm_info *info = &this_cpu(percpu_mm_info);
-
- deferred_ops = info->deferred_ops;
- info->deferred_ops = 0;
-
- if ( deferred_ops & (DOP_FLUSH_ALL_TLBS|DOP_FLUSH_TLB) )
- {
- if ( deferred_ops & DOP_FLUSH_ALL_TLBS )
- flush_tlb_mask(&d->domain_dirty_cpumask);
- else
- flush_tlb_local();
- }
-
- /*
- * Do this after flushing TLBs, to ensure we see fresh LDT mappings
- * via the linear pagetable mapping.
- */
- if ( deferred_ops & DOP_RELOAD_LDT )
- (void)map_ldt_shadow_page(0);
-
- if ( unlikely(info->foreign != NULL) )
- {
- rcu_unlock_domain(info->foreign);
- info->foreign = NULL;
- }
-}
-
-static int set_foreigndom(domid_t domid)
-{
- struct domain *e, *d = current->domain;
- struct percpu_mm_info *info = &this_cpu(percpu_mm_info);
- int okay = 1;
-
- ASSERT(info->foreign == NULL);
+static struct domain *get_pg_owner(domid_t domid)
+{
+ struct domain *pg_owner = NULL, *curr = current->domain;
if ( likely(domid == DOMID_SELF) )
+ {
+ pg_owner = rcu_lock_domain(curr);
goto out;
-
- if ( unlikely(domid == d->domain_id) )
+ }
+
+ if ( unlikely(domid == curr->domain_id) )
{
MEM_LOG("Cannot specify itself as foreign domain");
- okay = 0;
- }
- else if ( unlikely(paging_mode_translate(d)) )
+ goto out;
+ }
+
+ if ( unlikely(paging_mode_translate(curr)) )
{
MEM_LOG("Cannot mix foreign mappings with translated domains");
- okay = 0;
- }
- else switch ( domid )
+ goto out;
+ }
+
+ switch ( domid )
{
case DOMID_IO:
- info->foreign = rcu_lock_domain(dom_io);
+ pg_owner = rcu_lock_domain(dom_io);
break;
case DOMID_XEN:
- if (!IS_PRIV(d)) {
+ if ( !IS_PRIV(curr) )
+ {
MEM_LOG("Cannot set foreign dom");
- okay = 0;
break;
}
- info->foreign = rcu_lock_domain(dom_xen);
+ pg_owner = rcu_lock_domain(dom_xen);
break;
default:
- if ( (e = rcu_lock_domain_by_id(domid)) == NULL )
+ if ( (pg_owner = rcu_lock_domain_by_id(domid)) == NULL )
{
MEM_LOG("Unknown domain '%u'", domid);
- okay = 0;
break;
}
- if ( !IS_PRIV_FOR(d, e) )
+ if ( !IS_PRIV_FOR(curr, pg_owner) )
{
MEM_LOG("Cannot set foreign dom");
- okay = 0;
- rcu_unlock_domain(e);
- break;
- }
- info->foreign = e;
+ rcu_unlock_domain(pg_owner);
+ pg_owner = NULL;
+ }
break;
}
out:
- return okay;
+ return pg_owner;
+}
+
+static void put_pg_owner(struct domain *pg_owner)
+{
+ rcu_unlock_domain(pg_owner);
}
static inline int vcpumask_to_pcpumask(
@@ -2642,6 +2597,7 @@ int do_mmuext_op(
struct page_info *page;
struct vcpu *curr = current;
struct domain *d = curr->domain;
+ struct domain *pg_owner;
if ( unlikely(count & MMU_UPDATE_PREEMPTED) )
{
@@ -2658,7 +2614,7 @@ int do_mmuext_op(
goto out;
}
- if ( !set_foreigndom(foreigndom) )
+ if ( (pg_owner = get_pg_owner(foreigndom)) == NULL )
{
rc = -ESRCH;
goto out;
@@ -2680,7 +2636,7 @@ int do_mmuext_op(
}
gmfn = op.arg1.mfn;
