Since I know several people here have been following the development of Rust concerning object lifetime and memory safety vs. (segmented) stacks, I figured this might be interesting:

http://cmr.github.io/blog/2013/10/21/on-stack-safety/

It's a new blog post by Corey Richardson, discussing how to mitigate memory corruption due to stack overflows.

David

David Nadlinger:

> http://cmr.github.io/blog/2013/10/21/on-stack-safety/
>
> It's a new blog post by Corey Richardson, discussing how to mitigate memory corruption due to stack overflows.

(Sorry for the late answer).

Very nice article, it presents a interesting hybrid strategy. It seems Rust designers have understood some of the disadvantages of the segmented stacks Walter wrote about.

D used to print a "stack overflow", and it was a nice thing. Is D using a a single guard page after the stack of a thread? If this is true can't D intercept the page fault and print the that error message again?

Bye,
bearophile

On Tuesday, 5 November 2013 at 15:25:21 UTC, bearophile wrote:
> David Nadlinger:
>
>> http://cmr.github.io/blog/2013/10/21/on-stack-safety/
>>
>> It's a new blog post by Corey Richardson, discussing how to mitigate memory corruption due to stack overflows.
>
> Very nice article, it presents a interesting hybrid strategy. It seems Rust designers have understood some of the disadvantages of the segmented stacks Walter wrote about.


They removed segmented stacks completely now.

"The new runtime does not implement segmented stacks and it never will."

https://mail.mozilla.org/pipermail/rust-dev/2013-November/006314.html

On 11/5/13 7:50 AM, qznc wrote:
> On Tuesday, 5 November 2013 at 15:25:21 UTC, bearophile wrote:
>> David Nadlinger:
>>
>>> http://cmr.github.io/blog/2013/10/21/on-stack-safety/
>>>
>>> It's a new blog post by Corey Richardson, discussing how to mitigate
>>> memory corruption due to stack overflows.
>>
>> Very nice article, it presents a interesting hybrid strategy. It seems
>> Rust designers have understood some of the disadvantages of the
>> segmented stacks Walter wrote about.
>
>
> They removed segmented stacks completely now.
>
> "The new runtime does not implement segmented stacks and it never will."
>
> https://mail.mozilla.org/pipermail/rust-dev/2013-November/006314.html

Well now at least we'll get bearophile off of our backs touting segmented stacks, using Rust as the sole argument they're good :o).

Andrei

On 11/5/13, bearophile <bearophileHUGS@lycos.com> wrote:
> D used to print a "stack overflow", and it was a nice thing.

Yes I miss this too. Issue 6088 which you've filed: http://d.puremagic.com/issues/show_bug.cgi?id=6088

Now whenever I see a really large negative number returned from the app I just assume it was a stack overflow.

qznc:

> They removed segmented stacks completely now.

And it seems the annotation system too was not appreciated:

> Avoiding this overhead would require an elaborate and inherently unsafe system of annotation (#8822), increasing a burden on the FFI interface.

Bye,
bearophile

On Tue, 2013-11-05 at 08:12 -0800, Andrei Alexandrescu wrote: […]
> Well now at least we'll get bearophile off of our backs touting segmented stacks, using Rust as the sole argument they're good :o).

I guess we could gs/Rust/Go/g. Go uses segmented stacks and is proud of it. For Go it is likely a good architecture, D and Rust have different features and needs.

-- 
Russel. ============================================================================= Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder@ekiga.net 41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel@winder.org.uk London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder

On 11/6/13 4:47 AM, Russel Winder wrote:
> On Tue, 2013-11-05 at 08:12 -0800, Andrei Alexandrescu wrote:
> […]
>> Well now at least we'll get bearophile off of our backs touting
>> segmented stacks, using Rust as the sole argument they're good :o).
>
> I guess we could gs/Rust/Go/g. Go uses segmented stacks and is proud of
> it.

Go is also abandoning segmented stacks: http://goo.gl/fkCvXd

Walter deserves all credit for calling this one.


Andrei

On 11/6/2013 10:33 AM, Andrei Alexandrescu wrote:
> Walter deserves all credit for calling this one.

I'm right twice a day!

On 06.11.2013 23:06, Walter Bright wrote:
> On 11/6/2013 10:33 AM, Andrei Alexandrescu wrote:
>> Walter deserves all credit for calling this one.
>
> I'm right twice a day!
>
If you ignore the B, you're right all the time!

-- 
  Simen