⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 generic.h

📁 very comprehensive example, windows WDM driver development ex The use of drive technology can s
💻 H
字号:
/* generic.h -- anything you don't #undef at the end remains in effect.   The ONLY things that go in here are generic indicator flags; it's up   to your programs to declare and call things based on those flags.   You should only need to make changes via a minimal system-specific section   at the end of this file.  To build a new section, rip through this and   check everything it mentions on your platform, and #undef that which needs   it.  If you generate a system-specific section you didn't find in here,   please mail me a copy so I can update the "master".   I realize I'm probably inventing another pseudo-standard here, but   goddamnit, everybody ELSE has already, and I can't include all of their   hairball schemes too.  HAVE_xx conforms to the gnu/autoconf usage and   seems to be the most common format.  In fact, I dug a lot of these out   of autoconf and tried to common them all together using "stupidh" to   collect data from platforms.   In disgust...  _H*  940910, 941115, 950511.  Pseudo-version: 1.3   Updated 951104 with many patches from netcat feedback, and properly   closed a lot of slop in open-ended comments: version 1.4   960217 + nextstep: version 1.5*/#ifndef GENERIC_H		/* only run through this once */#define GENERIC_H/* =============================== *//* System calls, lib routines, etc *//* =============================== *//* How does your system declare malloc, void or char?  Usually void, but go   ask the SunOS people why they had to be different... */#define VOID_MALLOC/* notably from fwtk/firewall.h: posix locking? */#define HAVE_FLOCK		/* otherwise it's lockf() *//* if you don't have setsid(), you might have setpgrp(). */#define HAVE_SETSID/* random() is generally considered better than rand() */#define HAVE_RANDOM/* the srand48/lrand48/etc family is s'posedly even better */#define HAVE_RAND48/* bmc@telebase and others have suggested these macros if a box *does* have   rand48.  Will consider for later if we're doing something that really   requires stronger random numbers, but netcat and such certainly doesn't.#define srandom(seed) srand48((long) seed)#define random()      lrand48() *//* if your machine doesn't have lstat(), it should have stat() [dos...] */#define HAVE_LSTAT/* different kinds of term ioctls.  How to recognize them, very roughly:   sysv/POSIX_ME_HARDER:  termio[s].h, struct termio[s], tty.c_*[]   bsd/old stuff:  sgtty.h, ioctl(TIOCSETP), sgttyb.sg_*, tchars.t_*  */#define HAVE_TERMIOS/* dbm vs ndbm */#define HAVE_NDBM/* extended utmp/wtmp stuff.  MOST machines still do NOT have this SV-ism */#define UTMPX/* some systems have nice() which takes *relative* values... [resource.h] */#define HAVE_SETPRIORITY/* a sysvism, I think, but ... */#define HAVE_SYSINFO/* ============= *//* Include files *//* ============= *//* Presence of these can be determined via a script that sniffs them   out if you aren't sure.  See "stupidh"... *//* stdlib comes with most modern compilers, but ya never know */#define HAVE_STDLIB_H/* not on a DOS box! */#define HAVE_UNISTD_H/* stdarg is a weird one */#define HAVE_STDARG_H/* dir.h or maybe ndir.h otherwise. */#define HAVE_DIRENT_H/* string or strings */#define HAVE_STRINGS_H/* if you don't have lastlog.h, what you want might be in login.h */#define HAVE_LASTLOG_H/* predefines for _PATH_various */#define HAVE_PATHS_H/* some SV-flavors break select stuff out separately */#define HAVE_SELECT_H/* assorted others */#define HAVE_PARAM_H		/* in sys/ */#define HAVE_SYSMACROS_H	/* in sys/ */#define HAVE_TTYENT_H		/* securetty et al *//* ==================== *//* Still maybe have to do something about the following, if it's even   worth it.  