saneex.c: try / catch / finally based on setjmp / longjmp (C99) faster than standard C ++ exceptions ¹

While writing this purely technical article, Habr managed to turn into the local office of WHO and now I am even ashamed to publish it ... but the hope is warming in the soul that IT people have not yet fled and she will find her reader. Or not?




I was always admired by the standard C library, and C itself - with all its minimalism, it still blows the spirit of those very first red-eyed hackers . The draft of the first official standard (ANSI C, aka C89, aka ANS X3.159-1989, later, C90 and IEC 9899: 1990) defines 145 functions and macros, of which about 25 are variations (due to the lack of in the language of overloads), and 26 are purely mathematical. K&R in the second edition ² cites 114 functions (plus mathematical), considering the rest as exotic. In the draft³ C11 there are already 348 functions, but more than a hundred are mathematics, and 90 more are “overloads”. Now let's look at Boost, where there are only 160 libraries . Chur me ...


And among this hundred and a half functions there were always: signal processing, variational functions (which reached interpreted PHP 25 years later, and in Delphi, which was booming at one time, they are still missing) and about 50 string functions like printf () ( mmm ... JavaScript), strftime () (...) and scanf () (a cheap alternative to regulars).



And there were always setjmp () / longjmp () , which allow you to implement the exception mechanism familiar to other languages, without going beyond the scope of portable C. Here's talk about them - the Quake World , stacks, registers, assemblers and other materiel , and cherry is a curious statistic ( spoiler : Visual Studio is not constant, as a March hare, and throw saneex.c in half the time of all).



Hidden text

¹ .


² , . 270 , 80 — . , . K&R — , .


³ , ANSI ISO , . .


⁴ , «» TinyURL, URL , і . , -. urlex.org.


Table of contents:



,  — setjmp()/longjmp(), setjmp.h, «SJLJ» ( , ). C89 , -, , (  — ,  — ).


, , , zzeng. , , , ¹, , , ( , ), .


¹ CException  — 60 , , , ANSI C, finally , .


,  — , , , , , « ». (, , - «» , errno .  — , .)


, :


  • , - , ( ,  — , if (error) return -1;)
  • -  — , ,

. , .



setjmp()/longjmp()



, --


, longjmp() — goto, setjmp() — goto run-time. , «goto ». , , , goto,  — , goto . , - , (  — «try»).


, , , , setjmp.h ^W ? , setjmp() , (, , ). , , , . .


, , setjmp() — fork() POSIX . , *nix’ API WinAPI,  — , .  — « , ?»… .



, ,  — , () .  — ( Intel’ — ),  — ( Intel’ — -, ). :


void sub(int s) {
  char buf[256];
  sub(2);
}

int main(int m) {
  sub(1);
}

 — tcc (Tiny C Compiler) - . . tcc . sub() ( Intel, ):


sub     esp, 100h       ;     
mov     eax, 2          ;  
push    eax
call    sub_401000      ;  sub()
add     esp, 4          ;     (= cdecl)

:



 — ( Intel… , ). ESP (RSP x86_64). setjmp() ESP/RSP, , jmp_buf, . longjmp() ( ) — , , setjmp(), (). , «undo» (, ).


setjmp() jmp FEF8h (FDF0h .. — ) , :


void sub(int s) {
  char buf[256];
  jmp_buf jmp;
  setjmp(jmp);
  sub(2);
}

, , ™:




  •  — , , ! — , , , setjmp() , longjmp() , setjmp()


, -, clobbering


. :


#include <stdlib.h>
#include <stdio.h>
#include <setjmp.h>

int main(void) {
  int i;
  jmp_buf jmp;
  i = rand();
  if (setjmp(jmp) == 0) {
    i = rand();
    printf("%d\n", i);
    longjmp(jmp, 1);
  } else {
    printf("%d\n", i);
  }
}

: ?


: . -!


