This pull request primarily focuses on modifications to the ubpf_vm package, specifically the translate function in ubpf_jit_x86_64.c and the ubpf_exec function in ubpf_vm.c. The changes are centered around the handling of ALU32 instructions. In both functions, additional code has been added to check if an instruction is an ALU32 instruction and perform certain operations if it is.
Here are the key changes:
vm/ubpf_jit_x86_64.c: In the translate function, a check has been added to identify if an instruction is an ALU32 instruction. If it is, the target register is truncated to 32 bits using the emit_truncate_u32 function.
vm/ubpf_vm.c: Similar to the translate function, the ubpf_exec function now checks if an instruction is an ALU32 instruction. If it is, the destination register (inst.dst) is bitwise ANDed with UINT32_MAX to ensure it only retains the lower 32 bits of the value.
Resolves: #452
This pull request primarily focuses on modifications to the
ubpf_vmpackage, specifically thetranslatefunction inubpf_jit_x86_64.cand theubpf_execfunction inubpf_vm.c. The changes are centered around the handling of ALU32 instructions. In both functions, additional code has been added to check if an instruction is an ALU32 instruction and perform certain operations if it is.Here are the key changes:
vm/ubpf_jit_x86_64.c: In thetranslatefunction, a check has been added to identify if an instruction is an ALU32 instruction. If it is, the target register is truncated to 32 bits using theemit_truncate_u32function.vm/ubpf_vm.c: Similar to thetranslatefunction, theubpf_execfunction now checks if an instruction is an ALU32 instruction. If it is, the destination register (inst.dst) is bitwise ANDed withUINT32_MAXto ensure it only retains the lower 32 bits of the value.