From: Min Fanlei minfanlei@wxiat.com
Sunway inclusion category: bugfix bugzilla: https://gitee.com/openeuler/kernel/issues/I56WV8
--------------------------------
Signed-off-by: Min Fanlei minfanlei@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/kernel/dup_print.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/arch/sw_64/kernel/dup_print.c b/arch/sw_64/kernel/dup_print.c index ac0a95d4d30b..fae543625f62 100644 --- a/arch/sw_64/kernel/dup_print.c +++ b/arch/sw_64/kernel/dup_print.c @@ -6,6 +6,7 @@ #include <linux/delay.h> #include <linux/spinlock.h> #include <linux/uaccess.h> +#include <asm/chip3_io.h>
#ifdef CONFIG_SW64_RRK
@@ -20,7 +21,7 @@ unsigned long sw64_printk_offset; * For output the kernel message on the console * with full-system emulator. */ -#define QEMU_PRINTF_BUFF_BASE (0x805000040000ULL | PAGE_OFFSET) +#define QEMU_PRINTF_BUFF_BASE (IO_BASE | MCU_BASE | 0x40000UL | PAGE_OFFSET)
int sw64_printk(const char *fmt, va_list args) {
From: Mao Minkai maominkai@wxiat.com
Sunway inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I56TYY
--------------------------------
Mark pci reserved region and memmap reserved region as MEMBLOCK_NOMAP instead of removing them from memblock, so they can be managed by using memblock api.
Signed-off-by: Mao Minkai maominkai@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/kernel/setup.c | 22 ++++++++-------------- 1 file changed, 8 insertions(+), 14 deletions(-)
diff --git a/arch/sw_64/kernel/setup.c b/arch/sw_64/kernel/setup.c index 2784ea5aa0ae..4f735f44ccc3 100644 --- a/arch/sw_64/kernel/setup.c +++ b/arch/sw_64/kernel/setup.c @@ -383,7 +383,7 @@ void __init process_memmap(void) } else { pr_info("reserved memmap region [mem %#018llx-%#018llx]\n", base, base + size - 1); - ret = memblock_remove(base, size); + ret = memblock_mark_nomap(base, size); if (ret) pr_err("reserve memmap region [mem %#018llx-%#018llx] failed\n", base, base + size - 1); @@ -399,7 +399,7 @@ void __init process_memmap(void) } else { pr_info("pci memmap region [mem %#018llx-%#018llx]\n", base, base + size - 1); - ret = memblock_remove(base, size); + ret = memblock_mark_nomap(base, size); if (ret) pr_err("reserve memmap region [mem %#018llx-%#018llx] failed\n", base, base + size - 1); @@ -496,7 +496,6 @@ insert_ram_resource(u64 start, u64 end, bool reserved)
static int __init request_standard_resources(void) { - int i; struct memblock_region *mblk;
extern char _text[], _etext[]; @@ -504,17 +503,12 @@ static int __init request_standard_resources(void) extern char __bss_start[], __bss_stop[];
for_each_mem_region(mblk) { - insert_ram_resource(mblk->base, mblk->base + mblk->size - 1, 0); - } - - for (i = 0; i < memmap_nr; i++) { - switch (memmap_map[i].type) { - case memmap_crashkernel: - break; - default: - insert_ram_resource(memmap_map[i].addr, - memmap_map[i].addr + memmap_map[i].size - 1, 1); - } + if (!memblock_is_nomap(mblk)) + insert_ram_resource(mblk->base, + mblk->base + mblk->size - 1, 0); + else + insert_ram_resource(mblk->base, + mblk->base + mblk->size - 1, 1); }
code_resource.start = __pa_symbol(_text);
From: Mao Minkai maominkai@wxiat.com
Sunway inclusion category: performance bugzilla: https://gitee.com/openeuler/kernel/issues/I56XPR
--------------------------------
Use jump label instead of normal if-else statement for running modes to improve performance.
