Skip to contentSkip to conversation

mtdchar_write_ioctl(): prevent integer overflow, use kvmalloc()

PATCH Applied Michał Kępień 3 participants 7 messages May 16, 2022 – Jun 9, 2022 Memory technology devices (MTD)

lore
Share & tools
lore
#0 lore raw

cover mtdchar_write_ioctl(): prevent integer overflow, use kvmalloc()

This patch series addresses two flaws in mtdchar_write_ioctl() found by Richard Weinberger while he was reviewing v3 of the patch series proposing a new MEMREAD ioctl [1].

Richard, I included a Suggested-by tag crediting you in both of the patches in this series. I hope that is okay with you, but please let me know if it isn't.

Once this patch series is reviewed and accepted, I will apply similar fixes in v4 of the MEMREAD ioctl patch series.

[1] https://lists.infradead.org/pipermail/linux-mtd/2022-February/091276.html

Michał Kępień (2):
mtdchar: prevent integer overflow in a safety check
mtdchar: use kvmalloc() for potentially large allocations

 drivers/mtd/mtdchar.c      | 13 ++++++++-----
 include/uapi/mtd/mtd-abi.h |  4 ++--
 2 files changed, 10 insertions(+), 7 deletions(-)
Signature · 7 lines
2.36.1



______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
#1 lore raw

PATCH 1/2 mtdchar: prevent integer overflow in a safety check

Commit 6420ac0af95d ("mtdchar: prevent unbounded allocation in MEMWRITE ioctl") added a safety check to mtdchar_write_ioctl() which attempts to ensure that the write request sent by user space does not extend beyond the MTD device's size. However, that check contains an addition of two struct mtd_write_req fields, 'start' and 'len', both of which are u64 variables. The result of that addition can overflow, allowing the safety check to be bypassed.

The arguably simplest fix - changing the data types of the relevant struct mtd_write_req fields - is not feasible as it would break user space.

Fix by making mtdchar_write_ioctl() truncate the value provided by user space in the 'len' field of struct mtd_write_req, so that only the lower 32 bits of that field are used, preventing the overflow.

While the 'ooblen' field of struct mtd_write_req is not currently used in any similarly flawed safety check, also truncate it to 32 bits, for consistency with the 'len' field and with other MTD routines handling OOB data.

Update include/uapi/mtd/mtd-abi.h accordingly.

---
 drivers/mtd/mtdchar.c      | 3 +++
 include/uapi/mtd/mtd-abi.h | 4 ++--
 2 files changed, 5 insertions(+), 2 deletions(-)
drivers/mtd/mtdchar.c +3 -0
index d0f9c4b0285c..b2700f8467ff 100644
@@ -615,6 +615,9 @@ static int mtdchar_write_ioctl(struct mtd_info *mtd,
 	if (!usr_oob)
 		req.ooblen = 0;
 
+	req.len &= 0xffffffff;
+	req.ooblen &= 0xffffffff;
+
 	if (req.start + req.len > mtd->size)
 		return -EINVAL;
 
include/uapi/mtd/mtd-abi.h +2 -2
index b869990c2db2..890d9e5b76d7 100644
@@ -69,8 +69,8 @@ enum {
  * struct mtd_write_req - data structure for requesting a write operation
  *
  * @start:	start address
- * @len:	length of data buffer
- * @ooblen:	length of OOB buffer
+ * @len:	length of data buffer (only lower 32 bits are used)
+ * @ooblen:	length of OOB buffer (only lower 32 bits are used)
  * @usr_data:	user-provided data buffer
  * @usr_oob:	user-provided OOB buffer
  * @mode:	MTD mode (see "MTD operation modes")
Signature · 7 lines
2.36.1



______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
#2 lore raw

Re: [PATCH 1/2] mtdchar: prevent integer overflow in a safety check

----- Ursprüngliche Mail -----

Quoted 41 lines

Von: "Michał Kępień" <[email protected]>
An: "Miquel Raynal" <[email protected]>, "richard" <[email protected]>, "Vignesh Raghavendra" <[email protected]>
CC: "linux-mtd" <[email protected]>, "linux-kernel" <[email protected]>
Gesendet: Montag, 16. Mai 2022 09:06:00
Betreff: [PATCH 1/2] mtdchar: prevent integer overflow in a safety check

Commit 6420ac0af95d ("mtdchar: prevent unbounded allocation in MEMWRITE ioctl") added a safety check to mtdchar_write_ioctl() which attempts to ensure that the write request sent by user space does not extend beyond the MTD device's size. However, that check contains an addition of two struct mtd_write_req fields, 'start' and 'len', both of which are u64 variables. The result of that addition can overflow, allowing the safety check to be bypassed.

The arguably simplest fix - changing the data types of the relevant struct mtd_write_req fields - is not feasible as it would break user space.

Fix by making mtdchar_write_ioctl() truncate the value provided by user space in the 'len' field of struct mtd_write_req, so that only the lower 32 bits of that field are used, preventing the overflow.

While the 'ooblen' field of struct mtd_write_req is not currently used in any similarly flawed safety check, also truncate it to 32 bits, for consistency with the 'len' field and with other MTD routines handling OOB data.

Update include/uapi/mtd/mtd-abi.h accordingly.

