CFM-68K Runtime Enabler Question...
Absinthe
I KNOW that I am risking 'losing data' when I use a program that requires
the Enabler (AOL 3.0) -- But does that mean simply that I may lose any
information not already saved to disk (in any program) because the Enabler
may cause a random crash? (which is something I can effectively guard
against to my satisfaction...)
OR, does it mean that by simply having it floating around in my extensions
folder--whether using programs that call upon it, or not--I run the risk of
some greater, subtler, more insidious form of corruption that can foul up
my entire life (or system)?
Enquiring minds wish to understand...
Tony Nelson
In article <absinthicum-ya02358...@news.earthlink.net>,
absin...@borealis.net (Absinthe) wrote:
Apple's Tech Note 1084, "Running CFM-68K Code at Interrupt Time: Is Your
Code at Risk?", says that the problem occurs if any CFM-68K code runs at
interrupt time (it can trash register A5 in whatever was interrupted), so
it can't affect anything if no CFM-68K code is running. Apple's new
Laserwriter driver (8.4?) uses CFM-68K (but I don't know if it has any
interrupt-level CFM-68K code) so you could potentially have the problem
anytime you printed.
My understanding is that 3rd party apps don't have this problem.
What you would see if you ran such code, as an app or for printing, is
occasional unexplained crashes, which could cause data loss in the app that
was crashed. Note that the app that is running at the time of the
interrupt is what crashes, which may not be where the CFM-68K code is.
IMO, Apple overreacted and caused unnecessary panic, and CFM-68K is now
dead forever. What self-respecting sysadmin would install any CFM-68K
stuff now?
____________________________________________________________________
TonyN.:' to...@tiac.net
'
David Schmitt
Tony Nelson wrote:
>
> Apple's Tech Note 1084, "Running CFM-68K Code at Interrupt Time: Is Your
> Code at Risk?", says that the problem occurs if any CFM-68K code runs at
> interrupt time (it can trash register A5 in whatever was interrupted), so
> it can't affect anything if no CFM-68K code is running. Apple's new
> Laserwriter driver (8.4?) uses CFM-68K (but I don't know if it has any
> interrupt-level CFM-68K code) so you could potentially have the problem
> anytime you printed.
>
> My understanding is that 3rd party apps don't have this problem.
>
> What you would see if you ran such code, as an app or for printing, is
> occasional unexplained crashes, which could cause data loss in the app that
> was crashed. Note that the app that is running at the time of the
> interrupt is what crashes, which may not be where the CFM-68K code is.
>
> IMO, Apple overreacted and caused unnecessary panic, and CFM-68K is now
> dead forever. What self-respecting sysadmin would install any CFM-68K
> stuff now?
> ____________________________________________________________________
> TonyN.:' to...@tiac.net
> '
This interpretation is not correct. The CFM-68K Runtime Enabler can only
be tripped by a CFM-dependent application. But tripping the bug doesn't
cause an immediate crash: it only places a bad value in one of the
address registers. Only when an application -- CFM-dependent or not --
then tries to use that address can there be problems. So, any
application CAN be affected by the bug (crash, data loss, etc.) without
actually using CFM code. Since a bad value is being written to an
address register, ANYTHING can happen.
Because of this danger, we decided it was best to recommend that
customers disable the extension -- the costs could be high, and it is
impossible to say "just don't do <whatever>" in a CFM-dependent
application. Boy, would we have loved to be able not have to make the
recommendation we did.
I would also counter that we did not overreact. After all, if we didn't
make this recommendation, what would we say to a customer who lost
important data? That we didn't think it was worth letting them know
about the bug and our recommended course of action? Would it perhaps be
better had we said nothing, only to say afterwards, "Sorry, we knew
about that but didn't want to overreact"? Or maybe we should have said,
"There's a bug, but just deal with it?"
No, instead we made it very clear that there is a bug, and what we
recommend customers do to avoid it.
I would think that a self-respecting sysadmin, as well as customers,
would appreciate this straightforward, honest response.
--
David Schmitt
Mac OS Product Marketing Manager
Apple Computer, Inc.
Charles W. Snyder, Jr.
Do you have any idea when the fix will be released?
--
Charles W. Snyder, Jr. (cs...@rmii.com)
"Nothing rings finer than a barbershop seventh"
David Schmitt
Charles W. Snyder, Jr. wrote:
>
>
> Do you have any idea when the fix will be released?
