They defined a stable ABI for kernel modules and hardware support will be done primarily via out-of-tree modules targeting a stable ABI for an LTS branch. ABI can't be at all stable across LTS versions. They won't convince Linux to do anything like that any time soon.
Conversation
github.com/GrapheneOS/ker (just that, no userspace code was dropped) was replaced with literally thousands of lines of code in userspace across multiple components to dynamically manage eBPF rules. What they mean by upstream first is making about a dozen sacrifices like that.
1
2
A particularly painful example is that they switched from ashmem to memfd despite losing the ability to unpin/pin pages similarly to MEM_RESET and MEM_RESET_UNDO on Linux where you can tell the kernel it can free that memory if there's memory pressure but you can ask for it back.
1
1
This was used by web browsers to implement volatile caching aware of memory pressure on Android as they do on Windows since the kernel takes care of it and purges the caches. No such thing on non-Android Linux and now it will be gone on Android, increasing browser memory use.
1
1
They got close enough to not having any required Android common kernel code for them to decide to make these final dozen sacrifices to be able to directly use kernel.org kernels with backported bug fixes, performance improvements and in some cases actual features.
2
1
Replying to
I think there still some active development for proactive reclaim / pageout? But maybe not the part where you are able to undo that...
1
Replying to
I helped get MADV_FREE landed and that's a form of lazy free designed for use with malloc/free and similar systems. It does result in the weird property of uninitialized malloc memory being able to change from non-zero data to zero data, but that's how GCC/LLVM already handle it.
1
1
It doesn't drop the pages once writes happen to them via hardware support for detecting that they became dirty. On hardware without support, it can just map to MADV_DONTNEED. Use case for Android ASHMEM_UNPIN/ASHMEM_PIN or Windows MEM_RESET/MEM_RESET_UNDO is mostly different.
1
You generate data you need, use it, and then ASHMEM_UNPIN/MEM_RESET. Later, when you need it again, you do ASHMEM_PIN/MEM_RESET_UNDO. Between those, the kernel was able to drop the memory if there was memory pressure. Extremely useful for a lot of often large in-memory caches.
1
It's possible to poorly mimic this by setting up callbacks that are wired up to be called when the kernel reports there's memory pressure. It's far better if the kernel has been told that it can drop the pages so that it can drop them immediately including not swapping them out.
1
Android has never used swap but does use zram and more recently daily zram writeback where pages that have been in zram for a whole day without being needed are written out on a daily basis. It marks 1 hour after boot, then write out 23 hours later + marks again, then every 24h.