modules: fix freebsd32_modstat on big endian platforms

[khng]

Summary

The layout of modspecific_t on both little endian and big endian are as
follows:

|0|1|2|3|4|5|6|7|
+-------+-------+
|uintval|       |
+-------+-------+
|ulongval       |
+-------+-------+

For the following code snippet:

CP(mod->data, data32, longval);
CP(mod->data, data32, ulongval);

It only takes care of little endian platforms that it truncates the
highest 32bit automatically. However on big endian platforms it takes
the highest 32bit instead. This eventually returns a garbage syscall
number to the 32bit userland.

Since modspecific_t's usage currently is for the use of syscall modules,
we only initialize modspecific32_t with uintval. Now on both BE and LE
64-bit platforms it always pick up the first 4 bytes.

Sponsored by: Juniper Networks, Inc.

[delphij]

sys/kern/kern_module.c

525–526

Can you add a compile time assert that sizeof(data32) == sizeof(data32.uintval) so we don't need to zero data32?

(In fact I don't really like this approach, but it's probably the lesser evil without breaking ABI further. It would be much better if data is defined as a fixed width type on all platforms, but that would mean we would need v3 call.)

[markj]

Seems ok to me Xin's suggestion incorporated.

[markj]

In D40814#930219, @markj wrote:

Seems ok to me Xin's suggestion incorporated.

Well, thinking about it a bit further, I'd also add a comment explaining why we only copy the one field.