>
> --
> Charles W. Snyder, Jr. (cs...@rmii.com)
> "Nothing rings finer than a barbershop seventh"
We are not making any comments about when we expect to deliver a fix
until our engineering team has a solution in hand. Why? Because until
they know for certain that their approach works, any attempt at setting
a release date would be speculation, and we only want to give our
customers information they can rely on.
This is a difficult bug to fix, and it is taking time. Believe me, we
understand the frustration our 68k customers have at being unable to use
the new AOL 3.0 and IE 3.0 (and AOL and Microsoft have also been kindly
reminding us!). All I can do is ask that you bear with us, and also
thank you for your continued patience. We hope to be able to provide
more information soon.
Phil Stripling
Absinthe <absin...@borealis.net> wrote:
: I KNOW that I am risking 'losing data' when I use a program that requires
: the Enabler (AOL 3.0) -- But does that mean simply that I may lose any
: information not already saved to disk (in any program) because the Enabler
: may cause a random crash? (which is something I can effectively guard
: against to my satisfaction...)
: OR, does it mean that by simply having it floating around in my extensions
: folder--whether using programs that call upon it, or not--I run the risk of
: some greater, subtler, more insidious form of corruption that can foul up
: my entire life (or system)?
My understanding is that leaving it floating around in your extensions
folder is not a problem. However, leaving it available for _use_ by that
unexpected program that activates the CFM _may_ result in data loss.
I'm unclear myself on just what "data loss" encompasses. If it's loss of
what's not been saved to disk, who cares. I'm surfing, nothing is lost
that's irreplacable. But if "data loss" means corruption of existing
data, and who knows which data will be corrutped, then I'm not a happy
surfer. I've deleted the extension because I'm not willing to chance
corrupted data as opposed to just losing what hasn't been changed.
Proceed at your own risk. Apple is pretty severe in its warnings.
Phil
Greg McPherson
Phil Stripling (phi...@crl13.crl.com) wrote:
: Absinthe <absin...@borealis.net> wrote:
: : I KNOW that I am risking 'losing data' when I use a program that requires
: : the Enabler (AOL 3.0) -- But does that mean simply that I may lose any
: : information not already saved to disk (in any program) because the Enabler
: : may cause a random crash? (which is something I can effectively guard
: : against to my satisfaction...)
: unexpected program that activates the CFM _may_ result in data loss.
As a CyberDog 68K user, I've experienced the CFM crash many times.
Often it will hang your whole system. Sometines CyberDog would survive
and crash ANOTHER program. Either way, you better not have been writing
some CodeWarrior code, because suddenly a hours work is down the drain.
Anyway, it IS a serious bug, but it doesnt affect all CFM programs.
I've seen some companies (such as Metrowerks) point out that the bug does
not affect them. I have no idea if AOL 3.0 is affected. Ask them.
greg
Tony Nelson
> Tony Nelson wrote:
> >
> > Apple's Tech Note 1084, "Running CFM-68K Code at Interrupt Time: Is Your
> > Code at Risk?", says that the problem occurs if any CFM-68K code runs at
> > interrupt time (it can trash register A5 in whatever was interrupted), so
> > it can't affect anything if no CFM-68K code is running. ...
>
> This interpretation is not correct. The CFM-68K Runtime Enabler can only
> be tripped by a CFM-dependent application. But tripping the bug doesn't
> cause an immediate crash: it only places a bad value in one of the
> address registers. ...
Although this fellow works at Apple, and is also in a position to know
better, he is not correct according to other contacts (programmers) within
Apple. TN 1084 is correct, and all that can happen is a system crash, and
the only data loss is what you haven't saved.
If you have been experiencing crashes using CFM-68K, stop using the apps or
whatever that call CFM-68K at interrupt time (if only there were a list).
If you have not been experiencing crashes, you are probably using safe
apps.
Metrowerks has stated in c.s.m.p.cw that the Codewarrior Debugger for 68K
is safe.
____________________________________________________________________
TonyN.:' to...@tiac.net
'
Nevin :-] Liber
In article <5bfle5$7ce$1...@aggedor.rmit.edu.au>, rs...@minyos.its.rmit.EDU.AU
(Greg McPherson) wrote:
> As a CyberDog 68K user, I've experienced the CFM crash many times.
> Often it will hang your whole system. Sometines CyberDog would survive
> and crash ANOTHER program.
And the way that you traced this problem to the CFM-68K design flaw is??
