Skip to contentSkip to conversation

malloc: Fix issue with calloc memory possibly being non-zero

PATCH New Kumar Gala 3 participants 4 messages Nov 15, 2010 – Nov 16, 2010

lore
Share & tools
lore

Kumar Gala

author
#0 lore raw

PATCH 1/1 malloc: Fix issue with calloc memory possibly being non-zero

Since we set #define MORECORE_CLEARS 1, the code assumes 'sbrk' always returns zero'd out memory. However since its possible that free() returns memory ack to sbrk() via malloc_trim we could possible get non-zero'd memory from sbrk() if it allocates back memory that was previously freed to it.

There are two possible solutions to this problem.
1. change #define MORECORE_CLEARS 0
2. memset to zero memory returned to sbrk.

We go with the second since the sbrk being called to free up memory should be pretty rare.

The following code problems an example test to show the issue. This test code was inserted right after the call to mem_malloc_init(). we could possible get non-zero'd memory from sbrk() if it allocates back memory that was previously freed to it.

There are two possible solutions to this problem.
1. change #define MORECORE_CLEARS 0
2. memset to zero memory returned to sbrk.

We go with the second since the sbrk being called to free up memory should be pretty rare.

The following code problems an example test to show the issue. This test code was inserted right after the call to mem_malloc_init().

...

       u8 *p2;
       int i;

       printf("MALLOC TEST\n");
       p1 = malloc(135176);
       printf("P1 = %p\n", p1);
       memset(p1, 0xab, 135176);

       free(p1);
       p2 = calloc(4097, 1);
       printf("P2 = %p %p\n", p2, p2 + 4097);

       for (i = 0; i < 4097; i++) {
	       if (p2[i] != 0)
		       printf("miscompare at byte %d got %x\n", i, p2[i]);

       free(p2);
       printf("END MALLOC TEST\n\n");

...

---
 common/dlmalloc.c |    7 +++++++
 1 files changed, 7 insertions(+), 0 deletions(-)
common/dlmalloc.c +7 -0
index 4871f4b..e9bab09 100644
@@ -1511,6 +1511,13 @@ void *sbrk(ptrdiff_t increment)
 	ulong old = mem_malloc_brk;
 	ulong new = old + increment;
 
+	/*
+	 * if we are giving memory back make sure we clear it out since
+	 * we set MORECORE_CLEARS to 1
+	 */
+	if (increment < 0)
+		memset((void *)new, 0, -increment);
+
 	if ((new < mem_malloc_start) || (new > mem_malloc_end))
 		return (void *)MORECORE_FAILURE;
 
Signature · 1 line
1.7.2.3
#1 lore raw

PATCH 1/1 malloc: Fix issue with calloc memory possibly being non-zero

Quoted 1 line · On Monday, November 15, 2010 15:51:14 Kumar Gala wrote:

returns memory ack to sbrk() via malloc_trim we could possible get

"ack" -> "back"

Quoted 1 line

The following code problems an example test to show the issue.

i dont understand this sentence

Quoted 8 lines

we
could possible get non-zero'd memory from sbrk() if it allocates back memory that was previously freed to it.

There are two possible solutions to this problem.
1. change #define MORECORE_CLEARS 0
2. memset to zero memory returned to sbrk.

Quoted 5 lines

We go with the second since the sbrk being called to free up memory should be pretty rare.

The following code problems an example test to show the issue. This test code was inserted right after the call to mem_malloc_init().

this looks like the changelog is partially duplicated
-mike
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
Url : http://lists.denx.de/pipermail/u-boot/attachments/20101115/1238cc42/attachment.pgp

#2 lore raw

PATCH 1/1 malloc: Fix issue with calloc memory possibly being non-zero

In reply to Kumar Gala

Dear Kumar Gala,

Quoted 5 lines · In message <[email protected]> you…

Since we set #define MORECORE_CLEARS 1, the code assumes 'sbrk' always returns zero'd out memory. However since its possible that free() returns memory ack to sbrk() via malloc_trim we could possible get non-zero'd memory from sbrk() if it allocates back memory that was previously freed to it.

I confirm that the test case works fine here, too. So:

But the commit message needs serious rework before this can get applied.

Best regards,

Wolfgang Denk

Signature · 4 lines
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
If ignorance is bliss, why aren't there more happy people?
author
#3 lore raw

PATCH 1/1 malloc: Fix issue with calloc memory possibly being non-zero

Quoted 9 lines · Wolfgang Denk

Dear Kumar Gala,

Quoted 5 lines · In message <[email protected]> you…

Since we set #define MORECORE_CLEARS 1, the code assumes 'sbrk' always returns zero'd out memory. However since its possible that free() returns memory ack to sbrk() via malloc_trim we could possible get non-zero'd memory from sbrk() if it allocates back memory that was previously freed to it.

Quoted 7 lines

I confirm that the test case works fine here, too. So:

But the commit message needs serious rework before this can get applied.

Oops, will fix and repost.

- k