Skip to contentSkip to conversation

fs: prevent double-free on an error path in core dumper

PATCH Oleg Nesterov 3 participants 3 messages Sep 11, 2012 – Sep 12, 2012 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw

Denys, let's not discuss this offlist... add lkml/cc's

Quoted 6 lines · Denys Vlasenko

If elf_note_info_init fails to allocate memory for info->fields, it frees already alloceted stuff and returns error to its caller, fill_note_info. Which in turn returns error to its caller, elf_core_dump. Which jumps to cleanup label and calls free_note_info, which will happily try to free all info->fields again. BOOM.

This is !CORE_DUMP_USE_REGSET case, please mention this in the changelog. And s/fs:/coredump:/ in the subject ;)

OK, I think you are right, but unless I missed something there is a better fix.

Quoted 21 lines

This is the fix.

---
 fs/binfmt_elf.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletions(-)
fs/binfmt_elf.c +1 -1
index 760d7f5..b2b7034 100644
@@ -1617,7 +1617,7 @@ static void free_note_info(struct elf_note_info *info)
 	kfree(info->psinfo.data);
 }
 
-#else
+#else /* !CORE_DUMP_USE_REGSET */
 
 /* Here is the structure in which status of each thread is captured. */
 struct elf_thread_status

fs/binfmt_elf.c static int elf_note_info_init(struct elf_note_info *info) · line 1721

Quoted 4 lines
fs/binfmt_elf.c +1 -0
@@ -1718,6 +1718,8 @@ static int elf_note_info_init(struct elf_note_info *info)
 	kfree(info->psinfo);
  notes_free:
 	kfree(info->notes);
+	/* The caller will try to clean us up again, must set ptrs to NULL */

Exactly. The caller will do free_note_info() and info was zeroed at the start. See below

fs/binfmt_elf.c static int elf_note_info_init(struct elf_note_info *info) · line 1722

Quoted 1 line
fs/binfmt_elf.c +1 -0
+	memset(info, 0, sizeof(*info));

Yes, this will fix the problem.

But, again, the caller does free_note_info(), so why elf_note_info_init() tries to handle the kmalloc failures? Afaics, we can simplify the code and fix the bug.

What do you think about the patch below?

Oleg.

fs/binfmt_elf.c +4 -15
@@ -1696,30 +1696,19 @@ static int elf_note_info_init(struct elf
 		return 0;
 	info->psinfo = kmalloc(sizeof(*info->psinfo), GFP_KERNEL);
 	if (!info->psinfo)
-		goto notes_free;
+		return 0;
 	info->prstatus = kmalloc(sizeof(*info->prstatus), GFP_KERNEL);
 	if (!info->prstatus)
-		goto psinfo_free;
+		return 0;
 	info->fpu = kmalloc(sizeof(*info->fpu), GFP_KERNEL);
 	if (!info->fpu)
-		goto prstatus_free;
+		return 0;
 #ifdef ELF_CORE_COPY_XFPREGS
 	info->xfpu = kmalloc(sizeof(*info->xfpu), GFP_KERNEL);
 	if (!info->xfpu)
-		goto fpu_free;
+		return 0;
 #endif
 	return 1;
-#ifdef ELF_CORE_COPY_XFPREGS
- fpu_free:
-	kfree(info->fpu);
-#endif
- prstatus_free:
-	kfree(info->prstatus);
- psinfo_free:
-	kfree(info->psinfo);
- notes_free:
-	kfree(info->notes);
-	return 0;
 }
 
 static int fill_note_info(struct elfhdr *elf, int phdrs,
#2 lore raw
Quoted 6 lines · Oleg Nesterov

But, again, the caller does free_note_info(), so why elf_note_info_init()
tries to handle the kmalloc failures? Afaics, we can simplify the code and fix the bug.

What do you think about the patch below?

Looks reasonable and neat. :)