On Wed, Mar 20, 2019 at 01:56:41PM +0100, Andrzej Pietrasiewicz wrote: > Don't actually allocate anything if userspace enqueues a zero-length > buffer. Otherwise vmalloc of zero-sized area will be attempted in > ffs_build_sg_list(). > > Signed-off-by: Andrzej Pietrasiewicz <[email protected]> > --- > drivers/usb/gadget/function/f_fs.c | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/drivers/usb/gadget/function/f_fs.c > b/drivers/usb/gadget/function/f_fs.c > index 08357c5dd026..5cecfe3e9395 100644 > --- a/drivers/usb/gadget/function/f_fs.c > +++ b/drivers/usb/gadget/function/f_fs.c > @@ -797,6 +797,9 @@ static void *ffs_build_sg_list(struct sg_table *sgt, > size_t sz) > static inline void *ffs_alloc_buffer(struct ffs_io_data *io_data, > size_t data_len) > { > + if (!data_len) > + return ZERO_SIZE_PTR; > + > if (io_data->use_sg) > return ffs_build_sg_list(&io_data->sgt, data_len); > > @@ -805,7 +808,7 @@ static inline void *ffs_alloc_buffer(struct ffs_io_data > *io_data, > > static inline void ffs_free_buffer(struct ffs_io_data *io_data) > { > - if (!io_data->buf) > + if (ZERO_OR_NULL_PTR(io_data->buf)) > return; > > if (io_data->use_sg) {
Are you sure this is ok? zero length packets mean something in the USB protocol :) What is the problem of trying to allocate a buffer of 0 bytes? thanks, greg k-h
