Re: Build problems in kdd.c with xen-4.14.0-rc4

[Tim Deegan <tim@xxxxxxx>]

Hi Michael,

Thanks for ther report!

At 23:21 +0100 on 30 Jun (1593559296), Michael Young wrote:
> I get the following errors when trying to build xen-4.14.0-rc4
> 
> kdd.c: In function 'kdd_tx':
> kdd.c:754:15: error: array subscript 16 is above array bounds of 
> 'uint8_t[16]' {aka 'unsigned char[16]'} [-Werror=array-bounds]
>    754 |         s->txb[len++] = 0xaa;
>        |         ~~~~~~^~~~~~~
> kdd.c:82:17: note: while referencing 'txb'
>     82 |         uint8_t txb[sizeof (kdd_hdr)];           /* Marshalling area 
> for tx */
>        |                 ^~~
> kdd.c: In function 'kdd_break':
> kdd.c:819:19: error: array subscript 16 is above array bounds of 
> 'uint8_t[16]' {aka 'unsigned char[16]'} [-Werror=array-bounds]

Oh dear.  The fix for the last kdd bug seems to have gone wrong
somewhere.  The patch I posted has:

-        uint8_t txb[sizeof (kdd_hdr) + 65536];   /* Marshalling area for tx */
+        uint8_t txb[sizeof (kdd_pkt)];           /* Marshalling area for tx */

but as applied in master it's:

-        uint8_t txb[sizeof (kdd_hdr) + 65536];   /* Marshalling area for tx */
+        uint8_t txb[sizeof (kdd_hdr)];           /* Marshalling area for tx */

i.e. the marshalling area is only large enough for a header and GCC
is correctly complaining about that.

Wei, it looks like you committed this patch - can you figure out what
happened to it please?

Cheers,

Tim.