Hi Arnd,
+CC: Kees Cook
On Mon. 20 Sep 2021 at 21:30, Arnd Bergmann <arnd@xxxxxxxxxx> wrote:
> From: Arnd Bergmann <arnd@xxxxxxxx>
>
> gcc complains when writing into a zero-length array:
>
> drivers/net/can/usb/etas_es58x/es581_4.c: In function 'es581_4_tx_can_msg':
> drivers/net/can/usb/etas_es58x/es581_4.c:374:42: warning: array subscript 65535 is outside the bounds of an interior zero-length array 'u8[0]' {aka 'unsigned char[]'} [-Wzero-length-bounds]
> 374 | tx_can_msg = (typeof(tx_can_msg))&es581_4_urb_cmd->raw_msg[msg_len];
> | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> In file included from drivers/net/can/usb/etas_es58x/es58x_core.h:21,
> from drivers/net/can/usb/etas_es58x/es581_4.c:15:
> drivers/net/can/usb/etas_es58x/es581_4.h:195:20: note: while referencing 'raw_msg'
> 195 | u8 raw_msg[0];
> | ^~~~~~~
> CC [M] drivers/net/can/usb/etas_es58x/es58x_fd.o
> drivers/net/can/usb/etas_es58x/es58x_fd.c: In function 'es58x_fd_tx_can_msg':
> drivers/net/can/usb/etas_es58x/es58x_fd.c:360:42: warning: array subscript 65535 is outside the bounds of an interior zero-length array 'u8[0]' {aka 'unsigned char[]'} [-Wzero-length-bounds]
> 360 | tx_can_msg = (typeof(tx_can_msg))&es58x_fd_urb_cmd->raw_msg[msg_len];
> | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> In file included from drivers/net/can/usb/etas_es58x/es58x_core.h:22,
> from drivers/net/can/usb/etas_es58x/es58x_fd.c:17:
> drivers/net/can/usb/etas_es58x/es58x_fd.h:222:20: note: while referencing 'raw_msg'
> 222 | u8 raw_msg[0];
> | ^~~~~~~
>
> The solution is usually to use a flexible-array member the struct, but
> we can't directly have that inside of a union, nor can it be the only
> member of a struct, so add a dummy struct with another zero-length
> member to get the intended behavior.
>
> If someone has a better workaround, let me know and I can send a new
> patch, as this version is rather ugly.
Actually, there is one. Kees Cook introduced a new macro,
DECLARE_FLEX_ARRAY(), to do this in a more elegant way:
https://lkml.org/lkml/2021/8/27/524
The same series also fixes the warning in the etas_es58x driver:
https://lkml.org/lkml/2021/8/27/523
So we only need to wait for Kees's series to get merged :)
Yours sincerely,
Vincent Mailhol