I just grepped a lot of these out of various code, without   looking them up yet:#define HAVE_EINPROGRESS#define HAVE_F_SETOWNHAVE_FILIO_H ... fionbio, fiosetown, etc... will need for hairier  select loops.#define HAVE_SETENV ... now *there's* a hairy one; **environ is portable#define BIG_ENDIAN/little_endian ... *please* try to avoid this stupidity   and LSBFIRST/MSBFIRST#define HAVE_GETUSERSHELL ... you could always pull it out of getpwent()#define HAVE_SETE[UG]ID ... lib or syscall, it varies on diff platforms#define HAVE_STRCHR ... should actually be handled by string/strings#define HAVE_PSTAT#define HAVE_ST_BLKSIZE ... a stat() thing?#define HAVE_IP_TOS#define HAVE_STRFTIME ... screw this, we'll just INCLUDE one for lame   old boxes that don't have it [sunos 3.x, early 4.x?]#define HAVE_VFPRINTF#define HAVE_SHADOW_PASSWD  ... in its multitudinous schemes?? ... how   about sumpin' like #define SHADOW_PASSWD_TYPE ... could get grody.   ... looks like sysv /etc/shadow, getspent() family is common.#define SIG*  ... what a swamp, punt for now; should all be in signal.h#define HAVE_STRCSPN  ... see larry wall's comment in the fwtk regex code#define ULTRIX_AUTH  ... bwahaha.#define HAVE_YP  or NIS or whatever you wanna call it this weekrandomness about VARARGS??--- later stuff to be considered ---#define UINT4 ... u-int on alpha/osf, i.e. __alpha/__osf__, ulong elsewhere?   dont name it that, though, it'll conflict with extant .h files like md5randomness about machine/endian.h, machine/inline.h -- bsdi, net/2randomness about _PATH_WTMP vs WTMP_FILE and where they even live!!#define HAVE_SYS_ERRLIST ... whether it's in stdio.h or not [bsd 4.4]--- still more stuff#define HAVE_SETENV#define _PATH_UTMP vs UTMP_FILE, a la deslogind?!#define HAVE_DAEMON#define HAVE_INETADDR  [vixie bind?]lseek: SEEK_SET vs L_SET and associated lossage [epi-notes, old 386Mach]bsdi: ioctl_compat.h ?--- takin' some ifdefs from CNS krb:F_GETOWN/F_SETOWNCRAY: long = 8 bytes, etc  [class with alpha?]CGETENTSIGINFOSIGTSTP SIGTTOU SIGWINCHSPX?SYSV_TERMIO -- covered elsewhere, I hopeTIOCEXT TIOCFLUSH TIOC[GS]WINSIZ NEWINIT: something about init cleaning up dead login processes [telnet?]PARENT_DOES_UTMP, too  [telnet]VDISCARDVEOL/VEOL2/VLNEXT VREPRINT -- termios stuff?, and related...STREAMSPTY/STREAMSPTYEMAF_INET/AF_UNSPEC, PF_*ECHOCTL/ECHOKEF_ULOCK [?!]setpgrp/getpgrp() ONEARG business..HAVE_ALLOCAHAVE_GETUTENTHAVE_SYS_SELECT_H  [irix!]HAVE_DIRENT  [old 386mach has *direct.h*!]HAVE_SIGSETHAVE_VFORK_H and HAVE_VFORKHAVE_VHANGUPHAVE_VSPRINTFHAVE_IPTOS_*HAVE_STRCASECMP, STRNCASECMPHAVE_SYS_FCNTL_HHAVE_SYS_TIME_HHAVE_UTIMESNOTTYENT [?]HAVE_FCHMODHAVE_GETUSERSHELLHAVE_SIGCONTEXT  [stack hair, very machine-specific]YYLINENO?POSIX_SIGNALSPOSIX_TERMIOSSETPROCTITLE -- breaks some places, like fbsd sendmailSIG* -- actual signal names?  some are missingSIOCGIFCONFSO_BROADCASTSHMEM  [krb tickets]VARARGS, or HAVE_VARARGSCBAUD... and B300, B9600, etc etcHAVE_BZERO  vs  memset/memcpyHAVE_SETVBUFHAVE_STRDUPHAVE_GETENVHAVE_STRSAVEHAVE_STBLKSIZE  [stat?]HAVE_STREAM_H -- in sys/, ref sendmail 8.7 for IP_SRCROUTEFCHMODINITGROUPS -- most machines seem to *have*SETREUIDSNPRINTFSETPGRP semantics bsd vs. sys5 styleThere's also the issue about WHERE various .h files live, sys/ or otherwise.There's a BIG swamp lurking where network code of any sort lives.*//* ======================== *//* System-specific sections *//* ======================== *//* By turning OFF various bits of the above,  you can customize for   a given platform.  