3.16.80-rc1 review patch. If anyone has any objections, please let me know.
------------------
From: "Gustavo A. R. Silva" gustavo@embeddedor.com
commit b987b66ac3a2bc2f7b03a0ba48a07dc553100c07 upstream.
It seems that the right variable to use in this case is *i*, instead of *n*, otherwise there is an undefined behavior when right shifiting by more than 31 bits when multiplying n by 8; notice that *n* can take values equal or greater than 4 (4, 8, 16, ...).
Also, notice that under the current conditions (bl = 3), we are skiping the handling of bytes 3, 7, 31... So, fix this by updating this logic and limit *bl* up to 4 instead of up to 3.
This fix is based on function udc_stuff_fifo().
Addresses-Coverity-ID: 1454834 ("Bad bit shift operation") Fixes: 24a28e428351 ("USB: gadget driver for LPC32xx") Signed-off-by: Gustavo A. R. Silva gustavo@embeddedor.com Link: https://lore.kernel.org/r/20191014191830.GA10721@embeddedor Signed-off-by: Greg Kroah-Hartman gregkh@linuxfoundation.org [bwh: Backported to 3.16: adjust filename] Signed-off-by: Ben Hutchings ben@decadent.org.uk --- drivers/usb/gadget/lpc32xx_udc.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
--- a/drivers/usb/gadget/lpc32xx_udc.c +++ b/drivers/usb/gadget/lpc32xx_udc.c @@ -1228,11 +1228,11 @@ static void udc_pop_fifo(struct lpc32xx_ tmp = readl(USBD_RXDATA(udc->udp_baseaddr));
bl = bytes - n; - if (bl > 3) - bl = 3; + if (bl > 4) + bl = 4;
for (i = 0; i < bl; i++) - data[n + i] = (u8) ((tmp >> (n * 8)) & 0xFF); + data[n + i] = (u8) ((tmp >> (i * 8)) & 0xFF); } break;