Re: [PATCH bpf-next 8/9] bpf: Introduce bpf register BPF_REG_PARAMS

[Yonghong Song <yonghong.song@linux.dev>]

On 4/21/26 5:42 PM, Alexei Starovoitov wrote:
> On Tue Apr 21, 2026 at 5:09 PM PDT, Yonghong Song wrote:
>>
>> On 4/21/26 3:10 PM, Alexei Starovoitov wrote:
>>> On Tue Apr 21, 2026 at 10:20 AM PDT, Yonghong Song wrote:
>>>>    
>>>>    /* Kernel hidden auxiliary/helper register. */
>>>> -#define BPF_REG_AX		MAX_BPF_REG
>>>> -#define MAX_BPF_EXT_REG		(MAX_BPF_REG + 1)
>>>> +#define BPF_REG_PARAMS		MAX_BPF_REG
>>>> +#define BPF_REG_AX		(MAX_BPF_REG + 1)
>>>> +#define MAX_BPF_EXT_REG		(MAX_BPF_REG + 2)
>>> ...
>>>
>>>> --- a/tools/testing/selftests/bpf/progs/verifier_bpf_fastcall.c
>>>> +++ b/tools/testing/selftests/bpf/progs/verifier_bpf_fastcall.c
>>>> @@ -630,13 +630,13 @@ __xlated("...")
>>>>    __xlated("4: r0 = &(void __percpu *)(r0)")
>>>>    __xlated("...")
>>>>    /* may_goto expansion starts */
>>>> -__xlated("6: r11 = *(u64 *)(r10 -24)")
>>>> -__xlated("7: if r11 == 0x0 goto pc+6")
>>>> -__xlated("8: r11 -= 1")
>>>> -__xlated("9: if r11 != 0x0 goto pc+2")
>>>> -__xlated("10: r11 = -24")
>>>> +__xlated("6: r12 = *(u64 *)(r10 -24)")
>>>> +__xlated("7: if r12 == 0x0 goto pc+6")
>>>> +__xlated("8: r12 -= 1")
>>>> +__xlated("9: if r12 != 0x0 goto pc+2")
>>>> +__xlated("10: r12 = -24")
>>> maybe shift it to r15 right away, so we don't need to touch this code
>>> if/when true r12 is introduced?
>> We can do this. Do you think the following is okay:
>>
>> diff --git a/include/linux/filter.h b/include/linux/filter.h
>> index b77d0b06db6e..fe7b6b943ea4 100644
>> --- a/include/linux/filter.h
>> +++ b/include/linux/filter.h
>> @@ -59,9 +59,9 @@ struct ctl_table_header;
>>    
>>    /* Kernel hidden auxiliary/helper register. */
>>    #define BPF_REG_PARAMS         MAX_BPF_REG
>> -#define BPF_REG_AX             (MAX_BPF_REG + 1)
>> -#define MAX_BPF_EXT_REG                (MAX_BPF_REG + 2)
>> +#define MAX_BPF_EXT_REG                (MAX_BPF_REG + 1)
>>    #define MAX_BPF_JIT_REG                MAX_BPF_EXT_REG
>> +#define BPF_REG_AX             (MAX_BPF_REG + 4)
>>    
>>    /* unused opcode to mark special call to bpf_tail_call() helper */
>>    #define BPF_TAIL_CALL  0xf0
>> diff --git a/kernel/bpf/core.c b/kernel/bpf/core.c
>> index ae10b9ca018d..3ad286ef3085 100644
>> --- a/kernel/bpf/core.c
>> +++ b/kernel/bpf/core.c
>> @@ -1299,8 +1299,8 @@ static int bpf_jit_blind_insn(const struct bpf_insn *from,
>>           u32 imm_rnd = get_random_u32();
>>           s16 off;
>>    
>> -       BUILD_BUG_ON(BPF_REG_PARAMS + 2 != MAX_BPF_JIT_REG);
>> -       BUILD_BUG_ON(BPF_REG_AX + 1 != MAX_BPF_JIT_REG);
>> +       BUILD_BUG_ON(BPF_REG_PARAMS + 1 != MAX_BPF_JIT_REG);
>> +       BUILD_BUG_ON(BPF_REG_AX != MAX_BPF_JIT_REG + 3);
> Ohh. We have:
> static unsigned int PROG_NAME(stack_size)(const void *ctx, const struct bpf_insn *insn) \
> { \
>          u64 stack[stack_size / sizeof(u64)]; \
>          u64 regs[MAX_BPF_EXT_REG] = {}; \
>
>
> Please double check that BPF_REG_AX is not used by the interpreter.
>
> but it's starting to feel that my suggestion was premature.
> Probably better to keep this patchset as-is.

Okay, let me keep the current one.