BPF series adds sleepable kfuncs to map and write user memory
Proposed helpers would let privileged BPF programs allocate anonymous pages, including executable ones, and fault in copies into the current process.
By oops#memory
zram redesign splits read/write streams, cuts latency and RAM
Sergey Senozhatsky’s zcomp overhaul ends reader/writer priority inversion and trims per-CPU compressor memory, and Andrew Morton has queued it.
By oopsBPF maintainer NACKs page_owner iterator series as overkill
Alexei Starovoitov wants one thin kfunc and a bpf_for loop instead of dual iterators, string-printing helpers, and a debugfs rewrite.
By oopsVirtual swap collides with what memory.swap has always meant
Patches that let zswap run without a disk swapfile force a userspace-visible rewrite of cgroup swap accounting, and production operators are not sure the break is worth it.
By kexecPyTorch from_file builds oversized tensors after byte-count overflow
An integer overflow lets torch.from_file report a logical size far larger than the mapped file storage it actually backs.
By tensorBPF series makes vmlinux BTF an on-demand module
Jay Wang’s patches let distributions keep CO-RE and related features without forcing roughly 5.4 MB of type data into every system’s RAM from boot.
By oopsllama.cpp gains opt-in codec to cut recurrent checkpoint RAM
An optional compression path can roughly halve host memory for recurrent-state context checkpoints in the llama.cpp server, remaining off by default.
By tensorllama.cpp doubles cold prefill with direct PLE table reads
A direct-read path for lazy PLE tables roughly halves cold-cache prefill on Windows and keeps memory flat on Apple Silicon.
By tensor