, gcc. -O0, , :


; int main(void) {
  push    ebp             ;  ( stack frame)
  mov     ebp, esp        ; EBP     ESP (  )
  sub     esp, E0h
  ...
  call    _rand           ;    EAX
  mov     [ebp-D4h], eax  ;  i = rand();  i   (EBP-D4h)
  ...
; if (... == 0) {         ;  setjmp()      
  call    _rand
  mov     [ebp-D4h], eax  ;  i = rand();  
; printf("%d\n", i);
  mov     eax, [ebp-D4h]  ;  i    
  mov     esi, eax
  lea     edi, format     ;   "%d\n"
  mov     eax, 0
  call    _printf
  ...
; } else {                ;    setjmp()  
  mov     eax, [ebp-D4h]  ;   i,    
  mov     esi, eax
  lea     edi, format     ; "%d\n"
  mov     eax, 0
  call    _printf

, i ( EBP - D4h). , :


  • 256 char int jmp_buf, 4 200 , 20 - , 224 (E0h) 100h,
  • ESP setjmp() FFF8h - E0h = FF18h ( FEF8h), jmp
    • , ,
  • i, i ( FF18h)
  • longjmp() FF18h, , i , - , (jmp), main() ( )
    • ESP , longjmp() , main()

-O1, :


;   stack frame  ,   ESP 
  sub     esp, E8h
  ...
  call    _rand
  mov     [esp+E8h-DCh], eax  ; i = rand();  ,    -O0
  ...
; -O1 - ,   else  , 
; if (setjmp() == 0) ( - ),  
;  ;       
; if (... == 0) {
  call    _rand
  mov     esi, eax            ; !  i  
; printf("%d\n", i);
  lea     edi, format     ; "%d\n"
  mov     eax, 0
  call    _printf
  ...
; } else {
  mov     esi, [esp+E8h-DCh]  ; !  i  
  lea     edi, format     ; "%d\n"
  mov     eax, 0
  call    _printf

, -O1 gcc :


test.c:6:11: warning: variable ‘i’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]

? i , ( if) gcc, i printf(), ESI, ( ESI printf(), . ABI, . 22 — RDI (format), RSI (i), …). - :


  • ESP + E8h - DCh rand()
  • ESI
  • printf() ( ) ()
  • longjmp() , ,
  • printf() ( else) , , ,
    • ESI, (, printf() longjmp())


, :


stack[i] = rand();          // i = rand();   (1)
if (setjmp(jmp) == 0) {
  ESI = rand();             // i = rand();   (2)
  printf("%d\n", ESI);      //   (2)
  longjmp(jmp, 1);          // 
} else {
  printf("%d\n", stack[i]); //   (1)
  //     :
  printf("%d\n", ESI);      //  ,   -
                            // "" ( printf()  longjmp())
}

, , gcc rand() ESI ( -O3). SO , x86_64 ( ) , EAX. ? , gcc printf() else longjmp(), rand() printf() — , i .


 — .



volatile


« » — volatile ( — «»). , , , :


volatile int i;

-O1 if:


; :
  call    _rand
  mov     esi, eax
; :
  call    _rand
  mov     [rsp+E8h-DCh], eax
  mov     esi, [rsp+E8h-DCh]
;    :
  call    _rand
  mov     esi, eax
  mov     [rsp+E8h-DCh], eax

, ():


if (setjmp(jmp) == 0) {
  ESI = stack[i] = rand();


IRL


,  — -volatile , SJLJ .  — , SJLJ :



, ,  — , « », , , - (). , DrMefistO.


Quake World WinMain(), jmp_buf, , « continue»:


// WinQuake/host.c
jmp_buf         host_abortserver;

void Host_EndGame (char *message, ...)
{
  ...

  if (cls.demonum != -1)
    CL_NextDemo ();
  else
    CL_Disconnect ();

  longjmp (host_abortserver, 1);
}

void Host_Error (char *error, ...)
{
  ...

  if (cls.state == ca_dedicated)
    Sys_Error ("Host_Error: %s\n",string);  // dedicated servers exit

  CL_Disconnect ();
  cls.demonum = -1;

  inerror = false;

  longjmp (host_abortserver, 1);
}

void _Host_Frame (float time)
{
  static double           time1 = 0;
  static double           time2 = 0;
  static double           time3 = 0;
  int                     pass1, pass2, pass3;

  if (setjmp (host_abortserver) )
    return;                 // something bad happened, or the server disconnected

