Re: [PATCH bpf-next v2 4/6] bpf, arm64: Impelment bpf_arch_text_poke() for arm64

On 4/15/2022 10:34 AM, Hou Tao wrote:

Hi,

On 4/15/2022 12:22 AM, Xu Kuohai wrote:

...

Impelment bpf_arch_text_poke() for arm64, so bpf trampoline code can use it to replace nop with jump, or replace jump with nop.

Signed-off-by: Xu Kuohai xukuohai@huawei.com Acked-by: Song Liu songliubraving@fb.com

---

arch/arm64/net/bpf_jit_comp.c | 52 +++++++++++++++++++++++++++++++++++ 1 file changed, 52 insertions(+)

diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c index 8ab4035dea27..1a1c3ea75ee2 100644 --- a/arch/arm64/net/bpf_jit_comp.c +++ b/arch/arm64/net/bpf_jit_comp.c @@ -9,6 +9,7 @@ #include <linux/bitfield.h> #include <linux/bpf.h> +#include <linux/memory.h> #include <linux/filter.h> #include <linux/printk.h> #include <linux/slab.h> @@ -18,6 +19,7 @@ #include <asm/cacheflush.h> #include <asm/debug-monitors.h> #include <asm/insn.h> +#include <asm/patching.h> #include <asm/set_memory.h> #include "bpf_jit.h" @@ -1529,3 +1531,53 @@ void bpf_jit_free_exec(void *addr) { return vfree(addr); }

+static int gen_branch_or_nop(enum aarch64_insn_branch_type type, void *ip,

• 	     void *addr, u32 *insn)
  

+{

• if (!addr)

• *insn = aarch64_insn_gen_nop();
  

• else

• *insn = aarch64_insn_gen_branch_imm((unsigned long)ip,
  

• 				    (unsigned long)addr,
  

• 				    type);
  

• return *insn != AARCH64_BREAK_FAULT ? 0 : -EFAULT;

+}

+int bpf_arch_text_poke(void *ip, enum bpf_text_poke_type poke_type,

•        void *old_addr, void *new_addr)
  

+{

• int ret;
• u32 old_insn;
• u32 new_insn;
• u32 replaced;
• enum aarch64_insn_branch_type branch_type;

In bpf_arch_text_poke() of x86, it disables the poking of kernel module, can you explain why it is OK to do so in arm64 ? Because there is no test cases for fentry on linux kernel module, could you please add some tests for it ?

Oops, I forget to check this condition. It's not safe to patch a ko without ko unloading disabled.

For arm64, the fentry is only patched by ftrace since the nop instruciton to be instrumented is not the first instruction, so bpf_text_poke() fails when comparing the old instruction (pointed to by the "old_addr") with the nop. Since the nop in fentry is reserved by ftrace, I dont think it's reasonable to patch the nop by another interface not provided by ftrace.

Besides, for long jumps outside the range of 128MB, a single branch instruction is not sufficient, perhaps we could use ftrace trampoline or some other method to support long jumps.

...

• if (poke_type == BPF_MOD_CALL)

• branch_type = AARCH64_INSN_BRANCH_LINK;
  

• else

• branch_type = AARCH64_INSN_BRANCH_NOLINK;
  

• if (gen_branch_or_nop(branch_type, ip, old_addr, &old_insn) < 0)

• return -EFAULT;
  

• if (gen_branch_or_nop(branch_type, ip, new_addr, &new_insn) < 0)

• return -EFAULT;
  

• mutex_lock(&text_mutex);
• if (aarch64_insn_read(ip, &replaced)) {

• ret = -EFAULT;
  

• goto out;
  

• }
• if (replaced != old_insn) {

• ret = -EFAULT;
  

• goto out;
  

• }
• ret = aarch64_insn_patch_text_nosync((void *)ip, new_insn);

+out:

• mutex_unlock(&text_mutex);
• return ret;

+}

.