[v2] bpf: s390: add JIT support for multi-function programs

This adds support for bpf-to-bpf function calls in the s390 JIT
compiler. The JIT compiler converts the bpf call instructions to
native branch instructions. After a round of the usual passes, the
start addresses of the JITed images for the callee functions are
known. Finally, to fixup the branch target addresses, we need to
perform an extra pass.

Because of the address range in which JITed images are allocated on
s390, the offsets of the start addresses of these images from
__bpf_call_base are as large as 64 bits. So, for a function call,
the imm field of the instruction cannot be used to determine the
callee's address. Use bpf_jit_get_func_addr() helper instead.

The patch borrows a lot from:

8c11ea5ce13d bpf, arm64: fix getting subprog addr from aux for calls
e2c95a61656d bpf, ppc64: generalize fetching subprog into bpf_jit_get_func_addr
8484ce8306f9 bpf: powerpc64: add JIT support for multi-function programs

(including the commit message).

test_verifier (5.3-rc6 with CONFIG_BPF_JIT_ALWAYS_ON=y):

without patch:
Summary: 1501 PASSED, 0 SKIPPED, 47 FAILED

with patch:
Summary: 1540 PASSED, 0 SKIPPED, 8 FAILED

Signed-off-by: Yauheni Kaliuta <yauheni.kaliuta@redhat.com>
---

V1->V2:

- mention CONFIG_BPF_JIT_ALWAYS_ON=y in the commit message;
- in case of bpf_jit_get_func_addr() error return -1 to be consistent
  with the bpf_jit_insn() default branch return value.

---
 arch/s390/net/bpf_jit_comp.c | 63 +++++++++++++++++++++++++++++-------
 1 file changed, 52 insertions(+), 11 deletions(-)

> Am 27.08.2019 um 16:53 schrieb Yauheni Kaliuta <yauheni.kaliuta@redhat.com>:
> 
> @@ -1316,7 +1327,9 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
> {
> 	struct bpf_prog *tmp, *orig_fp = fp;
> 	struct bpf_binary_header *header;
> +	struct s390_jit_data *jit_data;
> 	bool tmp_blinded = false;
> +	bool extra_pass = false;
> 	struct bpf_jit jit;
> 	int pass;
> 
> @@ -1335,6 +1348,22 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
> 		fp = tmp;
> 	}
> 
> +	jit_data = fp->aux->jit_data;
> +	if (!jit_data) {
> +		jit_data = kzalloc(sizeof(*jit_data), GFP_KERNEL);
> +		if (!jit_data) {
> +			fp = orig_fp;
> +			goto out;
> +		}
> +		fp->aux->jit_data = jit_data;
> +	}
> +	if (jit_data->ctx.addrs) {
> +		jit = jit_data->ctx;
> +		header = jit_data->header;
> +		extra_pass = true;
> +		goto skip_init_ctx;
> +	}
> +

I've noticed that I'm getting the following warning, presumably because
of the added goto skip_init_ctx:

linux/arch/s390/net/bpf_jit_comp.c: In function 'bpf_int_jit_compile':
linux/arch/s390/net/bpf_jit_comp.c:1406:3: warning: 'pass' may be used uninitialized in this function [-Wmaybe-uninitialized]
   bpf_jit_dump(fp->len, jit.size, pass, jit.prg_buf);

Maybe set the initial value of pass to 1?

> Am 27.08.2019 um 18:39 schrieb Ilya Leoshkevich <iii@linux.ibm.com>:
> 
>> Am 27.08.2019 um 16:53 schrieb Yauheni Kaliuta <yauheni.kaliuta@redhat.com>:
>> 
>> @@ -1316,7 +1327,9 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
>> {
>> 	struct bpf_prog *tmp, *orig_fp = fp;
>> 	struct bpf_binary_header *header;
>> +	struct s390_jit_data *jit_data;
>> 	bool tmp_blinded = false;
>> +	bool extra_pass = false;
>> 	struct bpf_jit jit;
>> 	int pass;
>> 
>> @@ -1335,6 +1348,22 @@ struct bpf_prog *bpf_int_jit_compile(struct bpf_prog *fp)
>> 		fp = tmp;
>> 	}
>> 
>> +	jit_data = fp->aux->jit_data;
>> +	if (!jit_data) {
>> +		jit_data = kzalloc(sizeof(*jit_data), GFP_KERNEL);
>> +		if (!jit_data) {
>> +			fp = orig_fp;
>> +			goto out;
>> +		}
>> +		fp->aux->jit_data = jit_data;
>> +	}
>> +	if (jit_data->ctx.addrs) {
>> +		jit = jit_data->ctx;
>> +		header = jit_data->header;
>> +		extra_pass = true;
>> +		goto skip_init_ctx;
>> +	}
>> +
> 
> I've noticed that I'm getting the following warning, presumably because
> of the added goto skip_init_ctx:
> 
> linux/arch/s390/net/bpf_jit_comp.c: In function 'bpf_int_jit_compile':
> linux/arch/s390/net/bpf_jit_comp.c:1406:3: warning: 'pass' may be used uninitialized in this function [-Wmaybe-uninitialized]
>   bpf_jit_dump(fp->len, jit.size, pass, jit.prg_buf);
> 
> Maybe set the initial value of pass to 1?

... or save pass in s390_jit_data and do the following?

	pass = jit_data->pass + 1;
	goto skip_init_ctx;

I've also noticed that

	$ git show --format=email | scripts/checkpatch.pl --strict

complains - could you please fix these warnings?


I've finally reviewed the actual logic of the patch, and it looks good
to me.  The integration with the common code is somewhat involved, but
it follows the established architecture.