[v7,07/13] LoongArch: Linux Syscall Interface

  ---
 sysdeps/loongarch/abort-instr.h               |   2 +
 sysdeps/loongarch/hp-timing.h                 |  42 +++
 .../unix/sysv/linux/loongarch/arch-syscall.h  | 301 ++++++++++++++++
 sysdeps/unix/sysv/linux/loongarch/clone.S     |  96 ++++++
 sysdeps/unix/sysv/linux/loongarch/clone3.S    |  83 +++++
 sysdeps/unix/sysv/linux/loongarch/sysdep.S    |  53 +++
 sysdeps/unix/sysv/linux/loongarch/sysdep.h    | 320 ++++++++++++++++++
 sysdeps/unix/sysv/linux/loongarch/vfork.S     |  50 +++
 8 files changed, 947 insertions(+)
 create mode 100644 sysdeps/loongarch/abort-instr.h
 create mode 100644 sysdeps/loongarch/hp-timing.h
 create mode 100644 sysdeps/unix/sysv/linux/loongarch/arch-syscall.h
 create mode 100644 sysdeps/unix/sysv/linux/loongarch/clone.S
 create mode 100644 sysdeps/unix/sysv/linux/loongarch/clone3.S
 create mode 100644 sysdeps/unix/sysv/linux/loongarch/sysdep.S
 create mode 100644 sysdeps/unix/sysv/linux/loongarch/sysdep.h
 create mode 100644 sysdeps/unix/sysv/linux/loongarch/vfork.S
  

LGTM, thanks.