- mfn = gmfn_to_mfn(FOREIGNDOM, gmfn);
+ mfn = gmfn_to_mfn(pg_owner, gmfn);
if ( mfn == INVALID_MFN )
{
MEM_LOG("Bad gmfn_to_mfn");
@@ -2705,7 +2661,7 @@ int do_mmuext_op(
goto pin_page;
case MMUEXT_PIN_L4_TABLE:
- if ( is_pv_32bit_domain(FOREIGNDOM) )
+ if ( is_pv_32bit_domain(pg_owner) )
break;
type = PGT_l4_page_table;
@@ -2718,10 +2674,10 @@ int do_mmuext_op(
if ( (op.cmd - MMUEXT_PIN_L1_TABLE) > (CONFIG_PAGING_LEVELS - 1) )
break;
- if ( paging_mode_refcounts(FOREIGNDOM) )
+ if ( paging_mode_refcounts(pg_owner) )
break;
- rc = get_page_and_type_from_pagenr(mfn, type, FOREIGNDOM, 0, 1);
+ rc = get_page_and_type_from_pagenr(mfn, type, pg_owner, 0, 1);
okay = !rc;
if ( unlikely(!okay) )
{
@@ -2745,14 +2701,14 @@ int do_mmuext_op(
paging_mark_dirty(d, mfn);
/* We can race domain destruction (domain_relinquish_resources). */
- if ( unlikely(this_cpu(percpu_mm_info).foreign != NULL) )
+ if ( unlikely(pg_owner != d) )
{
int drop_ref;
- spin_lock(&FOREIGNDOM->page_alloc_lock);
- drop_ref = (FOREIGNDOM->is_dying &&
+ spin_lock(&pg_owner->page_alloc_lock);
+ drop_ref = (pg_owner->is_dying &&
test_and_clear_bit(_PGT_pinned,
&page->u.inuse.type_info));
- spin_unlock(&FOREIGNDOM->page_alloc_lock);
+ spin_unlock(&pg_owner->page_alloc_lock);
if ( drop_ref )
put_page_and_type(page);
}
@@ -2789,7 +2745,6 @@ int do_mmuext_op(
case MMUEXT_NEW_BASEPTR:
okay = new_guest_cr3(mfn);
- this_cpu(percpu_mm_info).deferred_ops &= ~DOP_FLUSH_TLB;
break;
#ifdef __x86_64__
@@ -2826,7 +2781,7 @@ int do_mmuext_op(
#endif
case MMUEXT_TLB_FLUSH_LOCAL:
- this_cpu(percpu_mm_info).deferred_ops |= DOP_FLUSH_TLB;
+ flush_tlb_local();
break;
case MMUEXT_INVLPG_LOCAL:
@@ -2838,7 +2793,7 @@ int do_mmuext_op(
case MMUEXT_TLB_FLUSH_MULTI:
case MMUEXT_INVLPG_MULTI:
{
- cpumask_t pmask;
+ cpumask_t pmask;
if ( unlikely(vcpumask_to_pcpumask(d, op.arg2.vcpumask, &pmask)) )
{
@@ -2853,7 +2808,7 @@ int do_mmuext_op(
}
case MMUEXT_TLB_FLUSH_ALL:
- this_cpu(percpu_mm_info).deferred_ops |= DOP_FLUSH_ALL_TLBS;
+ flush_tlb_mask(&d->domain_dirty_cpumask);
break;
case MMUEXT_INVLPG_ALL:
@@ -2893,13 +2848,12 @@ int do_mmuext_op(
(curr->arch.guest_context.ldt_base != ptr) )
{
invalidate_shadow_ldt(curr, 0);
- this_cpu(percpu_mm_info).deferred_ops |= DOP_FLUSH_TLB;
+ flush_tlb_local();
curr->arch.guest_context.ldt_base = ptr;
curr->arch.guest_context.ldt_ents = ents;
load_LDT(curr);
- this_cpu(percpu_mm_info).deferred_ops &= ~DOP_RELOAD_LDT;
if ( ents != 0 )
- this_cpu(percpu_mm_info).deferred_ops |= DOP_RELOAD_LDT;
+ (void)map_ldt_shadow_page(0);
}
break;
}
@@ -2909,7 +2863,7 @@ int do_mmuext_op(
unsigned char *ptr;
okay = !get_page_and_type_from_pagenr(mfn, PGT_writable_page,
- FOREIGNDOM, 0, 0);
+ pg_owner, 0, 0);
if ( unlikely(!okay) )
{
MEM_LOG("Error while clearing mfn %lx", mfn);
