Subject: Re: panic in vmapbuf
To: None <current-users@sun-lamp.cs.berkeley.edu,>
From: None <mycroft@gnu.ai.mit.edu>
List: current-users
Date: 06/26/1994 01:54:53
It happens for the following reason:

The pages you're reading into are zero pages.  Normally, vslock() forces
them to be preallocated, and thus vmapbuf() doesn't find any null page
frames and barf.

However, when the I/O request is > MAXPHYS, the minphys routine truncates
the request.  The call to vslock() uses the new size, but the call to
vmapbuf() is still using the old (larger) size.  Thus, vslock() didn't
allocate enough zero pages for vmapbuf() in this case.

Obviously this code needs to be thought out a bit more.


------------------------------------------------------------------------------