Ian Abbott <[email protected]> writes: > On 08/07/2025 13:37, Edward Adam Davis wrote: >> The irq number 0x2166 passed by the reproducer is too large and is not >> within the supported range [2-7, 10-12, 14, or 15], which triggers the oob. >> Fixes: ad7a370c8be4 (“staging: comedi: aio_iiro_16: add command >> support for change of state detection”) >> Reported-by: [email protected] >> Closes: <https://syzkaller.appspot.com/bug?extid=f1bb7e4ea47ea12b535c> >> Signed-off-by: Edward Adam Davis <[email protected]> >> — >> drivers/comedi/drivers/aio_iiro_16.c | 3 ++- >> 1 file changed, 2 insertions(+), 1 deletion(-) >> diff –git a/drivers/comedi/drivers/aio_iiro_16.c >> b/drivers/comedi/drivers/aio_iiro_16.c >> index b00fab0b89d4..e43730f00c8b 100644 >> — a/drivers/comedi/drivers/aio_iiro_16.c >> +++ b/drivers/comedi/drivers/aio_iiro_16.c >> @@ -177,7 +177,8 @@ static int aio_iiro_16_attach(struct comedi_device *dev, >> * Digital input change of state interrupts are optionally supported >> * using IRQ 2-7, 10-12, 14, or 15. >> */ >> - if ((1 << it->options[1]) & 0xdcfc) { >> + if (it->options[1] > 1 && it->options[1] < 16 && >> + (1 << it->options[1]) & 0xdcfc) { >> ret = request_irq(it->options[1], aio_iiro_16_cos, 0, >> dev->board_name, dev); >> if (ret == 0) > > The patch is fine apart from the misspelling of “invalid” in the subject > line, but I’d already submitted a patch for this before syzbot detected > it: > > <https://lore.kernel.org/lkml/[email protected]/>
#syz fix: comedi: aio_iiro_16: Fix bit shift out of bounds