Reviewed-by: Adhemerval Zanella  <adhemerval.zanella@linaro.org>

On 18/07/22 22:20, caiyinyu wrote:
> ---
>  sysdeps/loongarch/abort-instr.h               |   2 +
>  sysdeps/loongarch/hp-timing.h                 |  42 +++
>  .../unix/sysv/linux/loongarch/arch-syscall.h  | 301 ++++++++++++++++
>  sysdeps/unix/sysv/linux/loongarch/clone.S     |  96 ++++++
>  sysdeps/unix/sysv/linux/loongarch/clone3.S    |  83 +++++
>  sysdeps/unix/sysv/linux/loongarch/sysdep.S    |  53 +++
>  sysdeps/unix/sysv/linux/loongarch/sysdep.h    | 320 ++++++++++++++++++
>  sysdeps/unix/sysv/linux/loongarch/vfork.S     |  50 +++
>  8 files changed, 947 insertions(+)
>  create mode 100644 sysdeps/loongarch/abort-instr.h
>  create mode 100644 sysdeps/loongarch/hp-timing.h
>  create mode 100644 sysdeps/unix/sysv/linux/loongarch/arch-syscall.h
>  create mode 100644 sysdeps/unix/sysv/linux/loongarch/clone.S
>  create mode 100644 sysdeps/unix/sysv/linux/loongarch/clone3.S
>  create mode 100644 sysdeps/unix/sysv/linux/loongarch/sysdep.S
>  create mode 100644 sysdeps/unix/sysv/linux/loongarch/sysdep.h
>  create mode 100644 sysdeps/unix/sysv/linux/loongarch/vfork.S
> 
> diff --git a/sysdeps/loongarch/abort-instr.h b/sysdeps/loongarch/abort-instr.h
> new file mode 100644
> index 0000000000..46d3ad0871
> --- /dev/null
> +++ b/sysdeps/loongarch/abort-instr.h
> @@ -0,0 +1,2 @@
> +/* An instruction which should crash any program is a breakpoint.  */
> +#define ABORT_INSTRUCTION asm ("break 0")
> diff --git a/sysdeps/loongarch/hp-timing.h b/sysdeps/loongarch/hp-timing.h
> new file mode 100644
> index 0000000000..af1da81024
> --- /dev/null
> +++ b/sysdeps/loongarch/hp-timing.h
> @@ -0,0 +1,42 @@
> +/* High precision, low overhead timing functions.
> +   Copyright (C) 2022 Free Software Foundation, Inc.
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library; if not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +#ifndef _HP_TIMING_H
> +#define _HP_TIMING_H 1
> +
> +/* We always assume having the timestamp register.  */
> +#define HP_TIMING_AVAIL (1)
> +#define HP_SMALL_TIMING_AVAIL (1)
> +
> +/* We indeed have inlined functions.  */
> +#define HP_TIMING_INLINE (1)
> +
> +/* We use 64bit values for the times.  */
> +typedef unsigned long long int hp_timing_t;
> +
> +/* Read the stable counter.  */
> +#define HP_TIMING_NOW(Var) \
> +  ({ \
> +    unsigned long long int _count; \
> +    asm volatile ("rdtime.d\t%0,$r0" : "=r" (_count)); \
> +    (Var) = _count; \
> +  })
> +
> +#include <hp-timing-common.h>
> +
> +#endif /* hp-timing.h */
> diff --git a/sysdeps/unix/sysv/linux/loongarch/arch-syscall.h b/sysdeps/unix/sysv/linux/loongarch/arch-syscall.h
> new file mode 100644
> index 0000000000..6bb3c8adbc
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/loongarch/arch-syscall.h
> @@ -0,0 +1,301 @@
> +/* AUTOGENERATED by update-syscall-lists.py.  */
> +#define __NR_accept 202
> +#define __NR_accept4 242
> +#define __NR_acct 89
> +#define __NR_add_key 217
> +#define __NR_adjtimex 171
> +#define __NR_bind 200
> +#define __NR_bpf 280
> +#define __NR_brk 214
> +#define __NR_capget 90
> +#define __NR_capset 91
> +#define __NR_chdir 49
> +#define __NR_chroot 51
> +#define __NR_clock_adjtime 266
> +#define __NR_clock_getres 114
> +#define __NR_clock_gettime 113
> +#define __NR_clock_nanosleep 115
> +#define __NR_clock_settime 112
> +#define __NR_clone 220
> +#define __NR_clone3 435
> +#define __NR_close 57
> +#define __NR_close_range 436
> +#define __NR_connect 203
> +#define __NR_copy_file_range 285
> +#define __NR_delete_module 106
> +#define __NR_dup 23
> +#define __NR_dup3 24
> +#define __NR_epoll_create1 20
> +#define __NR_epoll_ctl 21
> +#define __NR_epoll_pwait 22
> +#define __NR_epoll_pwait2 441
> +#define __NR_eventfd2 19
> +#define __NR_execve 221
> +#define __NR_execveat 281
> +#define __NR_exit 93
> +#define __NR_exit_group 94
> +#define __NR_faccessat 48
> +#define __NR_faccessat2 439
> +#define __NR_fadvise64 223
> +#define __NR_fallocate 47
> +#define __NR_fanotify_init 262
> +#define __NR_fanotify_mark 263
> +#define __NR_fchdir 50
> +#define __NR_fchmod 52
> +#define __NR_fchmodat 53
> +#define __NR_fchown 55
> +#define __NR_fchownat 54
> +#define __NR_fcntl 25
> +#define __NR_fdatasync 83
> +#define __NR_fgetxattr 10
> +#define __NR_finit_module 273
> +#define __NR_flistxattr 13
> +#define __NR_flock 32
> +#define __NR_fremovexattr 16
> +#define __NR_fsconfig 431
> +#define __NR_fsetxattr 7
> +#define __NR_fsmount 432
> +#define __NR_fsopen 430
> +#define __NR_fspick 433
> +#define __NR_fstatfs 44
> +#define __NR_fsync 82
> +#define __NR_ftruncate 46
> +#define __NR_futex 98
> +#define __NR_futex_waitv 449
> +#define __NR_get_mempolicy 236
> +#define __NR_get_robust_list 100
> +#define __NR_getcpu 168
> +#define __NR_getcwd 17
> +#define __NR_getdents64 61
> +#define __NR_getegid 177
> +#define __NR_geteuid 175
> +#define __NR_getgid 176
> +#define __NR_getgroups 158
> +#define __NR_getitimer 102
> +#define __NR_getpeername 205
> +#define __NR_getpgid 155
> +#define __NR_getpid 172
> +#define __NR_getppid 173
> +#define __NR_getpriority 141
> +#define __NR_getrandom 278
> +#define __NR_getresgid 150
> +#define __NR_getresuid 148
> +#define __NR_getrusage 165
> +#define __NR_getsid 156
> +#define __NR_getsockname 204
> +#define __NR_getsockopt 209
> +#define __NR_gettid 178
> +#define __NR_gettimeofday 169
> +#define __NR_getuid 174
> +#define __NR_getxattr 8
> +#define __NR_init_module 105
> +#define __NR_inotify_add_watch 27
> +#define __NR_inotify_init1 26
> +#define __NR_inotify_rm_watch 28
> +#define __NR_io_cancel 3
> +#define __NR_io_destroy 1
> +#define __NR_io_getevents 4
> +#define __NR_io_pgetevents 292
> +#define __NR_io_setup 0
> +#define __NR_io_submit 2
> +#define __NR_io_uring_enter 426
> +#define __NR_io_uring_register 427
> +#define __NR_io_uring_setup 425
> +#define __NR_ioctl 29
> +#define __NR_ioprio_get 31
> +#define __NR_ioprio_set 30
> +#define __NR_kcmp 272
> +#define __NR_kexec_file_load 294
> +#define __NR_kexec_load 104
> +#define __NR_keyctl 219
> +#define __NR_kill 129
> +#define __NR_landlock_add_rule 445
> +#define __NR_landlock_create_ruleset 444
> +#define __NR_landlock_restrict_self 446
> +#define __NR_lgetxattr 9
> +#define __NR_linkat 37
> +#define __NR_listen 201
> +#define __NR_listxattr 11
> +#define __NR_llistxattr 12
> +#define __NR_lookup_dcookie 18
> +#define __NR_lremovexattr 15
> +#define __NR_lseek 62
> +#define __NR_lsetxattr 6
> +#define __NR_madvise 233
> +#define __NR_mbind 235
> +#define __NR_membarrier 283
> +#define __NR_memfd_create 279
> +#define __NR_migrate_pages 238
> +#define __NR_mincore 232
> +#define __NR_mkdirat 34
> +#define __NR_mknodat 33
> +#define __NR_mlock 228
> +#define __NR_mlock2 284
> +#define __NR_mlockall 230
> +#define __NR_mmap 222
> +#define __NR_mount 40
> +#define __NR_mount_setattr 442
> +#define __NR_move_mount 429
> +#define __NR_move_pages 239
> +#define __NR_mprotect 226
> +#define __NR_mq_getsetattr 185
> +#define __NR_mq_notify 184
> +#define __NR_mq_open 180
> +#define __NR_mq_timedreceive 183
> +#define __NR_mq_timedsend 182
> +#define __NR_mq_unlink 181
> +#define __NR_mremap 216
> +#define __NR_msgctl 187
> +#define __NR_msgget 186
> +#define __NR_msgrcv 188
> +#define __NR_msgsnd 189
> +#define __NR_msync 227
> +#define __NR_munlock 229
> +#define __NR_munlockall 231
> +#define __NR_munmap 215
> +#define __NR_name_to_handle_at 264
> +#define __NR_nanosleep 101
> +#define __NR_nfsservctl 42
