[PATCH] RDMA/cxgb3: Fix unintended sign extension

[PATCH] RDMA/cxgb3: Fix unintended sign extension

[Colin King]

From: Colin Ian King <colin.king@canonical.com>

In the expression "utx_len << 28", utx_len starts as u8, but is promoted
to a signed int, then sign-extended to u64.  If utx_len is 0xf8 or greater
then the sign extension will set all the upper bits of utx_cmd which is
probably not what was intended.  Cast to utx_len to u64  to avoid the sign
extension.

Detected by CoverityScan, CID#138764 ("Unintended sign extension")

Fixes: b038ced7b370 ("RDMA/cxgb3: Add driver for Chelsio T3 RNIC")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
---
 drivers/infiniband/hw/cxgb3/cxio_hal.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.c b/drivers/infiniband/hw/cxgb3/cxio_hal.c
index dcb4bba522ba..c5cb80ccc6d6 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.c
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.c
@@ -647,7 +647,7 @@ static int cxio_hal_ctrl_qp_write_mem(struct cxio_rdev *rdev_p, u32 addr,
 		wqe += (sizeof(struct t3_bypass_wr) >> 3);
 		utx_cmd = (T3_UTX_MEM_WRITE << 28) | (addr + i * 3);
 		utx_cmd <<= 32;
-		utx_cmd |= (utx_len << 28) | ((utx_len << 2) + 1);
+		utx_cmd |= ((u64)utx_len << 28) | ((utx_len << 2) + 1);
 		*wqe = cpu_to_be64(utx_cmd);
 		wqe++;
 		copy_data = (u8 *) data + i * 96;
-- 
2.19.1

Re: [PATCH] RDMA/cxgb3: Fix unintended sign extension

[Steve Wise]

On 10/25/2018 9:31 AM, Colin King wrote:
> From: Colin Ian King <colin.king@canonical.com>
> 
> In the expression "utx_len << 28", utx_len starts as u8, but is promoted
> to a signed int, then sign-extended to u64.  If utx_len is 0xf8 or greater
> then the sign extension will set all the upper bits of utx_cmd which is
> probably not what was intended.  Cast to utx_len to u64  to avoid the sign
> extension.
> 
> Detected by CoverityScan, CID#138764 ("Unintended sign extension")
> 
> Fixes: b038ced7b370 ("RDMA/cxgb3: Add driver for Chelsio T3 RNIC")
> Signed-off-by: Colin Ian King <colin.king@canonical.com>
> ---
>  drivers/infiniband/hw/cxgb3/cxio_hal.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.c b/drivers/infiniband/hw/cxgb3/cxio_hal.c
> index dcb4bba522ba..c5cb80ccc6d6 100644
> --- a/drivers/infiniband/hw/cxgb3/cxio_hal.c
> +++ b/drivers/infiniband/hw/cxgb3/cxio_hal.c
> @@ -647,7 +647,7 @@ static int cxio_hal_ctrl_qp_write_mem(struct cxio_rdev *rdev_p, u32 addr,
>  		wqe += (sizeof(struct t3_bypass_wr) >> 3);
>  		utx_cmd = (T3_UTX_MEM_WRITE << 28) | (addr + i * 3);
>  		utx_cmd <<= 32;
> -		utx_cmd |= (utx_len << 28) | ((utx_len << 2) + 1);
> +		utx_cmd |= ((u64)utx_len << 28) | ((utx_len << 2) + 1);
>  		*wqe = cpu_to_be64(utx_cmd);
>  		wqe++;
>  		copy_data = (u8 *) data + i * 96;
> 

Acked-by: Steve Wise <swise@chelsio.com>

RE: [PATCH] RDMA/cxgb3: Fix unintended sign extension

[David Laight]

From: Colin King
> Sent: 25 October 2018 15:32
> 
> From: Colin Ian King <colin.king@canonical.com>
> 
> In the expression "utx_len << 28", utx_len starts as u8, but is promoted
> to a signed int, then sign-extended to u64.  If utx_len is 0xf8 or greater
> then the sign extension will set all the upper bits of utx_cmd which is
> probably not what was intended.  Cast to utx_len to u64  to avoid the sign
> extension.

RTFC...
utx_len is only ever 1, 2 or 3.
The 'problem' would arise if utx_len << 28 set the high bit.
This can only happen if utx_len is more than 7 (NFI where 0xf8 came from).

In any case the best fix is to use 'unsigned int' for wr_len and utx_len.
There is no point making local variable (or functions parameters/results)
smaller that 'int' unless you explicitly want the arithmetic to wrap.

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)