interruptible

2003-12-17 Thread glen herrmannsfeldt
Normally, you only need to worry about access due to
other CPU's or channels.

For an interruptible instruction, you also need to worry
about access by other tasks on the same CPU.

In the olden days, the interval timer at address 80 could
be updated with an MVC instruction.  The new value was
stored at 84, and and 8 byte MVC move from 80 to 76 would
update the timer and return the previous value without
losing any counts.

MVCL would not necessarily prevent the loss of counts.

-- glen


Re: interruptible

2003-12-17 Thread Phil Payne
 In the olden days, the interval timer at address 80 could
 be updated with an MVC instruction.  The new value was
 stored at 84, and and 8 byte MVC move from 80 to 76 would
 update the timer and return the previous value without
 losing any counts.

I was never too sure about this.  On the 360/50 under heavy channel load I'm sure I 
used to
miss ticks.

--
  Phil Payne
  http://www.isham-research.com
  +44 7785 302 803


Re: interruptible

2003-12-17 Thread Ulrich Weigand
glen herrmannsfeldt wrote:

In the olden days, the interval timer at address 80 could
be updated with an MVC instruction.  The new value was
stored at 84, and and 8 byte MVC move from 80 to 76 would
update the timer and return the previous value without
losing any counts.

MVCL would not necessarily prevent the loss of counts.

I would suggest to consult the Principles of Operation, which has a whole
section (more than 10 pages) on exactly that topic:

5.13 Sequence of Storage References

and in particular

5.13.9 Storage Operand Consistency


Mit freundlichen Gruessen / Best Regards

Ulrich Weigand

--
  Dr. Ulrich Weigand
  Linux for S/390 Design  Development
  IBM Deutschland Entwicklung GmbH, Schoenaicher Str. 220, 71032 Boeblingen
  Phone: +49-7031/16-3727   ---   Email: [EMAIL PROTECTED]


Anyone using OpenLDAP with SLES8? Question with Passwords

2003-12-17 Thread Eric Sammons
If anyone out there is using LDAP v2.1 with SLES8 for their authentication
I am looking for assistance.  I seem to have everything configured and
working just fine except I seem to have a problem with the forcing of a
password change.  If you execute a passwd --help you will see that you can
force a password change for a user on next login with the -e argument.  I
execute this and get Change to login shell failed.  I am able as a user
and / or root to change a users password by typing passwd.  It seems
though the -e flag is causing issues.

Anyone have an idea what I need to do or how to address this?  And at this
time I am not running Kerberos with my LDAP, everything is done in the
LDAP.


thanks!
Eric Sammons
(804)697-3925
FRIT - Unix Systems


Re: FW: IBM iSource -- U.S. Announcements

2003-12-17 Thread Adam Thornton
On Wed, 2003-12-17 at 18:05, Alan Cox wrote:
 Wouldn't this depend on the price per instance and the number of
 instances, as well as the availability of both options ?

Well, I believe that there are few (note: I'm *not* saying no) good
reasons to run Linux/390 in any context other than VM.  And, so, no,
because if you're using Linux/390 in a cost-effective manner you're
almost certainly in a situation where you're using multiple instances
per CPU.

I suppose in theory the per-instance price could be low enough that this
wasn't a problem, but then, how do you ensure license compliance?  One
of the great things about VM is the ease with which you can set up and
then destroy instances.  How do you enforce a per-instance license in
that scenario, anyway?

Adam