[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Xen-devel] [PATCH 1/6] mmu: Introduce XENMEM_claim_pages (subop of memory ops).



From: Dan Magenheimer <dan.magenheimer@xxxxxxxxxx>

When guests memory consumption is volatile (multiple guests
ballooning up/down) we are presented with the problem of
being able to determine exactly how much memory there is
for allocation of new guests without negatively impacting
existing guests. Note that the existing models (xapi, xend)
drive the memory consumption from the tool-stack and assume
that the guest will eventually hit the memory target. Other
models, such as the dynamic memory utilized by tmem, do this
differently - the guest drivers the memory consumption (up
to the d->max_pages ceiling). With dynamic memory model, the
guest frequently can balloon up and down as it sees fit.
This presents the problem to the toolstack that it does not
know atomically how much free memory there is (as the information
gets stale the moment the d->tot_pages information is provided
to the tool-stack), and hence when starting a guest can fail
during the memory creation process. Especially if the process
is done in parallel. In a nutshell what we need is a atomic
value of all domains tot_pages during the allocation of guests.
Naturally holding a lock for such a long time is unacceptable.
Hence the goal of this hypercall is to attempt to atomically and very
quickly determine if there are sufficient pages available in the
system and, if so, "set aside" that quantity of pages for future
allocations by that domain.  Unlike an existing hypercall such as
increase_reservation or populate_physmap, specific physical
pageframes are not assigned to the domain because this
cannot be done sufficiently quickly (especially for very large
allocations in an arbitrarily fragmented system) and so the
existing mechanisms result in classic time-of-check-time-of-use
(TOCTOU) races.  One can think of claiming as similar to a
"lazy" allocation, but subsequent hypercalls are required
to do the actual physical pageframe allocation.

Note that one of effects of this hypercall is that from the
perspective of other running guests -  suddenly there is
a new guest occupying X amount of pages. This means that when
we try to balloon up they will hit the system-wide ceiling of
available free memory (if the total sum of the existing d->max_pages
>= host memory). This is OK - as that is part of the overcommit.
What we DO NOT want to do is dictate their ceiling should be
(d->max_pages) as that is risky and can lead to guests OOM-ing.
It is something the guest needs to figure out.

In order for a toolstack to "get" information about whether
a domain has a claim and, if so, how large, and also for
the toolstack to measure the total system-wide claim, a
second subop has been added and exposed through domctl
and libxl (see "xen: XENMEM_claim_pages: xc").

== Alternative solutions ==
There has been a variety of discussion whether the problem
hypercall is solving can be done in user-space, such as:
 - For all the existing guest, set their d->max_pages temporarily
   to d->tot_pages and create the domain. This forces those
   domains to stay at their current consumption level (fyi, this is what
   the tmem freeze call is doing). The disadvantage of this is
   that needlessly forces the guests to stay at the memory usage
   instead of allowing it to decide the optimal target.
 - Account only using d->max_pages of how much free memory there is.
   This ignores ballooning changes and any over-commit scenario. This
   is similar to the scenario where the sum of all d->max_pages (and
   the one to be allocated now) on the host is smaller than the available
   free memory. As such it ignores the over-commit problem.
 - Provide a ring/FIFO along with event channel to notify an userspace
   daemon of guests memory consumption. This daemon can then provide
   up-to-date information to the toolstack of how much free memory
   there is. This duplicates what the hypervisor is already doing and
   introduced latency issues and catching breath for the toolstack as there
   might be millions of these updates on heavily used machine. There might
   not be any quiescent state ever and the toolstack will heavily consume
   CPU cycles and not ever provide up-to-date information.

It has been noted that this claim mechanism solves the
underlying problem (slow failure of domain creation) for
a large class of domains but not all, specifically not
handling (but also not making the problem worse for) PV
domains that specify the "superpages" flag, and 32-bit PV
domains on large RAM systems.  These will be addressed at a
later time.

Code overview:

Though the hypercall simply does arithmetic within locks,
some of the semantics in the code may be a bit subtle.

The key variables (d->unclaimed_pages and total_unclaimed_pages)
starts at zero if no claim has yet been staked for any domain.
(Perhaps a better name is "claimed_but_not_yet_possessed" but that's
a bit unwieldy.)  If no claim hypercalls are executed, there
should be no impact on existing usage.