---
drivers/mtd/mtdchar.c | 3 +++
include/uapi/mtd/mtd-abi.h | 4 ++--

2 files changed, 5 insertions(+), 2 deletions(-)
drivers/mtd/mtdchar.c +0 -0
index d0f9c4b0285c..b2700f8467ff 100644

drivers/mtd/mtdchar.c static int mtdchar_write_ioctl(struct mtd_info *mtd, · line 620

Quoted 6 lines
drivers/mtd/mtdchar.c +3 -0
@@ -615,6 +615,9 @@ static int mtdchar_write_ioctl(struct mtd_info *mtd,
 	if (!usr_oob)
 		req.ooblen = 0;
 
+	req.len &= 0xffffffff;
+	req.ooblen &= 0xffffffff;
+

Yeah, I think it is reasonable to limit write requests to 4GiB.

Thanks,
//richard

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

maintainer Applied
#3 lore raw

Re: [PATCH 1/2] mtdchar: prevent integer overflow in a safety check

In reply to Michał Kępień
Quoted 21 lines · On Mon, 2022-05-16 at 07:06:00 UTC, =?utf-8?b?TWljaGHFgiBLxJlwaWXFhA==?= wrote:

Commit 6420ac0af95d ("mtdchar: prevent unbounded allocation in MEMWRITE ioctl") added a safety check to mtdchar_write_ioctl() which attempts to ensure that the write request sent by user space does not extend beyond the MTD device's size. However, that check contains an addition of two struct mtd_write_req fields, 'start' and 'len', both of which are u64 variables. The result of that addition can overflow, allowing the safety check to be bypassed.

The arguably simplest fix - changing the data types of the relevant struct mtd_write_req fields - is not feasible as it would break user space.

Fix by making mtdchar_write_ioctl() truncate the value provided by user space in the 'len' field of struct mtd_write_req, so that only the lower 32 bits of that field are used, preventing the overflow.

While the 'ooblen' field of struct mtd_write_req is not currently used in any similarly flawed safety check, also truncate it to 32 bits, for consistency with the 'len' field and with other MTD routines handling OOB data.

Quoted 4 lines

Update include/uapi/mtd/mtd-abi.h accordingly.

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/next, thanks.

Miquel

#4 lore raw

PATCH 2/2 mtdchar: use kvmalloc() for potentially large allocations

In reply to Michał Kępień

mtdchar_write_ioctl() calls kmalloc() with the 'size' argument set to the smaller of two values: the write request's data/OOB length provided by user space and the erase block size of the MTD device. If the latter is large, kmalloc() may not be able to serve such allocation requests. Use kvmalloc() instead. Correspondingly, replace kfree() calls with kvfree() calls.

---
 drivers/mtd/mtdchar.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
drivers/mtd/mtdchar.c +5 -5
index b2700f8467ff..05860288a7af 100644
@@ -623,16 +623,16 @@ static int mtdchar_write_ioctl(struct mtd_info *mtd,
 
 	datbuf_len = min_t(size_t, req.len, mtd->erasesize);
 	if (datbuf_len > 0) {
-		datbuf = kmalloc(datbuf_len, GFP_KERNEL);
+		datbuf = kvmalloc(datbuf_len, GFP_KERNEL);
 		if (!datbuf)
 			return -ENOMEM;
 	}
 
 	oobbuf_len = min_t(size_t, req.ooblen, mtd->erasesize);
 	if (oobbuf_len > 0) {
-		oobbuf = kmalloc(oobbuf_len, GFP_KERNEL);
+		oobbuf = kvmalloc(oobbuf_len, GFP_KERNEL);
 		if (!oobbuf) {
-			kfree(datbuf);
+			kvfree(datbuf);
 			return -ENOMEM;
 		}
 	}
@@ -682,8 +682,8 @@ static int mtdchar_write_ioctl(struct mtd_info *mtd,
 		usr_oob += ops.oobretlen;
 	}
 
-	kfree(datbuf);
-	kfree(oobbuf);
+	kvfree(datbuf);
+	kvfree(oobbuf);
 
 	return ret;
 }
Signature · 7 lines
2.36.1



______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
maintainer Acked-by on v1 2/2
#5 lore raw

Re: [PATCH 2/2] mtdchar: use kvmalloc() for potentially large allocations

----- Ursprüngliche Mail -----

Quoted 13 lines

Von: "Michał Kępień" <[email protected]>
An: "Miquel Raynal" <[email protected]>, "richard" <[email protected]>, "Vignesh Raghavendra" <[email protected]>
CC: "linux-mtd" <[email protected]>, "linux-kernel" <[email protected]>
Gesendet: Montag, 16. Mai 2022 09:06:01
Betreff: [PATCH 2/2] mtdchar: use kvmalloc() for potentially large allocations

mtdchar_write_ioctl() calls kmalloc() with the 'size' argument set to the smaller of two values: the write request's data/OOB length provided by user space and the erase block size of the MTD device. If the latter is large, kmalloc() may not be able to serve such allocation requests. Use kvmalloc() instead. Correspondingly, replace kfree() calls with kvfree() calls.

Quoted 2 lines

Looks good to me.

Thanks,
//richard

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

maintainer Applied
#6 lore raw

Re: [PATCH 2/2] mtdchar: use kvmalloc() for potentially large allocations

In reply to Michał Kępień
Quoted 10 lines · On Mon, 2022-05-16 at 07:06:01 UTC, =?utf-8?b?TWljaGHFgiBLxJlwaWXFhA==?= wrote:

mtdchar_write_ioctl() calls kmalloc() with the 'size' argument set to the smaller of two values: the write request's data/OOB length provided by user space and the erase block size of the MTD device. If the latter is large, kmalloc() may not be able to serve such allocation requests. Use kvmalloc() instead. Correspondingly, replace kfree() calls with kvfree() calls.

Applied to https://git.kernel.org/pub/scm/linux/kernel/git/mtd/linux.git mtd/next, thanks.

Miquel