Message ID | 20230913153413.1446068-5-pulehui@huaweicloud.com (mailing list archive) |
---|---|
State | Superseded |
Delegated to: | BPF |
Headers | show |
Series | Zbb support and code simplification for RV64 JIT | expand |
On Wed, Sep 13, 2023 at 11:34:11PM +0800, Pu Lehui wrote: > From: Pu Lehui <pulehui@huawei.com> > > Add necessary Zbb instructions introduced by [0] to reduce code size and > improve performance of RV64 JIT. At the same time, a helper is added to > check whether the CPU supports Zbb instructions. > > [0] https://github.com/riscv/riscv-bitmanip/releases/download/1.0.0/bitmanip-1.0.0-38-g865e7a7.pdf > > Signed-off-by: Pu Lehui <pulehui@huawei.com> > --- > arch/riscv/net/bpf_jit.h | 26 ++++++++++++++++++++++++++ > 1 file changed, 26 insertions(+) > > diff --git a/arch/riscv/net/bpf_jit.h b/arch/riscv/net/bpf_jit.h > index 8e0ef4d08..7ee59d1f6 100644 > --- a/arch/riscv/net/bpf_jit.h > +++ b/arch/riscv/net/bpf_jit.h > @@ -18,6 +18,11 @@ static inline bool rvc_enabled(void) > return IS_ENABLED(CONFIG_RISCV_ISA_C); > } > > +static inline bool rvzbb_enabled(void) > +{ > + return IS_ENABLED(CONFIG_RISCV_ISA_ZBB); > +} I dunno much about bpf, so passing question that may be a bit obvious: Is this meant to be a test as to whether the kernel binary is built with support for the extension, or whether the underlying platform is capable of executing zbb instructions. Sorry if that would be obvious to a bpf aficionado, context I have here is the later user and the above rvc_enabled() test, which functions differently to Zbb and so doesn't really help me. Thanks, Conor. > + > enum { > RV_REG_ZERO = 0, /* The constant value 0 */ > RV_REG_RA = 1, /* Return address */ > @@ -727,6 +732,27 @@ static inline u16 rvc_swsp(u32 imm8, u8 rs2) > return rv_css_insn(0x6, imm, rs2, 0x2); > } > > +/* RVZBB instrutions. */ > +static inline u32 rvzbb_sextb(u8 rd, u8 rs1) > +{ > + return rv_i_insn(0x604, rs1, 1, rd, 0x13); > +} > + > +static inline u32 rvzbb_sexth(u8 rd, u8 rs1) > +{ > + return rv_i_insn(0x605, rs1, 1, rd, 0x13); > +} > + > +static inline u32 rvzbb_zexth(u8 rd, u8 rs) > +{ > + return rv_i_insn(0x80, rs, 4, rd, __riscv_xlen == 64 ? 0x3b : 0x33); > +} > + > +static inline u32 rvzbb_rev8(u8 rd, u8 rs) > +{ > + return rv_i_insn(__riscv_xlen == 64 ? 0x6b8 : 0x698, rs, 5, rd, 0x13); > +} > + > /* > * RV64-only instructions. > * > -- > 2.25.1 > > > _______________________________________________ > linux-riscv mailing list > linux-riscv@lists.infradead.org > http://lists.infradead.org/mailman/listinfo/linux-riscv
On Wed, Sep 13, 2023 at 05:23:48PM +0100, Conor Dooley wrote: > On Wed, Sep 13, 2023 at 11:34:11PM +0800, Pu Lehui wrote: > > From: Pu Lehui <pulehui@huawei.com> > > > > Add necessary Zbb instructions introduced by [0] to reduce code size and > > improve performance of RV64 JIT. At the same time, a helper is added to > > check whether the CPU supports Zbb instructions. > > > > [0] https://github.com/riscv/riscv-bitmanip/releases/download/1.0.0/bitmanip-1.0.0-38-g865e7a7.pdf > > > > Signed-off-by: Pu Lehui <pulehui@huawei.com> > > --- > > arch/riscv/net/bpf_jit.h | 26 ++++++++++++++++++++++++++ > > 1 file changed, 26 insertions(+) > > > > diff --git a/arch/riscv/net/bpf_jit.h b/arch/riscv/net/bpf_jit.h > > index 8e0ef4d08..7ee59d1f6 100644 > > --- a/arch/riscv/net/bpf_jit.h > > +++ b/arch/riscv/net/bpf_jit.h > > @@ -18,6 +18,11 @@ static inline bool rvc_enabled(void) > > return IS_ENABLED(CONFIG_RISCV_ISA_C); > > } > > > > +static inline bool rvzbb_enabled(void) > > +{ > > + return IS_ENABLED(CONFIG_RISCV_ISA_ZBB); > > +} > > I dunno much about bpf, so passing question that may be a bit obvious: > Is this meant to be a test as to whether the kernel binary is built with > support for the extension, or whether the underlying platform is capable > of executing zbb instructions. > > Sorry if that would be obvious to a bpf aficionado, context I have here > is the later user and the above rvc_enabled() test, which functions > differently to Zbb and so doesn't really help me. FTR, I got an off-list reply about this & it is meant to be a check as to whether the underlying platform supports the extension. The current test here is insufficient for that. Thanks, Conor.