> +#define __NR_open_by_handle_at 265
> +#define __NR_open_tree 428
> +#define __NR_openat 56
> +#define __NR_openat2 437
> +#define __NR_perf_event_open 241
> +#define __NR_personality 92
> +#define __NR_pidfd_getfd 438
> +#define __NR_pidfd_open 434
> +#define __NR_pidfd_send_signal 424
> +#define __NR_pipe2 59
> +#define __NR_pivot_root 41
> +#define __NR_pkey_alloc 289
> +#define __NR_pkey_free 290
> +#define __NR_pkey_mprotect 288
> +#define __NR_ppoll 73
> +#define __NR_prctl 167
> +#define __NR_pread64 67
> +#define __NR_preadv 69
> +#define __NR_preadv2 286
> +#define __NR_prlimit64 261
> +#define __NR_process_madvise 440
> +#define __NR_process_mrelease 448
> +#define __NR_process_vm_readv 270
> +#define __NR_process_vm_writev 271
> +#define __NR_pselect6 72
> +#define __NR_ptrace 117
> +#define __NR_pwrite64 68
> +#define __NR_pwritev 70
> +#define __NR_pwritev2 287
> +#define __NR_quotactl 60
> +#define __NR_quotactl_fd 443
> +#define __NR_read 63
> +#define __NR_readahead 213
> +#define __NR_readlinkat 78
> +#define __NR_readv 65
> +#define __NR_reboot 142
> +#define __NR_recvfrom 207
> +#define __NR_recvmmsg 243
> +#define __NR_recvmsg 212
> +#define __NR_remap_file_pages 234
> +#define __NR_removexattr 14
> +#define __NR_renameat2 276
> +#define __NR_request_key 218
> +#define __NR_restart_syscall 128
> +#define __NR_rseq 293
> +#define __NR_rt_sigaction 134
> +#define __NR_rt_sigpending 136
> +#define __NR_rt_sigprocmask 135
> +#define __NR_rt_sigqueueinfo 138
> +#define __NR_rt_sigreturn 139
> +#define __NR_rt_sigsuspend 133
> +#define __NR_rt_sigtimedwait 137
> +#define __NR_rt_tgsigqueueinfo 240
> +#define __NR_sched_get_priority_max 125
> +#define __NR_sched_get_priority_min 126
> +#define __NR_sched_getaffinity 123
> +#define __NR_sched_getattr 275
> +#define __NR_sched_getparam 121
> +#define __NR_sched_getscheduler 120
> +#define __NR_sched_rr_get_interval 127
> +#define __NR_sched_setaffinity 122
> +#define __NR_sched_setattr 274
> +#define __NR_sched_setparam 118
> +#define __NR_sched_setscheduler 119
> +#define __NR_sched_yield 124
> +#define __NR_seccomp 277
> +#define __NR_semctl 191
> +#define __NR_semget 190
> +#define __NR_semop 193
> +#define __NR_semtimedop 192
> +#define __NR_sendfile 71
> +#define __NR_sendmmsg 269
> +#define __NR_sendmsg 211
> +#define __NR_sendto 206
> +#define __NR_set_mempolicy 237
> +#define __NR_set_mempolicy_home_node 450
> +#define __NR_set_robust_list 99
> +#define __NR_set_tid_address 96
> +#define __NR_setdomainname 162
> +#define __NR_setfsgid 152
> +#define __NR_setfsuid 151
> +#define __NR_setgid 144
> +#define __NR_setgroups 159
> +#define __NR_sethostname 161
> +#define __NR_setitimer 103
> +#define __NR_setns 268
> +#define __NR_setpgid 154
> +#define __NR_setpriority 140
> +#define __NR_setregid 143
> +#define __NR_setresgid 149
> +#define __NR_setresuid 147
> +#define __NR_setreuid 145
> +#define __NR_setsid 157
> +#define __NR_setsockopt 208
> +#define __NR_settimeofday 170
> +#define __NR_setuid 146
> +#define __NR_setxattr 5
> +#define __NR_shmat 196
> +#define __NR_shmctl 195
> +#define __NR_shmdt 197
> +#define __NR_shmget 194
> +#define __NR_shutdown 210
> +#define __NR_sigaltstack 132
> +#define __NR_signalfd4 74
> +#define __NR_socket 198
> +#define __NR_socketpair 199
> +#define __NR_splice 76
> +#define __NR_statfs 43
> +#define __NR_statx 291
> +#define __NR_swapoff 225
> +#define __NR_swapon 224
> +#define __NR_symlinkat 36
> +#define __NR_sync 81
> +#define __NR_sync_file_range 84
> +#define __NR_syncfs 267
> +#define __NR_sysinfo 179
> +#define __NR_syslog 116
> +#define __NR_tee 77
> +#define __NR_tgkill 131
> +#define __NR_timer_create 107
> +#define __NR_timer_delete 111
> +#define __NR_timer_getoverrun 109
> +#define __NR_timer_gettime 108
> +#define __NR_timer_settime 110
> +#define __NR_timerfd_create 85
> +#define __NR_timerfd_gettime 87
> +#define __NR_timerfd_settime 86
> +#define __NR_times 153
> +#define __NR_tkill 130
> +#define __NR_truncate 45
> +#define __NR_umask 166
> +#define __NR_umount2 39
> +#define __NR_uname 160
> +#define __NR_unlinkat 35
> +#define __NR_unshare 97
> +#define __NR_userfaultfd 282
> +#define __NR_utimensat 88
> +#define __NR_vhangup 58
> +#define __NR_vmsplice 75
> +#define __NR_wait4 260
> +#define __NR_waitid 95
> +#define __NR_write 64
> +#define __NR_writev 66
> diff --git a/sysdeps/unix/sysv/linux/loongarch/clone.S b/sysdeps/unix/sysv/linux/loongarch/clone.S
> new file mode 100644
> index 0000000000..c73245be82
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/loongarch/clone.S
> @@ -0,0 +1,96 @@
> +/* The clone syscall wrapper.
> +   Copyright (C) 2022 Free Software Foundation, Inc.
> +
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library.  If not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +/* clone() is even more special than fork() as it mucks with stacks
> +   and invokes a function in the right context after its all over.  */
> +
> +#include <sys/asm.h>
> +#include <sysdep.h>
> +#define _ERRNO_H  1
> +#include <bits/errno.h>
> +#include <tls.h>
> +#include "tcb-offsets.h"
> +
> +/* int clone(int (*fn)(void *arg), void *child_stack, int flags, void *arg,
> +   void *parent_tidptr, void *tls, void *child_tidptr) */
> +
> +ENTRY (__clone)
> +
> +	/* Align stack to 16.  */
> +	BSTRINS		a1, zero, 3, 0
> +
> +	/* Sanity check arguments.  */
> +	beqz		a0, L (invalid) /* No NULL function pointers.  */
> +	beqz		a1, L (invalid) /* No NULL stack pointers.  */
> +
> +	ADDI 		a1, a1, -16 /* Reserve argument save space.  */
> +	REG_S		a0, a1, 0   /* Save function pointer.  */
> +	REG_S		a3, a1, SZREG   /* Save argument pointer.  */
> +
> +	/* The syscall expects the args to be in different slots.  */
> +	or		a0, a2, zero
> +	or		a2, a4, zero
> +	or		a3, a6, zero
> +	or		a4, a5, zero
> +
> +	/* Do the system call.  */
> +	LI		a7,__NR_clone
> +	syscall		0
> +
> +	blt		a0, zero ,L (error)
> +	beqz		a0,L (thread_start)
> +
> +	/* Successful return from the parent.  */
> +	ret
> +
> +L (invalid):
> +	LI		a0, -EINVAL
> +
> +	/* Something bad happened -- no child created.  */
> +L (error):
> +	b		__syscall_error
> +
> +END (__clone)
> +
> +/* Load up the arguments to the function.  Put this block of code in
> +   its own function so that we can terminate the stack trace with our
> +   debug info.  */
> +ENTRY (__thread_start)
> +L (thread_start):
> +
> +/* Terminate call stack by noting ra is undefined.  Use a dummy
> +   .cfi_label to force starting the FDE.  */
> +	.cfi_label .Ldummy
> +	cfi_undefined (1)
> +
> +	/* Restore the arg for user's function.  */
> +	REG_L		a1, sp, 0   /* Function pointer.  */
> +	REG_L		a0, sp, SZREG   /* Argument pointer.  */
> +
> +	/* Call the user's function.  */
> +	jirl		ra, a1, 0
> +
> +	/* Call exit with the function's return value.  */
> +	LI		a7, __NR_exit
> +	syscall		0
> +
> +	END (__thread_start)
> +
> +libc_hidden_def (__clone)
> +weak_alias (__clone, clone)
> diff --git a/sysdeps/unix/sysv/linux/loongarch/clone3.S b/sysdeps/unix/sysv/linux/loongarch/clone3.S
> new file mode 100644
> index 0000000000..38be4c809b
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/loongarch/clone3.S
> @@ -0,0 +1,83 @@
> +/* The clone3 syscall wrapper.
> +   Copyright (C) 2022 Free Software Foundation, Inc.
> +
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library.  If not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +/* clone3() is even more special than fork() as it mucks with stacks
