[Info-vax] September 6, 2016 - new Roadmap and State of the Port updates now on VSI website

[Simon Clubley]

On 2016-09-07, clairgrant71 at gmail.com <clairgrant71 at gmail.com> wrote:
> On Wednesday, September 7, 2016 at 2:51:30 PM UTC-4, Simon Clubley wrote:
>> 
>> Thank you Clair (and Paul for the links).
>> 
>> Some very interesting reading indeed.
>> 
>
> A few answers...
>

Thanks, Clair.

>
>> From the Roadmap:
>> 
>> I see we are about 18 months to 2 years away from an Early Adopters
>> x86-64 release although the new file system looks like it's going to
>> show up earlier in IA64 (but with the current 32 bit size limitations).
>> 
>
> The new file system is really two pieces of work: 1) improving performance and
> 2) removing the 2TB volume limit. The latter is dependent on another
> project which we call 64bLBN which increases the size of the logical block number.
> This work is in progress now but must be in a major, i.e x.0, release. But the
> performance aspect will be mostly completed earlier so we see no reason not to
> get it out when it is ready.
>

Thanks; I hadn't fully realised there was a distinct two stage
implementation approach with the new file system.

>> I wonder what "Enhanced Application Isolation and Management" means ?
>> Is it SELinux MAC type security or more a chroot/jail type security ?
>>
>
> MAC is not what I have in mind. I wanted to get away from the term "containers",
> too much baggage, too many preconceived notions, but expanding and
> adding to some existing VMS features should get us a more managed application
> environment. Right, very vague, but that's the starting point for now.
>  

It will be interesting to see what emerges here.

>  
>> I also like the additional emphasis on security in years to come.
>> 
>
> That was intentional. The list of projects will get bigger.
>

Good. That's nice to see.

>> 
>> With the two processor mode situation, are you far enough along with
>> the port to get a feeling for the overheads of emulating executive mode
>> when it comes to switching page tables on the user mode -> executive
>> mode -> kernel mode -> executive mode -> user mode path while using RMS ?
>> 
>> I'm wondering just how painful TLB invalidation has proven to be in
>> real life when you are switching page tables. I know that was an early
>> concern of mine but others pointed out that x86-64 had features (the
>> PCID support) which I was unaware of and which could reduce the pain
>> dramatically.
>> 
>
> Too early for any of that. This was a pain to do on Itanium and we know the
> details will be different on x86. We'll save our energy for the real thing.
>

Understood. I'll be patient. :-)

Thanks once again for the documents. I think the time spent on them
is well worth it because they give the VMS community a strong sense
of how things are progressing with the port (and your future plans)
instead of just being kept hanging.

Simon.

-- 
Simon Clubley, clubley at remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world