  ...
}

// QW/client/sys_win.c
int WINAPI WinMain (...)
{
  ...

  while (1)
  {
    ...
    newtime = Sys_DoubleTime ();
    time = newtime - oldtime;
    Host_Frame (time);
    oldtime = newtime;
  }

  /* return success of application */
  return TRUE;
}



,  — . , setjmp() glibc , . :


  • , , SJLJ/saneex.c
  • -…- (, ) , -  — ,
  • ,  — zero-cost exceptions (, , zero-cost try), try, () — goto , , , «» , -,

«» zero-cost exceptions , volatile-, , ( longjmp()). , :



, saneex.c zero-cost (  — ), setjmp(), ? , ?  — .




« », main() 100 try/catch throw().


C:


#include <stdio.h>
#include <time.h>
#include "saneex.h"

int main(void) {
  for (int i = 0; i < 100000; i++) {
    try {
      //  ("" = ):
      throw(msgex("A quick fox jumped over a red dog and a nyancat was spawned"));
      //  ("" = ):
      time(NULL);
    } catchall {
      fprintf(stderr, "%s\n", curex().message);
    } endtry
  }
}

++ ( , cerr << fprintf()):


#include <iostream>
#include <vector>
#include <stdexcept>
#include <time.h>

int main() {
  std::vector<int> vec{ 3, 4, 3, 1 };

  for (int i = 0; i < 100000; i++) {
    try {
      //  ("" = ):
      int i{ vec.at(4) };
      //  ("" = ):
      time(NULL);
    }
    catch (std::out_of_range & e) {
      // <<  fprintf()     25-50%
      //std::cerr << "Accessing a non-existent element: " << e.what() << '\n';
      fprintf(stderr, "%s\n", e.what());
    }
    catch (std::exception & e) {
      //std::cerr << "Exception thrown: " << e.what() << '\n';
      fprintf(stderr, "%s\n", e.what());
    }
    catch (...) {
      //std::cerr << "Some fatal error\n";
      fprintf(stderr, "Some fatal error");
    }
  }

  return 0;
}

( 64-):


  • Windows 10 2019 LTSC PowerShell Measure-Command { test.exe 2>$null }
  • Live CD Ubuntu time

Windows __try/__except, :


#include <windows.h>
#include <stdio.h>
#include <vector>

int filterExpression(EXCEPTION_POINTERS* ep) {
  ep->ContextRecord->Eip += 8;
  return EXCEPTION_EXECUTE_HANDLER;
}

int main() {
  static int zero;
  for (int i = 0; i < 100000; i++) {
    __try {
      zero = 1 / zero;
      __asm {
        nop
        nop
        nop
        nop
        nop
        nop
        nop
      }
      printf("Past the exception.\n");
    }
    __except (filterExpression(GetExceptionInformation())) {
      printf("Handler called.\n");
    }
  }
}

 — , :


error C2712: Cannot use __try in functions that require object unwinding

, , . 1100-1300 (Debug Release, x86) — , VS, , g++.




№                     ()¹           saneex 

1.  VS 2019 v16.0.0   Debug     x64   saneex.c        9713  / 8728  = 1.1    1.8 / 1.8
2.  VS 2019 v16.0.0   Debug     x64   saneex.c       95    / 46    = 2      4.5 / 2.3
3.  VS 2019 v16.0.0   Debug     x64   C++             5449  / 4750² = 1.6
4.  VS 2019 v16.0.0   Debug     x64   C++            21    / 20    = 1
5.  VS 2019 v16.0.0   Release   x64   saneex.c        8542³ / 182   = 47     1.8 / 0.4
6.  VS 2019 v16.0.0   Release   x64   saneex.c       80³   / 23    = 3.5    8   / 1.8
7.  VS 2019 v16.0.0   Release   x64   C++             4669³ / 420   = 11
8.  VS 2019 v16.0.0   Release   x64   C++            10³   / 13    = 0.8
9.  gcc 9.2.1         -O0       x64   saneex.c        71    / 351   = 0.2    0.2 / 0.6
10. gcc 9.2.1         -O0       x64   saneex.c       6     / 39    = 0.2    1.5 / 1.1
11. g++ 9.2.1         -O0       x64   C++             378   / 630   = 0.6
12. g++ 9.2.1         -O0       x64   C++            4     / 37    = 0.1
13. gcc 9.2.1         -O3       x64   saneex.c        66    / 360   = 0.2    0.2 / 0.6
14. gcc 9.2.1         -O3       x64   saneex.c       5     / 23    = 0.2    1   / 0.6
15. g++ 9.2.1         -O3       x64   C++             356   / 605   = 0.6
16. g++ 9.2.1         -O3       x64   C++            5     / 38    = 0.1

