man2/syscall.2

   1 .\" Copyright (c) 1980, 1991, 1993
   2 .\"     The Regents of the University of California.  All rights reserved.
   3 .\"
   4 .\" %%%LICENSE_START(BSD_4_CLAUSE_UCB)
   5 .\" Redistribution and use in source and binary forms, with or without
   6 .\" modification, are permitted provided that the following conditions
   7 .\" are met:
   8 .\" 1. Redistributions of source code must retain the above copyright
   9 .\"    notice, this list of conditions and the following disclaimer.
  10 .\" 2. Redistributions in binary form must reproduce the above copyright
  11 .\"    notice, this list of conditions and the following disclaimer in the
  12 .\"    documentation and/or other materials provided with the distribution.
  13 .\" 3. All advertising materials mentioning features or use of this software
  14 .\"    must display the following acknowledgement:
  15 .\"     This product includes software developed by the University of
  16 .\"     California, Berkeley and its contributors.
  17 .\" 4. Neither the name of the University nor the names of its contributors
  18 .\"    may be used to endorse or promote products derived from this software
  19 .\"    without specific prior written permission.
  20 .\"
  21 .\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  22 .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  23 .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  24 .\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  25 .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  26 .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  27 .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  28 .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  29 .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  30 .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  31 .\" SUCH DAMAGE.
  32 .\" %%%LICENSE_END
  33 .\"
  34 .\"     @(#)syscall.2   8.1 (Berkeley) 6/16/93
  35 .\"
  36 .\"
  37 .\" 2002-03-20  Christoph Hellwig <hch@infradead.org>
  38 .\"     - adopted for Linux
  39 .\" 2015-01-17, Kees Cook <keescook@chromium.org>
  40 .\"     Added mips and arm64.
  41 .\"
  42 .TH SYSCALL 2 2020-02-09 "Linux" "Linux Programmer's Manual"
  43 .SH NAME
  44 syscall \- indirect system call
  45 .SH SYNOPSIS
  46 .nf
  47 .B #include <unistd.h>
  48 .BR "#include <sys/syscall.h>   "  "/* For SYS_xxx definitions */"
  49 .PP
  50 .BI "long syscall(long " number ", ...);"
  51 .fi
  52 .PP
  53 .in -4n
  54 Feature Test Macro Requirements for glibc (see
  55 .BR feature_test_macros (7)):
  56 .in
  57 .BR syscall ():
  58 .PD 0
  59 .ad l
  60 .RS 4
  61 .TP 4
  62 Since glibc 2.19:
  63 _DEFAULT_SOURCE
  64 .TP
  65 Before glibc 2.19:
  66 _BSD_SOURCE || _SVID_SOURCE
  67 .RE
  68 .ad
  69 .PD
  70 .SH DESCRIPTION
  71 .BR syscall ()
  72 is a small library function that invokes
  73 the system call whose assembly language
  74 interface has the specified
  75 .I number
  76 with the specified arguments.
  77 Employing
  78 .BR syscall ()
  79 is useful, for example,
  80 when invoking a system call that has no wrapper function in the C library.
  81 .PP
  82 .BR syscall ()
  83 saves CPU registers before making the system call,
  84 restores the registers upon return from the system call,
  85 and stores any error returned by the system call in
  86 .BR errno (3).
  87 .PP
  88 Symbolic constants for system call numbers can be found in the header file
  89 .IR <sys/syscall.h> .
  90 .SH RETURN VALUE
  91 The return value is defined by the system call being invoked.
  92 In general, a 0 return value indicates success.
  93 A \-1 return value indicates an error,
  94 and an error number is stored in
  95 .IR errno .
  96 .SH NOTES
  97 .BR syscall ()
  98 first appeared in
  99 4BSD.
 100 .SS Architecture-specific requirements
 101 Each architecture ABI has its own requirements on how
 102 system call arguments are passed to the kernel.
 103 For system calls that have a glibc wrapper (e.g., most system calls),
 104 glibc handles the details of copying arguments to the right registers
 105 in a manner suitable for the architecture.
