🚀 go-pugleaf

I ran into this panic in Oct 1988 when I wrote a device driver that allows
a file  containing an image of a filesystem to be "mounted". Whenever I
use the "blockdevice" I have no problems at all. But when I use the "raw"
device extensively (several fsck(8) passes) the system panics with the
message:

vmunix: panic: hat_unload - pmg not kept

The routine hat_unload() is part of the new vm stuff in SunOS 4.0x. "Hat"
means hardware adress translation; it is used as part of the new type
caching for shared text and filesystem buffering. 

On the SUG conference on dec 1988 in Miami I asked a unix guru from sun
about that panic, and he told me that my driver must be wrong. But I
double checked my driver and I am shure that my driver is correct.

My opinion is, that the panic is a result of sun's vm stuff is doing
something wrong with the keepcount. But as I can only repeat it with my
driver I cannot report it to sun's hotline.

So everyone who has had this panic by only using sun code, please report
it to sun's "hotline" I hope they will beleve yout that's a bug.

        J. Schilling 
        H. Berthold AG 
        Teltowkanalstr. 1-4 
        D 1000 Berlin 46 
        +49 30  7795 - 400 


joerg@berthold.de 
... tub!berthold!joerg 
... unido!berthold!joerg