Re: Our compiler

[Sanzo Miyazawa <miyazawa@smlab.sci.gunma-u.ac.jp>]

>From lprylli@ens-lyon.fr Sun May  9 23:50:24 1999
>Sanzo Miyazawa writes:
> > 
> > 
> > os:	slink 2.1
> > pc:	ruffian (UX board)
> > 
> > Compilations of kernel-source (2.2.7-ac2) frequently failed
> > in the wide range of sources becasue of "internal error".
> > 
> > But "internal error" does not mean the presence of
> > bugs in those sources but in egcc, because those sources could be compiled
> > without any trouble.
>
>The other cause of such errrors are hardware problems, which seems to
>be quite probable in your case given the presence of "machine checks".
>
> > May  6 12:01:57 sma660 kernel:  +110 6 11fff9fc0
>
>I think the DCPERR registers equal to 6 means there was a parity error
>in the data cache. That could explains the internal errors, did you
>try any overcloking? If not you may try either to underclock a bit, or
>maybe switch to SRM which could have different hardware settings.
>
>Regards,
>
>Loic
>

Thanks for your mail.

My alpha is 21164A 660MHz made by Samsung; I think that
this one is not a clock-up cpu but one shipped by Samsung.

Since then, no such errors have occurred.
I replaced the kernel, so that these errors mighy be caused by the 
kernel.   
(Right now I am using 2.2.9; the previous versions are 2.2.5 and 2.0.35.)


Thanks you again.