Bug 1379139 - Instead of not recycling huge chunks, zero them. r=njn

It turns out that not recycling some kinds of chunk can lead to the
recycle queue being starved in some scenarios. When that happens, we end
up mmap()ing new memory, but that turns out to be significantly slower.

So instead of not recycling huge chunks, we force-clean them, before
madvising so that the pages can still be reclaimed in case of memory
pressure.

--HG--
extra : rebase_source : 2dbd028daca92c9cd7c8079eb3dc5a0cfa06495b

memory/mozjemalloc/mozjemalloc.cpp

@@ -1924,7 +1924,7 @@ chunk_alloc_mmap(size_t size, size_t alignment)
 }
 
 bool
-pages_purge(void *addr, size_t length)
+pages_purge(void *addr, size_t length, bool force_zero)
 {
 	bool unzeroed;
 
@@ -1932,6 +1932,10 @@ pages_purge(void *addr, size_t length)
 	pages_decommit(addr, length);
 	unzeroed = false;
 #else
+#  ifndef MOZ_MEMORY_LINUX
+	if (force_zero)
+		memset(addr, 0, length);
+#  endif
 #  ifdef MOZ_MEMORY_WINDOWS
 	/*
 	* The region starting at addr may have been allocated in multiple calls
@@ -1947,14 +1951,14 @@ pages_purge(void *addr, size_t length)
 		length -= pages_size;
 		pages_size = std::min(length, chunksize);
 	}
-	unzeroed = true;
+	unzeroed = !force_zero;
 #  else
 #    ifdef MOZ_MEMORY_LINUX
 #      define JEMALLOC_MADV_PURGE MADV_DONTNEED
 #      define JEMALLOC_MADV_ZEROS true
 #    else /* FreeBSD and Darwin. */
 #      define JEMALLOC_MADV_PURGE MADV_FREE
-#      define JEMALLOC_MADV_ZEROS false
+#      define JEMALLOC_MADV_ZEROS force_zero
 #    endif
 	int err = madvise(addr, length, JEMALLOC_MADV_PURGE);
 	unzeroed = (JEMALLOC_MADV_ZEROS == false || err != 0);
@@ -2121,13 +2125,7 @@ chunk_record(extent_tree_t *chunks_szad, extent_tree_t *chunks_ad, void *chunk,
 	bool unzeroed;
 	extent_node_t *xnode, *node, *prev, *xprev, key;
 
-	unzeroed = pages_purge(chunk, size);
-
-	/* If purge doesn't zero the chunk, only record arena chunks or
-	 * previously recycled chunks. */
-	if (unzeroed && type != ARENA_CHUNK && type != RECYCLED_CHUNK) {
-		return;
-	}
+	unzeroed = pages_purge(chunk, size, type == HUGE_CHUNK);
 
 	/*
 	 * Allocate a node before acquiring chunks_mtx even though it might not