Arm Exploring IO_uring For Graphics Drivers For Better Performance & Synchronization

[debrouxl]

Eh, why not, as long as it remains optional and the Linux kernel remains fully functional without io_uring and its steady stream of serious security vulnerabilities already mentioned by other commenters in this thread. Performance isn't everything.

[oleid]

It's all logic bugs in a highly concurrent subsystem. This might be one of the times where Rust genuinely won't help much (I'm obviously not Jens Axboe though, so this is me talking out of my ass).

Maybe, maybe not. I recall people having a hard time wrapping the io_uring API safely in rust. But these days you can simply use tokio_iouring in rust and profit from io_uring

[varikonniemi]

It's all logic bugs in a highly concurrent subsystem. This might be one of the times where Rust genuinely won't help much (I'm obviously not Jens Axboe though, so this is me talking out of my ass).

AFAIK Rust should also help in making safe concurrency of which multithreading is only one aspect.

[intelfx]

When google has paid out a million dollars in uring bug bounties before removing it from android it seems that it's just too hard for a team of humans to wrap their head around to make it work properly. Maybe this is a case where Rust rewrite would be an actual requirement to make it work properly.

It's all logic bugs in a highly concurrent subsystem. This might be one of the times where Rust genuinely won't help much (I'm obviously not Jens Axboe though, so this is me talking out of my ass).

[varikonniemi]

Given my personal experience with IO_uring on the file systems we had it going on... nothing but bad will happen here. We had to disable IO_uring in the end to keep the file system driver from crapping out in production.

Nothing ever corrupted (thankfully) but we could not get it run stably (past a few weeks at a time).

When google has paid out a million dollars in uring bug bounties before removing it from android it seems that it's just too hard for a team of humans to wrap their head around to make it work properly. Maybe this is a case where Rust rewrite would be an actual requirement to make it work properly.

[oleid]

Given my personal experience with IO_uring on the file systems we had it going on... nothing but bad will happen here. We had to disable IO_uring in the end to keep the file system driver from crapping out in production.

Nothing ever corrupted (thankfully) but we could not get it run stably (past a few weeks at a time).

Interesting. What programming language/library/framework combination did you use? Was it bare io_uring?

I was wondering because depending on what you're using, you don't have to program io_uring directly, just use something that uses io_uring under the hood. That is usually quite easy.

[Mark Rose]

Given my personal experience with IO_uring on the file systems we had it going on... nothing but bad will happen here. We had to disable IO_uring in the end to keep the file system driver from crapping out in production.

Nothing ever corrupted (thankfully) but we could not get it run stably (past a few weeks at a time).

Have you reported the bug?

[zexelon]

Given my personal experience with IO_uring on the file systems we had it going on... nothing but bad will happen here. We had to disable IO_uring in the end to keep the file system driver from crapping out in production.

Nothing ever corrupted (thankfully) but we could not get it run stably (past a few weeks at a time).

[coder]

Does this mean we're now going to have a string of articles about some Axboe-type achieving ever greater numbers of draw calls per second?
🤩

[aviallon]

IO_uring go brrrr