 106 However, when using
 107 .BR syscall ()
 108 to make a system call,
 109 the caller might need to handle architecture-dependent details;
 110 this requirement is most commonly encountered on certain 32-bit architectures.
 111 .PP
 112 For example, on the ARM architecture Embedded ABI (EABI), a
 113 64-bit value (e.g.,
 114 .IR "long long" )
 115 must be aligned to an even register pair.
 116 Thus, using
 117 .BR syscall ()
 118 instead of the wrapper provided by glibc,
 119 the
 120 .BR readahead (2)
 121 system call would be invoked as follows on the ARM architecture with the EABI
 122 in little endian mode:
 123 .PP
 124 .in +4n
 125 .EX
 126 syscall(SYS_readahead, fd, 0,
 127         (unsigned int) (offset & 0xFFFFFFFF),
 128         (unsigned int) (offset >> 32),
 129         count);
 130 .EE
 131 .in
 132 .PP
 133 Since the offset argument is 64 bits, and the first argument
 134 .RI ( fd )
 135 is passed in
 136 .IR r0 ,
 137 the caller must manually split and align the 64-bit value
 138 so that it is passed in the
 139 .IR r2 / r3
 140 register pair.
 141 That means inserting a dummy value into
 142 .I r1
 143 (the second argument of 0).
 144 Care also must be taken so that the split follows endian conventions
 145 (according to the C ABI for the platform).
 146 .PP
 147 Similar issues can occur on MIPS with the O32 ABI,
 148 on PowerPC and parisc with the 32-bit ABI, and on Xtensa.
 149 .\" Mike Frysinger: this issue ends up forcing MIPS
 150 .\" O32 to take 7 arguments to syscall()
 151 .PP
 152 .\" See arch/parisc/kernel/sys_parisc.c.
 153 Note that while the parisc C ABI also uses aligned register pairs,
 154 it uses a shim layer to hide the issue from user space.
 155 .PP
 156 The affected system calls are
 157 .BR fadvise64_64 (2),
 158 .BR ftruncate64 (2),
 159 .BR posix_fadvise (2),
 160 .BR pread64 (2),
 161 .BR pwrite64 (2),
 162 .BR readahead (2),
 163 .BR sync_file_range (2),
 164 and
 165 .BR truncate64 (2).
 166 .PP
 167 .\" You need to look up the syscalls directly in the kernel source to see if
 168 .\" they should be in this list.  For example, look at fs/read_write.c and
 169 .\" the function signatures that do:
 170 .\" ..., unsigned long, pos_l, unsigned long, pos_h, ...
 171 .\" If they use off_t, then they most likely do not belong in this list.
 172 This does not affect syscalls that manually split and assemble 64-bit values
 173 such as
 174 .BR _llseek (2),
 175 .BR preadv (2),
 176 .BR preadv2 (2),
 177 .BR pwritev (2),
 178 and
 179 .BR pwritev2 (2).
 180 Welcome to the wonderful world of historical baggage.
 181 .SS Architecture calling conventions
 182 Every architecture has its own way of invoking and passing arguments to the
 183 kernel.
 184 The details for various architectures are listed in the two tables below.
 185 .PP
 186 The first table lists the instruction used to transition to kernel mode
 187 (which might not be the fastest or best way to transition to the kernel,
 188 so you might have to refer to
 189 .BR vdso (7)),
 190 the register used to indicate the system call number,
 191 the register(s) used to return the system call result,
 192 and the register used to signal an error.
 193 .if t \{\
 194 .ft CW
 195 \}
 196 .TS
 197 l2      l2      l2      l2      l1      l2      l.