Signed-off-by: Mao Minkai maominkai@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/include/asm/hw_init.h | 26 +++++++------------------- arch/sw_64/kernel/setup.c | 19 +++++++++++++++---- 2 files changed, 22 insertions(+), 23 deletions(-)
diff --git a/arch/sw_64/include/asm/hw_init.h b/arch/sw_64/include/asm/hw_init.h index 9a56590ef653..f3d75e0fda7a 100644 --- a/arch/sw_64/include/asm/hw_init.h +++ b/arch/sw_64/include/asm/hw_init.h @@ -2,6 +2,7 @@ #ifndef _ASM_SW64_HW_INIT_H #define _ASM_SW64_HW_INIT_H #include <linux/numa.h> +#include <linux/jump_label.h>
#define MMSIZE __va(0x2040)
@@ -96,26 +97,13 @@ static inline bool icache_is_vivt_no_ictag(void) return (cpu_desc.arch_var == 0x3 && cpu_desc.arch_rev == 0x1); }
-enum RUNMODE { - HOST_MODE = 0, - GUEST_MODE = 1, - EMUL_MODE = 2, -}; +DECLARE_STATIC_KEY_TRUE(run_mode_host_key); +DECLARE_STATIC_KEY_FALSE(run_mode_guest_key); +DECLARE_STATIC_KEY_FALSE(run_mode_emul_key);
-static inline bool is_in_host(void) -{ - return !cpu_desc.run_mode; -} - -static inline bool is_in_guest(void) -{ - return cpu_desc.run_mode == GUEST_MODE; -} - -static inline bool is_guest_or_emul(void) -{ - return !!cpu_desc.run_mode; -} +#define is_in_host() static_branch_likely(&run_mode_host_key) +#define is_in_guest() static_branch_unlikely(&run_mode_guest_key) +#define is_guest_or_emul() !static_branch_likely(&run_mode_host_key)
#define CPU_SW3231 0x31 #define CPU_SW831 0x32 diff --git a/arch/sw_64/kernel/setup.c b/arch/sw_64/kernel/setup.c index 4f735f44ccc3..27f59fe0d1b1 100644 --- a/arch/sw_64/kernel/setup.c +++ b/arch/sw_64/kernel/setup.c @@ -133,6 +133,9 @@ static struct resource bss_resource = { struct cpuinfo_sw64 cpu_data[NR_CPUS]; EXPORT_SYMBOL(cpu_data);
+DEFINE_STATIC_KEY_TRUE(run_mode_host_key); +DEFINE_STATIC_KEY_FALSE(run_mode_guest_key); +DEFINE_STATIC_KEY_FALSE(run_mode_emul_key); struct cpu_desc_t cpu_desc; struct socket_desc_t socket_desc[MAX_NUMSOCKETS]; int memmap_nr; @@ -639,10 +642,18 @@ static void __init setup_cpu_info(void) cpu_desc.arch_rev = CPUID_ARCH_REV(val); cpu_desc.pa_bits = CPUID_PA_BITS(val); cpu_desc.va_bits = CPUID_VA_BITS(val); - cpu_desc.run_mode = HOST_MODE;
- if (*(unsigned long *)MMSIZE) - cpu_desc.run_mode = GUEST_MODE; + if (*(unsigned long *)MMSIZE) { + pr_info("run mode: guest\n"); + static_branch_disable(&run_mode_host_key); + static_branch_enable(&run_mode_guest_key); + static_branch_disable(&run_mode_emul_key); + } else { + pr_info("run mode: host\n"); + static_branch_enable(&run_mode_host_key); + static_branch_disable(&run_mode_guest_key); + static_branch_disable(&run_mode_emul_key); + }
for (i = 0; i < VENDOR_ID_MAX; i++) { val = cpuid(GET_VENDOR_ID, i); @@ -747,6 +758,7 @@ void __init sw64_kvm_reserve(void) void __init setup_arch(char **cmdline_p) { + jump_label_init(); setup_cpu_info(); sw64_chip->fixup(); sw64_chip_init->fixup(); @@ -754,7 +766,6 @@ setup_arch(char **cmdline_p) show_socket_mem_layout(); sw64_chip_init->early_init.setup_core_start(&core_start);
- jump_label_init(); setup_sched_clock(); #ifdef CONFIG_GENERIC_SCHED_CLOCK sw64_sched_clock_init();
From: Cui Mingrui cuimingrui@wxiat.com
Sunway inclusion category: performance bugzilla: https://gitee.com/openeuler/kernel/issues/I56W9F
--------------------------------
Function csum_partial_copy_from_user() copy data from userspace and return checksum result. It used to read 8 bytes data from userspace every time, calculate checksum and store it to the destination, which slowed down the copying process. This patch moves the copying process to the beginning. This patch also rewrites do_csum and move some common codes to header file to avoid duplication.