¹ Windows 7 SP1 x64 VS 2017 v15.9.17 gcc cygwin.


² : fprintf() cerr <<, 3 : 1386/1527 .


³ VS , .


… :


  • / «» VS.  — .
  • cerr << fprintf() VS 3-4 ( 3). ?
  • try throw — (4-28 100 ).
  • «» Debug VS, saneex.c , ( 2.3 VS, 5 gcc/g++), try throw — , . !

… . !


use-case — try throw (« , »), setjmp(), , , , . , , setjmp() OpenBSD (1.45) — Solaris. 2005 . «» — , .


, …



 — saneex.c


, :


  • Visual Studio
  • , throw() , finally catch ( )
  • ( static)
  • - (__thread/_Thread_local)
  • public domain (CC0)


GitHub. , . saneex-demo.c :


01.    #include <stdio.h>
02.    #include "saneex.h"
03.
04.    int main(void) {
05.      sxTag = "SaneC's Exceptions Demo";
06.
07.      try {
08.        printf("Enter a message to fail with: [] [1] [2] [!] ");
09.
10.        char msg[50];
11.        thrif(!fgets(msg, sizeof(msg), stdin), "fgets() error");
12.
13.        int i = strlen(msg) - 1;
14.        while (i >= 0 && msg[i] <= ' ') { msg[i--] = 0; }
15.
16.        if (msg[0]) {
17.          errno = atoi(msg);
18.          struct SxTraceEntry e = newex();
19.          e = sxprintf(e, "Your message: %s", msg);
20.          e.uncatchable = msg[0] == '!';
21.          throw(e);
22.        }
23.
24.        puts("End of try body");
25.
26.      } catch (1) {
27.        puts("Caught in catch (1)");
28.        sxPrintTrace();
29.
30.      } catch (2) {
31.        puts("Caught in catch (2)");
32.        errno = 123;
33.        rethrow(msgex("calling rethrow() with code 123"));
34.
35.      } catchall {
36.        printf("Caught in catchall, message is: %s\n", curex().message);
37.
38.      } finally {
39.        puts("Now in finally");
40.
41.      } endtry
42.
43.      puts("End of main()");
44.    }

, :


  •  — , :

End of try body
Now in finally
End of main()

  • , , (1), catch (1) (26.), :

Caught in catch (1)
Your message: 1 hello, habr!
    ...at saneex-demo.c:18, code 1
Now in finally
End of main()

  • , (30.), ( , ) (33.), :

Caught in catch (2)
Now in finally

Uncaught exception (code 123) - terminating. Tag: SaneC's Exceptions Demo
Your message: 2 TM! kak tam blok4ain?
    ...at saneex-demo.c:18, code 2
calling rethrow() with code 123
    ...at saneex-demo.c:33, code 123
rethrown by ENDTRY
    ...at saneex-demo.c:41, code 123

  • !, «» (uncatchable; 20.) — try , ( catch, finally),  — abort():

Caught in catch (1)
Your message: ! it is a good day to die
    ...UNCATCHABLE at saneex-demo.c:18, code 0
Now in finally

Uncaught exception (code 0) - terminating. Tag: SaneC's Exceptions Demo
Your message: ! it is a good day to die
    ...UNCATCHABLE at saneex-demo.c:18, code 0
UNCATCHABLE rethrown by ENDTRY
    ...at saneex-demo.c:41, code 0

  • , , catchall (35.), :

Caught in catchall, message is: Your message: 3 we need more gold
Now in finally
End of main()


«»



. , ( MSVC¹), C11 (TLS):


#define SX_THREAD_LOCAL _Thread_local

¹ Microsoft - open source, , , 8 , .


sxTag (05.) — , stderr.  — (__DATE__ __TIME__).