When a claim is successfully staked by a domain, it is like a
watermark but there is no record kept of the size of the claim.
Instead, d->unclaimed_pages is set to the difference between
d->tot_pages and the claim.  When d->tot_pages increases or decreases,
d->unclaimed_pages atomically decreases or increases.  Once
d->unclaimed_pages reaches zero, the claim is satisfied and
d->unclaimed pages stays at zero -- unless a new claim is
subsequently staked.

The systemwide variable total_unclaimed_pages is always the sum
of d->unclaimed_pages, across all domains.  A non-domain-
specific heap allocation will fail if total_unclaimed_pages
exceeds free (plus, on tmem enabled systems, freeable) pages.

Claim semantics could be modified by flags.  The initial
implementation had three flag, which discerns whether the
caller would like tmem freeable pages to be considered
in determining whether or not the claim can be successfully
staked. This in later patches was removed and there are no
flags.

A claim can be cancelled by requesting a claim with the
number of pages being zero.

A second subop returns the total outstanding claimed pages
systemwide.

Note: Save/restore/migrate may need to be modified,
else it can be documented that all claims are cancelled.

==
This patch of the proposed XENMEM_claim_pages hypercall/subop, takes
into account review feedback from Jan and Keir and IanC and Matthew Daley,
plus some fixes found via runtime debugging.

v12:
 - Dropped flags per Time suggestion.

v11:
 - Altered names and some of the code assumptions per Tim
 - Added Acked-by Keir
v9->v10:
 - Altered the API a bit (new flags, etc).

v8->v9:
- Drop snippets already merged by Keir
  (26269:21a5b181f8ad and 26270:03cb71bc32f9)

v7->v8:
- Replace missing tot_pages adjustment [JBeulich]
- Combine domain_inc/dec_tot_pages into one function [JBeulich]
- Added a few more clarifying comments [djm]

v6->v7:
- Need co-existence with low_mem_virq [JBeulich]

v5->v6:
- Fix post-inc problem [mattjd]

v4->v5:
- Split tools part into separate patch [JBeulich]
- Minor coding style fixups [JBeulich]
- Use rcu_lock_domain_by_id, not _target version [JBeulich]

v3->v4: (please ignore v3)
- Process error, sent stale patch, sorry [djm]

v2->v3:
- Add per-domain and global "get" for unclaimed info [JBeulich]
- New hypercall(s) should fail for unpriv callers [IanC]
- Non-zero extent_order disallowed [JBeulich]
- Remove bonehead ASSERTs [JBeulich]
- Need not hold heaplock for decrease case too [JBeulich]
- More descriptive failure return values [JBeulich]
- Don't allow a claim to exceed max_pages [IanC]
- Subops must be in correct ifdef block in memory.c [keir]

v1->v2:
- Add reset-to-zero page claim in domain_kill [JBeulich]
- Proper handling of struct passed to hypercall [JBeulich]
- Fix alloc_heap_pages when a domain has a claim [JBeulich]
- Need not hold heap_lock if !d->unclaimed_pages [keir]
- Fix missed tot_pages call in donate_page [djm]
- Remove domain_reset_unclaimed_pages; use set with zero [djm]
- Bugfixes found through testing in set_unclaimed [djm]
- More comments in code [djm]
- Code formatting fixes [djm]
===

Acked-by: Tim Deegan <tim@xxxxxxx>
Acked-by: Keir Fraser <keir@xxxxxxx>
Signed-off-by: Dan Magenheimer <dan.magenheimer@xxxxxxxxxx>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
---
 xen/common/domain.c         |   1 +
 xen/common/domctl.c         |   1 +
 xen/common/memory.c         |  33 +++++++++++++
 xen/common/page_alloc.c     | 116 +++++++++++++++++++++++++++++++++++++++++++-
 xen/include/public/domctl.h |   3 +-
 xen/include/public/memory.h |  35 ++++++++++++-
 xen/include/xen/mm.h        |   3 ++
 xen/include/xen/sched.h     |   1 +
 8 files changed, 188 insertions(+), 5 deletions(-)

