> hi,
> 
> could you give me some more information about how UBC makes your shark
> crash more often?  I'd like to fix any problems, but there's not much
> I can do if people don't even report problems they experience.
> 
> -Chuck
> 
> 
> On Thu, Feb 15, 2001 at 02:06:26PM +0000, Richard Earnshaw wrote:
> > Yeah!  My sources are a bit out-of-date (cos the latest UBC code makes my 
> > shark crash more often -- it does it anyway, but less often with the old 
> > code).


I don't think it was UBC that caused the problem (it happens on 1.5I 
kernels as well, which predate the UBC merge).  It's just that with the 
heavier disk activity I was seeing with the UBC build I tried, the crashes 
happened sufficiently more often to make the machine unusable.  Machine 
only has 32M memory and is trying to link a large program (eg the new 
gcc-3.0 C++ shared libraries).

Note the kernel I was testing was a 1.5Q from around 10 January, there 
have been changes since then that might alleviate the problems, but I 
haven't had chance to try them.

Symptoms are that the machine locks solid (I think it's disk IO related, 
but it could be soft-dep or any of a number of other things).  I can't get 
any backtrace out of the machine since the console is dead.  Certainly the 
crash seems to happen most when there is heavy disk io involved.  I've 
seen reports that this might pre-date 1.5 (see PR port-arm32/11921 and 
port-arm32/9859).

R.