SxTraceEntry ( stack trace).  — (struct SxTraceEntry) {...}:


  • newex() — ; __FILE__, __LINE__ = errno ( , fgets(); 11.)
    • 1 1 ( setjmp() 0 ), catch (0)
  • msgex(m) — newex(), ( )
  • exex(m, e) — msgex(), ; free() :

try {
  TimeoutException *e = malloc(sizeof(*e));
  e->elapsed = timeElapsed;
  e->limit = MAX_TIMEOUT;
  errno = 146;
  throw(exex("Connection timed out", e));
} catch (146) {
  printf("%s after %d\n", curex().message,
    //   void *SxTraceEntry.extra:
    ((TimeoutException *) curex().extra)->elapsed);
} endtry

, , designated initializers C99 ( Visual Studio 2013+):


throw( (struct SxTraceEntry) {.message = "kaboom!"} );

:


  • throw(e) — SxTraceEntry
  • rethrow(e) — throw(), stack trace; catch/catchall
  • thrif(x, m) — ; if (x) SxTraceEntry x + m «»
  • thri(x) — thrif(), m

«»  — fgets() (11.), . fgets() ( EOF: ./a.out </dev/null), . :


thri(read(0xBaaD, buf, nbyte));
// errno = 9, "Bad file descriptor"
// Assertion error: read(0xBaaD, buf, nbyte);


… « »


( !):


  • endtry — ( try)
    • , , , try {, endtry
  • return try endtry — , ; PR
    • , goto , - ? </sarcasm>

«», volatile. «»  — (. longjmp()), , try, catch/catchall/finally endtry, volatile. . :


int foo = 1;
try {
  foo = 2;
  //    foo
} catchall {
  //    !
} finally {
  //   !
} endtry
//   !

volatile :


volatile int foo = 1;
try {
  ...


:


- () :


  • struct SxTryContext — try,  — , jmp_buf ; , :

try {
  try {
    //  
  } endtry
} endtry

  • struct SxTraceEntry — stack trace, , ; , try:

try {         //  SxTryContext
  try {       //  SxTryContext
              //  SxTraceEntry
    throw(msgex(" !"));
              //  SxTraceEntry
  } catchall {
              //  SxTraceEntry
    rethrow(msgex("   !"));
              //  SxTraceEntry (*)
  } endtry
} endtry

rethrow() throw(), SxTraceEntry (*) ,  — (stack trace ). , sxAddTraceEntry(e).


try and other structural elements are macros (- your K.O. ). Brackets { }after them are optional. As a result, all this boils down to the following pseudo-code:


try {                             int _sxLastJumpCode = setjmp(add_context()¹);
                                  bool handled = false;
                                  if (_sxLastJumpCode == 0) {
  throw(msgex("Mama mia!"));        clearTrace();
                                    sxAddTraceEntry(msgex(...));
                                    if (count_contexts() == 0) {
                                      fprintf(stderr, "Shurik, vs propalo!");
                                      sxPrintTrace();
                                      exit(curex().code);
                                    } else {
                                      longjmp(top_context());
                                    }
} catch (9000) {                  } else if (_sxLastJumpCode == 9000) {
                                    handled = true;
} catchall {                      } else {
                                    handled = true;
} finally {                       }
                                  //    finally { }
} endtry                          remove_context();
                                  if (!handled) {
                                    //    throw()
                                  }

¹ The names with _ are not used in the library, they are abstractions.


I think, after detailed explanations of how SJLJ works, something else is unnecessary to comment here, so let me take a leave and give the floor to you already.


All Articles