On 2023/9/14 21:02, Conor Dooley wrote: > On Wed, Sep 13, 2023 at 05:23:48PM +0100, Conor Dooley wrote: >> On Wed, Sep 13, 2023 at 11:34:11PM +0800, Pu Lehui wrote: >>> From: Pu Lehui <pulehui@huawei.com> >>> >>> Add necessary Zbb instructions introduced by [0] to reduce code size and >>> improve performance of RV64 JIT. At the same time, a helper is added to >>> check whether the CPU supports Zbb instructions. >>> >>> [0] https://github.com/riscv/riscv-bitmanip/releases/download/1.0.0/bitmanip-1.0.0-38-g865e7a7.pdf >>> >>> Signed-off-by: Pu Lehui <pulehui@huawei.com> >>> --- >>> arch/riscv/net/bpf_jit.h | 26 ++++++++++++++++++++++++++ >>> 1 file changed, 26 insertions(+) >>> >>> diff --git a/arch/riscv/net/bpf_jit.h b/arch/riscv/net/bpf_jit.h >>> index 8e0ef4d08..7ee59d1f6 100644 >>> --- a/arch/riscv/net/bpf_jit.h >>> +++ b/arch/riscv/net/bpf_jit.h >>> @@ -18,6 +18,11 @@ static inline bool rvc_enabled(void) >>> return IS_ENABLED(CONFIG_RISCV_ISA_C); >>> } >>> >>> +static inline bool rvzbb_enabled(void) >>> +{ >>> + return IS_ENABLED(CONFIG_RISCV_ISA_ZBB); >>> +} >> >> I dunno much about bpf, so passing question that may be a bit obvious: >> Is this meant to be a test as to whether the kernel binary is built with >> support for the extension, or whether the underlying platform is capable >> of executing zbb instructions. >> >> Sorry if that would be obvious to a bpf aficionado, context I have here >> is the later user and the above rvc_enabled() test, which functions >> differently to Zbb and so doesn't really help me. > > FTR, I got an off-list reply about this & it is meant to be a check as > to whether the underlying platform supports the extension. The current > test here is insufficient for that. > Thanks Conor for explain me lot about the difference between Compressed instructions and Zbb instructions. As the compressed instructions are a build-time option, while the Zbb is runtime detected. We need to add additional runtime detection as show bellow: riscv_has_extension_likely(RISCV_ISA_EXT_ZBB) will patch this suggestion to the next version. Thanks, Lehui. > Thanks, > Conor.
diff --git a/arch/riscv/net/bpf_jit.h b/arch/riscv/net/bpf_jit.h index 8e0ef4d08..7ee59d1f6 100644 --- a/arch/riscv/net/bpf_jit.h +++ b/arch/riscv/net/bpf_jit.h @@ -18,6 +18,11 @@ static inline bool rvc_enabled(void) return IS_ENABLED(CONFIG_RISCV_ISA_C); } +static inline bool rvzbb_enabled(void) +{ + return IS_ENABLED(CONFIG_RISCV_ISA_ZBB); +} + enum { RV_REG_ZERO = 0, /* The constant value 0 */ RV_REG_RA = 1, /* Return address */ @@ -727,6 +732,27 @@ static inline u16 rvc_swsp(u32 imm8, u8 rs2) return rv_css_insn(0x6, imm, rs2, 0x2); } +/* RVZBB instrutions. */ +static inline u32 rvzbb_sextb(u8 rd, u8 rs1) +{ + return rv_i_insn(0x604, rs1, 1, rd, 0x13); +} + +static inline u32 rvzbb_sexth(u8 rd, u8 rs1) +{ + return rv_i_insn(0x605, rs1, 1, rd, 0x13); +} + +static inline u32 rvzbb_zexth(u8 rd, u8 rs) +{ + return rv_i_insn(0x80, rs, 4, rd, __riscv_xlen == 64 ? 0x3b : 0x33); +} + +static inline u32 rvzbb_rev8(u8 rd, u8 rs) +{ + return rv_i_insn(__riscv_xlen == 64 ? 0x6b8 : 0x698, rs, 5, rd, 0x13); +} + /* * RV64-only instructions. *