diff --git a/xen/common/domain.c b/xen/common/domain.c
index b360de1..64ee29d 100644
--- a/xen/common/domain.c
+++ b/xen/common/domain.c
@@ -507,6 +507,7 @@ int domain_kill(struct domain *d)
         evtchn_destroy(d);
         gnttab_release_mappings(d);
         tmem_destroy(d->tmem);
+        domain_set_outstanding_pages(d, 0);
         d->tmem = NULL;
         /* fallthrough */
     case DOMDYING_dying:
diff --git a/xen/common/domctl.c b/xen/common/domctl.c
index b7f6619..c98e99c 100644
--- a/xen/common/domctl.c
+++ b/xen/common/domctl.c
@@ -154,6 +154,7 @@ void getdomaininfo(struct domain *d, struct 
xen_domctl_getdomaininfo *info)
 
     info->tot_pages         = d->tot_pages;
     info->max_pages         = d->max_pages;
+    info->outstanding_pages = d->outstanding_pages;
     info->shr_pages         = atomic_read(&d->shr_pages);
     info->paged_pages       = atomic_read(&d->paged_pages);
     info->shared_info_frame = mfn_to_gmfn(d, virt_to_mfn(d->shared_info));
diff --git a/xen/common/memory.c b/xen/common/memory.c
index 08550ef..3cfc1e3 100644
--- a/xen/common/memory.c
+++ b/xen/common/memory.c
@@ -712,6 +712,39 @@ long do_memory_op(unsigned long cmd, 
XEN_GUEST_HANDLE_PARAM(void) arg)
         break;
     }
 
+    case XENMEM_claim_pages:
+        if ( !IS_PRIV(current->domain) )
+            return -EPERM;
+
+        if ( copy_from_guest(&reservation, arg, 1) )
+            return -EFAULT;
+
+        if ( !guest_handle_is_null(reservation.extent_start) )
+            return -EINVAL;
+
+        if ( reservation.extent_order != 0 )
+            return -EINVAL;
+
+        if ( reservation.mem_flags != 0 )
+            return -EINVAL;
+
+        d = rcu_lock_domain_by_id(reservation.domid);
+        if ( d == NULL )
+            return -EINVAL;
+
+        rc = domain_set_outstanding_pages(d, reservation.nr_extents);
+
+        rcu_unlock_domain(d);
+
+        break;
+
+    case XENMEM_get_outstanding_pages:
+        if ( !IS_PRIV(current->domain) )
+            return -EPERM;
+
+        rc = get_outstanding_claims();
+        break;
+
     default:
         rc = arch_memory_op(op, arg);
         break;
diff --git a/xen/common/page_alloc.c b/xen/common/page_alloc.c
index 9e9fb15..aefef29 100644
--- a/xen/common/page_alloc.c
+++ b/xen/common/page_alloc.c
@@ -252,11 +252,114 @@ static long midsize_alloc_zone_pages;
 #define MIDSIZE_ALLOC_FRAC 128
 
 static DEFINE_SPINLOCK(heap_lock);
+static long outstanding_claims; /* total outstanding claims by all domains */
 
