summaryrefslogtreecommitdiffstats
path: root/drivers/media/Kconfig
diff options
context:
space:
mode:
authorMauro Carvalho Chehab <mchehab+samsung@kernel.org>2018-08-07 06:49:26 -0400
committerMauro Carvalho Chehab <mchehab+samsung@kernel.org>2018-08-08 10:57:14 -0400
commit845b978a871bff3707eee611b32e4be0b9a94dd2 (patch)
tree3807a64684108d9172c53bcf1c14229413c05331 /drivers/media/Kconfig
parentd9a71866944967b8622c493ef1b570ccb74e45ae (diff)
downloadlinux-845b978a871bff3707eee611b32e4be0b9a94dd2.tar.bz2
media: rtl28xxu: be sure that it won't go past the array size
smatch warns that the RC query code could go past the array size: drivers/media/usb/dvb-usb-v2/rtl28xxu.c:1757 rtl2832u_rc_query() error: buffer overflow 'buf' 128 <= 130 drivers/media/usb/dvb-usb-v2/rtl28xxu.c:1758 rtl2832u_rc_query() error: buffer overflow 'buf' 128 <= 130 The driver logic gets the length of the IR RX buffer with: ret = rtl28xxu_rd_reg(d, IR_RX_BC, &buf[0]); ... len = buf[0]; In thesis, this could range between 0 and 255 [1]. While this should never happen in practice, due to hardware limits, smatch is right when it complains about that, as there's nothing at the logic that would prevent it. So, if for whatever reason, buf[0] gets filled by rtl28xx read functions with a value bigger than 128, it will go past the array. So, add an explicit check. [1] I've no idea why smatch thinks that the maximum value is 130. I double-checked the code several times. Was unable to find any reason for assuming 130. Perhaps smatch is not properly parsing u8 here? Fixes: b5cbaa43a676 ("[media] rtl28xx: initial support for rtl2832u") Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Diffstat (limited to 'drivers/media/Kconfig')
0 files changed, 0 insertions, 0 deletions