 198 Arch/ABI        Instruction     System  Ret     Ret     Error   Notes
 199                 call #  val     val2
 200 _
 201 alpha   callsys v0      v0      a4      a3      1, 6
 202 arc     trap0   r8      r0      -       -
 203 arm/OABI        swi NR  -       r0      -       -       2
 204 arm/EABI        swi 0x0 r7      r0      r1      -
 205 arm64   svc #0  w8      x0      x1      -
 206 blackfin        excpt 0x0       P0      R0      -       -
 207 i386    int $0x80       eax     eax     edx     -
 208 ia64    break 0x100000  r15     r8      r9      r10     1, 6
 209 m68k    trap #0 d0      d0      -       -
 210 microblaze      brki r14,8      r12     r3      -       -
 211 mips    syscall v0      v0      v1      a3      1, 6
 212 nios2   trap    r2      r2      -       r7
 213 parisc  ble 0x100(%sr2, %r0)    r20     r28     -       -
 214 powerpc sc      r0      r3      -       r0      1
 215 powerpc64       sc      r0      r3      -       cr0.SO  1
 216 riscv   ecall   a7      a0      a1      -
 217 s390    svc 0   r1      r2      r3      -       3
 218 s390x   svc 0   r1      r2      r3      -       3
 219 superh  trap #0x17      r3      r0      r1      -       4, 6
 220 sparc/32        t 0x10  g1      o0      o1      psr/csr 1, 6
 221 sparc/64        t 0x6d  g1      o0      o1      psr/csr 1, 6
 222 tile    swint1  R10     R00     -       R01     1
 223 x86-64  syscall rax     rax     rdx     -       5
 224 x32     syscall rax     rax     rdx     -       5
 225 xtensa  syscall a2      a2      -       -
 226 .TE
 227 .PP
 228 Notes:
 229 .IP [1] 4
 230 On a few architectures,
 231 a register is used as a boolean
 232 (0 indicating no error, and \-1 indicating an error) to signal that the
 233 system call failed.
 234 The actual error value is still contained in the return register.
 235 On sparc, the carry bit
 236 .RI ( csr )
 237 in the processor status register
 238 .RI ( psr )
 239 is used instead of a full register.
 240 On powerpc64, the summary overflow bit
 241 .RI ( SO )
 242 in field 0 of the condition register
 243 .RI ( cr0 )
 244 is used.
 245 .IP [2]
 246 .I NR
 247 is the system call number.
 248 .IP [3]
 249 For s390 and s390x,
 250 .I NR
 251 (the system call number) may be passed directly with
 252 .I "svc\ NR"
 253 if it is less than 256.
 254 .IP [4]
 255 On SuperH, the trap number controls the maximum number of arguments passed.
 256 A
 257 .IR "trap\ #0x10"
 258 can be used with only 0-argument system calls, a
 259 .IR "trap\ #0x11"
 260 can be used with 0- or 1-argument system calls,
 261 and so on up to
 262 .IR "trap #0x17"
 263 for 7-argument system calls.
 264 .IP [5]
 265 The x32 ABI shares syscall table with x86-64 ABI, but there are some
 266 nuances:
 267 .RS
 268 .IP \(bu 3
 269 In order to indicate that a system call is called under the x32 ABI,
 270 an additional bit,
 271 .BR __X32_SYSCALL_BIT ,
 272 is bitwise-ORed with the system call number.
 273 The ABI used by a process affects some process behaviors,
 274 including signal handling or system call restarting.
 275 .IP \(bu
 276 Since x32 has different sizes for
 277 .I long
 278 and pointer types, layouts of some (but not all;
 279 .I struct timeval
 280 or
 281 .I struct rlimit
 282 are 64-bit, for example) structures are different.
 283 In order to handle this,
 284 additional system calls are added to the system call table,
 285 starting from number 512
 286 (without the
 287 .BR __X32_SYSCALL_BIT ).
 288 For example,
 289 .B __NR_readv
 290 is defined as 19 for the x86-64 ABI and as
 291 .IR __X32_SYSCALL_BIT " | " \fB515\fP
 292 for the x32 ABI.
 293 Most of these additional system calls are actually identical
 294 to the system calls used for providing i386 compat.
 295 There are some notable exceptions, however, such as
 296 .BR preadv2 (2),
 297 which uses
 298 .I struct iovec
 299 entities with 4-byte pointers and sizes ("compat_iovec" in kernel terms),
 300 but passes an 8-byte
 301 .I pos
 302 argument in a single register and not two, as is done in every other ABI.
 303 .RE
 304 .IP [6]
 305 Some architectures
 306 (namely, Alpha, IA-64, MIPS, SuperH, sparc/32, and sparc/64)
 307 use an additional register ("Retval2" in the above table)
 308 to pass back a second return value from the
 309 .BR pipe (2)
 310 system call;
 311 Alpha uses this technique in the architecture-specific
 312 .BR getxpid (2),
 313 .BR getxuid (2),
 314 and
 315 .BR getxgid (2)
 316 system calls as well.
 317 Other architectures do not use the second return value register
 318 in the system call interface, even if it is defined in the System V ABI.
 319 .if t \{\
 320 .in
 321 .ft P
 322 \}
 323 .PP
 324 The second table shows the registers used to pass the system call arguments.
 325 .if t \{\
 326 .ft CW
 327 \}
 328 .TS
 329 l       l2      l2      l2      l2      l2      l2      l2      l.
 330 Arch/ABI        arg1    arg2    arg3    arg4    arg5    arg6    arg7    Notes
 331 _
 332 alpha   a0      a1      a2      a3      a4      a5      -
 333 arc     r0      r1      r2      r3      r4      r5      -
 334 arm/OABI        r0      r1      r2      r3      r4      r5      r6
 335 arm/EABI        r0      r1      r2      r3      r4      r5      r6
 336 arm64   x0      x1      x2      x3      x4      x5      -
 337 blackfin        R0      R1      R2      R3      R4      R5      -
 338 i386    ebx     ecx     edx     esi     edi     ebp     -
 339 ia64    out0    out1    out2    out3    out4    out5    -
 340 m68k    d1      d2      d3      d4      d5      a0      -
 341 microblaze      r5      r6      r7      r8      r9      r10     -
 342 mips/o32        a0      a1      a2      a3      -       -       -       1
 343 mips/n32,64     a0      a1      a2      a3      a4      a5      -
 344 nios2   r4      r5      r6      r7      r8      r9      -
 345 parisc  r26     r25     r24     r23     r22     r21     -
 346 powerpc r3      r4      r5      r6      r7      r8      r9
 347 powerpc64       r3      r4      r5      r6      r7      r8      -
 348 riscv   a0      a1      a2      a3      a4      a5      -
 349 s390    r2      r3      r4      r5      r6      r7      -
 350 s390x   r2      r3      r4      r5      r6      r7      -
 351 superh  r4      r5      r6      r7      r0      r1      r2
 352 sparc/32        o0      o1      o2      o3      o4      o5      -
 353 sparc/64        o0      o1      o2      o3      o4      o5      -
 354 tile    R00     R01     R02     R03     R04     R05     -
 355 x86-64  rdi     rsi     rdx     r10     r8      r9      -
 356 x32     rdi     rsi     rdx     r10     r8      r9      -
 357 xtensa  a6      a3      a4      a5      a8      a9      -
 358 .TE
 359 .PP
 360 Notes:
 361 .IP [1] 4
 362 The mips/o32 system call convention passes
 363 arguments 5 through 8 on the user stack.
 364 .if t \{\
 365 .in
 366 .ft P
 367 \}
 368 .PP
 369 Note that these tables don't cover the entire calling convention\(emsome
 370 architectures may indiscriminately clobber other registers not listed here.
 371 .SH EXAMPLES
 372 .EX
 373 #define _GNU_SOURCE
 374 #include <unistd.h>
 375 #include <sys/syscall.h>
 376 #include <sys/types.h>
 377 #include <signal.h>
 378
 379 int
 380 main(int argc, char *argv[])
 381 {
 382     pid_t tid;
 383
 384     tid = syscall(SYS_gettid);
 385     syscall(SYS_tgkill, getpid(), tid, SIGHUP);
 386 }
 387 .EE
 388 .SH SEE ALSO
 389 .BR _syscall (2),
 390 .BR intro (2),
 391 .BR syscalls (2),
 392 .BR errno (3),
 393 .BR vdso (7)