io_uring

“The wins on the book for io-uring are incredible. Massive. It looks like a miracle.”

“io_uring is incredibly fast for async I/O but is also a massive security disaster in the Linux kernel.”

“I am typically getting double the perf just switching to uring.”

“The benefits of io_uring are limited without a commitment to designing your own schedulers.”

“This is where io_uring is really important because it lets one thread schedule I/O to satisfy one request and resume processing a different request.”

“io_uring lets that same thread continue to handle other requests which themselves might generate more I/O that gets amortized.”

“io_uring is significantly faster for certain workloads, but you can't expect that merely running a syscall via io_uring will somehow magically make it faster.”

“Do this with io_uring with the preadv syscall. It'll be the same or faster”

“using mmap and just telling uring to read form there to beat anything else”

“io_uring has been one of the greatest sources of exploits for the Linux kernel in years.”

“use io_uring as reference and cancellation in io_uring is fairly simple”

“Async I/O (and later just everything async) converged on io_uring.”

“A lot of the socket featureset of io_uring seems available in AF_XDP”

“genuinely a step up from the overlapped I/O APIs of Windows, as it avoids eating syscall transition overhead in many cases”

“"io_uring" (2019) have advanced the state of the art clearly beyond what already existed”

“at least now we have some decent answers (like io_uring on Linux.)”

“Linux support generic non-conventional kernel bypass for memory, storage and compute with io_uring, eBPF”

“Newer system calls like io_uring are doing it right.”

“there are robust low-syscall paths for io/latency sensitive operations with DPDK, io_uring, and futex”

“it's nice for things like SPI which have no user space non-blocking API”

“The catch is that its kernel opt-in and disabled just about everywhere for security reasons.”

“At the level you'd be exposing io_uring (internals) via external http requests; it's security is perfectly fine.”

“if syscalls are expensive, queue them up in a buffer with one syscall to transfer the thread to the os”

“io-uring is in a different league imo. I'd say that it's second only to user namespaces in terms of attack surface.”

“io-uring is a security nightmare. Constant privescs and a powerful primitive for syscall smuggling.”

“using APIs io_uring and grand central dispatch is basically the whole point of all this async stuff”

“io_uring is still significantly better than doing threadpool nvme io”

“hackers having fun with io_uring coming up with all sorts of patterns”

“io_uring largely eliminates this issue by decoupling submission from syscalls”

“the list of vulnerabilities in io_uring over the past 12 months is pretty staggering to me”

“it continues to be such a consistent source of bugs - many with serious security implications”

“goes into using eBPF and IO_URING to collect system stats instead of all the huge overhead of userspace calls”

“io_uring supports submitting openat requests, which sounds like what you want.”

“I admire io_uring. I appreciate the fact that it exists and continues despite the security problems”

“io_uring syscalls are blocked by default in Docker, because io_uring is a large surface area for attacks”

“Shows that io_uring really brings extra performance when properly used”

“io_uring is a large surface area for attacks, and this has proven to be a real problem in practice”

“This explained to me very simply what the benefits are in a way no article I've read before has.”

“Io-uring has it's fair amount of CVEs ; I'm wondering if people are checking these out”

“We found the async I/O overhead to be negligible compared to the memory gains”