@@ -2933,8 +2887,8 @@ int do_mmuext_op(
unsigned char *dst;
unsigned long src_mfn;
- src_mfn = gmfn_to_mfn(FOREIGNDOM, op.arg2.src_mfn);
- okay = get_page_from_pagenr(src_mfn, FOREIGNDOM);
+ src_mfn = gmfn_to_mfn(pg_owner, op.arg2.src_mfn);
+ okay = get_page_from_pagenr(src_mfn, pg_owner);
if ( unlikely(!okay) )
{
MEM_LOG("Error while copying from mfn %lx", src_mfn);
@@ -2942,7 +2896,7 @@ int do_mmuext_op(
}
okay = !get_page_and_type_from_pagenr(mfn, PGT_writable_page,
- FOREIGNDOM, 0, 0);
+ pg_owner, 0, 0);
if ( unlikely(!okay) )
{
put_page(mfn_to_page(src_mfn));
@@ -2985,7 +2939,7 @@ int do_mmuext_op(
__HYPERVISOR_mmuext_op, "hihi",
uops, (count - i) | MMU_UPDATE_PREEMPTED, pdone, foreigndom);
- process_deferred_ops();
+ put_pg_owner(pg_owner);
perfc_add(num_mmuext_ops, i);
@@ -3012,7 +2966,7 @@ int do_mmu_update(
struct page_info *page;
int rc = 0, okay = 1, i = 0;
unsigned int cmd, done = 0, pt_dom;
- struct domain *d = current->domain, *pt_owner = d;
+ struct domain *d = current->domain, *pt_owner = d, *pg_owner;
struct vcpu *v = current;
struct domain_mmap_cache mapcache;
@@ -3053,7 +3007,7 @@ int do_mmu_update(
}
}
- if ( !set_foreigndom((uint16_t)foreigndom) )
+ if ( (pg_owner = get_pg_owner((uint16_t)foreigndom)) == NULL )
{
rc = -ESRCH;
goto out;
@@ -3088,7 +3042,7 @@ int do_mmu_update(
*/
case MMU_NORMAL_PT_UPDATE:
case MMU_PT_UPDATE_PRESERVE_AD:
- rc = xsm_mmu_normal_update(d, FOREIGNDOM, req.val);
+ rc = xsm_mmu_normal_update(d, pg_owner, req.val);
if ( rc )
break;
@@ -3115,7 +3069,8 @@ int do_mmu_update(
{
l1_pgentry_t l1e = l1e_from_intpte(req.val);
okay = mod_l1_entry(va, l1e, mfn,
- cmd == MMU_PT_UPDATE_PRESERVE_AD, v);
+ cmd == MMU_PT_UPDATE_PRESERVE_AD, v,
+ pg_owner);
}
break;
case PGT_l2_page_table:
@@ -3174,13 +3129,13 @@ int do_mmu_update(
if ( rc )
break;
- if ( unlikely(!get_page_from_pagenr(mfn, FOREIGNDOM)) )
+ if ( unlikely(!get_page_from_pagenr(mfn, pg_owner)) )
{
MEM_LOG("Could not get page for mach->phys update");
break;
}
- if ( unlikely(paging_mode_translate(FOREIGNDOM)) )
+ if ( unlikely(paging_mode_translate(pg_owner)) )
{
MEM_LOG("Mach-phys update on auto-translate guest");
break;
@@ -3189,7 +3144,7 @@ int do_mmu_update(
set_gpfn_from_mfn(mfn, gpfn);
okay = 1;
- paging_mark_dirty(FOREIGNDOM, mfn);
+ paging_mark_dirty(pg_owner, mfn);
put_page(mfn_to_page(mfn));
break;
@@ -3215,7 +3170,7 @@ int do_mmu_update(
__HYPERVISOR_mmu_update, "hihi",
ureqs, (count - i) | MMU_UPDATE_PREEMPTED, pdone, foreigndom);
- process_deferred_ops();
+ put_pg_owner(pg_owner);
domain_mmap_cache_destroy(&mapcache);
@@ -3717,8 +3672,8 @@ int steal_page(
return -1;
}
-int do_update_va_mapping(unsigned long va, u64 val64,
- unsigned long flags)
+static int __do_update_va_mapping(
