elf: limit max map count to safe value

With ELF, at generating coredump, some more headers other than used
vmas are added.

When max_map_count == 65536, a core generated by following kinds of
code can be unreadable because the number of ELF's program header is
written in 16bit in Ehdr (please see elf.h) and the number overflows.

==
	... = mmap(); (munmap, mprotect, etc...)
	if (failed)
		abort();
==

This can happen in mmap/munmap/mprotect/etc...which calls split_vma().

I think 65536 is not safe as _default_ and reduce it to 65530 is good
for avoiding unexpected corrupted core.

Anyway, max_map_count can be enlarged by sysctl if a user is brave..

Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Jakub Jelinek <jakub@redhat.com>
Acked-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

2 files changed, 18 insertions, 3 deletions

diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c
index 9fa212b014a..f1867900e45 100644
--- a/fs/binfmt_elf.c
+++ b/fs/binfmt_elf.c
@@ -1929,7 +1929,10 @@ static int elf_core_dump(long signr, struct pt_regs *regs, struct file *file, un
 	elf = kmalloc(sizeof(*elf), GFP_KERNEL);
 	if (!elf)
 		goto out;
-	
+	/*
+	 * The number of segs are recored into ELF header as 16bit value.
+	 * Please check DEFAULT_MAX_MAP_COUNT definition when you modify here.
+	 */
 	segs = current->mm->map_count;
 #ifdef ELF_CORE_EXTRA_PHDRS
 	segs += ELF_CORE_EXTRA_PHDRS;
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 4d075426988..0085d758d64 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -349,8 +349,20 @@ extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
 struct nsproxy;
 struct user_namespace;
 
-/* Maximum number of active map areas.. This is a random (large) number */
-#define DEFAULT_MAX_MAP_COUNT	65536
+/*
+ * Default maximum number of active map areas, this limits the number of vmas
+ * per mm struct. Users can overwrite this number by sysctl but there is a
+ * problem.
+ *
+ * When a program's coredump is generated as ELF format, a section is created
+ * per a vma. In ELF, the number of sections is represented in unsigned short.
+ * This means the number of sections should be smaller than 65535 at coredump.
+ * Because the kernel adds some informative sections to a image of program at
+ * generating coredump, we need some margin. The number of extra sections is
+ * 1-3 now and depends on arch. We use "5" as safe margin, here.
+ */
+#define MAPCOUNT_ELF_CORE_MARGIN	(5)
+#define DEFAULT_MAX_MAP_COUNT	(USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
 
 extern int sysctl_max_map_count;