 unsigned long domain_adjust_tot_pages(struct domain *d, long pages)
 {
+    long dom_before, dom_after, dom_claimed, sys_before, sys_after;
+
     ASSERT(spin_is_locked(&d->page_alloc_lock));
-    return d->tot_pages += pages;
+    d->tot_pages += pages;
+
+    /*
+     * can test d->claimed_pages race-free because it can only change
+     * if d->page_alloc_lock and heap_lock are both held, see also
+     * domain_set_outstanding_pages below
+     */
+    if ( !d->outstanding_pages )
+        goto out;
+
+    spin_lock(&heap_lock);
+    /* adjust domain outstanding pages; may not go negative */
+    dom_before = d->outstanding_pages;
+    dom_after = dom_before - pages;
+    BUG_ON(dom_before < 0);
+    dom_claimed = dom_after < 0 ? 0 : dom_after;
+    d->outstanding_pages = dom_claimed;
+    /* flag accounting bug if system outstanding_claims would go negative */
+    sys_before = outstanding_claims;
+    sys_after = sys_before - (dom_before - dom_claimed);
+    BUG_ON(sys_after < 0);
+    outstanding_claims = sys_after;
+    spin_unlock(&heap_lock);
+
+out:
+    return d->tot_pages;
+}
+
+int domain_set_outstanding_pages(struct domain *d, unsigned long pages)
+{
+    int ret = -ENOMEM;
+    unsigned long claim, avail_pages;
+
+    /*
+     * take the domain's page_alloc_lock, else all d->tot_page adjustments
+     * must always take the global heap_lock rather than only in the much
+     * rarer case that d->outstanding_pages is non-zero
+     */
+    spin_lock(&d->page_alloc_lock);
+    spin_lock(&heap_lock);
+
+    /* pages==0 means "unset" the claim. */
+    if ( pages == 0 )
+    {
+        outstanding_claims -= d->outstanding_pages;
+        d->outstanding_pages = 0;
+        ret = 0;
+        goto out;
+    }
+
+    /* only one active claim per domain please */
+    if ( d->outstanding_pages )
+    {
+        ret = -EINVAL;
+        goto out;
+    }
+
+    /* disallow a claim not exceeding current tot_pages or above max_pages */
+    if ( (pages <= d->tot_pages) || (pages > d->max_pages) )
+    {
+        ret = -EINVAL;
+        goto out;
+    }
+
+    /* how much memory is available? */
+    avail_pages = total_avail_pages;
+
+    /* Note: The usage of claim means that allocation from a guest *might*
+     * have to come from freeable memory. Using free memory is always better, 
if
+     * it is available, than using freeable memory.
+     *
+     * But that is OK as once the claim has been made, it still can take 
minutes
+     * before the claim is fully satisfied. Tmem can make use of the unclaimed
+     * pages during this time (to store ephemeral/freeable pages only,
+     * not persistent pages).
+     */
+    avail_pages += tmem_freeable_pages();
+    avail_pages -= outstanding_claims;
+
+    /*
+     * Note, if domain has already allocated memory before making a claim
+     * then the claim must take tot_pages into account
+     */
+    claim = pages - d->tot_pages;
+    if ( claim > avail_pages )
+        goto out;
+
+    /* yay, claim fits in available memory, stake the claim, success! */
+    d->outstanding_pages = claim;
+    outstanding_claims += d->outstanding_pages;
+    ret = 0;
+
+out:
+    spin_unlock(&heap_lock);
+    spin_unlock(&d->page_alloc_lock);
+    return ret;
+}
+
+long get_outstanding_claims(void)
+{
+    return outstanding_claims;
 }
 
 static unsigned long init_node_heap(int node, unsigned long mfn,
@@ -397,7 +500,7 @@ static void __init setup_low_mem_virq(void)
 static void check_low_mem_virq(void)
 {
     unsigned long avail_pages = total_avail_pages +
-        (opt_tmem ? tmem_freeable_pages() : 0);
+        (opt_tmem ? tmem_freeable_pages() : 0) - outstanding_claims;
 
     if ( unlikely(avail_pages <= low_mem_virq_th) )
     {
@@ -466,6 +569,15 @@ static struct page_info *alloc_heap_pages(
     spin_lock(&heap_lock);
 
     /*
+     * Claimed memory is considered unavailable unless the request
+     * is made by a domain with sufficient unclaimed pages.
+     */
+    if ( (outstanding_claims + request >
+          total_avail_pages + tmem_freeable_pages()) &&
+          (d == NULL || d->outstanding_pages < request) )
+        goto not_found;
+
+    /*
      * TMEM: When available memory is scarce due to tmem absorbing it, allow
      * only mid-size allocations to avoid worst of fragmentation issues.
      * Others try tmem pools then fail.  This is a workaround until all
diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
index deb19db..113b8dc 100644
--- a/xen/include/public/domctl.h
+++ b/xen/include/public/domctl.h
@@ -36,7 +36,7 @@
 #include "grant_table.h"
 #include "hvm/save.h"
 
-#define XEN_DOMCTL_INTERFACE_VERSION 0x00000008
+#define XEN_DOMCTL_INTERFACE_VERSION 0x00000009
 