--
Nevin ":-)" Liber <mailto:ne...@CS.Arizona.EDU> (520) 293-2799
<http://www.cs.arizona.edu/people/nevin/>
David Schmitt
Tony Nelson wrote:
>
>
> Although this fellow works at Apple, and is also in a position to know
> better, he is not correct according to other contacts (programmers) within
> Apple. TN 1084 is correct, and all that can happen is a system crash, and
> the only data loss is what you haven't saved.
>
> If you have been experiencing crashes using CFM-68K, stop using the apps or
> whatever that call CFM-68K at interrupt time (if only there were a list).
> If you have not been experiencing crashes, you are probably using safe
> apps.
>
> Metrowerks has stated in c.s.m.p.cw that the Codewarrior Debugger for 68K
> is safe.
> ____________________________________________________________________
> TonyN.:' to...@tiac.net
> '
I want to avoid getting into a back and forth sparring over this, so
here is my final statement on the matter of whether or not I am telling
the truth:
I've had dozens of conversations with engineers about this. I
reconfirmed my statements this morning with the lead engineer on the
team working to fix the bug (the same engineer who wrote the
aforementioned tech note). My statements on CFM have been correct, and I
continue to stand by them.
At this point, all I can say is: ask yourself why we would want to make
this information public and deal with the myriad questions from
customers and developers. Make no mistake: our statements on CFM have
been made to ensure our customers understand the problem and the risks
and to recommend a course of action that will ensure they avoid the
risks.
Mr. Nelson obviously thinks otherwise, and is entitled to his opinion.
The rest of you can decide who to believe.
--
David Schmitt
Mac OS Product Marketing Consiparacy Manager
Apple Computer, Inc.
Glen Yates
In article <5be04t$6...@nexp.crl.com>, Phil Stripling <phi...@crl13.crl.com>
wrote:
> Absinthe <absin...@borealis.net> wrote:
> : I KNOW that I am risking 'losing data' when I use a program that requires
> : the Enabler (AOL 3.0) -- But does that mean simply that I may lose any
> : information not already saved to disk (in any program) because the Enabler
> : may cause a random crash? (which is something I can effectively guard
> : against to my satisfaction...)
>
> : OR, does it mean that by simply having it floating around in my extensions
> : folder--whether using programs that call upon it, or not--I run the risk of
> : some greater, subtler, more insidious form of corruption that can foul up
> : my entire life (or system)?
>
> My understanding is that leaving it floating around in your extensions
> folder is not a problem. However, leaving it available for _use_ by that
> unexpected program that activates the CFM _may_ result in data loss.
>
> I'm unclear myself on just what "data loss" encompasses. If it's loss of
> what's not been saved to disk, who cares. I'm surfing, nothing is lost
> that's irreplacable. But if "data loss" means corruption of existing
> data, and who knows which data will be corrutped, then I'm not a happy
> surfer. I've deleted the extension because I'm not willing to chance
> corrupted data as opposed to just losing what hasn't been changed.
>
> Proceed at your own risk. Apple is pretty severe in its warnings.
Apple is severe in their warnings for a good reason, let me relate my
recent experience with CFM-68k:
Let me preface by saying that I've used CFM-68k for several months with no
problems, but the only programs that called on it were Laserwriter 8.4 and
Mac Runtime for JAVA. Note that MRJ was used sparingly, while I print about
once a day.
Then last week - for some stupid reason - I decided to download Micro$oft
Internet Explorer 3, which within the first hour of use, while trying to
download something, crashed hard. After rebooting I could no longer access
the network, I tried to open the TCP/IP control panel but it wouldn't open.
I looked in the extensions folder and noticed that several files were
missing, especially all the Open Transport libraries.
Well, I ran Disk First Aid and it reported problems that it could not fix,
so I tried Norton Utilities (2.0 not 3.x) which ended up totally trashing
my HD. I ended up reformatting and restoring from a backup.
In short, I would be very leary of CFM-68k, and if I did use it again,
would only use it with LaserWriter 8.4, nothing else.
--
Glen Yates
Software Engineer -- To Reply remove .NOSPAM from my address
Tony Nelson
In article <glen_yates-ya024080...@news.saic.com>,
glen_...@cpqm.mail.NOSPAM.saic.com (Glen Yates) wrote:
> Then last week - for some stupid reason - I decided to download Micro$oft
> Internet Explorer 3, which within the first hour of use, while trying to
> download something, crashed hard.
What does this have to do with CFM-68K?