> +   and invokes a function in the right context after its all over.  */
> +
> +#include <sys/asm.h>
> +#include <sysdep.h>
> +#define _ERRNO_H  1
> +#include <bits/errno.h>
> +#include <tls.h>
> +#include "tcb-offsets.h"
> +
> +/* int clone3(struct clone_args *cl_args, size_t size,
> +   int (*func)(void *arg), void *arg); */
> +
> +ENTRY (__clone3)
> +
> +	/* Sanity check arguments.  */
> +	beqz		a0, L (invalid)	/* No NULL cl_args pointer.  */
> +	beqz		a2, L (invalid)	/* No NULL function pointer.  */
> +
> +	/* Do the system call.  */
> +	LI		a7, __NR_clone3
> +	syscall		0
> +
> +	blt		a0, zero ,L (error)
> +	beqz		a0, L (thread_start3)
> +
> +	/* Successful return from the parent.  */
> +	ret
> +
> +L (invalid):
> +	LI		a0, -EINVAL
> +
> +	/* Something bad happened -- no child created.  */
> +L (error):
> +	b		__syscall_error
> +
> +END (__clone3)
> +
> +/* Load up the arguments to the function.  Put this block of code in
> +   its own function so that we can terminate the stack trace with our
> +   debug info.  */
> +ENTRY (__thread_start3)
> +L (thread_start3):
> +
> +/* Terminate call stack by noting ra is undefined.  Use a dummy
> +   .cfi_label to force starting the FDE.  */
> +	.cfi_label .Ldummy
> +	cfi_undefined (1)
> +
> +	/* Align stack to 16.  */
> +	BSTRINS		sp, zero, 3, 0
> +
> +	/* Set up arguments for the function call.  */
> +	move		a0, a3		/* Argument.  */
> +	jirl		ra, a2, 0	/* Call function.  */
> +
> +	/* Call exit with the function's return value.  */
> +	LI		a7, __NR_exit
> +	syscall		0
> +
> +	END (__thread_start3)
> +
> +libc_hidden_def (__clone3)
> +weak_alias (__clone3, clone3)
> diff --git a/sysdeps/unix/sysv/linux/loongarch/sysdep.S b/sysdeps/unix/sysv/linux/loongarch/sysdep.S
> new file mode 100644
> index 0000000000..a4db2a97b5
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/loongarch/sysdep.S
> @@ -0,0 +1,53 @@
> +/* syscall error handlers
> +   Copyright (C) 2022 Free Software Foundation, Inc.
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library.  If not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +#include <sysdep.h>
> +
> +#if IS_IN (libc)
> +# define errno __libc_errno
> +#endif
> +
> +ENTRY (__syscall_error)
> +/* Fall through to __syscall_set_errno */
> +END (__syscall_error)
> +
> +/* Non-standard calling convention: argument in a0, return address in t0,
> +   and clobber only t1.
> + */
> +ENTRY (__syscall_set_errno)
> +
> +/* We got here because a0 < 0, but only codes in the range [-4095, -1]
> +   represent errors.  Otherwise, just return the result normally.
> + */
> +	li.d		t1, -4096
> +	bgeu		t1, a0, L (out)
> +	sub.w		a0, zero, a0
> +#if RTLD_PRIVATE_ERRNO
> +	la		t1, rtld_errno
> +#elif defined(__PIC__)
> +	la.tls.ie	t1, errno
> +	add.d		t1, tp, t1
> +#else
> +	la.tls.le	t1, errno
> +	add.d		t1, tp, t1
> +#endif
> +	st.w		a0, t1, 0
> +	li.d		a0, -1
> +L (out):
> +	ret
> +END (__syscall_set_errno)
> diff --git a/sysdeps/unix/sysv/linux/loongarch/sysdep.h b/sysdeps/unix/sysv/linux/loongarch/sysdep.h
> new file mode 100644
> index 0000000000..c586df819c
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/loongarch/sysdep.h
> @@ -0,0 +1,320 @@
> +/* Assembly macros for LoongArch.
> +   Copyright (C) 2022 Free Software Foundation, Inc.
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public
> +   License as published by the Free Software Foundation; either
> +   version 2.1 of the License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library.  If not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +#ifndef _LINUX_LOONGARCH_SYSDEP_H
> +#define _LINUX_LOONGARCH_SYSDEP_H 1
> +
> +#include <sysdeps/unix/sysv/linux/generic/sysdep.h>
> +#include <tls.h>
> +
> +#ifdef __ASSEMBLER__
> +
> +#include <sys/asm.h>
> +#define ret jirl zero, ra, 0
> +#define L(label) .L##label
> +
> +/* Performs a system call, handling errors by setting errno.  Linux indicates
> +   errors by setting a0 to a value between -1 and -4095.  */
> +#undef PSEUDO
> +#define PSEUDO(name, syscall_name, args) \
> +  ENTRY (name); \
> +  li.d a7, SYS_ify (syscall_name); \
> +  syscall 0; \
> +  li.d a7, -4096; \
> +  bltu a7, a0, .Lsyscall_error##name;
> +
> +#undef PSEUDO_END
> +#define PSEUDO_END(sym) \
> +  SYSCALL_ERROR_HANDLER (sym); \
> +  ret; \
> +  END (sym);
> +
> +#if !IS_IN(libc)
> +#if RTLD_PRIVATE_ERRNO
> +
> +#define SYSCALL_ERROR_HANDLER(name) \
> +  .Lsyscall_error##name : la t0, rtld_errno; \
> +  sub.w a0, zero, a0; \
> +  st.w a0, t0, 0; \
> +  li.d a0, -1;
> +
> +#else
> +
> +#define SYSCALL_ERROR_HANDLER(name) \
> +  .Lsyscall_error##name : la.tls.ie t0, errno; \
> +  add.d t0, tp, t0; \
> +  sub.w a0, zero, a0; \
> +  st.w a0, t0, 0; \
> +  li.d a0, -1;
> +
> +#endif
> +#else
> +
> +#define SYSCALL_ERROR_HANDLER(name) .Lsyscall_error##name : b __syscall_error;
> +
> +#endif
> +
> +/* Performs a system call, not setting errno.  */
> +#undef PSEUDO_NEORRNO
> +#define PSEUDO_NOERRNO(name, syscall_name, args) \
> +  ENTRY (name); \
> +  li.d a7, SYS_ify (syscall_name); \
> +  syscall 0;
> +
> +#undef PSEUDO_END_NOERRNO
> +#define PSEUDO_END_NOERRNO(name) END (name);
> +
> +#undef ret_NOERRNO
> +#define ret_NOERRNO ret
> +
> +/* Perfroms a system call, returning the error code.  */
> +#undef PSEUDO_ERRVAL
> +#define PSEUDO_ERRVAL(name, syscall_name, args) \
> +  PSEUDO_NOERRNO (name, syscall_name, args); \
> +  slli.d a0, a0, 32; \
> +  srai.d a0, a0, 32; /* sign_ext */ \
> +  sub.d a0, zero, a0;
> +
> +#undef PSEUDO_END_ERRVAL
> +#define PSEUDO_END_ERRVAL(name) END (name);
> +
> +#undef ret_ERRVAL
> +#define ret_ERRVAL ret
> +
> +#endif /* __ASSEMBLER__ */
> +
> +/* In order to get __set_errno() definition in INLINE_SYSCALL.  */
> +#ifndef __ASSEMBLER__
> +#include <errno.h>
> +#endif
> +
> +#include <sysdeps/unix/sysdep.h>
> +
> +#undef SYS_ify
> +#define SYS_ify(syscall_name) __NR_##syscall_name
> +
> +#ifndef __ASSEMBLER__
> +
> +#define VDSO_NAME "LINUX_2.6"
> +#define VDSO_HASH 61765110
> +
> +/* List of system calls which are supported as vsyscalls.  */
> +#define HAVE_CLOCK_GETRES64_VSYSCALL "__vdso_clock_getres"
> +#define HAVE_CLOCK_GETTIME64_VSYSCALL "__vdso_clock_gettime"
> +#define HAVE_GETCPU_VSYSCALL "__vdso_getcpu"
> +
> +#define HAVE_CLONE3_WRAPPER 1
> +
> +#define INTERNAL_SYSCALL(name, nr, args...) \
> +  internal_syscall##nr (SYS_ify (name), args)
> +
> +#define INTERNAL_SYSCALL_NCS(number, nr, args...) \
> +  internal_syscall##nr (number, args)
> +
> +#define internal_syscall0(number, dummy...) \
> +  ({ \
> +    long int _sys_result; \
> +\
> +    { \
> +      register long int __a7 asm ("$a7") = number; \
> +      register long int __a0 asm ("$a0"); \
> +      __asm__ volatile ("syscall 0\n\t" \
> +		       : "=r"(__a0) \
> +		       : "r"(__a7) \
> +		       : __SYSCALL_CLOBBERS); \
> +      _sys_result = __a0; \
> +    } \
> +    _sys_result; \
> +  })
> +
> +#define internal_syscall1(number, arg0) \
> +  ({ \
> +    long int _sys_result; \
> +\
> +    { \
> +      long int _arg0 = (long int) (arg0); \
> +      register long int __a7 asm ("$a7") = number; \
> +      register long int __a0 asm ("$a0") = _arg0; \
> +      __asm__ volatile ("syscall 0\n\t" \
> +		       : "+r"(__a0) \
> +		       : "r"(__a7) \
> +		       : __SYSCALL_CLOBBERS); \
> +      _sys_result = __a0; \
> +    } \
> +    _sys_result; \
> +  })
> +
> +#define internal_syscall2(number, arg0, arg1) \
> +  ({ \
> +    long int _sys_result; \
> +\
> +    { \
