Re: kernel structures in 2.0.29->2.0.30

Chel van Gennip (linux@vangennip.nl)
Sun, 27 Apr 1997 12:18:26 +0100 (WETDST)


Doug Ledford <dledford@dialnet.net> wrote:
>> I suspect the syn fixes and network changes may have broken the binary only
>> drivers from ETinc for their WAN card. Fingers crossed it didnt as it shouldnt
>> be referencing socket fields that moved unless they are doing magic I dont
>> know about.
>
>At least with my kernel configuration, it survived. As did the module from
>Equinox.
>

This could be from linux/include/asm-i386/semaphore.h:
===============================
struct semaphore {
int count;
- int waiting;
+ int waking;
+ int lock ; /* to make waking testing atomic */
struct wait_queue * wait;
};

-#define MUTEX ((struct semaphore) { 1, 0, NULL })
-#define MUTEX_LOCKED ((struct semaphore) { 0, 0, NULL })
+#define MUTEX ((struct semaphore) { 1, 0, 0, NULL })
+#define MUTEX_LOCKED ((struct semaphore) { 0, 0, 0, NULL })
==============================
These kind of changes are killing for binary compatibility. For the Equinox
module recompilation at Equinox is needed to solve this problem.

Chel