____________________________________________________________________
TonyN.:' to...@tiac.net
'
David Schmitt
Peter Kerr wrote:
>
> 1. <http://devworld.apple.com/dev/technotes/tn/tn1084.html>
> URL as found at <http://17.126.23.20/dev/technotes/cfm/cfm-1.html>
> returns:
> The URL you requested was not found on this server
>
> So have Apple pulled that technote now? for re-writing?
> or is it just my web connection?
>
> 2. If the only CFM compliant/requiring app that I run is LW8.4.1
> then I shouldn't have trouble, because nothing else will call CFM-68K
> at interrupt time, will it?
>
> I have been running LW8.4.1 with CFM-68K for over 6 weeks now
> on a LC3 system 7.5.3, and LW8.4.1 has not crashed yet, and even if
> it did all I think I would lose, data-wise, would be the current
> print job.
>
> Oh, I do make a habit of stripping all "fat" apps thru Fat-Free
> to make them 68K only code when I install. Perhaps this is why
> I haven't crashed yet?
>
> --
> Peter Kerr
I just checked the link as well and it is indeed broken. *That* will
bring out the conspiracy theorists! I will check in to it. Looks like
Devworld rearranged its site and the page got orphaned.
In any case, running one CFM-based program does not make you immune from
the bug, and stripping out PPC code doesn't change anything either.
Sorry to be the bearer of unpleasant news.
--
David Schmitt
Mac OS Product Marketing Manager
Apple Computer, Inc.
Peter Kerr
> Tony Nelson wrote:
> > If you have been experiencing crashes using CFM-68K, stop using the apps or
> > whatever that call CFM-68K at interrupt time (if only there were a list).
> > If you have not been experiencing crashes, you are probably using safe
> > apps.
> >
1. <http://devworld.apple.com/dev/technotes/tn/tn1084.html>
URL as found at <http://17.126.23.20/dev/technotes/cfm/cfm-1.html>
returns:
The URL you requested was not found on this server
So have Apple pulled that technote now? for re-writing?
or is it just my web connection?
2. If the only CFM compliant/requiring app that I run is LW8.4.1
then I shouldn't have trouble, because nothing else will call CFM-68K
at interrupt time, will it?
I have been running LW8.4.1 with CFM-68K for over 6 weeks now
on a LC3 system 7.5.3, and LW8.4.1 has not crashed yet, and even if
it did all I think I would lose, data-wise, would be the current
print job.
Oh, I do make a habit of stripping all "fat" apps thru Fat-Free
to make them 68K only code when I install. Perhaps this is why
I haven't crashed yet?
--
Peter Kerr bodger
School of Music chandler
University of Auckland NZ neo-Luddite
David Schmitt
AOL 3.0 requires the CFM-68K Runtime Enabler extension on 68k computers.
Michael Thews
Tony Nelson <to...@tiac.net> wrote:
> In article <glen_yates-ya024080...@news.saic.com>,
> glen_...@cpqm.mail.NOSPAM.saic.com (Glen Yates) wrote:
>
> > Then last week - for some stupid reason - I decided to download Micro$oft
> > Internet Explorer 3, which within the first hour of use, while trying to
> > download something, crashed hard.
>
> What does this have to do with CFM-68K?
> ____________________________________________________________________
> TonyN.:' to...@tiac.net
> '
MSIE 3.0 uses and requires the CFM-68k (on 680x0 machines only, of
course).
Michael
--
Dr. Michael Thews
University of Mainz - FB 03 - 55099 Mainz - Germany
"People who like sausages or the law shouldn't be allowed
to watch either one being made."
Greg McPherson
Nevin ":-]" Liber (ne...@CS.Arizona.EDU) wrote:
: In article <5bfle5$7ce$1...@aggedor.rmit.edu.au>, rs...@minyos.its.rmit.EDU.AU
The CFM bug occurs when CFM is used at interrupt time.
A typical use of interrupts is to signal a timeout... if an attempt to
connect to a site has taken too long.
The often when one of these "crash another program" bugs occurs CyberDog
will have beeped and popped up a "Could not connect" message.
You have to be running MacsBug by the way, or the entire Mac crashes.
greg
Glen Yates
In article <tonyn-ya02348000...@news.tiac.net>, to...@tiac.net
(Tony Nelson) wrote:
> In article <glen_yates-ya024080...@news.saic.com>,
> glen_...@cpqm.mail.NOSPAM.saic.com (Glen Yates) wrote:
>
> > Then last week - for some stupid reason - I decided to download Micro$oft
> > Internet Explorer 3, which within the first hour of use, while trying to
> > download something, crashed hard.
