[PATCH v2 0/8] Implement EBPF on powerpc32

From: Christophe Leroy <christophe.leroy@csgroup.eu>
To: Benjamin Herrenschmidt <benh@kernel.crashing.org>,
	Paul Mackerras <paulus@samba.org>,
	Michael Ellerman <mpe@ellerman.id.au>,
	ast@kernel.org, daniel@iogearbox.net, andrii@kernel.org,
	kafai@fb.com, songliubraving@fb.com, yhs@fb.com,
	john.fastabend@gmail.com, kpsingh@chromium.org,
	naveen.n.rao@linux.ibm.com, sandipan@linux.ibm.com
Cc: linux-kernel@vger.kernel.org, linuxppc-dev@lists.ozlabs.org,
	netdev@vger.kernel.org, bpf@vger.kernel.org
Subject: [PATCH v2 0/8] Implement EBPF on powerpc32
Date: Mon, 22 Mar 2021 16:37:45 +0000 (UTC)	[thread overview]
Message-ID: <cover.1616430991.git.christophe.leroy@csgroup.eu> (raw)

This series implements extended BPF on powerpc32. For the implementation
details, see the patch before the last.

The following operations are not implemented:

		case BPF_ALU64 | BPF_DIV | BPF_X: /* dst /= src */
		case BPF_ALU64 | BPF_MOD | BPF_X: /* dst %= src */
		case BPF_STX | BPF_XADD | BPF_DW: /* *(u64 *)(dst + off) += src */

The following operations are only implemented for power of two constants:

		case BPF_ALU64 | BPF_MOD | BPF_K: /* dst %= imm */
		case BPF_ALU64 | BPF_DIV | BPF_K: /* dst /= imm */

Below are the results on a powerpc 885:
- with the patch, with and without bpf_jit_enable
- without the patch, with bpf_jit_enable (ie with CBPF)

With the patch, with bpf_jit_enable = 1 :

[   60.826529] test_bpf: Summary: 378 PASSED, 0 FAILED, [354/366 JIT'ed]
[   60.832505] test_bpf: test_skb_segment: Summary: 2 PASSED, 0 FAILED

With the patch, with bpf_jit_enable = 0 :

[   75.186337] test_bpf: Summary: 378 PASSED, 0 FAILED, [0/366 JIT'ed]
[   75.192325] test_bpf: test_skb_segment: Summary: 2 PASSED, 0 FAILED

Without the patch, with bpf_jit_enable = 1 :

[  186.112429] test_bpf: Summary: 371 PASSED, 7 FAILED, [119/366 JIT'ed]

Couldn't run test_progs because it doesn't build (clang 11 crashes during the build).

Changes in v2:
- Simplify 16 bits swap
- Rework tailcall, use stack for tailcall counter
- Fix handling of BPF_REG_FP:
  - must be handler like any other register allthough only the lower 32 bits part is used as a pointer.
  - r18 was TMP_REG, r17/r18 become de BPF_REG_FP
  - r31 was BPF_REG_FP, it is now TMP_REG
- removed bpf_jit32.h
- Reorder register allocation dynamically to use the volatile registers as much as possible when not doing function calls (last patch - new)

Christophe Leroy (8):
  powerpc/bpf: Remove classical BPF support for PPC32
  powerpc/bpf: Change register numbering for bpf_set/is_seen_register()
  powerpc/bpf: Move common helpers into bpf_jit.h
  powerpc/bpf: Move common functions into bpf_jit_comp.c
  powerpc/bpf: Change values of SEEN_ flags
  powerpc/asm: Add some opcodes in asm/ppc-opcode.h for PPC32 eBPF
  powerpc/bpf: Implement extended BPF on PPC32
  powerpc/bpf: Reallocate BPF registers to volatile registers when
    possible on PPC32

 Documentation/admin-guide/sysctl/net.rst |    2 +-
 arch/powerpc/Kconfig                     |    3 +-
 arch/powerpc/include/asm/ppc-opcode.h    |   12 +
 arch/powerpc/net/Makefile                |    6 +-
 arch/powerpc/net/bpf_jit.h               |   61 ++
 arch/powerpc/net/bpf_jit32.h             |  139 ---
 arch/powerpc/net/bpf_jit64.h             |   21 +-
 arch/powerpc/net/bpf_jit_asm.S           |  226 -----
 arch/powerpc/net/bpf_jit_comp.c          |  782 ++++-----------
 arch/powerpc/net/bpf_jit_comp32.c        | 1095 ++++++++++++++++++++++
 arch/powerpc/net/bpf_jit_comp64.c        |  295 +-----
 11 files changed, 1372 insertions(+), 1270 deletions(-)
 delete mode 100644 arch/powerpc/net/bpf_jit32.h
 delete mode 100644 arch/powerpc/net/bpf_jit_asm.S
 create mode 100644 arch/powerpc/net/bpf_jit_comp32.c

-- 
2.25.0