but people are seeing problems between builds of 3.4a. someone should
compare the 5.00 (or something pre-5.36) build of 3.4a to 5.67 and see if
there's any difference there.
------------------------------------------------------------------------
| Jeff Horwitz University of Michigan |
| [EMAIL PROTECTED] Ann Arbor |
| http://www-personal.umich.edu/~jhorwitz ITD Login Service |
------------------------------------------------------------------------
On Fri, 12 Mar 1999 09:57:42 -0500, Ted Anderson <[EMAIL PROTECTED]> said:
> I compared the 3.4 and 3.5 sources and found this difference in the
> ubik_client structure:
> /* 3.4 */
> /* per-client structure for ubik */
> struct ubik_client {
> short initializationState; /* ubik client init state */
> short states[MAXSERVERS]; /* state bits */
> struct rx_connection *conns[MAXSERVERS];
> };
>
> and
>
> /* 3.5 */
> /* per-client structure for ubik */
> struct ubik_client {
> short initializationState; /* ubik client init state */
> short states[MAXSERVERS]; /* state bits */
> struct rx_connection *conns[MAXSERVERS];
> int32 syncSite;
> #ifdef AFS_PTHREAD_ENV
> pthread_mutex_t cm;
> #endif
> };
>
> Very likely the core dump is caused by the output structure getting
> trashed because it isn't large enough to include the new 'cm' member.
>
> Ted
>