>
> What does this have to do with CFM-68K?
Internet Explorer 3 uses CFM-68k, and the crash was traced to CFM-68k use
in the same way that Greg McPherson (in a previous post) traced crashes
attributable to Cyberdogs CFM-68k use.
Peter Kerr
In article <p.kerr-1601...@news.auckland.ac.nz>,
p.k...@auckland.ac.nz (Peter Kerr) wrote:
> I have been running LW8.4.1 with CFM-68K for over 6 weeks now
> on a LC3 system 7.5.3, and LW8.4.1 has not crashed yet, and even if
> it did all I think I would lose, data-wise, would be the current
> print job.
Partial retraction:
In article <32D878...@apple.com>, dsch...@apple.com wrote:
> The CFM-68K Runtime Enabler can only
> be tripped by a CFM-dependent application. But tripping the bug doesn't
> cause an immediate crash: it only places a bad value in one of the
> address registers. ...
I now believe I may have a reproducible CFM-dependent crash: only if I
attempt to recover a news article after a search on DejaNews using
MacWeb 1.1.1E it goes off into Hyperspace (invalid instruction at xxxx,
MacsBug shows that part of memory to be all FFFF)
Now I am not running NetScrape, PhotoSlop, MS Word 6.6.6.
I am happy with Word 4, Claris Works 4, a whole slew of communications tools,
music notation, sound synthesis and digital editing apps, AppleShare via a
Novell server and TCP thru a unix proxy.
I am still confident that I can give up DejaNews for the benefit of LW8.4
Dr Terri Allen
>> Tony Nelson wrote:
>> > If you have been experiencing crashes using CFM-68K, stop using the apps or
>> > whatever that call CFM-68K at interrupt time (if only there were a list).
> if only ^^^^^^^^^^^^^^^^^^^^^^^^^
>> > If you have not been experiencing crashes, you are probably using safe
>> > apps.
>> >
>
>1. <http://devworld.apple.com/dev/technotes/tn/tn1084.html>
>URL as found at <http://17.126.23.20/dev/technotes/cfm/cfm-1.html>
>returns:
>The URL you requested was not found on this server
>
>So have Apple pulled that technote now? for re-writing?
>or is it just my web connection?
>
>2. If the only CFM compliant/requiring app that I run is LW8.4.1
>then I shouldn't have trouble, because nothing else will call CFM-68K
>at interrupt time, will it?
>
>I have been running LW8.4.1 with CFM-68K for over 6 weeks now
>on a LC3 system 7.5.3, and LW8.4.1 has not crashed yet, and even if
>it did all I think I would lose, data-wise, would be the current
>print job.
>
>Oh, I do make a habit of stripping all "fat" apps thru Fat-Free
>to make them 68K only code when I install. Perhaps this is why
>I haven't crashed yet?
>
>--
>Peter Kerr bodger
Peter
I also have continued to use the CFM-68K with LW 8.4.1 and 7.5.5 I have not
used Fat Free. No crashes, no problems....
Terri Allen
Austin and Repat Medical Centre, Uni of Melbourne, Australia
Greg McPherson
Dr Terri Allen (te...@austin.unimelb.edu.au) wrote:
: In article <p.kerr-1601...@news.auckland.ac.nz>,
: p.k...@auckland.ac.nz (Peter Kerr) wrote:
:
: >2. If the only CFM compliant/requiring app that I run is LW8.4.1
: >then I shouldn't have trouble, because nothing else will call CFM-68K
: >at interrupt time, will it?
: >I have been running LW8.4.1 with CFM-68K for over 6 weeks now
: >on a LC3 system 7.5.3, and LW8.4.1 has not crashed yet, and even if
: >it did all I think I would lose, data-wise, would be the current
: >print job.
: I also have continued to use the CFM-68K with LW 8.4.1 and 7.5.5 I have not
: used Fat Free. No crashes, no problems....
Maybe interrupts are rare in MW 8.4.1.
Try this: Start a print job, then rip out your ethernet/localtalk cable
so that the job will eventually fail.
Did this crash your mac? If so, then interrupts are being used to deal
with bad connections.
Of course your mac might not crash, it might just go subtly mad.
greg
Edsilberg
I don't now the answer to your questions, but after over 50 forced crashes
in 3 weeks, I am ready to call it quits on AOL and AOL 3.0 till the
problem gets fixed: funny things are starting to happen to my other
programs!!