> +      long int _arg0 = (long int) (arg0); \
> +      long int _arg1 = (long int) (arg1); \
> +      register long int __a7 asm ("$a7") = number; \
> +      register long int __a0 asm ("$a0") = _arg0; \
> +      register long int __a1 asm ("$a1") = _arg1; \
> +      __asm__ volatile ("syscall 0\n\t" \
> +		       : "+r"(__a0) \
> +		       : "r"(__a7), "r"(__a1) \
> +		       : __SYSCALL_CLOBBERS); \
> +      _sys_result = __a0; \
> +    } \
> +    _sys_result; \
> +  })
> +
> +#define internal_syscall3(number, arg0, arg1, arg2) \
> +  ({ \
> +    long int _sys_result; \
> +\
> +    { \
> +      long int _arg0 = (long int) (arg0); \
> +      long int _arg1 = (long int) (arg1); \
> +      long int _arg2 = (long int) (arg2); \
> +      register long int __a7 asm ("$a7") = number; \
> +      register long int __a0 asm ("$a0") = _arg0; \
> +      register long int __a1 asm ("$a1") = _arg1; \
> +      register long int __a2 asm ("$a2") = _arg2; \
> +      __asm__ volatile ("syscall 0\n\t" \
> +		       : "+r"(__a0) \
> +		       : "r"(__a7), "r"(__a1), "r"(__a2) \
> +		       : __SYSCALL_CLOBBERS); \
> +      _sys_result = __a0; \
> +    } \
> +    _sys_result; \
> +  })
> +
> +#define internal_syscall4(number, arg0, arg1, arg2, arg3) \
> +  ({ \
> +    long int _sys_result; \
> +\
> +    { \
> +      long int _arg0 = (long int) (arg0); \
> +      long int _arg1 = (long int) (arg1); \
> +      long int _arg2 = (long int) (arg2); \
> +      long int _arg3 = (long int) (arg3); \
> +      register long int __a7 asm ("$a7") = number; \
> +      register long int __a0 asm ("$a0") = _arg0; \
> +      register long int __a1 asm ("$a1") = _arg1; \
> +      register long int __a2 asm ("$a2") = _arg2; \
> +      register long int __a3 asm ("$a3") = _arg3; \
> +      __asm__ volatile ("syscall 0\n\t" \
> +		       : "+r"(__a0) \
> +		       : "r"(__a7), "r"(__a1), "r"(__a2), "r"(__a3) \
> +		       : __SYSCALL_CLOBBERS); \
> +      _sys_result = __a0; \
> +    } \
> +    _sys_result; \
> +  })
> +
> +#define internal_syscall5(number, arg0, arg1, arg2, arg3, arg4) \
> +  ({ \
> +    long int _sys_result; \
> +\
> +    { \
> +      long int _arg0 = (long int) (arg0); \
> +      long int _arg1 = (long int) (arg1); \
> +      long int _arg2 = (long int) (arg2); \
> +      long int _arg3 = (long int) (arg3); \
> +      long int _arg4 = (long int) (arg4); \
> +      register long int __a7 asm ("$a7") = number; \
> +      register long int __a0 asm ("$a0") = _arg0; \
> +      register long int __a1 asm ("$a1") = _arg1; \
> +      register long int __a2 asm ("$a2") = _arg2; \
> +      register long int __a3 asm ("$a3") = _arg3; \
> +      register long int __a4 asm ("$a4") = _arg4; \
> +      __asm__ volatile ("syscall 0\n\t" \
> +		       : "+r"(__a0) \
> +		       : "r"(__a7), "r"(__a1), "r"(__a2), \
> +			 "r"(__a3), "r"(__a4) \
> +		       : __SYSCALL_CLOBBERS); \
> +      _sys_result = __a0; \
> +    } \
> +    _sys_result; \
> +  })
> +
> +#define internal_syscall6(number, arg0, arg1, arg2, arg3, arg4, arg5) \
> +  ({ \
> +    long int _sys_result; \
> +\
> +    { \
> +      long int _arg0 = (long int) (arg0); \
> +      long int _arg1 = (long int) (arg1); \
> +      long int _arg2 = (long int) (arg2); \
> +      long int _arg3 = (long int) (arg3); \
> +      long int _arg4 = (long int) (arg4); \
> +      long int _arg5 = (long int) (arg5); \
> +      register long int __a7 asm ("$a7") = number; \
> +      register long int __a0 asm ("$a0") = _arg0; \
> +      register long int __a1 asm ("$a1") = _arg1; \
> +      register long int __a2 asm ("$a2") = _arg2; \
> +      register long int __a3 asm ("$a3") = _arg3; \
> +      register long int __a4 asm ("$a4") = _arg4; \
> +      register long int __a5 asm ("$a5") = _arg5; \
> +      __asm__ volatile ("syscall 0\n\t" \
> +		       : "+r"(__a0) \
> +		       : "r"(__a7), "r"(__a1), "r"(__a2), "r"(__a3), \
> +			 "r"(__a4), "r"(__a5) \
> +		       : __SYSCALL_CLOBBERS); \
> +      _sys_result = __a0; \
> +    } \
> +    _sys_result; \
> +  })
> +
> +#define internal_syscall7(number, arg0, arg1, arg2, arg3, arg4, arg5, arg6) \
> +  ({ \
> +    long int _sys_result; \
> +\
> +    { \
> +      long int _arg0 = (long int) (arg0); \
> +      long int _arg1 = (long int) (arg1); \
> +      long int _arg2 = (long int) (arg2); \
> +      long int _arg3 = (long int) (arg3); \
> +      long int _arg4 = (long int) (arg4); \
> +      long int _arg5 = (long int) (arg5); \
> +      long int _arg6 = (long int) (arg6); \
> +      register long int __a7 asm ("$a7") = number; \
> +      register long int __a0 asm ("$a0") = _arg0; \
> +      register long int __a1 asm ("$a1") = _arg1; \
> +      register long int __a2 asm ("$a2") = _arg2; \
> +      register long int __a3 asm ("$a3") = _arg3; \
> +      register long int __a4 asm ("$a4") = _arg4; \
> +      register long int __a5 asm ("$a5") = _arg5; \
> +      register long int __a6 asm ("$a6") = _arg6; \
> +      __asm__ volatile ("syscall 0\n\t" \
> +		       : "+r"(__a0) \
> +		       : "r"(__a7), "r"(__a1), "r"(__a2), "r"(__a3), \
> +			 "r"(__a4), "r"(__a5), "r"(__a6) \
> +		       : __SYSCALL_CLOBBERS); \
> +      _sys_result = __a0; \
> +    } \
> +    _sys_result; \
> +  })
> +
> +#define __SYSCALL_CLOBBERS \
> +  "$t0", "$t1", "$t2", "$t3", "$t4", "$t5", "$t6", "$t7", "$t8", "memory"
> +
> +extern long int __syscall_error (long int neg_errno);
> +
> +#endif /* ! __ASSEMBLER__ */
> +
> +/* Pointer mangling is not supported.  */
> +#define PTR_MANGLE(var) (void) (var)
> +#define PTR_DEMANGLE(var) (void) (var)
> +
> +#endif /* linux/loongarch/sysdep.h */
> diff --git a/sysdeps/unix/sysv/linux/loongarch/vfork.S b/sysdeps/unix/sysv/linux/loongarch/vfork.S
> new file mode 100644
> index 0000000000..fcc84e3549
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/loongarch/vfork.S
> @@ -0,0 +1,50 @@
> +/* vfork for Linux.
> +   Copyright (C) 2022 Free Software Foundation, Inc.
> +
> +   This file is part of the GNU C Library.
> +
> +   The GNU C Library is free software; you can redistribute it and/or
> +   modify it under the terms of the GNU Lesser General Public License as
> +   published by the Free Software Foundation; either version 2.1 of the
> +   License, or (at your option) any later version.
> +
> +   The GNU C Library is distributed in the hope that it will be useful,
> +   but WITHOUT ANY WARRANTY; without even the implied warranty of
> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> +   Lesser General Public License for more details.
> +
> +   You should have received a copy of the GNU Lesser General Public
> +   License along with the GNU C Library; if not, see
> +   <https://www.gnu.org/licenses/>.  */
> +
> +#include <sysdep.h>
> +#define _ERRNO_H  1
> +#include <bits/errno.h>
> +
> +/* Clone the calling process, but without copying the whole address space.
> +   The calling process is suspended until the new process exits or is
> +   replaced by a call to `execve'.  Return -1 for errors, 0 to the new process,
> +   and the process ID of the new process to the old process.  */
> +
> +ENTRY (__vfork)
> +
> +	li.d	a0, 0x4111 /* CLONE_VM | CLONE_VFORK | SIGCHLD */
> +	add.d   a1, zero, sp
> +
> +	/* Do the system call.  */
> +	li.d	a7, __NR_clone
> +	syscall	0
> +
> +	blt	a0, zero ,L (error)
> +
> +	ret
> +
> +L (error):
> +	b	__syscall_error
> +
> +END (__vfork)
> +
> +libc_hidden_def (__vfork)
> +
> +weak_alias (__vfork, vfork)
> +strong_alias (__vfork, __libc_vfork)
  

