linux/arch/mips/net/bpf_jit.h

82 lines
2.1 KiB
C
Raw Normal View History

/* SPDX-License-Identifier: GPL-2.0-only */
/*
* Just-In-Time compiler for BPF filters on MIPS
*
* Copyright (c) 2014 Imagination Technologies Ltd.
* Author: Markos Chandras <markos.chandras@imgtec.com>
*/
#ifndef BPF_JIT_MIPS_OP_H
#define BPF_JIT_MIPS_OP_H
/* Registers used by JIT */
#define MIPS_R_ZERO 0
#define MIPS_R_V0 2
#define MIPS_R_A0 4
#define MIPS_R_A1 5
#define MIPS_R_T4 12
#define MIPS_R_T5 13
#define MIPS_R_T6 14
#define MIPS_R_T7 15
#define MIPS_R_S0 16
#define MIPS_R_S1 17
#define MIPS_R_S2 18
#define MIPS_R_S3 19
#define MIPS_R_S4 20
#define MIPS_R_S5 21
#define MIPS_R_S6 22
#define MIPS_R_S7 23
#define MIPS_R_SP 29
#define MIPS_R_RA 31
/* Conditional codes */
#define MIPS_COND_EQ 0x1
#define MIPS_COND_GE (0x1 << 1)
#define MIPS_COND_GT (0x1 << 2)
#define MIPS_COND_NE (0x1 << 3)
#define MIPS_COND_ALL (0x1 << 4)
/* Conditionals on X register or K immediate */
#define MIPS_COND_X (0x1 << 5)
#define MIPS_COND_K (0x1 << 6)
#define r_ret MIPS_R_V0
/*
* Use 2 scratch registers to avoid pipeline interlocks.
* There is no overhead during epilogue and prologue since
* any of the $s0-$s6 registers will only be preserved if
* they are going to actually be used.
*/
MIPS: BPF: Introduce BPF ASM helpers This commit introduces BPF ASM helpers for MIPS and MIPS64 kernels. The purpose of this patch is to twofold: 1) We are now able to handle negative offsets instead of either falling back to the interpreter or to simply not do anything and bail out. 2) Optimize reads from the packet header instead of calling the C helpers Because of this patch, we are now able to get rid of quite a bit of code in the JIT generation process by using MIPS optimized assembly code. The new assembly code makes the test_bpf testsuite happy with all 60 test passing successfully compared to the previous implementation where 2 tests were failing. Doing some basic analysis in the results between the old implementation and the new one we can obtain the following summary running current mainline on an ER8 board (+/- 30us delta is ignored to prevent noise from kernel scheduling or IRQ latencies): Summary: 22 tests are faster, 7 are slower and 47 saw no improvement with the most notable improvement being the tcpdump tests. The 7 tests that seem to be a bit slower is because they all follow the slow path (bpf_internal_load_pointer_neg_helper) which is meant to be slow so that's not a problem. Signed-off-by: Markos Chandras <markos.chandras@imgtec.com> Cc: netdev@vger.kernel.org Cc: "David S. Miller" <davem@davemloft.net> Cc: Alexei Starovoitov <ast@plumgrid.com> Cc: Daniel Borkmann <dborkman@redhat.com> Cc: Hannes Frederic Sowa <hannes@stressinduktion.org> Cc: linux-kernel@vger.kernel.org Cc: linux-mips@linux-mips.org Cc: netdev@vger.kernel.org Patchwork: http://patchwork.linux-mips.org/patch/10530/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2015-06-04 10:56:16 +00:00
#define r_skb_hl MIPS_R_S0 /* skb header length */
#define r_skb_data MIPS_R_S1 /* skb actual data */
#define r_off MIPS_R_S2
#define r_A MIPS_R_S3
#define r_X MIPS_R_S4
#define r_skb MIPS_R_S5
#define r_M MIPS_R_S6
MIPS: BPF: Introduce BPF ASM helpers This commit introduces BPF ASM helpers for MIPS and MIPS64 kernels. The purpose of this patch is to twofold: 1) We are now able to handle negative offsets instead of either falling back to the interpreter or to simply not do anything and bail out. 2) Optimize reads from the packet header instead of calling the C helpers Because of this patch, we are now able to get rid of quite a bit of code in the JIT generation process by using MIPS optimized assembly code. The new assembly code makes the test_bpf testsuite happy with all 60 test passing successfully compared to the previous implementation where 2 tests were failing. Doing some basic analysis in the results between the old implementation and the new one we can obtain the following summary running current mainline on an ER8 board (+/- 30us delta is ignored to prevent noise from kernel scheduling or IRQ latencies): Summary: 22 tests are faster, 7 are slower and 47 saw no improvement with the most notable improvement being the tcpdump tests. The 7 tests that seem to be a bit slower is because they all follow the slow path (bpf_internal_load_pointer_neg_helper) which is meant to be slow so that's not a problem. Signed-off-by: Markos Chandras <markos.chandras@imgtec.com> Cc: netdev@vger.kernel.org Cc: "David S. Miller" <davem@davemloft.net> Cc: Alexei Starovoitov <ast@plumgrid.com> Cc: Daniel Borkmann <dborkman@redhat.com> Cc: Hannes Frederic Sowa <hannes@stressinduktion.org> Cc: linux-kernel@vger.kernel.org Cc: linux-mips@linux-mips.org Cc: netdev@vger.kernel.org Patchwork: http://patchwork.linux-mips.org/patch/10530/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2015-06-04 10:56:16 +00:00
#define r_skb_len MIPS_R_S7
#define r_s0 MIPS_R_T4 /* scratch reg 1 */
#define r_s1 MIPS_R_T5 /* scratch reg 2 */
#define r_tmp_imm MIPS_R_T6 /* No need to preserve this */
#define r_tmp MIPS_R_T7 /* No need to preserve this */
#define r_zero MIPS_R_ZERO
#define r_sp MIPS_R_SP
#define r_ra MIPS_R_RA
MIPS: BPF: Introduce BPF ASM helpers This commit introduces BPF ASM helpers for MIPS and MIPS64 kernels. The purpose of this patch is to twofold: 1) We are now able to handle negative offsets instead of either falling back to the interpreter or to simply not do anything and bail out. 2) Optimize reads from the packet header instead of calling the C helpers Because of this patch, we are now able to get rid of quite a bit of code in the JIT generation process by using MIPS optimized assembly code. The new assembly code makes the test_bpf testsuite happy with all 60 test passing successfully compared to the previous implementation where 2 tests were failing. Doing some basic analysis in the results between the old implementation and the new one we can obtain the following summary running current mainline on an ER8 board (+/- 30us delta is ignored to prevent noise from kernel scheduling or IRQ latencies): Summary: 22 tests are faster, 7 are slower and 47 saw no improvement with the most notable improvement being the tcpdump tests. The 7 tests that seem to be a bit slower is because they all follow the slow path (bpf_internal_load_pointer_neg_helper) which is meant to be slow so that's not a problem. Signed-off-by: Markos Chandras <markos.chandras@imgtec.com> Cc: netdev@vger.kernel.org Cc: "David S. Miller" <davem@davemloft.net> Cc: Alexei Starovoitov <ast@plumgrid.com> Cc: Daniel Borkmann <dborkman@redhat.com> Cc: Hannes Frederic Sowa <hannes@stressinduktion.org> Cc: linux-kernel@vger.kernel.org Cc: linux-mips@linux-mips.org Cc: netdev@vger.kernel.org Patchwork: http://patchwork.linux-mips.org/patch/10530/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
2015-06-04 10:56:16 +00:00
#ifndef __ASSEMBLY__
/* Declare ASM helpers */
#define DECLARE_LOAD_FUNC(func) \
extern u8 func(unsigned long *skb, int offset); \
extern u8 func##_negative(unsigned long *skb, int offset); \
extern u8 func##_positive(unsigned long *skb, int offset)
DECLARE_LOAD_FUNC(sk_load_word);
DECLARE_LOAD_FUNC(sk_load_half);
DECLARE_LOAD_FUNC(sk_load_byte);
#endif
#endif /* BPF_JIT_MIPS_OP_H */