7852ebc088
Before llvm20, (void)__sync_fetch_and_add(...) always generates locked xadd insns. In linux kernel upstream discussion [1], it is found that for arm64 architecture, the original semantics of (void)__sync_fetch_and_add(...), i.e., __atomic_fetch_add(...), is preferred in order for jit to emit proper native barrier insns. In llvm commits [2] and [3], (void)__sync_fetch_and_add(...) will generate the following insns: - for cpu v1/v2: locked xadd insns to keep backward compatibility - for cpu v3/v4: __atomic_fetch_add() insns To ensure proper barrier semantics for (void)__sync_fetch_and_add(...), cpu v3/v4 is recommended. This patch enables cpu=v3 as the default cpu version. For users wanting to use cpu v1, -mcpu=v1 needs to be explicitly added to clang/llc command line. [1] https://lore.kernel.org/bpf/ZqqiQQWRnz7H93Hc@google.com/T/#mb68d67bc8f39e35a0c3db52468b9de59b79f021f [2] https://github.com/llvm/llvm-project/pull/101428 [3] https://github.com/llvm/llvm-project/pull/106494
41 lines
1.1 KiB
LLVM
41 lines
1.1 KiB
LLVM
; RUN: llc < %s -march=bpf -mcpu=v1 -verify-machineinstrs | FileCheck %s
|
|
; Source Code:
|
|
; struct loc_prog {
|
|
; unsigned int ip;
|
|
; int len;
|
|
; };
|
|
; int exec_prog(struct loc_prog *prog) {
|
|
; if (prog->ip < prog->len) {
|
|
; int x = prog->ip;
|
|
; if (x < 3)
|
|
; prog->ip += 2;
|
|
; }
|
|
; return 3;
|
|
; }
|
|
; Compilation flag:
|
|
; clang -target bpf -O2 -S -emit-llvm t.c
|
|
|
|
%struct.loc_prog = type { i32, i32 }
|
|
|
|
; Function Attrs: nofree norecurse nounwind willreturn
|
|
define dso_local i32 @exec_prog(ptr nocapture %prog) local_unnamed_addr {
|
|
entry:
|
|
%0 = load i32, ptr %prog, align 4
|
|
%len = getelementptr inbounds %struct.loc_prog, ptr %prog, i64 0, i32 1
|
|
%1 = load i32, ptr %len, align 4
|
|
%cmp = icmp ult i32 %0, %1
|
|
%cmp2 = icmp slt i32 %0, 3
|
|
%or.cond = and i1 %cmp2, %cmp
|
|
; CHECK: r{{[0-9]+}} <<= 32
|
|
; CHECK: r{{[0-9]+}} s>>= 32
|
|
br i1 %or.cond, label %if.then3, label %if.end5
|
|
|
|
if.then3: ; preds = %entry
|
|
%add = add nsw i32 %0, 2
|
|
store i32 %add, ptr %prog, align 4
|
|
br label %if.end5
|
|
|
|
if.end5: ; preds = %if.then3, %entry
|
|
ret i32 3
|
|
}
|