Wait. I almost missed this, glad I took an extra look before starting 
afternoon's $DAY_JOB work...

On 2022/7/19 09:20, caiyinyu wrote:
> |+#define VDSO_NAME "LINUX_2.6" +#define VDSO_HASH 61765110|

|According to Linux sources [1], this should be "LINUX_5.10" instead, 
and the hash should be 182947696. The current value is for the old world 
and will lead to the vDSO being essentially wasted.
|

|To the Release Manager: do this need a proper patch, or if this can be 
fixed up in the branch pending inclusion? I can prepare one in short 
order if that's the proper way.
|

|[1]: 
https://github.com/torvalds/linux/blob/v5.19-rc8/arch/loongarch/vdso/vdso.lds.S#L59
|
  

On 2022/7/27 13:27, WANG Xuerui wrote:
> Wait. I almost missed this, glad I took an extra look before starting 
> afternoon's $DAY_JOB work...
>
> On 2022/7/19 09:20, caiyinyu wrote:
>> |+#define VDSO_NAME "LINUX_2.6" +#define VDSO_HASH 61765110|
>
> |According to Linux sources [1], this should be "LINUX_5.10" instead, 
> and the hash should be 182947696. The current value is for the old 
> world and will lead to the vDSO being essentially wasted.
> |
>
> |To the Release Manager: do this need a proper patch, or if this can 
> be fixed up in the branch pending inclusion? I can prepare one in 
> short order if that's the proper way.
> |
>
> |[1]: 
> https://github.com/torvalds/linux/blob/v5.19-rc8/arch/loongarch/vdso/vdso.lds.S#L59
> |
>
>
Ah. I just saw the branch merged in master.

But a quick vdsotest shows the libc version's performance on par with 
correct vDSO calls:

clock-gettime-monotonic: syscall: 183 nsec/call
clock-gettime-monotonic:    libc: 44 nsec/call
clock-gettime-monotonic:    vdso: 34 nsec/call
clock-getres-monotonic: syscall: 152 nsec/call
clock-getres-monotonic:    libc: 19 nsec/call
clock-getres-monotonic:    vdso: 14 nsec/call
clock-gettime-monotonic-coarse: syscall: 172 nsec/call
clock-gettime-monotonic-coarse:    libc: 39 nsec/call
clock-gettime-monotonic-coarse:    vdso: 29 nsec/call
clock-getres-monotonic-coarse: syscall: 154 nsec/call
clock-getres-monotonic-coarse:    libc: 18 nsec/call
clock-getres-monotonic-coarse:    vdso: 14 nsec/call
clock-gettime-monotonic-raw: syscall: 181 nsec/call
clock-gettime-monotonic-raw:    libc: 45 nsec/call
clock-gettime-monotonic-raw:    vdso: 36 nsec/call
clock-getres-monotonic-raw: syscall: 151 nsec/call
clock-getres-monotonic-raw:    libc: 19 nsec/call
clock-getres-monotonic-raw:    vdso: 14 nsec/call
clock-gettime-tai: syscall: 187 nsec/call
clock-gettime-tai:    libc: 44 nsec/call
clock-gettime-tai:    vdso: 34 nsec/call
clock-getres-tai: syscall: 151 nsec/call
clock-getres-tai:    libc: 19 nsec/call
clock-getres-tai:    vdso: 14 nsec/call
clock-gettime-boottime: syscall: 184 nsec/call
clock-gettime-boottime:    libc: 44 nsec/call
clock-gettime-boottime:    vdso: 34 nsec/call
clock-getres-boottime: syscall: 151 nsec/call
clock-getres-boottime:    libc: 19 nsec/call
clock-getres-boottime:    vdso: 14 nsec/call
clock-gettime-realtime: syscall: 181 nsec/call
clock-gettime-realtime:    libc: 44 nsec/call
clock-gettime-realtime:    vdso: 34 nsec/call
clock-getres-realtime: syscall: 151 nsec/call
clock-getres-realtime:    libc: 19 nsec/call
clock-getres-realtime:    vdso: 14 nsec/call
clock-gettime-realtime-coarse: syscall: 179 nsec/call
clock-gettime-realtime-coarse:    libc: 40 nsec/call
clock-gettime-realtime-coarse:    vdso: 29 nsec/call
clock-getres-realtime-coarse: syscall: 154 nsec/call
clock-getres-realtime-coarse:    libc: 18 nsec/call
clock-getres-realtime-coarse:    vdso: 14 nsec/call
getcpu: syscall: 134 nsec/call
getcpu:    libc: 20 nsec/call
getcpu:    vdso: 13 nsec/call
gettimeofday: syscall: 171 nsec/call
gettimeofday:    libc: 47 nsec/call
gettimeofday:    vdso: 33 nsec/call

Which is getting interesting. (This may explain why nobody noticed 
before the merge.) I think we should still "fix" it, but at this point 
I'm starting to doubt whether it's a problem in the first place...
  

