Avoiding page reference-count overflows

Posted Apr 18, 2019 6:49 UTC (Thu) by willy (subscriber, #9762)
In reply to: Avoiding page reference-count overflows by mm7323
Parent article: Avoiding page reference-count overflows

You should write for The Register. Your ability to string together half-understood concepts with innuendo into a full-blown conspiracy theory would ensure your employment at a semi-mainstream news site dependent on sensationalistic clicks for a living. It is, however, unworthy of an LWN comment.

For the record, the original idea of using the negative reference count in this way was mine. I didn't consider switching to refcount_t for a second, not because I have a long-standing problem with Spender, but because it would be a ridiculous amount of change for a fix. And as my earlier comment said, it would have had significant downsides.

All the things you ask to be explained have already been explained in earlier responses; I'm not going to type them out again.

Avoiding page reference-count overflows

Posted Apr 19, 2019 7:44 UTC (Fri) by marcH (subscriber, #57642) [Link]

You can blame The Register for a number of things (even more if you take it seriously when it doesn't), but... "conspiracy theory", really? Citation needed.

Avoiding page reference-count overflows

Posted Apr 19, 2019 9:45 UTC (Fri) by mm7323 (subscriber, #87386) [Link] (1 responses)

You should write for The Register....

If you are going to tell me where I should write, may I respectfully suggest that as a leading linux kernel developer, perhaps you should read the CoC and at least choose a more respectful tone in your replies. This isn't the lkml and serves a more diverse audience. It's reasonable to expect different levels of understanding here.

It is, however, unworthy of an LWN comment.

That is just your opinion; I don't believe you speak on behalf of LWN.

it would be a ridiculous amount of change for a fix.

Having read and re-read the article and comments, that seems to be the crux of the argument for making a new mechanism. Apart from "refcount_t isn't the appropriate fix for this problem", I don't see where any critical technical limitation of refcount_t is stated.

Short-cutting the work to make this change might have been the best option here, but it's surely creating technical debt. For a "vulnerability [that] is sufficiently difficult to exploit that almost nobody should feel the need to rush out a kernel update", I hope it is worth it.

Thanks for all your hard work all the same

Avoiding page reference-count overflows

Posted Apr 22, 2019 23:17 UTC (Mon) by jschrod (subscriber, #1646) [Link]

> > It is, however, unworthy of an LWN comment.
> That is just your opinion; I don't believe you speak on behalf of LWN.

Of course, he didn't speak on behalf of LWN -- but he didn't claim to do so, too.

He spoke on behalf of us long-time LWN readers who don't want to see ad-hominem attacks or comments with random assumptions expressing negative connotations on a personal level, why kernel developers did implement a change in a certain way besides them explaining explicitly why it was done this way here in this LWN thread.

Please abstain with this kind of comments in the future. It's not welcome here.

Thank you.