public inbox for bpf@vger.kernel.org
 help / color / mirror / Atom feed
From: Ilya Leoshkevich <iii@linux.ibm.com>
To: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Cc: Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Andrii Nakryiko <andrii@kernel.org>,
	bpf@vger.kernel.org, Heiko Carstens <hca@linux.ibm.com>,
	Vasily Gorbik <gor@linux.ibm.com>
Subject: Re: [PATCH bpf-next 24/24] s390/bpf: Implement bpf_jit_supports_kfunc_call()
Date: Fri, 27 Jan 2023 12:36:32 +0100	[thread overview]
Message-ID: <2dd35469c9df5d6ab81d798467e13eab82b1d254.camel@linux.ibm.com> (raw)
In-Reply-To: <20230126012812.vqg3oktknpnvvssf@macbook-pro-6.dhcp.thefacebook.com>

On Wed, 2023-01-25 at 17:28 -0800, Alexei Starovoitov wrote:
> On Wed, Jan 25, 2023 at 10:38:17PM +0100, Ilya Leoshkevich wrote:
> > +
> > +               /* Sign-extend the kfunc arguments. */
> > +               if (insn->src_reg == BPF_PSEUDO_KFUNC_CALL) {
> > +                       m = bpf_jit_find_kfunc_model(fp, insn);
> > +                       if (!m)
> > +                               return -1;
> > +
> > +                       for (j = 0; j < m->nr_args; j++) {
> > +                               if (sign_extend(jit, BPF_REG_1 + j,
> > +                                               m->arg_size[j],
> > +                                               m->arg_flags[j]))
> > +                                       return -1;
> > +                       }
> > +               }
> 
> Is this because s390 doesn't have subregisters?
> Could you give an example where it's necessary?
> I'm guessing a bpf prog compiled with alu32 and operates on signed
> int
> that is passed into a kfunc that expects 'int' in 64-bit reg?

Precisely. The test added in 13/24 fails without this:

verify_success:PASS:skel 0 nsec                                       
verify_success:PASS:bpf_object__find_program_by_name 0 nsec           
verify_success:PASS:kfunc_call_test4 0 nsec                           
verify_success:FAIL:retval unexpected retval: actual 4294966065 !=
expected -1234                                                        
#94/10   kfunc_call/kfunc_call_test4:FAIL                    

Looking at the assembly:

; long noinline bpf_kfunc_call_test4(signed char a, short b, int c,
long d)
0000000000936a78 <bpf_kfunc_call_test4>:
  936a78:       c0 04 00 00 00 00       jgnop   936a78
<bpf_kfunc_call_test4>
; 	return (long)a + (long)b + (long)c + d;
  936a7e:       b9 08 00 45             agr     %r4,%r5
  936a82:       b9 08 00 43             agr     %r4,%r3
  936a86:       b9 08 00 24             agr     %r2,%r4
  936a8a:       c0 f4 00 1e 3b 27       jg      cfe0d8
<__s390_indirect_jump_r14>

As per the s390x ABI, bpf_kfunc_call_test4() has the right to assume
that a, b and c are sign-extended by the caller, which results in using
64-bit additions (agr) without any additional conversions.

On the JITed code side (without this hunk) we have:

; tmp = bpf_kfunc_call_test4(-3, -30, -200, -1000);
;        5:       b4 10 00 00 ff ff ff fd w1 = -3
   0x3ff7fdcdad4:       llilf   %r2,0xfffffffd
;        6:       b4 20 00 00 ff ff ff e2 w2 = -30
   0x3ff7fdcdada:       llilf   %r3,0xffffffe2
;        7:       b4 30 00 00 ff ff ff 38 w3 = -200
   0x3ff7fdcdae0:       llilf   %r4,0xffffff38
;       8:       b7 40 00 00 ff ff fc 18 r4 = -1000
   0x3ff7fdcdae6:       lgfi    %r5,-1000
   0x3ff7fdcdaec:       mvc     64(4,%r15),160(%r15)
   0x3ff7fdcdaf2:       lgrl    %r1,bpf_kfunc_call_test4@GOT
   0x3ff7fdcdaf8:       brasl   %r14,__s390_indirect_jump_r1

This first 3 llilfs are 32-bit loads, that need to be sign-extended
to 64 bits.