Signed-off-by: Cui Mingrui cuimingrui@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/include/asm/checksum.h | 58 +++++- arch/sw_64/lib/checksum.c | 136 +++++-------- arch/sw_64/lib/csum_partial_copy.c | 310 +++++------------------------ 3 files changed, 150 insertions(+), 354 deletions(-)
diff --git a/arch/sw_64/include/asm/checksum.h b/arch/sw_64/include/asm/checksum.h index 0bb933350dc6..284c1678f51e 100644 --- a/arch/sw_64/include/asm/checksum.h +++ b/arch/sw_64/include/asm/checksum.h @@ -4,9 +4,33 @@
#include <linux/in6.h>
+#define extll(x, y, z) \ + ({__asm__ __volatile__("extll %1, %2, %0" : "=r" (z) \ + : "r" (x), "r" (y)); }) + +#define exthl(x, y, z) \ + ({__asm__ __volatile__("exthl %1, %2, %0" : "=r" (z) \ + : "r" (x), "r" (y)); }) + +#define maskll(x, y, z) \ + ({__asm__ __volatile__("maskll %1, %2, %0" : "=r" (z) \ + : "r" (x), "r" (y)); }) + +#define maskhl(x, y, z) \ + ({__asm__ __volatile__("maskhl %1, %2, %0" : "=r" (z) \ + : "r" (x), "r" (y)); }) + +#define insll(x, y, z) \ + ({__asm__ __volatile__("insll %1, %2, %0" : "=r" (z) \ + : "r" (x), "r" (y)); }) + +#define inshl(x, y, z) \ + ({__asm__ __volatile__("inshl %1, %2, %0" : "=r" (z) \ + : "r" (x), "r" (y)); }) + /* - * This is a version of ip_compute_csum() optimized for IP headers, - * which always checksum on 4 octet boundaries. + * This is a version of ip_compute_csum() optimized for IP headers, + * which always checksum on 4 octet boundaries. */ extern __sum16 ip_fast_csum(const void *iph, unsigned int ihl);
@@ -55,7 +79,7 @@ __wsum csum_partial_copy_nocheck(const void *src, void *dst, int len); extern __sum16 ip_compute_csum(const void *buff, int len);
/* - * Fold a partial checksum without adding pseudo headers + * Fold a partial checksum without adding pseudo headers */
static inline __sum16 csum_fold(__wsum csum) @@ -71,4 +95,32 @@ static inline __sum16 csum_fold(__wsum csum) extern __sum16 csum_ipv6_magic(const struct in6_addr *saddr, const struct in6_addr *daddr, __u32 len, __u8 proto, __wsum sum); + +static inline unsigned short from64to16(unsigned long x) +{ + /* + * Using extract instructions is a bit more efficient + * than the original shift/bitmask version. + */ + + union { + unsigned long ul; + unsigned int ui[2]; + unsigned short us[4]; + } in_v, tmp_v, out_v; + + in_v.ul = x; + tmp_v.ul = (unsigned long)in_v.ui[0] + (unsigned long)in_v.ui[1]; + + /* + * Since the bits of tmp_v.sh[3] are going to always be zero, + * we don't have to bother to add that in. + */ + out_v.ul = (unsigned long)tmp_v.us[0] + (unsigned long)tmp_v.us[1] + + (unsigned long)tmp_v.us[2]; + + /* Similarly, out_v.us[2] is always zero for the final add. */ + return out_v.us[0] + out_v.us[1]; +} + #endif diff --git a/arch/sw_64/lib/checksum.c b/arch/sw_64/lib/checksum.c index 561bbac59f8d..1531b09cad11 100644 --- a/arch/sw_64/lib/checksum.c +++ b/arch/sw_64/lib/checksum.c @@ -7,31 +7,7 @@ #include <linux/module.h> #include <linux/string.h> #include <asm/byteorder.h> - -static inline unsigned short from64to16(unsigned long x) -{ - /* Using extract instructions is a bit more efficient - * than the original shift/bitmask version. - */ - - union { - unsigned long ul; - unsigned int ui[2]; - unsigned short us[4]; - } in_v, tmp_v, out_v; - - in_v.ul = x; - tmp_v.ul = (unsigned long) in_v.ui[0] + (unsigned long) in_v.ui[1]; - - /* Since the bits of tmp_v.sh[3] are going to always be zero, - *we don't have to bother to add that in. - */ - out_v.ul = (unsigned long) tmp_v.us[0] + (unsigned long) tmp_v.us[1] - + (unsigned long) tmp_v.us[2]; - - /* Similarly, out_v.us[2] is always zero for the final add. */ - return out_v.us[0] + out_v.us[1]; -} +#include <asm/checksum.h>
/* * computes the checksum of the TCP/UDP pseudo-header @@ -69,73 +45,61 @@ EXPORT_SYMBOL(csum_tcpudp_nofold);
/* * Do a 64-bit checksum on an arbitrary memory area.. - * - * This isn't a great routine, but it's not _horrible_ either. The - * inner loop could be unrolled a bit further, and there are better - * ways to do the carry, but this is reasonable. */ static inline unsigned long do_csum(const unsigned char *buff, int len) { - int odd, count; - unsigned long result = 0; - - if (len <= 0) - goto out; - odd = 1 & (unsigned long) buff; - if (odd) { - result = *buff << 8; - len--; - buff++; - } - count = len >> 1; /* nr of 16-bit words.. */ - if (count) { - if (2 & (unsigned long) buff) { - result += *(unsigned short *) buff; - count--; - len -= 2; - buff += 2; - } - count >>= 1; /* nr of 32-bit words.. */ - if (count) { - if (4 & (unsigned long) buff) { - result += *(unsigned int *) buff; - count--; - len -= 4; - buff += 4; - } - count >>= 1; /* nr of 64-bit words.. */ - if (count) { - unsigned long carry = 0; - - do { - unsigned long w = *(unsigned long *) buff; - - count--; - buff += 8; - result += carry; - result += w; - carry = (w > result); - } while (count); - result += carry; - result = (result & 0xffffffff) + (result >> 32); - } - if (len & 4) { - result += *(unsigned int *) buff; - buff += 4; - } + const unsigned long *dst = (unsigned long *)buff; + unsigned long doff = 7 & (unsigned long) dst; + unsigned long checksum = 0; + unsigned long word, patch; + unsigned long partial_dest, second_dest; + + len -= 8; + + if (!doff) { + while (len > 0) { + word = *dst; + checksum += word; + checksum += (checksum < word); + dst++; + len -= 8; } - if (len & 2) { - result += *(unsigned short *) buff; - buff += 2; + + len += 8; + word = *dst; + + if (len != 8) + maskll(word, len, word); + + checksum += word; + checksum += (checksum < word); + } else { + dst = (unsigned long *)((unsigned long)dst & (~7UL)); + word = *dst; + inshl(word, doff, partial_dest); + dst++; + + while (len >= 0) { + word = *dst; + insll(word, doff, second_dest); + patch = partial_dest | second_dest; + checksum += patch; + checksum += (checksum < patch); + inshl(word, doff, partial_dest); + dst++; + len -= 8; } + + len += 8; + word = *dst; + insll(word, doff, second_dest); + patch = partial_dest | second_dest; + maskll(patch, len, patch); + checksum += patch; + checksum += (checksum < patch); } - if (len & 1) - result += *buff; - result = from64to16(result); - if (odd) - result = ((result >> 8) & 0xff) | ((result & 0xff) << 8); -out: - return result; + + return from64to16(checksum); }
/* diff --git a/arch/sw_64/lib/csum_partial_copy.c b/arch/sw_64/lib/csum_partial_copy.c index 678d9aa78d15..f45d64631281 100644 --- a/arch/sw_64/lib/csum_partial_copy.c +++ b/arch/sw_64/lib/csum_partial_copy.c @@ -11,6 +11,7 @@ #include <linux/types.h> #include <linux/string.h> #include <linux/uaccess.h> +#include <asm/checksum.h>
#define ldl_u(x, y) \ @@ -37,25 +38,6 @@ static inline void sthl_u(unsigned long data, unsigned long *dst) *((char *)dst + 8 - doff + i) = *((char *)&data + 8 - doff + i); }
-#define extll(x, y, z) \ - __asm__ __volatile__("extll %1, %2, %0":"=r" (z):"r" (x), "r" (y)) - -#define exthl(x, y, z) \ - __asm__ __volatile__("exthl %1, %2, %0":"=r" (z):"r" (x), "r" (y)) - -#define maskll(x, y, z) \ - __asm__ __volatile__("maskll %1, %2, %0":"=r" (z):"r" (x), "r" (y)) - -#define maskhl(x, y, z) \ - __asm__ __volatile__("maskhl %1, %2, %0":"=r" (z):"r" (x), "r" (y)) - -#define insll(x, y, z) \ - __asm__ __volatile__("insll %1, %2, %0":"=r" (z):"r" (x), "r" (y)) - -#define inshl(x, y, z) \ - __asm__ __volatile__("inshl %1, %2, %0":"=r" (z):"r" (x), "r" (y)) - - #define __get_word(insn, x, ptr) \ ({ \ long __guu_err; \ @@ -71,286 +53,84 @@ static inline void sthl_u(unsigned long data, unsigned long *dst) __guu_err; \ })
-static inline unsigned short from64to16(unsigned long x) -{ - /* Using extract instructions is a bit more efficient - * than the original shift/bitmask version. - */ - - union { - unsigned long ul; - unsigned int ui[2]; - unsigned short us[4]; - } in_v, tmp_v, out_v; - - in_v.ul = x; - tmp_v.ul = (unsigned long) in_v.ui[0] + (unsigned long) in_v.ui[1]; - - /* Since the bits of tmp_v.sh[3] are going to always be zero, - * we don't have to bother to add that in. - */ - out_v.ul = (unsigned long) tmp_v.us[0] + (unsigned long) tmp_v.us[1] - + (unsigned long) tmp_v.us[2]; - - /* Similarly, out_v.us[2] is always zero for the final add. */ - return out_v.us[0] + out_v.us[1]; -} - -/* - * Ok. This isn't fun, but this is the EASY case. - */ -static inline unsigned long -csum_partial_cfu_aligned(const unsigned long __user *src, unsigned long *dst, - long len) -{ - unsigned long checksum = ~0U; - unsigned long carry = 0; - - while (len >= 0) { - unsigned long word; - - if (__get_word(ldl, word, src)) - return 0; - checksum += carry; - src++; - checksum += word; - len -= 8; - carry = checksum < word; - *dst = word; - dst++; - } - len += 8; - checksum += carry; - if (len) { - int i = 0; - unsigned long word; - - if (__get_word(ldl, word, src)) - return 0; - maskll(word, len, word); - checksum += word; - carry = checksum < word; - for (; i < len; i++) - *((char *)dst + i) = *((char *)&word + i); - checksum += carry; - } - return checksum; -} - -/* - * This is even less fun, but this is still reasonably - * easy. - */ static inline unsigned long csum_partial_cfu_dest_aligned(const unsigned long __user *src, - unsigned long *dst, unsigned long soff, long len) + unsigned long *dst, long len) { - unsigned long first; - unsigned long word, carry; - unsigned long lastsrc = 7+len+(unsigned long)src; - unsigned long checksum = ~0U; - - if (__get_word(ldl_u, first, src)) - return 0; - carry = 0; - while (len >= 0) { - unsigned long second; - - if (__get_word(ldl_u, second, src+1)) - return 0; - extll(first, soff, word); - len -= 8; - src++; - exthl(second, soff, first); - checksum += carry; - word |= first; - first = second; - checksum += word; - *dst = word; - dst++; - carry = checksum < word; - } - len += 8; - checksum += carry; - if (len) { - int i = 0; - unsigned long second; - - if (__get_word(ldl_u, second, lastsrc)) - return 0; - extll(first, soff, word); - exthl(second, soff, first); - word |= first; - maskll(word, len, word); - checksum += word; - carry = checksum < word; - for (; i < len; i++) - *((char *)dst + i) = *((char *)&word + i); - checksum += carry; - } - return checksum; -} - -/* - * This is slightly less fun than the above.. - */ -static inline unsigned long -csum_partial_cfu_src_aligned(const unsigned long __user *src, - unsigned long *dst, unsigned long doff, - long len, unsigned long partial_dest) -{ - unsigned long carry = 0; unsigned long word; - unsigned long second_dest; - int i; unsigned long checksum = ~0U; + int err = 0;
