* [PATCH 0/2] riscv: improve unaligned memory accesses @ 2021-09-13 12:19 Chen Huang 2021-09-13 12:19 ` [PATCH 1/2] riscv: support HAVE_EFFICIENT_UNALIGNED_ACCESS Chen Huang 2021-09-13 12:19 ` [PATCH 2/2] riscv: Support DCACHE_WORD_ACCESS Chen Huang 0 siblings, 2 replies; 4+ messages in thread From: Chen Huang @ 2021-09-13 12:19 UTC (permalink / raw) To: Paul Walmsley, Palmer Dabbelt, Albert Ou Cc: Chen Huang, Kefeng Wang, linux-riscv, linux-kernel The RISCV ISA can support unaligned memory accesses, so the patchset selects HAVE_EFFICIENT_UNALIGNED_ACCESS and supports DCACHE_WORD_ACCESS to improve the efficiency of unaligned memory accesses. Chen Huang (2): riscv: Kconfig: select HAVE_EFFICIENT_UNALIGNED_ACCESS riscv: Support DCACHE_WORD_ACCESS arch/riscv/Kconfig | 2 ++ arch/riscv/include/asm/word-at-a-time.h | 36 +++++++++++++++++++++++++ 2 files changed, 38 insertions(+) -- 2.18.0.huawei.25 ^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH 1/2] riscv: support HAVE_EFFICIENT_UNALIGNED_ACCESS 2021-09-13 12:19 [PATCH 0/2] riscv: improve unaligned memory accesses Chen Huang @ 2021-09-13 12:19 ` Chen Huang 2021-09-14 17:08 ` Darius Rad 2021-09-13 12:19 ` [PATCH 2/2] riscv: Support DCACHE_WORD_ACCESS Chen Huang 1 sibling, 1 reply; 4+ messages in thread From: Chen Huang @ 2021-09-13 12:19 UTC (permalink / raw) To: Paul Walmsley, Palmer Dabbelt, Albert Ou Cc: Chen Huang, Kefeng Wang, linux-riscv, linux-kernel The RISCV ISA can perform efficient unaligned memory accesses in hardware. This patch selects HAVE_EFFICIENT_UNALIGNED_ACCESS for that. Signed-off-by: Chen Huang <chenhuang5@huawei.com> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> --- arch/riscv/Kconfig | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig index aac669a6c3d8..6e70bf50b02a 100644 --- a/arch/riscv/Kconfig +++ b/arch/riscv/Kconfig @@ -81,6 +81,7 @@ config RISCV select HAVE_DEBUG_KMEMLEAK select HAVE_DMA_CONTIGUOUS if MMU select HAVE_EBPF_JIT if MMU + select HAVE_EFFICIENT_UNALIGNED_ACCESS select HAVE_FUNCTION_ERROR_INJECTION select HAVE_FUTEX_CMPXCHG if FUTEX select HAVE_GCC_PLUGINS -- 2.18.0.huawei.25 ^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH 1/2] riscv: support HAVE_EFFICIENT_UNALIGNED_ACCESS 2021-09-13 12:19 ` [PATCH 1/2] riscv: support HAVE_EFFICIENT_UNALIGNED_ACCESS Chen Huang @ 2021-09-14 17:08 ` Darius Rad 0 siblings, 0 replies; 4+ messages in thread From: Darius Rad @ 2021-09-14 17:08 UTC (permalink / raw) To: Chen Huang, Paul Walmsley, Palmer Dabbelt, Albert Ou Cc: Kefeng Wang, linux-riscv, linux-kernel On 9/13/21 8:19 AM, Chen Huang wrote: > The RISCV ISA can perform efficient unaligned memory accesses > in hardware. This patch selects HAVE_EFFICIENT_UNALIGNED_ACCESS > for that. > Not all implementations do, so it seems like this is not appropriate. > Signed-off-by: Chen Huang <chenhuang5@huawei.com> > Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> > --- > arch/riscv/Kconfig | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig > index aac669a6c3d8..6e70bf50b02a 100644 > --- a/arch/riscv/Kconfig > +++ b/arch/riscv/Kconfig > @@ -81,6 +81,7 @@ config RISCV > select HAVE_DEBUG_KMEMLEAK > select HAVE_DMA_CONTIGUOUS if MMU > select HAVE_EBPF_JIT if MMU > + select HAVE_EFFICIENT_UNALIGNED_ACCESS > select HAVE_FUNCTION_ERROR_INJECTION > select HAVE_FUTEX_CMPXCHG if FUTEX > select HAVE_GCC_PLUGINS > ^ permalink raw reply [flat|nested] 4+ messages in thread
