From 279038f15981bceccd67a18c8c36c44751a68d92 Mon Sep 17 00:00:00 2001
From: Douglas Katzman <dougk@google.com>
Date: Tue, 14 Feb 2017 09:56:01 -0500
Subject: [PATCH] Halve the number of syscalls in zero_pages_with_mmap on Linux

---
 src/runtime/coreparse.c | 11 +++++++++++
 src/runtime/gencgc.c    | 29 ++++++++++++++++++++++++-----
 2 files changed, 35 insertions(+), 5 deletions(-)

diff --git a/src/runtime/coreparse.c b/src/runtime/coreparse.c
index d58032432..ee29cd9c9 100644
--- a/src/runtime/coreparse.c
+++ b/src/runtime/coreparse.c
@@ -277,6 +277,10 @@ os_vm_address_t inflate_core_bytes(int fd, os_vm_offset_t offset,
 
 int merge_core_pages = -1;
 
+#ifdef LISP_FEATURE_LINUX
+os_vm_address_t anon_dynamic_space_start;
+#endif
+
 static void
 process_directory(int fd, lispobj *ptr, int count, os_vm_offset_t file_offset)
 {
@@ -369,6 +373,13 @@ process_directory(int fd, lispobj *ptr, int count, os_vm_offset_t file_offset)
              * space 0. (We checked above that for GENCGC,
              * addr==DYNAMIC_SPACE_START.) */
             current_dynamic_space = (lispobj *)addr;
+#ifdef LISP_FEATURE_LINUX
+            anon_dynamic_space_start = addr + len;
+            // This assertion is here because of the test in zero_pages_with_mmap()
+            // which trusts that if addr > anon_dynamic_space_start
+            // then addr did not come from any file mapping.
+            gc_assert((lispobj)anon_dynamic_space_start > STATIC_SPACE_END);
+#endif
             break;
         case STATIC_CORE_SPACE_ID:
             if (addr != (os_vm_address_t)STATIC_SPACE_START) {
diff --git a/src/runtime/gencgc.c b/src/runtime/gencgc.c
index 06d6cbb68..2fd93c985 100644
--- a/src/runtime/gencgc.c
+++ b/src/runtime/gencgc.c
@@ -635,11 +635,30 @@ void zero_pages_with_mmap(page_index_t start, page_index_t end) {
     gc_assert(length >= gencgc_release_granularity);
     gc_assert((length % gencgc_release_granularity) == 0);
 
-    os_invalidate(addr, length);
-    new_addr = os_validate(addr, length);
-    if (new_addr == NULL || new_addr != addr) {
-        lose("remap_free_pages: page moved, 0x%08x ==> 0x%08x",
-             start, new_addr);
+#ifdef LISP_FEATURE_LINUX
+    extern os_vm_address_t anon_dynamic_space_start;
+    // We use MADV_DONTNEED only on Linux due to differing semantics from BSD.
+    // Linux treats it as a demand that the memory be 0-filled, or refreshed
+    // from a file that backs the range. BSD takes it as a hint that you don't
+    // care if the memory has to brought in from swap when next accessed,
+    // i.e. it's not a request to make a user-visible alteration to memory.
+    // So in theory this can bring a page in from the core file, if we happen
+    // to hit a page that resides in the portion of memory mapped by coreparse.
+    // In practice this should not happen because objects from a core file can't
+    // become garbage. Except in save-lisp-and-die they can, and we must be
+    // cautious not to resurrect bytes that originally came from the file.
+    if ((os_vm_address_t)addr >= anon_dynamic_space_start) {
+        if (madvise(addr, length, MADV_DONTNEED) != 0)
+            lose("madvise failed\n");
+    } else
+#endif
+    {
+        os_invalidate(addr, length);
+        new_addr = os_validate(addr, length);
+        if (new_addr == NULL || new_addr != addr) {
+            lose("remap_free_pages: page moved, 0x%08x ==> 0x%08x",
+                 start, new_addr);
+        }
     }
 
     for (i = start; i <= end; i++) {
-- 
2.11.4.GIT