On 27/07/22 02:32, WANG Xuerui wrote:
> On 2022/7/27 13:27, WANG Xuerui wrote:
>> Wait. I almost missed this, glad I took an extra look before starting afternoon's $DAY_JOB work...
>>
>> On 2022/7/19 09:20, caiyinyu wrote:
>>> |+#define VDSO_NAME "LINUX_2.6" +#define VDSO_HASH 61765110|
>>
>> |According to Linux sources [1], this should be "LINUX_5.10" instead, and the hash should be 182947696. The current value is for the old world and will lead to the vDSO being essentially wasted.
>> |
>>
>> |To the Release Manager: do this need a proper patch, or if this can be fixed up in the branch pending inclusion? I can prepare one in short order if that's the proper way.
>> |
>>
>> |[1]: https://github.com/torvalds/linux/blob/v5.19-rc8/arch/loongarch/vdso/vdso.lds.S#L59
>> |
>>
>>
> Ah. I just saw the branch merged in master.
> 
> But a quick vdsotest shows the libc version's performance on par with correct vDSO calls:
> 
> clock-gettime-monotonic: syscall: 183 nsec/call
> clock-gettime-monotonic:    libc: 44 nsec/call
> clock-gettime-monotonic:    vdso: 34 nsec/call
> clock-getres-monotonic: syscall: 152 nsec/call
> clock-getres-monotonic:    libc: 19 nsec/call
> clock-getres-monotonic:    vdso: 14 nsec/call
> clock-gettime-monotonic-coarse: syscall: 172 nsec/call
> clock-gettime-monotonic-coarse:    libc: 39 nsec/call
> clock-gettime-monotonic-coarse:    vdso: 29 nsec/call
> clock-getres-monotonic-coarse: syscall: 154 nsec/call
> clock-getres-monotonic-coarse:    libc: 18 nsec/call
> clock-getres-monotonic-coarse:    vdso: 14 nsec/call
> clock-gettime-monotonic-raw: syscall: 181 nsec/call
> clock-gettime-monotonic-raw:    libc: 45 nsec/call
> clock-gettime-monotonic-raw:    vdso: 36 nsec/call
> clock-getres-monotonic-raw: syscall: 151 nsec/call
> clock-getres-monotonic-raw:    libc: 19 nsec/call
> clock-getres-monotonic-raw:    vdso: 14 nsec/call
> clock-gettime-tai: syscall: 187 nsec/call
> clock-gettime-tai:    libc: 44 nsec/call
> clock-gettime-tai:    vdso: 34 nsec/call
> clock-getres-tai: syscall: 151 nsec/call
> clock-getres-tai:    libc: 19 nsec/call
> clock-getres-tai:    vdso: 14 nsec/call
> clock-gettime-boottime: syscall: 184 nsec/call
> clock-gettime-boottime:    libc: 44 nsec/call
> clock-gettime-boottime:    vdso: 34 nsec/call
> clock-getres-boottime: syscall: 151 nsec/call
> clock-getres-boottime:    libc: 19 nsec/call
> clock-getres-boottime:    vdso: 14 nsec/call
> clock-gettime-realtime: syscall: 181 nsec/call
> clock-gettime-realtime:    libc: 44 nsec/call
> clock-gettime-realtime:    vdso: 34 nsec/call
> clock-getres-realtime: syscall: 151 nsec/call
> clock-getres-realtime:    libc: 19 nsec/call
> clock-getres-realtime:    vdso: 14 nsec/call
> clock-gettime-realtime-coarse: syscall: 179 nsec/call
> clock-gettime-realtime-coarse:    libc: 40 nsec/call
> clock-gettime-realtime-coarse:    vdso: 29 nsec/call
> clock-getres-realtime-coarse: syscall: 154 nsec/call
> clock-getres-realtime-coarse:    libc: 18 nsec/call
> clock-getres-realtime-coarse:    vdso: 14 nsec/call
> getcpu: syscall: 134 nsec/call
> getcpu:    libc: 20 nsec/call
> getcpu:    vdso: 13 nsec/call
> gettimeofday: syscall: 171 nsec/call
> gettimeofday:    libc: 47 nsec/call
> gettimeofday:    vdso: 33 nsec/call
> 
> Which is getting interesting. (This may explain why nobody noticed before the merge.) I think we should still "fix" it, but at this point I'm starting to doubt whether it's a problem in the first place...
> 

So does it this need to be fixed in the end? The numbers indicates that vDSO is indeed
being used.
  

On 2022/7/27 19:16, Adhemerval Zanella Netto wrote:
>
> On 27/07/22 02:32, WANG Xuerui wrote:
>> On 2022/7/27 13:27, WANG Xuerui wrote:
>>> Wait. I almost missed this, glad I took an extra look before starting afternoon's $DAY_JOB work...
>>>
>>> On 2022/7/19 09:20, caiyinyu wrote:
>>>> |+#define VDSO_NAME "LINUX_2.6" +#define VDSO_HASH 61765110|
>>> |According to Linux sources [1], this should be "LINUX_5.10" instead, and the hash should be 182947696. The current value is for the old world and will lead to the vDSO being essentially wasted.
>>> |
>>>
>>> |To the Release Manager: do this need a proper patch, or if this can be fixed up in the branch pending inclusion? I can prepare one in short order if that's the proper way.
>>> |
>>>
>>> |[1]: https://github.com/torvalds/linux/blob/v5.19-rc8/arch/loongarch/vdso/vdso.lds.S#L59
>>> |
>>>
>>>
>> Ah. I just saw the branch merged in master.
>>
>> But a quick vdsotest shows the libc version's performance on par with correct vDSO calls:
>>
>> clock-gettime-monotonic: syscall: 183 nsec/call
>> clock-gettime-monotonic:    libc: 44 nsec/call
>> clock-gettime-monotonic:    vdso: 34 nsec/call
>> clock-getres-monotonic: syscall: 152 nsec/call
>> clock-getres-monotonic:    libc: 19 nsec/call
>> clock-getres-monotonic:    vdso: 14 nsec/call
>> clock-gettime-monotonic-coarse: syscall: 172 nsec/call
>> clock-gettime-monotonic-coarse:    libc: 39 nsec/call
>> clock-gettime-monotonic-coarse:    vdso: 29 nsec/call
>> clock-getres-monotonic-coarse: syscall: 154 nsec/call
>> clock-getres-monotonic-coarse:    libc: 18 nsec/call
>> clock-getres-monotonic-coarse:    vdso: 14 nsec/call
>> clock-gettime-monotonic-raw: syscall: 181 nsec/call
>> clock-gettime-monotonic-raw:    libc: 45 nsec/call
>> clock-gettime-monotonic-raw:    vdso: 36 nsec/call
>> clock-getres-monotonic-raw: syscall: 151 nsec/call
>> clock-getres-monotonic-raw:    libc: 19 nsec/call
>> clock-getres-monotonic-raw:    vdso: 14 nsec/call
>> clock-gettime-tai: syscall: 187 nsec/call
>> clock-gettime-tai:    libc: 44 nsec/call
>> clock-gettime-tai:    vdso: 34 nsec/call
>> clock-getres-tai: syscall: 151 nsec/call
>> clock-getres-tai:    libc: 19 nsec/call
>> clock-getres-tai:    vdso: 14 nsec/call
>> clock-gettime-boottime: syscall: 184 nsec/call
>> clock-gettime-boottime:    libc: 44 nsec/call
>> clock-gettime-boottime:    vdso: 34 nsec/call
>> clock-getres-boottime: syscall: 151 nsec/call
>> clock-getres-boottime:    libc: 19 nsec/call
>> clock-getres-boottime:    vdso: 14 nsec/call
>> clock-gettime-realtime: syscall: 181 nsec/call
>> clock-gettime-realtime:    libc: 44 nsec/call
>> clock-gettime-realtime:    vdso: 34 nsec/call
>> clock-getres-realtime: syscall: 151 nsec/call
>> clock-getres-realtime:    libc: 19 nsec/call
>> clock-getres-realtime:    vdso: 14 nsec/call
>> clock-gettime-realtime-coarse: syscall: 179 nsec/call
>> clock-gettime-realtime-coarse:    libc: 40 nsec/call
>> clock-gettime-realtime-coarse:    vdso: 29 nsec/call
>> clock-getres-realtime-coarse: syscall: 154 nsec/call
>> clock-getres-realtime-coarse:    libc: 18 nsec/call
>> clock-getres-realtime-coarse:    vdso: 14 nsec/call
>> getcpu: syscall: 134 nsec/call
>> getcpu:    libc: 20 nsec/call
>> getcpu:    vdso: 13 nsec/call
>> gettimeofday: syscall: 171 nsec/call
>> gettimeofday:    libc: 47 nsec/call
>> gettimeofday:    vdso: 33 nsec/call
>>
>> Which is getting interesting. (This may explain why nobody noticed before the merge.) I think we should still "fix" it, but at this point I'm starting to doubt whether it's a problem in the first place...
>>
> So does it this need to be fixed in the end? The numbers indicates that vDSO is indeed
> being used.