* [PATCH 2/2] riscv: Support DCACHE_WORD_ACCESS 2021-09-13 12:19 [PATCH 0/2] riscv: improve unaligned memory accesses Chen Huang 2021-09-13 12:19 ` [PATCH 1/2] riscv: support HAVE_EFFICIENT_UNALIGNED_ACCESS Chen Huang @ 2021-09-13 12:19 ` Chen Huang 1 sibling, 0 replies; 4+ messages in thread From: Chen Huang @ 2021-09-13 12:19 UTC (permalink / raw) To: Paul Walmsley, Palmer Dabbelt, Albert Ou Cc: Chen Huang, Kefeng Wang, linux-riscv, linux-kernel This patch selects DCACHE_WORD_ACCESS on riscv and implements support for load_unaligned_zeropad. DCACHE_WORD_ACCESS uses the word-at-a-time API for optimised string comparisons in the vfs layer. Signed-off-by: Chen Huang <chenhuang5@huawei.com> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> --- arch/riscv/Kconfig | 1 + arch/riscv/include/asm/word-at-a-time.h | 36 +++++++++++++++++++++++++ 2 files changed, 37 insertions(+) diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig index 6e70bf50b02a..f6f0da0f436b 100644 --- a/arch/riscv/Kconfig +++ b/arch/riscv/Kconfig @@ -44,6 +44,7 @@ config RISCV select CLONE_BACKWARDS select CLINT_TIMER if !MMU select COMMON_CLK + select DCACHE_WORD_ACCESS select EDAC_SUPPORT select GENERIC_ARCH_TOPOLOGY if SMP select GENERIC_ATOMIC64 if !64BIT diff --git a/arch/riscv/include/asm/word-at-a-time.h b/arch/riscv/include/asm/word-at-a-time.h index 7c086ac6ecd4..0b77ce654f56 100644 --- a/arch/riscv/include/asm/word-at-a-time.h +++ b/arch/riscv/include/asm/word-at-a-time.h @@ -11,6 +11,8 @@ #include <linux/kernel.h> +#include <asm/asm.h> + struct word_at_a_time { const unsigned long one_bits, high_bits; }; @@ -45,4 +47,38 @@ static inline unsigned long find_zero(unsigned long mask) /* The mask we created is directly usable as a bytemask */ #define zero_bytemask(mask) (mask) +/* + * Load an unaligned word from kernel space. + * + * In the (very unlikely) case of the word being a page-crosser + * and the next page not being mapped, take the exception and + * return zeroes in the non-existing part. + */ +static inline unsigned long load_unaligned_zeropad(const void *addr) +{ + unsigned long ret, tmp; + + /* Load word from unaligned pointer addr */ + asm( + "1: " REG_L " %0, %3\n" + "2:\n" + " .section .fixup,\"ax\"\n" + " .balign 2\n" + "3: andi %1, %2, ~0x7\n" + " " REG_L " %0, (%1)\n" + " andi %1, %2, 0x7\n" + " slli %1, %1, 0x3\n" + " srl %0, %0, %1\n" + " jump 2b, %1\n" + " .previous\n" + " .section __ex_table,\"a\"\n" + " .balign " RISCV_SZPTR "\n" + " " RISCV_PTR " 1b, 3b\n" + " .previous" + : "=&r" (ret), "=&r" (tmp) + : "r" (addr), "m" (*(unsigned long *)addr)); + + return ret; +} + #endif /* _ASM_RISCV_WORD_AT_A_TIME_H */ -- 2.18.0.huawei.25 ^ permalink raw reply related [flat|nested] 4+ messages in thread
end of thread, other threads:[~2021-09-14 17:08 UTC | newest] Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2021-09-13 12:19 [PATCH 0/2] riscv: improve unaligned memory accesses Chen Huang 2021-09-13 12:19 ` [PATCH 1/2] riscv: support HAVE_EFFICIENT_UNALIGNED_ACCESS Chen Huang 2021-09-14 17:08 ` Darius Rad 2021-09-13 12:19 ` [PATCH 2/2] riscv: Support DCACHE_WORD_ACCESS Chen Huang
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).