[PATCH v1 0/2] Refine move_imm() and optimize memory load
From: Tiezhu Yang
Date: Mon Jul 27 2026 - 23:21:24 EST
This series is based on 7.2-rc5 and the following two series:
Fix and optimize BPF tailcall for LoongArch
https://lore.kernel.org/loongarch/20260707101508.32283-1-yangtiezhu@xxxxxxxxxxx/
LoongArch: BPF: clean up for build_insn() paths
https://lore.kernel.org/loongarch/20260711100956.4351-1-yangtiezhu@xxxxxxxxxxx/
The long-term goal is to eliminate frequent and redundant
zero-extensions and sign-extensions, but there are still
some issues to sort out. Here is the previous RFC patch:
LoongArch: BPF: Eliminate zero/sign extensions for 32-bit operations
https://lore.kernel.org/loongarch/20260711101656.7415-3-yangtiezhu@xxxxxxxxxxx/
For now, just send out the stable and verified patches
first.
Here are the test results of test_bpf:
fedora@linux:~$ sudo dmesg -c
fedora@linux:~$ sudo modprobe test_bpf
fedora@linux:~$ dmesg -t | grep Summary
test_bpf: Summary: 1061 PASSED, 0 FAILED, [1049/1049 JIT'ed]
test_bpf: test_tail_calls: Summary: 10 PASSED, 0 FAILED, [10/10 JIT'ed]
test_bpf: test_skb_segment: Summary: 2 PASSED, 0 FAILED
Tiezhu Yang (2):
LoongArch: BPF: Clean up and decouple is32 parameter from move_imm()
LoongArch: BPF: Eliminate zero-extension for memory load operations
arch/loongarch/net/bpf_jit.c | 123 ++++++++++++++++++++++-------------
arch/loongarch/net/bpf_jit.h | 7 +-
2 files changed, 79 insertions(+), 51 deletions(-)
--
2.42.0