Yes, we're ignoring the stock compiler predefines   and using our own plugged in via the Makefile. *//* DOS boxes, with MSC; you may need to adapt to a different compiler. *//* looks like later ones *do* have dirent.h, for example */#ifdef MSDOS#undef HAVE_FLOCK#undef HAVE_RANDOM#undef HAVE_LSTAT#undef HAVE_TERMIOS#undef UTMPX#undef HAVE_SYSINFO#undef HAVE_UNISTD_H#undef HAVE_DIRENT_H	/* unless you have the k00l little wrapper from L5!! */#undef HAVE_STRINGS_H#undef HAVE_LASTLOG_H#undef HAVE_PATHS_H#undef HAVE_PARAM_H#undef HAVE_SYSMACROS_H#undef HAVE_SELECT_H#undef HAVE_TTYENT_H#endif /* MSDOS *//* buglix 4.x; dunno about 3.x on down.  should be bsd4.2 */#ifdef ULTRIX#undef UTMPX#undef HAVE_PATHS_H#undef HAVE_SYSMACROS_H#undef HAVE_SELECT_H#endif /* buglix *//* some of this might still be broken on older sunoses */#ifdef SUNOS#undef VOID_MALLOC#undef UTMPX#undef HAVE_PATHS_H#undef HAVE_SELECT_H#endif /* sunos *//* "contact your vendor for a fix" */#ifdef SOLARIS/* has UTMPX */#undef HAVE_RANDOM#undef HAVE_SETPRIORITY#undef HAVE_STRINGS_H	/* this is genuinely the case, go figure */#undef HAVE_PATHS_H#undef HAVE_SELECT_H#undef HAVE_TTYENT_H#endif /* SOLARIS *//* whatever aix variant MIT had at the time; 3.2.x?? */#ifdef AIX#undef UTMPX#undef HAVE_LASTLOG_H#define HAVE_LOGIN_H	/* "special", in the educational sense */#endif /* aix *//* linux, which is trying as desperately as the gnu folks can to be   POSIXLY_CORRECT.  I think I'm gonna hurl... */#ifdef LINUX#undef UTMPX#undef HAVE_SYSINFO#undef HAVE_SELECT_H#undef HAVE_TTYENT_H#endif /* linux *//* irix 5.x; may not be correct for earlier ones */#ifdef IRIX/* wow, does irix really have everything?! */#endif /* irix *//* osf on alphas */#ifdef OSF#undef UTMPX#undef HAVE_SELECT_H#endif /* osf *//* they's some FUCKED UP paths in this one! */#ifdef FREEBSD#undef UTMPX#undef HAVE_SYSINFO#undef HAVE_LASTLOG_H#undef HAVE_SYSMACROS_H#undef HAVE_SELECT_H	/* actually a lie, but only for kernel */#endif /* freebsd *//* Originally from the sidewinder site, of all places, but subsequently   checked further under a more normal bsdi 2.0 */#ifdef BSDI#undef UTMPX#undef HAVE_LASTLOG_H#undef HAVE_SYSMACROS_H/* and their malloc.h was in sys/ ?! */#undef HAVE_SELECT_H#endif /* bsdi *//* netbsd/44lite, jives with amiga-netbsd from cactus */#ifdef NETBSD#undef UTMPX#undef HAVE_SYSINFO#undef HAVE_LASTLOG_H#undef HAVE_SELECT_H#endif /* netbsd *//* Hpux 9.0x, from BBN and various patches sent in */#ifdef HPUX#undef HAVE_RANDOM	/* but *does* have ?rand48 -- need to consider.. */#undef HAVE_UTMPX#undef HAVE_LASTLOG_H	/* has utmp/wtmp/btmp nonsense, and pututline() */#undef HAVE_PATHS_H#undef HAVE_SELECT_H#undef HAVE_TTYENT_H#endif /* hockeypux *//* Unixware [a loose definition of "unix", to be sure], 1.1.2 [at least]   from Brian Clapper.  He wasn't sure about 2.0... */#ifdef UNIXWARE/* has UTMPX */#undef HAVE_SETPRIORITY/* NOTE: UnixWare does provide the BSD stuff, in "/usr/ucbinclude" (headers)   and "/usr/ucblib" (libraries).  However, I've run into problems linking   stuff out of that version of the C library, when objects are also coming   out of the "regular" C library.  My advice: Avoid the BSD compatibility   stuff wherever possible.  Brian Clapper <bmc@telebase.com> */#undef HAVE_STRINGS_H#undef HAVE_PATHS_H#undef HAVE_TTYENT_H#endif /* UNIXWARE *//* A/UX 3.1.x from darieb@sandia.gov */#ifdef AUX#undef HAVE_RANDOM#undef HAVE_SELECT_H	/* xxx: untested */#endif /* a/ux *//* NeXTSTEP 3.2 motorola mudge@l0pht.com xxx should also work with   white hardware and Sparc/HPPA. Should work with 3.3 too as it's   4.3 / 4.4 bsd wrapped around mach */#ifdef NEXT#undef UTMPX#undef HAVE_SELECT_X#endif /* NeXTSTEP 3.2 motorola *//* Make some "generic" assumptions if all else fails */#ifdef GENERIC#undef HAVE_FLOCK#if defined(SYSV) && (SYSV < 4)  /* TW leftover: old SV doesnt have symlinks */#undef HAVE_LSTAT#endif /* old SYSV */#undef HAVE_TERMIOS#undef UTMPX#undef HAVE_PATHS_H#undef HAVE_SELECT_H#endif /* generic *//* ================ */#endif /* GENERIC_H */

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -