On Fri, Oct 04, 2024 at 12:33:13PM +0000, Pimyn@xxxxxxxxxxxxxxxxxx wrote: > Avoid kernel-usb-infoleak by initializing all memory written to device. > The buffer length uses 32bit alignment which might cause some buffer > data to be read without any initialization. > > Reported-by: syzbot+8f282cce71948071c335@xxxxxxxxxxxxxxxxxxxxxxxxx > Closes: https://lore.kernel.org/all/0000000000006f6622061eb52dba@xxxxxxxxxx/T/ > Signed-off-by: Pimyn Girgis <bemenboshra2001@xxxxxxxxx> (Google) > --- > drivers/usb/class/usbtmc.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/usb/class/usbtmc.c b/drivers/usb/class/usbtmc.c > index ffc9c6fdd7e1..d79a08dfb54d 100644 > --- a/drivers/usb/class/usbtmc.c > +++ b/drivers/usb/class/usbtmc.c > @@ -21,6 +21,7 @@ > #include <linux/usb.h> > #include <linux/compat.h> > #include <linux/usb/tmc.h> > +#include <linux/string.h> > > /* Increment API VERSION when changing tmc.h with new flags or ioctls > * or when changing a significant behavior of the driver. > @@ -1169,6 +1170,8 @@ static ssize_t usbtmc_generic_write(struct usbtmc_file_data *file_data, > * (size + 3 & ~3) rounds up and simplifies user code > */ > aligned = (this_part + 3) & ~3; > + /* Initialize the remaining part of the buffer */ > + memzero_explicit(buffer + this_part, aligned - this_part); Why "explicit"? Also, what about commit 625fa77151f0 ("USB: usbtmc: prevent kernel-usb-infoleak"), doesn't that solve the issue here? thanks, greg k-h