- if (len >= 0) { - if (__get_word(ldl, word, src)) - return 0; - checksum += carry; + err = __copy_from_user(dst, src, len+8); + while (len > 0) { + word = *dst; checksum += word; - carry = checksum < word; - stll_u(word, dst); - len -= 8; - src++; + checksum += (checksum < word); dst++; - - inshl(word, doff, partial_dest); - while (len >= 0) { - if (__get_word(ldl, word, src)) - return 0; - len -= 8; - insll(word, doff, second_dest); - checksum += carry; - stl_u(partial_dest | second_dest, dst); - src++; - checksum += word; - inshl(word, doff, partial_dest); - carry = checksum < word; - dst++; - } - sthl_u(word, dst - 1); + len -= 8; } len += 8; + word = *dst;
- if (__get_word(ldl, word, src)) - return 0; - maskll(word, len, word); - checksum += carry; + if (len != 8) + maskll(word, len, word); checksum += word; - carry = checksum < word; - for (i = 0; i < len; i++) - *((char *)dst + i) = *((char *)&word + i); + checksum += (checksum < word);
- checksum += carry; return checksum; }
-/* - * This is so totally un-fun that it's frightening. Don't - * look at this too closely, you'll go blind. - */ static inline unsigned long -csum_partial_cfu_unaligned(const unsigned long __user *src, - unsigned long *dst, unsigned long soff, unsigned long doff, - long len, unsigned long partial_dest) +csum_partial_cfu_dest_unaligned(const unsigned long __user *src, + unsigned long *dst, unsigned long doff, long len) { - unsigned long carry = 0; - unsigned long first; - unsigned long second, word; - unsigned long second_dest; - int i; + unsigned long word, patch; + unsigned long partial_dest, second_dest; unsigned long checksum = ~0U; + int err = 0;
- if (__get_word(ldl_u, first, src)) - return 0; - if (len >= 0) { - extll(first, soff, word); - if (__get_word(ldl_u, second, src+1)) - return 0; - exthl(second, soff, first); - word |= first; - checksum += carry; - checksum += word; - carry = checksum < word; - stll_u(word, dst); - sthl_u(word, dst); - len -= 8; - src++; - dst++; + err = __copy_from_user(dst, src, len+8); + + dst = (unsigned long *)((unsigned long)dst & (~7UL)); + word = *dst; + inshl(word, doff, partial_dest); + dst++;
- if (__get_word(ldl_u, first, src)) - return 0; - ldl_u(partial_dest, dst); - maskll(partial_dest, doff, partial_dest); - while (len >= 0) { - if (__get_word(ldl_u, second, src+1)) - return 0; - extll(first, soff, word); - checksum += carry; - len -= 8; - exthl(second, soff, first); - src++; - word |= first; - first = second; - insll(word, doff, second_dest); - checksum += word; - stl_u(partial_dest | second_dest, dst); - carry = checksum < word; - inshl(word, doff, partial_dest); - dst++; - } - sthl_u(word, dst - 1); + while (len >= 0) { + word = *dst; + insll(word, doff, second_dest); + patch = partial_dest | second_dest; + checksum += patch; + checksum += (checksum < patch); + inshl(word, doff, partial_dest); + dst++; + len -= 8; } - len += 8;
- checksum += carry; - if (__get_word(ldl_u, second, src+1)) - return 0; - extll(first, soff, word); - exthl(second, soff, first); - word |= first; - maskll(word, len, word); - checksum += word; - carry = checksum < word; - for (i = 0; i < len; i++) - *((char *)dst + i) = *((char *)&word + i); + len += 8; + word = *dst; + insll(word, doff, second_dest); + patch = partial_dest | second_dest; + maskll(patch, len, patch); + checksum += patch; + checksum += (checksum < patch);
- checksum += carry; return checksum; }
static __wsum __csum_and_copy(const void __user *src, void *dst, int len) { unsigned long checksum; - unsigned long soff = 7 & (unsigned long) src; unsigned long doff = 7 & (unsigned long) dst;
if (!doff) { - if (!soff) - checksum = csum_partial_cfu_aligned( - (const unsigned long __user *) src, - (unsigned long *) dst, len-8); - else - checksum = csum_partial_cfu_dest_aligned( - (const unsigned long __user *) src, - (unsigned long *) dst, - soff, len-8); + checksum = csum_partial_cfu_dest_aligned( + (const unsigned long __user *) src, + (unsigned long *) dst, len-8); } else { - unsigned long partial_dest; - - ldl_u(partial_dest, dst); - if (!soff) - checksum = csum_partial_cfu_src_aligned( - (const unsigned long __user *) src, - (unsigned long *) dst, - doff, len-8, partial_dest); - else - checksum = csum_partial_cfu_unaligned( - (const unsigned long __user *) src, - (unsigned long *) dst, - soff, doff, len-8, partial_dest); + checksum = csum_partial_cfu_dest_aligned( + (const unsigned long __user *) src, + (unsigned long *) dst, len-8); } return (__force __wsum)from64to16(checksum); }
From: Min Fanlei minfanlei@wxiat.com
Sunway inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I56WV8
--------------------------------
We distinguish the running mode of emulator from guest by EMUL_FLAG according to the related changes of qemu.
Signed-off-by: Min Fanlei minfanlei@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/chip/chip3/chip.c | 2 +- arch/sw_64/include/asm/hw_init.h | 3 +++ arch/sw_64/kernel/setup.c | 14 +++++++++++--- 3 files changed, 15 insertions(+), 4 deletions(-)
diff --git a/arch/sw_64/chip/chip3/chip.c b/arch/sw_64/chip/chip3/chip.c index adb4d325fc91..f4b17cfa1127 100644 --- a/arch/sw_64/chip/chip3/chip.c +++ b/arch/sw_64/chip/chip3/chip.c @@ -116,7 +116,7 @@ static int chip3_get_cpu_nums(void)
static unsigned long chip3_get_vt_node_mem(int nodeid) { - return *(unsigned long *)MMSIZE; + return *(unsigned long *)MMSIZE & MMSIZE_MASK; }
static unsigned long chip3_get_node_mem(int nodeid) diff --git a/arch/sw_64/include/asm/hw_init.h b/arch/sw_64/include/asm/hw_init.h index f3d75e0fda7a..ede8da6b0314 100644 --- a/arch/sw_64/include/asm/hw_init.h +++ b/arch/sw_64/include/asm/hw_init.h @@ -97,6 +97,9 @@ static inline bool icache_is_vivt_no_ictag(void) return (cpu_desc.arch_var == 0x3 && cpu_desc.arch_rev == 0x1); }
+#define EMUL_FLAG (0x1UL << 63) +#define MMSIZE_MASK (EMUL_FLAG - 1) + DECLARE_STATIC_KEY_TRUE(run_mode_host_key); DECLARE_STATIC_KEY_FALSE(run_mode_guest_key); DECLARE_STATIC_KEY_FALSE(run_mode_emul_key); diff --git a/arch/sw_64/kernel/setup.c b/arch/sw_64/kernel/setup.c index 27f59fe0d1b1..e5abef2dc593 100644 --- a/arch/sw_64/kernel/setup.c +++ b/arch/sw_64/kernel/setup.c @@ -74,6 +74,7 @@ #define DBGDCONT(args...) #endif
+ DEFINE_PER_CPU(unsigned long, hard_node_id) = { 0 };
#if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) @@ -644,10 +645,17 @@ static void __init setup_cpu_info(void) cpu_desc.va_bits = CPUID_VA_BITS(val);
if (*(unsigned long *)MMSIZE) { - pr_info("run mode: guest\n"); static_branch_disable(&run_mode_host_key); - static_branch_enable(&run_mode_guest_key); - static_branch_disable(&run_mode_emul_key); + if (*(unsigned long *)MMSIZE & EMUL_FLAG) { + pr_info("run mode: emul\n"); + static_branch_disable(&run_mode_guest_key); + static_branch_enable(&run_mode_emul_key); + + } else { + pr_info("run mode: guest\n"); + static_branch_enable(&run_mode_guest_key); + static_branch_disable(&run_mode_emul_key); + } } else { pr_info("run mode: host\n"); static_branch_enable(&run_mode_host_key);
From: Mao Minkai maominkai@wxiat.com
Sunway inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I56OLG
--------------------------------
There's no need to manually define MAX_ASN. Use calculated HARDWARE_ASN_MASK instead.
Signed-off-by: Mao Minkai maominkai@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/include/asm/mmu_context.h | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/arch/sw_64/include/asm/mmu_context.h b/arch/sw_64/include/asm/mmu_context.h index 6b2ab3224ec9..e3d7ae7c873e 100644 --- a/arch/sw_64/include/asm/mmu_context.h +++ b/arch/sw_64/include/asm/mmu_context.h @@ -48,7 +48,6 @@ __reload_thread(struct pcb_struct *pcb) */
#ifdef CONFIG_SUBARCH_C3B -#define MAX_ASN 1023 #define WIDTH_HARDWARE_ASN 10 #endif
@@ -89,7 +88,7 @@ __get_new_mm_context(struct mm_struct *mm, long cpu) unsigned long asn = cpu_last_asn(cpu); unsigned long next = asn + 1;
- if ((asn & HARDWARE_ASN_MASK) >= MAX_ASN) { + if ((asn & HARDWARE_ASN_MASK) >= HARDWARE_ASN_MASK) { tbiap(); next = (asn & ~HARDWARE_ASN_MASK) + ASN_FIRST_VERSION; }
Sunway inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I56OLG
--------------------------------
Signed-off-by: Gu Zitao guzitao@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/include/asm/pci.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/arch/sw_64/include/asm/pci.h b/arch/sw_64/include/asm/pci.h index c917009c9530..f1c291101811 100644 --- a/arch/sw_64/include/asm/pci.h +++ b/arch/sw_64/include/asm/pci.h @@ -42,7 +42,8 @@ struct pci_controller { DECLARE_BITMAP(piu_msiconfig, 256); int int_irq; /* For compatibility with current (as of July 2003) pciutils - and XFree86. Eventually will be removed. */ + * and XFree86. Eventually will be removed. + */ unsigned int need_domain_info; bool iommu_enable; struct sunway_iommu *pci_iommu;
From: He Sheng hesheng@wxiat.com
Sunway inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I56OLG
--------------------------------
They make no sense and should be removed.
Signed-off-by: He Sheng hesheng@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/include/uapi/asm/auxvec.h | 2 -- arch/sw_64/kernel/module.c | 4 ---- 2 files changed, 6 deletions(-)
diff --git a/arch/sw_64/include/uapi/asm/auxvec.h b/arch/sw_64/include/uapi/asm/auxvec.h index 59854f3ac501..5b554a593fc0 100644 --- a/arch/sw_64/include/uapi/asm/auxvec.h +++ b/arch/sw_64/include/uapi/asm/auxvec.h @@ -3,10 +3,8 @@ #define _UAPI_ASM_SW64_AUXVEC_H
/* Reserve these numbers for any future use of a VDSO. */ -#if 1 #define AT_SYSINFO 32 #define AT_SYSINFO_EHDR 33 -#endif
/* * More complete cache descriptions than AT_[DIU]CACHEBSIZE. If the diff --git a/arch/sw_64/kernel/module.c b/arch/sw_64/kernel/module.c index c75d8a2e4309..d2041afba7bb 100644 --- a/arch/sw_64/kernel/module.c +++ b/arch/sw_64/kernel/module.c @@ -7,11 +7,7 @@ #include <linux/kernel.h> #include <linux/slab.h>
-#if 0 -#define DEBUGP printk -#else #define DEBUGP(fmt...) -#endif
/* Allocate the GOT at the end of the core sections. */
From: Mao Minkai maominkai@wxiat.com
Sunway inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I56OLG
--------------------------------
There's no need to manually define MAX_VPN. Use calculated HARDWARE_VPN_MASK instead.
Signed-off-by: Mao Minkai maominkai@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/kvm/kvm-sw64.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/arch/sw_64/kvm/kvm-sw64.c b/arch/sw_64/kvm/kvm-sw64.c index 1481c3dbb211..e1f7a82fbc26 100644 --- a/arch/sw_64/kvm/kvm-sw64.c +++ b/arch/sw_64/kvm/kvm-sw64.c @@ -37,7 +37,6 @@ unsigned long sw64_kvm_last_vpn[NR_CPUS]; #define cpu_last_vpn(cpuid) sw64_kvm_last_vpn[cpuid]
#ifdef CONFIG_SUBARCH_C3B -#define MAX_VPN 255 #define WIDTH_HARDWARE_VPN 8 #endif
@@ -80,7 +79,7 @@ static unsigned long __get_new_vpn_context(struct kvm_vcpu *vcpu, long cpu) unsigned long vpn = cpu_last_vpn(cpu); unsigned long next = vpn + 1;
- if ((vpn & HARDWARE_VPN_MASK) >= MAX_VPN) { + if ((vpn & HARDWARE_VPN_MASK) >= HARDWARE_VPN_MASK) { tbia(); next = (vpn & ~HARDWARE_VPN_MASK) + VPN_FIRST_VERSION + 1; /* bypass 0 */ }
From: Mao Minkai maominkai@wxiat.com
Sunway inclusion category: feature bugzilla: https://gitee.com/openeuler/kernel/issues/I56OLG
--------------------------------
Signed-off-by: Mao Minkai maominkai@wxiat.com
Signed-off-by: Gu Zitao guzitao@wxiat.com --- arch/sw_64/include/asm/vga.h | 18 +++++------ arch/sw_64/include/uapi/asm/ioctls.h | 46 ++++++++++++++-------------- 2 files changed, 32 insertions(+), 32 deletions(-)
diff --git a/arch/sw_64/include/asm/vga.h b/arch/sw_64/include/asm/vga.h index 3ca5c397b946..28adb8b8b7f1 100644 --- a/arch/sw_64/include/asm/vga.h +++ b/arch/sw_64/include/asm/vga.h @@ -55,29 +55,29 @@ extern void scr_memcpyw(u16 *d, const u16 *s, unsigned int count);
extern struct pci_controller *pci_vga_hose;
-# define __is_port_vga(a) \ +#define __is_port_vga(a) \ (((a) >= 0x3b0) && ((a) < 0x3e0) && \ ((a) != 0x3b3) && ((a) != 0x3d3))
-# define __is_mem_vga(a) \ +#define __is_mem_vga(a) \ (((a) >= 0xa0000) && ((a) <= 0xc0000))
-# define FIXUP_IOADDR_VGA(a) do { \ +#define FIXUP_IOADDR_VGA(a) do { \ if (pci_vga_hose && __is_port_vga(a)) \ (a) += pci_vga_hose->io_space->start; \ } while (0)
-# define FIXUP_MEMADDR_VGA(a) do { \ +#define FIXUP_MEMADDR_VGA(a) do { \ if (pci_vga_hose && __is_mem_vga(a)) \ (a) += pci_vga_hose->mem_space->start; \ } while (0)
#else /* CONFIG_VGA_HOSE */ -# define pci_vga_hose 0 -# define __is_port_vga(a) 0 -# define __is_mem_vga(a) 0 -# define FIXUP_IOADDR_VGA(a) -# define FIXUP_MEMADDR_VGA(a) +#define pci_vga_hose 0 +#define __is_port_vga(a) 0 +#define __is_mem_vga(a) 0 +#define FIXUP_IOADDR_VGA(a) +#define FIXUP_MEMADDR_VGA(a) #endif /* CONFIG_VGA_HOSE */
#define VGA_MAP_MEM(x, s) ((unsigned long)ioremap(x, s)) diff --git a/arch/sw_64/include/uapi/asm/ioctls.h b/arch/sw_64/include/uapi/asm/ioctls.h index eab34173f222..b4b23b5d14e8 100644 --- a/arch/sw_64/include/uapi/asm/ioctls.h +++ b/arch/sw_64/include/uapi/asm/ioctls.h @@ -52,20 +52,20 @@ #define TIOCMBIS 0x5416 #define TIOCMBIC 0x5417 #define TIOCMSET 0x5418 -# define TIOCM_LE 0x001 -# define TIOCM_DTR 0x002 -# define TIOCM_RTS 0x004 -# define TIOCM_ST 0x008 -# define TIOCM_SR 0x010 -# define TIOCM_CTS 0x020 -# define TIOCM_CAR 0x040 -# define TIOCM_RNG 0x080 -# define TIOCM_DSR 0x100 -# define TIOCM_CD TIOCM_CAR -# define TIOCM_RI TIOCM_RNG -# define TIOCM_OUT1 0x2000 -# define TIOCM_OUT2 0x4000 -# define TIOCM_LOOP 0x8000 +#define TIOCM_LE 0x001 +#define TIOCM_DTR 0x002 +#define TIOCM_RTS 0x004 +#define TIOCM_ST 0x008 +#define TIOCM_SR 0x010 +#define TIOCM_CTS 0x020 +#define TIOCM_CAR 0x040 +#define TIOCM_RNG 0x080 +#define TIOCM_DSR 0x100 +#define TIOCM_CD TIOCM_CAR +#define TIOCM_RI TIOCM_RNG +#define TIOCM_OUT1 0x2000 +#define TIOCM_OUT2 0x4000 +#define TIOCM_LOOP 0x8000
#define TIOCGSOFTCAR 0x5419 #define TIOCSSOFTCAR 0x541A @@ -74,14 +74,14 @@ #define TIOCGSERIAL 0x541E #define TIOCSSERIAL 0x541F #define TIOCPKT 0x5420 -# define TIOCPKT_DATA 0 -# define TIOCPKT_FLUSHREAD 1 -# define TIOCPKT_FLUSHWRITE 2 -# define TIOCPKT_STOP 4 -# define TIOCPKT_START 8 -# define TIOCPKT_NOSTOP 16 -# define TIOCPKT_DOSTOP 32 -# define TIOCPKT_IOCTL 64 +#define TIOCPKT_DATA 0 +#define TIOCPKT_FLUSHREAD 1 +#define TIOCPKT_FLUSHWRITE 2 +#define TIOCPKT_STOP 4 +#define TIOCPKT_START 8 +#define TIOCPKT_NOSTOP 16 +#define TIOCPKT_DOSTOP 32 +#define TIOCPKT_IOCTL 64
#define TIOCNOTTY 0x5422 @@ -113,7 +113,7 @@ #define TIOCSERGSTRUCT 0x5458 /* For debugging only */ #define TIOCSERGETLSR 0x5459 /* Get line status register */ /* ioctl (fd, TIOCSERGETLSR, &result) where result may be as below */ -# define TIOCSER_TEMT 0x01 /* Transmitter physically empty */ +#define TIOCSER_TEMT 0x01 /* Transmitter physically empty */ #define TIOCSERGETMULTI 0x545A /* Get multiport config */ #define TIOCSERSETMULTI 0x545B /* Set multiport config */
-
Gu Zitao