Annotation of dietlibc/CHANGES, revision 1.77
1.74 fefe 1: 0.10:
2: update getpagesize.
3: add personality.
1.75 fefe 4: add "%*s" and "%*.*s" support to vsnprintf.
1.76 fefe 5: add putchar (apparently arm-linux-gcc will substitute printf("\n")
6: with putchar('\n') behind your back...?!
1.77 ! fefe 7: import include/scsi/ from glibc.
1.74 fefe 8:
1.73 fefe 9: 0.9:
1.65 fefe 10: fixed initgroups (oops)
1.66 fefe 11: Cleanup appending file open for stdio (thanks, James Antill)
12: imported vfprintf fix from James Antill.
1.67 fefe 13: add fnmatch
14: change index, rindex and bcmp as weak ELF symbols (were #defined before)
1.68 fefe 15: Olaf added character classes to *scanf.
1.69 fefe 16: Put object and library files into a separate directory (bin-$ARCH).
17: Add "cross" make target to make all architectures.
18: Add "diet" program and man page.
1.70 fefe 19: Add getgrnam, getgrgid
1.71 fefe 20: Add cfmakeraw (Thanks, Antonio Musumeci)
1.72 fefe 21: localtime won't call gettimeofday when the tzfile parser is used.
1.65 fefe 22:
1.64 fefe 23: 0.8:
1.34 fefe 24: vsnprintf(0,...) estimated the length of strings with a padding width
25: incorrectly.
1.35 fefe 26: new reboot() from Olaf.
1.36 fefe 27: added prototypes for getsockopt and setsockopt.
1.39 fefe 28: added experimental sem* and shm*
1.38 fefe 29: define EXIT_SUCCESS and EXIT_FAILURE
1.40 fefe 30: added setvbuf (only changes mode, not buffer)
1.41 fefe 31: added dietwarning.h (woohoo, linker warnings!) and used it to warn of
32: insecurity (system, sprintf) or bloat (printf and stdio).
1.42 fefe 33: added qsort() and bsearch(). Benchmark on 1000 random ints:
34: glibc: 679414 cycles qsort and 3352 cycles bsearch
35: diet libc: 300010 cycles qsort, 553 cycles bsearch
36: I have no idea why glibc is so slow.
1.43 fefe 37: added __isnan (references by libm.a from glibc).
38: added popen() and pclose(). Now the diet libc can link slsh (from
39: S-Lang!)
1.44 fefe 40: corrected usleep. It was more of an msleep until now.
1.45 fefe 41: corrected __dtostr (it looped when trying to format "0.0").
42: several fixes to stdio (thanks to S-Lang's excellent test suite!)
1.46 fefe 43: added ungetc (does it work?)
1.47 fefe 44: added inet_aton (yuck!)
1.50 fefe 45: added gethostbyname, gethostbyaddr, gethostbyname_r and gethostbyaddr_r
1.48 fefe 46: added h_errno and __h_errno_location
47: added inet_ntoa
1.49 fefe 48: added setpgrp and getpgrp, strsignal and execv (thanks, Olaf)
1.51 fefe 49: add bcopy, index, rindex macros
50: add isgraph and isxdigit
1.52 fefe 51: add strftime (ugh!!!)
1.53 olaf 52: Olaf fixed strftime, gmtime_r leap year bug.
1.54 fefe 53: Jeff Garzik contributed a spec file. I put it in contrib and hope
54: everbody knows what to do with it...
55: Olaf checked in a ton of libpthread stuff.
1.55 fefe 56: added mkstemp using /dev/urandom (harder to guess than getpid()).
1.56 fefe 57: add large file backwards compatibility support (so gzip will run on
58: kernel 2.2 and 2.0)
1.57 fefe 59: added shutdown
1.58 fefe 60: added nice
61: added pseudo-locale support for is* (they are now weak aliases,
62: isupper -> __isupper_ascii etc). So, for example, liblatin1 can
63: overwrite the default is* functions.
1.59 fefe 64: added liblatin1.a
1.60 fefe 65: added tzfile implementation (does it work for anyone else?)
1.61 fefe 66: added syslog(3) (thanks to Rene Müller!)
1.62 fefe 67: Jeff Garzik contributed getgr*, getpw* and getsp* in one tiny package.
68: Thanks!
1.63 fefe 69: merged duplicate code in execv and execvp into exec_lib.o
1.34 fefe 70:
1.21 fefe 71: 0.7.2:
72: added several prototypes to already implemented functions and
73: <sys/reboot.h>. Thanks for Abraham for pointing this out.