If striving for purity, the bug should definitely get fixed, but again I 
saw in sysdeps/unix/sysv/linux/dl-vdso.h that the dl_vdso_vsym makes use 
of dl_lookup_symbol_x, which states in its docstring:

 > Search loaded objects' symbol tables for a definition of the symbol 
UNDEF_NAME, perhaps with a requested version for the symbol.

I dug a little deeper but can't really verify the "perhaps" part, but 
judging from the doc alone the current behavior may well be expected. 
Maybe you or someone else more familiar with the details could enlighten 
me on this, but indeed vDSO must actually be effective for the 
order-of-magnitude improvement in numbers, hence the inconsistency 
should not be adversely affecting normal operation besides being a minor 
nuisance. We can fix it after 2.36 gets released then.
  

On 27/07/22 10:01, WANG Xuerui wrote:
> 
> On 2022/7/27 19:16, Adhemerval Zanella Netto wrote:
>>
>> On 27/07/22 02:32, WANG Xuerui wrote:
>>> On 2022/7/27 13:27, WANG Xuerui wrote:
>>>> Wait. I almost missed this, glad I took an extra look before starting afternoon's $DAY_JOB work...
>>>>
>>>> On 2022/7/19 09:20, caiyinyu wrote:
>>>>> |+#define VDSO_NAME "LINUX_2.6" +#define VDSO_HASH 61765110|
>>>> |According to Linux sources [1], this should be "LINUX_5.10" instead, and the hash should be 182947696. The current value is for the old world and will lead to the vDSO being essentially wasted.
>>>> |
>>>>
>>>> |To the Release Manager: do this need a proper patch, or if this can be fixed up in the branch pending inclusion? I can prepare one in short order if that's the proper way.
>>>> |
>>>>
>>>> |[1]: https://github.com/torvalds/linux/blob/v5.19-rc8/arch/loongarch/vdso/vdso.lds.S#L59
>>>> |
>>>>
>>>>
>>> Ah. I just saw the branch merged in master.
>>>
>>> But a quick vdsotest shows the libc version's performance on par with correct vDSO calls:
>>>
>>> clock-gettime-monotonic: syscall: 183 nsec/call
>>> clock-gettime-monotonic:    libc: 44 nsec/call
>>> clock-gettime-monotonic:    vdso: 34 nsec/call
>>> clock-getres-monotonic: syscall: 152 nsec/call
>>> clock-getres-monotonic:    libc: 19 nsec/call
>>> clock-getres-monotonic:    vdso: 14 nsec/call
>>> clock-gettime-monotonic-coarse: syscall: 172 nsec/call
>>> clock-gettime-monotonic-coarse:    libc: 39 nsec/call
>>> clock-gettime-monotonic-coarse:    vdso: 29 nsec/call
>>> clock-getres-monotonic-coarse: syscall: 154 nsec/call
>>> clock-getres-monotonic-coarse:    libc: 18 nsec/call
>>> clock-getres-monotonic-coarse:    vdso: 14 nsec/call
>>> clock-gettime-monotonic-raw: syscall: 181 nsec/call
>>> clock-gettime-monotonic-raw:    libc: 45 nsec/call
>>> clock-gettime-monotonic-raw:    vdso: 36 nsec/call
>>> clock-getres-monotonic-raw: syscall: 151 nsec/call
>>> clock-getres-monotonic-raw:    libc: 19 nsec/call
>>> clock-getres-monotonic-raw:    vdso: 14 nsec/call
>>> clock-gettime-tai: syscall: 187 nsec/call
>>> clock-gettime-tai:    libc: 44 nsec/call
>>> clock-gettime-tai:    vdso: 34 nsec/call
>>> clock-getres-tai: syscall: 151 nsec/call
>>> clock-getres-tai:    libc: 19 nsec/call
>>> clock-getres-tai:    vdso: 14 nsec/call
>>> clock-gettime-boottime: syscall: 184 nsec/call
>>> clock-gettime-boottime:    libc: 44 nsec/call
>>> clock-gettime-boottime:    vdso: 34 nsec/call
>>> clock-getres-boottime: syscall: 151 nsec/call
>>> clock-getres-boottime:    libc: 19 nsec/call
>>> clock-getres-boottime:    vdso: 14 nsec/call
>>> clock-gettime-realtime: syscall: 181 nsec/call
>>> clock-gettime-realtime:    libc: 44 nsec/call
>>> clock-gettime-realtime:    vdso: 34 nsec/call
>>> clock-getres-realtime: syscall: 151 nsec/call
>>> clock-getres-realtime:    libc: 19 nsec/call
>>> clock-getres-realtime:    vdso: 14 nsec/call
>>> clock-gettime-realtime-coarse: syscall: 179 nsec/call
>>> clock-gettime-realtime-coarse:    libc: 40 nsec/call
>>> clock-gettime-realtime-coarse:    vdso: 29 nsec/call
>>> clock-getres-realtime-coarse: syscall: 154 nsec/call
>>> clock-getres-realtime-coarse:    libc: 18 nsec/call
>>> clock-getres-realtime-coarse:    vdso: 14 nsec/call
>>> getcpu: syscall: 134 nsec/call
>>> getcpu:    libc: 20 nsec/call
>>> getcpu:    vdso: 13 nsec/call
>>> gettimeofday: syscall: 171 nsec/call
>>> gettimeofday:    libc: 47 nsec/call
>>> gettimeofday:    vdso: 33 nsec/call
>>>
>>> Which is getting interesting. (This may explain why nobody noticed before the merge.) I think we should still "fix" it, but at this point I'm starting to doubt whether it's a problem in the first place...
>>>
>> So does it this need to be fixed in the end? The numbers indicates that vDSO is indeed
>> being used.
> 
> If striving for purity, the bug should definitely get fixed, but again I saw in sysdeps/unix/sysv/linux/dl-vdso.h that the dl_vdso_vsym makes use of dl_lookup_symbol_x, which states in its docstring:
> 
>> Search loaded objects' symbol tables for a definition of the symbol UNDEF_NAME, perhaps with a requested version for the symbol.
> 
> I dug a little deeper but can't really verify the "perhaps" part, but judging from the doc alone the current behavior may well be expected. Maybe you or someone else more familiar with the details could enlighten me on this, but indeed vDSO must actually be effective for the order-of-magnitude improvement in numbers, hence the inconsistency should not be adversely affecting normal operation besides being a minor nuisance. We can fix it after 2.36 gets released then.
> 

Alright, if you do think this is required for 2.36 I think it should be ok
since is a localized change for loongarch.