> > +bool bpf_jit_supports_kfunc_call(void)
> > +{
> > +       return true;
> > +}
> 
> Timely :) Thanks for working it.


  reply	other threads:[~2023-01-27 11:38 UTC|newest]

Thread overview: 46+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-25 21:37 [PATCH bpf-next 00/24] Support bpf trampoline for s390x Ilya Leoshkevich
2023-01-25 21:37 ` [PATCH bpf-next 01/24] selftests/bpf: Fix liburandom_read.so linker error Ilya Leoshkevich
2023-01-26  1:07   ` Andrii Nakryiko
2023-01-26 13:30     ` Ilya Leoshkevich
2023-01-25 21:37 ` [PATCH bpf-next 02/24] selftests/bpf: Fix symlink creation error Ilya Leoshkevich
2023-01-25 21:37 ` [PATCH bpf-next 03/24] selftests/bpf: Fix fexit_stress on s390x Ilya Leoshkevich
2023-01-25 21:37 ` [PATCH bpf-next 04/24] selftests/bpf: Fix trampoline_count " Ilya Leoshkevich
2023-01-25 21:37 ` [PATCH bpf-next 05/24] selftests/bpf: Fix kfree_skb " Ilya Leoshkevich
2023-01-25 21:37 ` [PATCH bpf-next 06/24] selftests/bpf: Set errno when urand_spawn() fails Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 07/24] selftests/bpf: Fix decap_sanity_ns cleanup Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 08/24] selftests/bpf: Fix verify_pkcs7_sig on s390x Ilya Leoshkevich
2023-01-26  1:06   ` Andrii Nakryiko
2023-01-27 12:36     ` Ilya Leoshkevich
2023-01-27 17:26       ` Andrii Nakryiko
2023-01-25 21:38 ` [PATCH bpf-next 09/24] selftests/bpf: Fix xdp_do_redirect " Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 10/24] selftests/bpf: Fix cgrp_local_storage " Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 11/24] selftests/bpf: Check stack_mprotect() return value Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 12/24] selftests/bpf: Increase SIZEOF_BPF_LOCAL_STORAGE_ELEM on s390x Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 13/24] selftests/bpf: Add a sign-extension test for kfuncs Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 14/24] selftests/bpf: Fix test_lsm on s390x Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 15/24] selftests/bpf: Fix test_xdp_adjust_tail_grow2 " Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 16/24] selftests/bpf: Fix vmlinux test " Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 17/24] libbpf: Read usdt arg spec with bpf_probe_read_kernel() Ilya Leoshkevich
2023-01-26  0:26   ` Andrii Nakryiko
2023-01-26 11:41     ` Ilya Leoshkevich
2023-01-26 19:03       ` Andrii Nakryiko
2023-01-27 11:01         ` Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 18/24] s390/bpf: Fix a typo in a comment Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 19/24] s390/bpf: Add expoline to tail calls Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 20/24] s390/bpf: Implement bpf_arch_text_poke() Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 21/24] bpf: btf: Add BTF_FMODEL_SIGNED_ARG flag Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 22/24] s390/bpf: Implement arch_prepare_bpf_trampoline() Ilya Leoshkevich
2023-01-26  1:15   ` Andrii Nakryiko
2023-01-26 14:30     ` Ilya Leoshkevich
2023-01-26 19:06       ` Andrii Nakryiko
2023-01-27 11:15         ` Ilya Leoshkevich
2023-01-27 17:30           ` Andrii Nakryiko
2023-01-25 21:38 ` [PATCH bpf-next 23/24] s390/bpf: Implement bpf_jit_supports_subprog_tailcalls() Ilya Leoshkevich
2023-01-25 21:38 ` [PATCH bpf-next 24/24] s390/bpf: Implement bpf_jit_supports_kfunc_call() Ilya Leoshkevich
2023-01-26  1:28   ` Alexei Starovoitov
2023-01-27 11:36     ` Ilya Leoshkevich [this message]
2023-01-27 16:04       ` Alexei Starovoitov
2023-01-26  0:45 ` [PATCH bpf-next 00/24] Support bpf trampoline for s390x Andrii Nakryiko
2023-01-27 16:51   ` Ilya Leoshkevich
2023-01-27 17:24     ` Andrii Nakryiko
2023-01-27 22:50       ` Ilya Leoshkevich

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=2dd35469c9df5d6ab81d798467e13eab82b1d254.camel@linux.ibm.com \
    --to=iii@linux.ibm.com \
    --cc=alexei.starovoitov@gmail.com \
    --cc=andrii@kernel.org \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=gor@linux.ibm.com \
    --cc=hca@linux.ibm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox