diff options
Diffstat (limited to 'include/linux')
| -rw-r--r-- | include/linux/netdevice.h | 12 | ||||
| -rw-r--r-- | include/linux/reciprocal_div.h | 68 |
2 files changed, 72 insertions, 8 deletions
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index 2daf2fa6554f..4fa7f7a3f8b3 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -823,11 +823,8 @@ enum bpf_netdev_command { */ XDP_SETUP_PROG, XDP_SETUP_PROG_HW, - /* Check if a bpf program is set on the device. The callee should - * set @prog_attached to one of XDP_ATTACHED_* values, note that "true" - * is equivalent to XDP_ATTACHED_DRV. - */ XDP_QUERY_PROG, + XDP_QUERY_PROG_HW, /* BPF program for offload callbacks, invoked at program load time. */ BPF_OFFLOAD_VERIFIER_PREP, BPF_OFFLOAD_TRANSLATE, @@ -851,9 +848,8 @@ struct netdev_bpf { struct bpf_prog *prog; struct netlink_ext_ack *extack; }; - /* XDP_QUERY_PROG */ + /* XDP_QUERY_PROG, XDP_QUERY_PROG_HW */ struct { - u8 prog_attached; u32 prog_id; /* flags with which program was installed */ u32 prog_flags; @@ -3560,8 +3556,8 @@ struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev, typedef int (*bpf_op_t)(struct net_device *dev, struct netdev_bpf *bpf); int dev_change_xdp_fd(struct net_device *dev, struct netlink_ext_ack *extack, int fd, u32 flags); -void __dev_xdp_query(struct net_device *dev, bpf_op_t xdp_op, - struct netdev_bpf *xdp); +u32 __dev_xdp_query(struct net_device *dev, bpf_op_t xdp_op, + enum bpf_netdev_command cmd); int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb); int dev_forward_skb(struct net_device *dev, struct sk_buff *skb); diff --git a/include/linux/reciprocal_div.h b/include/linux/reciprocal_div.h index e031e9f2f9d8..585ce89c0f33 100644 --- a/include/linux/reciprocal_div.h +++ b/include/linux/reciprocal_div.h @@ -25,6 +25,9 @@ struct reciprocal_value { u8 sh1, sh2; }; +/* "reciprocal_value" and "reciprocal_divide" together implement the basic + * version of the algorithm described in Figure 4.1 of the paper. + */ struct reciprocal_value reciprocal_value(u32 d); static inline u32 reciprocal_divide(u32 a, struct reciprocal_value R) @@ -33,4 +36,69 @@ static inline u32 reciprocal_divide(u32 a, struct reciprocal_value R) return (t + ((a - t) >> R.sh1)) >> R.sh2; } +struct reciprocal_value_adv { + u32 m; + u8 sh, exp; + bool is_wide_m; +}; + +/* "reciprocal_value_adv" implements the advanced version of the algorithm + * described in Figure 4.2 of the paper except when "divisor > (1U << 31)" whose + * ceil(log2(d)) result will be 32 which then requires u128 divide on host. The + * exception case could be easily handled before calling "reciprocal_value_adv". + * + * The advanced version requires more complex calculation to get the reciprocal + * multiplier and other control variables, but then could reduce the required + * emulation operations. + * + * It makes no sense to use this advanced version for host divide emulation, + * those extra complexities for calculating multiplier etc could completely + * waive our saving on emulation operations. + * + * However, it makes sense to use it for JIT divide code generation for which + * we are willing to trade performance of JITed code with that of host. As shown + * by the following pseudo code, the required emulation operations could go down + * from 6 (the basic version) to 3 or 4. + * + * To use the result of "reciprocal_value_adv", suppose we want to calculate + * n/d, the pseudo C code will be: + * + * struct reciprocal_value_adv rvalue; + * u8 pre_shift, exp; + * + * // handle exception case. + * if (d >= (1U << 31)) { + * result = n >= d; + * return; + * } + * + * rvalue = reciprocal_value_adv(d, 32) + * exp = rvalue.exp; + * if (rvalue.is_wide_m && !(d & 1)) { + * // floor(log2(d & (2^32 -d))) + * pre_shift = fls(d & -d) - 1; + * rvalue = reciprocal_value_adv(d >> pre_shift, 32 - pre_shift); + * } else { + * pre_shift = 0; + * } + * + * // code generation starts. + * if (imm == 1U << exp) { + * result = n >> exp; + * } else if (rvalue.is_wide_m) { + * // pre_shift must be zero when reached here. + * t = (n * rvalue.m) >> 32; + * result = n - t; + * result >>= 1; + * result += t; + * result >>= rvalue.sh - 1; + * } else { + * if (pre_shift) + * result = n >> pre_shift; + * result = ((u64)result * rvalue.m) >> 32; + * result >>= rvalue.sh; + * } + */ +struct reciprocal_value_adv reciprocal_value_adv(u32 d, u8 prec); + #endif /* _LINUX_RECIPROCAL_DIV_H */ |