 /*
  * NB. xen_domctl.domain is an IN/OUT parameter for this operation.
@@ -95,6 +95,7 @@ struct xen_domctl_getdomaininfo {
     uint32_t flags;              /* XEN_DOMINF_* */
     uint64_aligned_t tot_pages;
     uint64_aligned_t max_pages;
+    uint64_aligned_t outstanding_pages;
     uint64_aligned_t shr_pages;
     uint64_aligned_t paged_pages;
     uint64_aligned_t shared_info_frame; /* GMFN of shared_info struct */
diff --git a/xen/include/public/memory.h b/xen/include/public/memory.h
index 1c5ca19..51d5254 100644
--- a/xen/include/public/memory.h
+++ b/xen/include/public/memory.h
@@ -68,6 +68,8 @@ struct xen_memory_reservation {
      *   IN:  GPFN bases of extents to populate with memory
      *   OUT: GMFN bases of extents that were allocated
      *   (NB. This command also updates the mach_to_phys translation table)
+     * XENMEM_claim_pages:
+     *   IN: must be zero
      */
     XEN_GUEST_HANDLE(xen_pfn_t) extent_start;
 
@@ -430,10 +432,39 @@ typedef struct xen_mem_sharing_op xen_mem_sharing_op_t;
 DEFINE_XEN_GUEST_HANDLE(xen_mem_sharing_op_t);
 
 /*
- * Reserve ops for future/out-of-tree "claim" patches (Oracle)
+ * Attempt to stake a claim for a domain on a quantity of pages
+ * of system RAM, but _not_ assign specific pageframes.  Only
+ * arithmetic is performed so the hypercall is very fast and need
+ * not be preemptible, thus sidestepping time-of-check-time-of-use
+ * races for memory allocation.  Returns 0 if the hypervisor page
+ * allocator has atomically and successfully claimed the requested
+ * number of pages, else non-zero.
+ *
+ * Any domain may have only one active claim.  When sufficient memory
+ * has been allocated to resolve the claim, the claim silently expires.
+ * Claiming zero pages effectively resets any outstanding claim and
+ * is always successful.
+ *
+ * Note that a valid claim may be staked even after memory has been
+ * allocated for a domain.  In this case, the claim is not incremental,
+ * i.e. if the domain's tot_pages is 3, and a claim is staked for 10,
+ * only 7 additional pages are claimed.
+ *
+ * Caller must be privileged or the hypercall fails.
  */
 #define XENMEM_claim_pages                  24
-#define XENMEM_get_unclaimed_pages          25
+
+/*
+ * XENMEM_claim_pages flags - the are no flags at this time.
+ * The zero value is appropiate.
+ */
+
+/*
+ * Get the number of pages currently claimed (but not yet "possessed")
+ * across all domains.  The caller must be privileged but otherwise
+ * the call never fails.
+ */
+#define XENMEM_get_outstanding_pages        25
 
 #endif /* defined(__XEN__) || defined(__XEN_TOOLS__) */
 
diff --git a/xen/include/xen/mm.h b/xen/include/xen/mm.h
index 2f701f5..28512fb 100644
--- a/xen/include/xen/mm.h
+++ b/xen/include/xen/mm.h
@@ -49,7 +49,10 @@ void free_xenheap_pages(void *v, unsigned int order);
 #define alloc_xenheap_page() (alloc_xenheap_pages(0,0))
 #define free_xenheap_page(v) (free_xenheap_pages(v,0))
 
+/* Claim handling */
 unsigned long domain_adjust_tot_pages(struct domain *d, long pages);
+int domain_set_outstanding_pages(struct domain *d, unsigned long pages);
+long get_outstanding_claims(void);
 
 /* Domain suballocator. These functions are *not* interrupt-safe.*/
 void init_domheap_pages(paddr_t ps, paddr_t pe);
diff --git a/xen/include/xen/sched.h b/xen/include/xen/sched.h
index e108436..569e76e 100644
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -242,6 +242,7 @@ struct domain
     struct page_list_head page_list;  /* linked list */
     struct page_list_head xenpage_list; /* linked list (size xenheap_pages) */
     unsigned int     tot_pages;       /* number of pages currently possesed */
+    unsigned int     outstanding_pages; /* pages claimed but not possessed  */
     unsigned int     max_pages;       /* maximum value for tot_pages        */
     atomic_t         shr_pages;       /* number of shared pages             */
     atomic_t         paged_pages;     /* number of paged-out pages          */
-- 
1.8.0.2


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


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.