public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* Bug in 2.4.0-test9 and test10 with sys_shmat()
@ 2000-11-16 16:57 Richard Jerrell
  2000-11-17 14:37 ` [Patch] " Christoph Rohland
  0 siblings, 1 reply; 2+ messages in thread
From: Richard Jerrell @ 2000-11-16 16:57 UTC (permalink / raw)
  To: linux-kernel

Sending -1 as the shmid to shmat will cause an oops.  2.2.16 caught this
with simple boundry checking, so replace the lines

if (!shm_sb || (shmid % SEQ_MULTIPLIER) == zero_id)
                return -EINVAL;

with

if (!shm_sb || shmid < 0 || (shmid % SEQ_MULTIPLIER) == zero_id)
                return -EINVAL;

Simple program to demonstrate the bug...

#include <sys/ipc.h>
#include <sys/shm.h>

int main(void) {
	shmat(-1,0,0);
	return 0;
}

Rich
jerrell@missioncriticallinux.com


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

^ permalink raw reply	[flat|nested] 2+ messages in thread

* [Patch] Re: Bug in 2.4.0-test9 and test10 with sys_shmat()
  2000-11-16 16:57 Bug in 2.4.0-test9 and test10 with sys_shmat() Richard Jerrell
@ 2000-11-17 14:37 ` Christoph Rohland
  0 siblings, 0 replies; 2+ messages in thread
From: Christoph Rohland @ 2000-11-17 14:37 UTC (permalink / raw)
  To: Linus Torvalds; +Cc: Richard Jerrell, linux-kernel

Hi Linus,

The attached patch fixes two things:

1) shmat should not oops on shmid < 0
2) I think the shm tables should be allocated with GFP_USER instead of
   GFP_KERNEL since these are user requests.

Greetings
		Christoph

--- 4-11-6/ipc/shm.c	Wed Oct  4 15:58:02 2000
+++ linux/ipc/shm.c	Fri Nov 17 13:47:29 2000
@@ -572,13 +572,13 @@
 	if (pages == 0)
 		return NULL;
 
-	ret = kmalloc ((dir+1) * sizeof(pte_t *), GFP_KERNEL);
+	ret = kmalloc ((dir+1) * sizeof(pte_t *), GFP_USER);
 	if (!ret)
 		goto nomem;
 
 	for (ptr = ret; ptr < ret+dir ; ptr++)
 	{
-		*ptr = (pte_t *)__get_free_page (GFP_KERNEL);
+		*ptr = (pte_t *)__get_free_page (GFP_USER);
 		if (!*ptr)
 			goto free;
 		init_ptes (*ptr, PTES_PER_PAGE);
@@ -586,7 +586,7 @@
 
 	/* The last one is probably not of PAGE_SIZE: we use kmalloc */
 	if (last) {
-		*ptr = kmalloc (last*sizeof(pte_t), GFP_KERNEL);
+		*ptr = kmalloc (last*sizeof(pte_t), GFP_USER);
 		if (!*ptr)
 			goto free;
 		init_ptes (*ptr, last);
@@ -724,7 +724,7 @@
 	struct shmid_kernel *shp;
 	pte_t		   **dir;
 
-	shp = (struct shmid_kernel *) kmalloc (sizeof (*shp) + namelen, GFP_KERNEL);
+	shp = (struct shmid_kernel *) kmalloc (sizeof (*shp) + namelen, GFP_USER);
 	if (!shp)
 		return ERR_PTR(-ENOMEM);
 
@@ -1202,7 +1202,7 @@
 	char   name[SHM_FMT_LEN+1];
 	void *user_addr;
 
-	if (!shm_sb || (shmid % SEQ_MULTIPLIER) == zero_id)
+	if (!shm_sb || shmid < 0 || (shmid % SEQ_MULTIPLIER) == zero_id)
 		return -EINVAL;
 
 	if ((addr = (ulong)shmaddr)) {
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2000-11-17 15:10 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2000-11-16 16:57 Bug in 2.4.0-test9 and test10 with sys_shmat() Richard Jerrell
2000-11-17 14:37 ` [Patch] " Christoph Rohland

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox