Patchwork [V2] xen-mapcache: Fix rlimit set size.

login
register
mail settings
Submitter Anthony PERARD
Date Aug. 4, 2011, 5:06 p.m.
Message ID <1312477562-15827-1-git-send-email-anthony.perard@citrix.com>
Download mbox | patch
Permalink /patch/108542/
State New
Headers show

Comments

Anthony PERARD - Aug. 4, 2011, 5:06 p.m.
Previously, the address space soft limit was set mcache_max_size. So,
before the mcache_max_size was reached by the mapcache, QEMU was killed
for overuse of the virtual address space.

This patch fix that by setting the soft limit the maximum than can have
QEMU. So the soft and hard limit are always set to RLIM_INFINITY if QEMU
is privileged.

In case QEMU is not run as root and the limit is too low, the maximum
mapcache size will be set the rlim_max - 80MB because observed that QEMU
use 75MB more than the maximum mapcache size after several empirical
tests.

Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>
---

Change:
  - This time, always set the rlimit to max or INFINITY if qemu is run as root.
  - Also, print a warning message if the rlimit can not be set to INFINITY.


 xen-mapcache.c |   27 ++++++++++++++++++++++-----
 1 files changed, 22 insertions(+), 5 deletions(-)
Stefano Stabellini - Aug. 12, 2011, 8:23 p.m.
On Thu, 4 Aug 2011, Anthony PERARD wrote:
> Previously, the address space soft limit was set mcache_max_size. So,
> before the mcache_max_size was reached by the mapcache, QEMU was killed
> for overuse of the virtual address space.
> 
> This patch fix that by setting the soft limit the maximum than can have
> QEMU. So the soft and hard limit are always set to RLIM_INFINITY if QEMU
> is privileged.
> 
> In case QEMU is not run as root and the limit is too low, the maximum
> mapcache size will be set the rlim_max - 80MB because observed that QEMU
> use 75MB more than the maximum mapcache size after several empirical
> tests.
> 
> Signed-off-by: Anthony PERARD <anthony.perard@citrix.com>

it looks good

Patch

diff --git a/xen-mapcache.c b/xen-mapcache.c
index 007136a..f698876 100644
--- a/xen-mapcache.c
+++ b/xen-mapcache.c
@@ -40,6 +40,13 @@ 
 #endif
 #define MCACHE_BUCKET_SIZE (1UL << MCACHE_BUCKET_SHIFT)
 
+/* This is the size of the virtual address space reserve to QEMU that will not
+ * be use by MapCache.
+ * From empirical tests I observed that qemu use 75MB more than the
+ * max_mcache_size.
+ */
+#define NON_MCACHE_MEMORY_SIZE (80 * 1024 * 1024)
+
 #define mapcache_lock()   ((void)0)
 #define mapcache_unlock() ((void)0)
 
@@ -92,15 +99,25 @@  void xen_map_cache_init(void)
     QTAILQ_INIT(&mapcache->locked_entries);
     mapcache->last_address_index = -1;
 
-    getrlimit(RLIMIT_AS, &rlimit_as);
-    if (rlimit_as.rlim_max < MCACHE_MAX_SIZE) {
-        rlimit_as.rlim_cur = rlimit_as.rlim_max;
+    if (geteuid() == 0) {
+        rlimit_as.rlim_cur = RLIM_INFINITY;
+        rlimit_as.rlim_max = RLIM_INFINITY;
+        mapcache->max_mcache_size = MCACHE_MAX_SIZE;
     } else {
-        rlimit_as.rlim_cur = MCACHE_MAX_SIZE;
+        getrlimit(RLIMIT_AS, &rlimit_as);
+        rlimit_as.rlim_cur = rlimit_as.rlim_max;
+
+        if (rlimit_as.rlim_max != RLIM_INFINITY) {
+            fprintf(stderr, "Warning: QEMU's maximum size of virtual memory is not infinity.\n");
+        }
+        if (rlimit_as.rlim_max < MCACHE_MAX_SIZE + NON_MCACHE_MEMORY_SIZE) {
+            mapcache->max_mcache_size = rlimit_as.rlim_max - NON_MCACHE_MEMORY_SIZE;
+        } else {
+            mapcache->max_mcache_size = MCACHE_MAX_SIZE;
+        }
     }
 
     setrlimit(RLIMIT_AS, &rlimit_as);
-    mapcache->max_mcache_size = rlimit_as.rlim_cur;
 
     mapcache->nr_buckets =
         (((mapcache->max_mcache_size >> XC_PAGE_SHIFT) +