[6034] in Athena Bugs

home help back first fref pref prev next nref lref last post

Re: why does xlogin touch the kernel?

daemon@ATHENA.MIT.EDU (Richard Basch)
Fri Sep 14 08:35:02 1990

Date: Fri, 14 Sep 90 08:34:41 -0400
To: jfc@MIT.EDU
Cc: bugs@ATHENA.MIT.EDU
From: Richard Basch <probe@MIT.EDU>


  [6031] daemon@ATHENA.MIT.EDU  Athena Bugs  09/14/90 01:49 (16 lines)
  Subject: why does xlogin touch the kernel?
  To: bugs@ATHENA.MIT.EDU
  Date: Fri, 14 Sep 90 01:49:15 EDT
  From: John Carr <jfc@ATHENA.MIT.EDU>

  I ran xlogin manually and this message was printed:

          Unable find all the symbols in /vmunix.

  Invoking xlogin should not require that vmunix be readable, writeable, or the
  boot kernel.  This looks like a message printed by the 6.4->7.0 update kernel
  patcher.  Xlogin should not touch the configuration or kernel except when run
  from init or toehold on a public workstation which needs an update
  (/srvd/auto_update should exit immediately on finding $PUBLIC or $AUTOUPDATE
  false).

  --[6031]--

Your analysis is incomplete... this hack is for 6.4 machines using 7.1
packs (6.4 is still somewhat supported).  As soon as 6.4 support is
dropped, the hack will be removed from xlogin (which calls
/usr/athena/lib/update/sys_patch).  Under 6.4, there was no hook to
update the syscall table, and xlogin would hang without the invocation
of this program.

In other words, this "bug" is actually a backwards-compatibility feature
that needs to stay around for a while...

-Richard

home help back first fref pref prev next nref lref last post