All of lore.kernel.org
 help / color / mirror / Atom feed
From: Boris Ostrovsky <boris.ostrovsky@oracle.com>
To: xen-devel@lists.xen.org
Cc: sstabellini@kernel.org, wei.liu2@citrix.com,
	George.Dunlap@eu.citrix.com, andrew.cooper3@citrix.com,
	ian.jackson@eu.citrix.com, tim@xen.org, julien.grall@arm.com,
	jbeulich@suse.com, Boris Ostrovsky <boris.ostrovsky@oracle.com>
Subject: [PATCH v7 3/9] mm: Extract allocation loop from alloc_heap_pages()
Date: Tue,  8 Aug 2017 17:45:01 -0400	[thread overview]
Message-ID: <1502228707-31883-4-git-send-email-boris.ostrovsky@oracle.com> (raw)
In-Reply-To: <1502228707-31883-1-git-send-email-boris.ostrovsky@oracle.com>

This will make code a bit more readable, especially with changes that
will be introduced in subsequent patches.

Signed-off-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Acked-by: Jan Beulich <jbeulich@suse.com>
---
 xen/common/page_alloc.c | 139 +++++++++++++++++++++++++++---------------------
 1 file changed, 77 insertions(+), 62 deletions(-)

diff --git a/xen/common/page_alloc.c b/xen/common/page_alloc.c
index b857a44..5920e87 100644
--- a/xen/common/page_alloc.c
+++ b/xen/common/page_alloc.c
@@ -697,22 +697,15 @@ static void page_list_add_scrub(struct page_info *pg, unsigned int node,
         page_list_add(pg, &heap(node, zone, order));
 }
 
-/* Allocate 2^@order contiguous pages. */
-static struct page_info *alloc_heap_pages(
-    unsigned int zone_lo, unsigned int zone_hi,
-    unsigned int order, unsigned int memflags,
-    struct domain *d)
+static struct page_info *get_free_buddy(unsigned int zone_lo,
+                                        unsigned int zone_hi,
+                                        unsigned int order, unsigned int memflags,
+                                        const struct domain *d)
 {
-    unsigned int i, j, zone = 0, nodemask_retry = 0, first_dirty;
     nodeid_t first_node, node = MEMF_get_node(memflags), req_node = node;
-    unsigned long request = 1UL << order;
+    nodemask_t nodemask = d ? d->node_affinity : node_online_map;
+    unsigned int j, zone, nodemask_retry = 0;
     struct page_info *pg;
-    nodemask_t nodemask = (d != NULL ) ? d->node_affinity : node_online_map;
-    bool_t need_tlbflush = 0;
-    uint32_t tlbflush_timestamp = 0;
-
-    /* Make sure there are enough bits in memflags for nodeID. */
-    BUILD_BUG_ON((_MEMF_bits - _MEMF_node) < (8 * sizeof(nodeid_t)));
 
     if ( node == NUMA_NO_NODE )
     {
@@ -728,34 +721,6 @@ static struct page_info *alloc_heap_pages(
     first_node = node;
 
     ASSERT(node < MAX_NUMNODES);
-    ASSERT(zone_lo <= zone_hi);
-    ASSERT(zone_hi < NR_ZONES);
-
-    if ( unlikely(order > MAX_ORDER) )
-        return NULL;
-
-    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()) &&
-          ((memflags & MEMF_no_refcount) ||
-           !d || 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
-     * post-dom0-creation-multi-page allocations can be eliminated.
-     */
-    if ( ((order == 0) || (order >= 9)) &&
-         (total_avail_pages <= midsize_alloc_zone_pages) &&
-         tmem_freeable_pages() )
-        goto try_tmem;
 
     /*
      * Start with requested node, but exhaust all node memory in requested 
@@ -767,17 +732,17 @@ static struct page_info *alloc_heap_pages(
         zone = zone_hi;
         do {
             /* Check if target node can support the allocation. */
-            if ( !avail[node] || (avail[node][zone] < request) )
+            if ( !avail[node] || (avail[node][zone] < (1UL << order)) )
                 continue;
 
             /* Find smallest order which can satisfy the request. */
             for ( j = order; j <= MAX_ORDER; j++ )
                 if ( (pg = page_list_remove_head(&heap(node, zone, j))) )
-                    goto found;
+                    return pg;
         } while ( zone-- > zone_lo ); /* careful: unsigned zone may wrap */
 
         if ( (memflags & MEMF_exact_node) && req_node != NUMA_NO_NODE )
-            goto not_found;
+            return NULL;
 
         /* Pick next node. */
         if ( !node_isset(node, nodemask) )
@@ -794,46 +759,96 @@ static struct page_info *alloc_heap_pages(
         {
             /* When we have tried all in nodemask, we fall back to others. */
             if ( (memflags & MEMF_exact_node) || nodemask_retry++ )
-                goto not_found;
+                return NULL;
             nodes_andnot(nodemask, node_online_map, nodemask);
             first_node = node = first_node(nodemask);
             if ( node >= MAX_NUMNODES )
-                goto not_found;
+                return NULL;
         }
     }
+}
+
+/* Allocate 2^@order contiguous pages. */
+static struct page_info *alloc_heap_pages(
+    unsigned int zone_lo, unsigned int zone_hi,
+    unsigned int order, unsigned int memflags,
+    struct domain *d)
+{
+    nodeid_t node;
+    unsigned int i, buddy_order, zone, first_dirty;
+    unsigned long request = 1UL << order;
+    struct page_info *pg;
+    bool need_tlbflush = false;
+    uint32_t tlbflush_timestamp = 0;
+
+    /* Make sure there are enough bits in memflags for nodeID. */
+    BUILD_BUG_ON((_MEMF_bits - _MEMF_node) < (8 * sizeof(nodeid_t)));
+
+    ASSERT(zone_lo <= zone_hi);
+    ASSERT(zone_hi < NR_ZONES);
+
+    if ( unlikely(order > MAX_ORDER) )
+        return NULL;
+
+    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()) &&
+          ((memflags & MEMF_no_refcount) ||
+           !d || d->outstanding_pages < request) )
+    {
+        spin_unlock(&heap_lock);
+        return NULL;
+    }
 
