PG_Zero: Faster Page Allocation Proposed For Linux By Zeroing Out Pages Ahead Of Time

[phoronix]

Phoronix: PG_Zero: Faster Page Allocation Proposed For Linux By Zeroing Out Pages Ahead Of Time

A set of patches sent out Sunday morning for "PG_zero" could provide much faster page allocation performance by the Linux kernel...

http://www.phoronix.com/scan.php?page=news_item&px=PG-Zero-Linux-Proposal

[dimko]

Interesting, thank you Michael!

[RussianNeuroMancer]

This is cool, but caches cause more issues during memory allocation for virtual machines - memory allocation fail and then virtual machine does not start at all: https://github.com/OpenNebula/one/is...ment-591329265

[milkylainen]

Hmm. I thought pages were zeroed/poisoned on request/configuration, not by default?
What pages are we talking about? Any? Userspace? Heap? Stack? Kernel?
Also, does this not break the general laziness principle?

[dimko]

This is cool, but caches cause more issues during memory allocation for virtual machines - memory allocation fail and then virtual machine does not start at all: https://github.com/OpenNebula/one/is...ment-591329265

That is price of of having improvement of existing technology, that sometime new things break.

[duby229]

That is price of of having improvement of existing technology, that sometime new things break.

That's a huge problem on linux. You update your system and then something unexpected breaks. That's not ok. It's not a price end users are willing to pay.

[Volta]

That's a huge problem on linux. You update your system and then something unexpected breaks. That's not ok. It's not a price end users are willing to pay.

That's the problem when stupid user upgrades something on its own. Kernel, for example. On windows it's stupid microsoft that was forcing broken upgrades on users. Every unreleased Kernel should be signed: not for morons. Smart users know the risk.

[duby229]

That's the problem when stupid user upgrades something on its own. Kernel, for example. On windows it's stupid microsoft that was forcing broken upgrades on users. Every unreleased Kernel should be signed: not for morons. Smart users know the risk.

So "smart" users are ok with broken systems? You must be so "smart"....

[Volta]

So "smart" users are ok with broken systems? You must be so "smart"....

If you cannot read I recommend to stay away from features not even included in Kernel yet. It seems Freedom is only for smart.

[AH88]

That's a huge problem on linux. You update your system and then something unexpected breaks. That's not ok. It's not a price end users are willing to pay.

It's strange how some people have been using Linux for over a decade without experiencing any such problems, while others seem to think it's a "huge problem". Really makes you think......

(Probably yet another "power user", who knows just enough to be dangerous)