arlad dies

[Love]

André Zúquete <avz at sabrina.inesc.pt> writes:

> |
> |I'm intressed in the values of size and entry->status.Length when it
> |crashes, also a backtrace from gdb would be valuable. Xfs and arlad trace
> |(started with ``fs arladeb all ; fs xfsdebug all'') might be useful.
> 
> In the particular case of the makedepend that I've tryed, the values are:
> 
> entry->status.Length 0
> size                 16

Just to add a summery for what we found out when looking deeper at this
problem:

It seems that the fileserver (afs3.3 1.68;Nbapi-3.3-1.68.compile.fixes 1.1;)
that André used was a weird one.

When the equivalent of a truncate(2) was done to a size N, the fileserver
returned a diffrent size. And since we assert()ed on that the fileserver
should return the same since as re truncated too, arlad abort()ed.

When the assert() was removed rx_Read returned -1 (RX_CALL_DEAD) during
data transfers.

Anyone know how good/bad 3.3 was ? That is before my time (well, at least I
didn't know about AFS then).

Love