- try_tmem:
-    /* Try to free memory from tmem */
-    if ( (pg = tmem_relinquish_pages(order, memflags)) != NULL )
+    /*
+     * 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
+     * post-dom0-creation-multi-page allocations can be eliminated.
+     */
+    if ( ((order == 0) || (order >= 9)) &&
+         (total_avail_pages <= midsize_alloc_zone_pages) &&
+         tmem_freeable_pages() )
     {
-        /* reassigning an already allocated anonymous heap page */
+        /* Try to free memory from tmem. */
+        pg = tmem_relinquish_pages(order, memflags);
         spin_unlock(&heap_lock);
         return pg;
     }
 
- not_found:
-    /* No suitable memory blocks. Fail the request. */
-    spin_unlock(&heap_lock);
-    return NULL;
+    pg = get_free_buddy(zone_lo, zone_hi, order, memflags, d);
+    if ( !pg )
+    {
+        /* No suitable memory blocks. Fail the request. */
+        spin_unlock(&heap_lock);
+        return NULL;
+    }
 
- found: 
+    node = phys_to_nid(page_to_maddr(pg));
+    zone = page_to_zone(pg);
+    buddy_order = PFN_ORDER(pg);
 
     first_dirty = pg->u.free.first_dirty;
 
     /* We may have to halve the chunk a number of times. */
-    while ( j != order )
+    while ( buddy_order != order )
     {
-        j--;
-        page_list_add_scrub(pg, node, zone, j,
-                            (1U << j) > first_dirty ?
+        buddy_order--;
+        page_list_add_scrub(pg, node, zone, buddy_order,
+                            (1U << buddy_order) > first_dirty ?
                             first_dirty : INVALID_DIRTY_IDX);
-        pg += 1U << j;
+        pg += 1U << buddy_order;
 
         if ( first_dirty != INVALID_DIRTY_IDX )
         {
             /* Adjust first_dirty */
-            if ( first_dirty >= 1U << j )
-                first_dirty -= 1U << j;
+            if ( first_dirty >= 1U << buddy_order )
+                first_dirty -= 1U << buddy_order;
             else
                 first_dirty = 0; /* We've moved past original first_dirty */
         }
-- 
1.8.3.1


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

  parent reply	other threads:[~2017-08-08 21:45 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-08-08 21:44 [PATCH v7 0/9] Memory scrubbing from idle loop Boris Ostrovsky
2017-08-08 21:44 ` [PATCH v7 1/9] mm: Clean up free_heap_pages() Boris Ostrovsky
2017-08-10 12:21   ` Wei Liu
2017-08-14 10:30   ` Jan Beulich
2017-08-08 21:45 ` [PATCH v7 2/9] mm: Place unscrubbed pages at the end of pagelist Boris Ostrovsky
2017-08-14 10:37   ` Jan Beulich
2017-08-14 14:29     ` Boris Ostrovsky
2017-08-15  8:18       ` Jan Beulich
2017-08-15 14:41         ` Boris Ostrovsky
2017-08-15 14:51           ` Jan Beulich
2017-08-15 14:52             ` Julien Grall
2017-08-15 15:03               ` Boris Ostrovsky
2017-08-15 15:08                 ` Jan Beulich
2017-08-14 11:16   ` Julien Grall
2017-08-08 21:45 ` Boris Ostrovsky [this message]
2017-08-08 21:45 ` [PATCH v7 4/9] mm: Scrub pages in alloc_heap_pages() if needed Boris Ostrovsky
2017-08-08 21:45 ` [PATCH v7 5/9] mm: Scrub memory from idle loop Boris Ostrovsky
2017-08-08 21:45 ` [PATCH v7 6/9] spinlock: Introduce spin_lock_cb() Boris Ostrovsky
2017-08-14 11:22   ` Julien Grall
2017-08-14 14:39     ` Boris Ostrovsky
2017-08-14 14:42       ` Julien Grall
2017-08-14 14:57         ` Boris Ostrovsky
2017-08-08 21:45 ` [PATCH v7 7/9] mm: Keep heap accessible to others while scrubbing Boris Ostrovsky
2017-08-14 10:38   ` Jan Beulich
2017-08-08 21:45 ` [PATCH v7 8/9] mm: Print number of unscrubbed pages in 'H' debug handler Boris Ostrovsky
2017-08-08 21:45 ` [PATCH v7 9/9] mm: Make sure pages are scrubbed Boris Ostrovsky

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=1502228707-31883-4-git-send-email-boris.ostrovsky@oracle.com \
    --to=boris.ostrovsky@oracle.com \
    --cc=George.Dunlap@eu.citrix.com \
    --cc=andrew.cooper3@citrix.com \
    --cc=ian.jackson@eu.citrix.com \
    --cc=jbeulich@suse.com \
    --cc=julien.grall@arm.com \
    --cc=sstabellini@kernel.org \
    --cc=tim@xen.org \
    --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.