Linux 6.13 Poised To Land Prep Patches Working Toward Proxy Execution

[phoronix]

Phoronix: Linux 6.13 Poised To Land Prep Patches Working Toward Proxy Execution

Years in the making has been the idea of Proxy Execution for the Linux kernel as a means of implementing priority inheritance by leveraging information from a task's scheduler context and its execution context. While the Proxy Execution patches themselves aren't yet queued for merging upstream, some prep patches look like they'll make it for the upcoming Linux 6.13 merge window...

https://www.phoronix.com/news/Linux-6.13-Prep-For-Proxy-Exec

[aviallon]

This is a very interesting approach to this very complex problem.

[waxhead]

I admit to not having read about the topic much but would not c groups be "just as good" of a solution?

[ATLief]

I admit to not having read about the topic much but would not c groups be "just as good" of a solution?

I’m similarly ignorant about Proxy Execution in particular, but my interpretation is that currently the CPU time that would have been allocated to p1 while it was blocking is instead evenly distributed among all eligible processes, whereas with PE, p2 would get p1’s resources. Since p1 is waiting for p2 to release the mutex, running p2 benefits both p1 and p2. As a result, p1 will get a more “fair” fraction of the resources even if it blocks a lot.

[DavidBrown]

This is a very interesting approach to this very complex problem.

It sounds more like an overly complex approach to an important problem. Basically, it's just "priority inheritance" that has been standard practice in real-time operating systems for decades (from long before Linux was conceived). The idea is that if a high priority task needs a resource that a low priority task has locked, the low priority task temporarily runs at the high priority until it has released the lock - that way the low priority task does not block the high priority task more than necessary.

Still, better late than never, and better this than the current solution based on luck.

[braiam]

Basically, it's just "priority inheritance" that has been standard practice in real-time operating systems for decades (from long before Linux was conceived).

In the video they note that it is also standard on Linux, but doesn't work for this particular problem. Timestamp 1:05 .

[DavidBrown]

In the video they note that it is also standard on Linux, but doesn't work for this particular problem. Timestamp 1:05 .

Thanks for the pointer. I didn't have the opportunity to look at the video properly earlier - hopefully I will see later on. I do realise that this sort of thing is hugely more complex on Linux than on a "typical" RTOS. But since Linux is also being used more and more in real-time systems, it's important that they get this right - and that it is simple enough that people can understand it and see that it is correct.

[NadiaEira]

Phoronix: Linux 6.13 Poised To Land Prep Patches Working Toward Proxy Execution

Years in the making has been the idea of Proxy Execution for the Linux kernel as a means of implementing priority inheritance by leveraging information from a task's scheduler context and its execution context. While the Proxy Execution patches themselves aren't yet queued for merging upstream, some prep patches look like they'll make it for the upcoming Linux 6.13 merge window...

cookie clicker

https://www.phoronix.com/news/Linux-...For-Proxy-Exec

This is a significant development for the Linux kernel, as the Proxy Execution concept has been in the works for years. The upcoming Linux 6.13 merge window looks promising with the prep patches geared towards Proxy Execution. This could potentially bring about improvements in priority inheritance by utilizing task scheduler and execution context information.