+ unsigned long va, u64 val64, unsigned long flags, struct domain *pg_owner)
{
l1_pgentry_t val = l1e_from_intpte(val64);
struct vcpu *v = current;
@@ -3731,7 +3686,7 @@ int do_update_va_mapping(unsigned long v
perfc_incr(calls_to_update_va);
- rc = xsm_update_va_mapping(d, FOREIGNDOM, val);
+ rc = xsm_update_va_mapping(d, pg_owner, val);
if ( rc )
return rc;
@@ -3754,7 +3709,7 @@ int do_update_va_mapping(unsigned long v
goto out;
}
- rc = mod_l1_entry(pl1e, val, gl1mfn, 0, v) ? 0 : -EINVAL;
+ rc = mod_l1_entry(pl1e, val, gl1mfn, 0, v, pg_owner) ? 0 : -EINVAL;
page_unlock(gl1pg);
put_page(gl1pg);
@@ -3769,32 +3724,24 @@ int do_update_va_mapping(unsigned long v
switch ( (bmap_ptr = flags & ~UVMF_FLUSHTYPE_MASK) )
{
case UVMF_LOCAL:
- this_cpu(percpu_mm_info).deferred_ops |= DOP_FLUSH_TLB;
+ flush_tlb_local();
break;
case UVMF_ALL:
- this_cpu(percpu_mm_info).deferred_ops |= DOP_FLUSH_ALL_TLBS;
+ flush_tlb_mask(&d->domain_dirty_cpumask);
break;
default:
- if ( this_cpu(percpu_mm_info).deferred_ops & DOP_FLUSH_ALL_TLBS )
- break;
rc = vcpumask_to_pcpumask(d, const_guest_handle_from_ptr(bmap_ptr,
void),
&pmask);
- if ( cpu_isset(smp_processor_id(), pmask) )
- this_cpu(percpu_mm_info).deferred_ops &= ~DOP_FLUSH_TLB;
flush_tlb_mask(&pmask);
break;
}
break;
case UVMF_INVLPG:
- if ( this_cpu(percpu_mm_info).deferred_ops & DOP_FLUSH_ALL_TLBS )
- break;
switch ( (bmap_ptr = flags & ~UVMF_FLUSHTYPE_MASK) )
{
case UVMF_LOCAL:
- if ( this_cpu(percpu_mm_info).deferred_ops & DOP_FLUSH_TLB )
- break;
if ( !paging_mode_enabled(d) ||
(paging_invlpg(v, va) != 0) )
flush_tlb_one_local(va);
@@ -3806,32 +3753,34 @@ int do_update_va_mapping(unsigned long v
rc = vcpumask_to_pcpumask(d, const_guest_handle_from_ptr(bmap_ptr,
void),
&pmask);
- if ( this_cpu(percpu_mm_info).deferred_ops & DOP_FLUSH_TLB )
- cpu_clear(smp_processor_id(), pmask);
flush_tlb_one_mask(&pmask, va);
break;
}
break;
}
- process_deferred_ops();
-
return rc;
+}
+
+int do_update_va_mapping(unsigned long va, u64 val64,
+ unsigned long flags)
+{
+ return __do_update_va_mapping(va, val64, flags, current->domain);
}
int do_update_va_mapping_otherdomain(unsigned long va, u64 val64,
unsigned long flags,
domid_t domid)
{
+ struct domain *pg_owner;
int rc;
- if ( !set_foreigndom(domid) )
+ if ( (pg_owner = get_pg_owner(domid)) == NULL )
return -ESRCH;
- rc = do_update_va_mapping(va, val64, flags);
-
- BUG_ON(this_cpu(percpu_mm_info).deferred_ops);
- process_deferred_ops(); /* only to clear foreigndom */
+ rc = __do_update_va_mapping(va, val64, flags, pg_owner);
+
+ put_pg_owner(pg_owner);
return rc;
}
_______________________________________________
Xen-changelog mailing list
Xen-changelog@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-changelog
|
 |
Lists.xenproject.org is hosted with RackSpace, monitoring our |