1.22 fefe 74: added prototypes for process group and dup syscalls. Thanks to
75: Abraham again. He also contributed an implementation for daemon(),
76: an ugly BSD hack. I decided to diversify the lib/ directory a
77: little and now created a libugly/ directory for stuff like system()
78: and daemon() ;-)
79: added libstdio and moved all that stdio crap from lib/ there. The
80: idea is to start separating now what we will separate for shared
81: libraries anyway.
1.26 fefe 82: optimized x86 unified-syscall by Olaf.
1.27 fefe 83: added limits.h, removed __ARCHBITS from endian.h (use __WORDSIZE from
84: limits.h instead)
85: added remove.c and enough cruft to limits.h to make it compile lzo,
86: lzop and bzip2.
1.28 fefe 87: added mktime, asctime_r, asctime and ctime (needed for gzip).
88: Someone will need to implement a reader for the glibc localtime file
1.31 fefe 89: format some day (man tzfile, ugh!) :-(
1.29 fefe 90: added unlink prototype to unistd.h.
91: added gmtime and gmtime_r.
1.31 fefe 92: fixed sparc unified syscall error handling (please upgrade!).
1.30 fefe 93: removed readdir glibc compatibility cruft.
1.31 fefe 94: removed stat glibc compatibility cruft.
1.32 fefe 95: oops, my fread was broken for pipes.
1.33 fefe 96: Olaf found a bug in pipe() for sparc, too.
1.21 fefe 97:
1.9 fefe 98: 0.7.1:
1.11 fefe 99: optimized away the static arrays from __ltostr and __lltostr.
100: optimized away the static arrays (and more!) from strtoul and strtoull.
101: added system (by Olaf).
1.12 fefe 102: added tolower and toupper
1.13 fefe 103: added getopt from Olaf (~600 bytes on x86!)
1.14 fefe 104: added clone for mips (by Olaf).
105: fixed longjmp and setjmp on sparc not to require glibc include files.
1.15 olaf 106: added pause for all (by Olaf).
1.17 fefe 107: added __attribute__ ((__const__)) to ctype.h
1.16 fefe 108: added isupper and islower.
109: added putc and putchar as macros.
1.17 fefe 110: added strcasecmp and strncasecmp (thanks to Abraham vd Merwe for
111: insightful discussions and patches I partially used for the last three)
1.18 fefe 112: added getpriority and setpriority (also by Abraham)
1.19 fefe 113: oops, I broke strtoul in a recent optimization attempt.
1.20 fefe 114: oi, setjmp and longjmp were broken on other platforms, too!
115: Thanks to Sébastien Côté for reporting this.
1.9 fefe 116:
117: 0.7:
1.6 olaf 118: clone support for i386,alpha,arm and sparc by Olaf.
1.1 cvs 119: Initial dynamic loading glue for i386, also by Olaf.
120: We don't have a dynamic loader yet, so this is not yet usable.
121: added readv, sigdelset, sigaltstack, sigfillset, sigismember, usleep,
122: vsprintf, writev, all by Olaf.
1.2 fefe 123: imported readdir fix from Guillaume. Mhh, can the kernel interface
124: really be this broken?
1.3 fefe 125: Found a kludgy way to detect whether the kernel includes define struct
126: stat64 (they also define STAT64_HAS_BROKEN_ST_INO). If this does
127: not work for you, please tell me!
1.4 fefe 128: Removed bogus __seek_types enum, it's a bunch of #defines now.
1.7 fefe 129: Fixed stdio line buffering. The code was already there but I forgot
130: to mark stdio and stdout as line buffered. This should really be
131: done dynamically using isatty, though.
1.8 fefe 132: Removed superfluous "if (1)" in execvp.c
1.1 cvs 133:
134: 0.6.14:
135: *scanf did not append the 0 byte for %s.
136: added creat and changed creat to call open directly instead of open64.
137: did for fread what 0.6.10 did for fwrite.
138: repaired buffered stdio a little (line buffering is still not
139: supported).
140: [All of these bugs have been reported by Guillaume. Thanks!]
141: added __pure__ to strchr and strrchr in <string.h>
142:
143: 0.6.13:
144: included COPYING to make clear that the diet libc is covered by the
145: GNU General Public License (and _not_ the LGPL). That means that
146: you need to obtain a license from me if you want to use the diet
147: libc in a proprietary program that you want to distribute.
148: The x86 unified syscall swapped arguments #4 and #5 and noone
149: noticed until now! Thanks to Guillaume Cottenceau for reporting
150: this!
151: The malloc had a subtle bug with small allocations that could cause
152: segfault. Guillaume reported it, Olaf fixed it.
153:
154: 0.6.12:
155: added assert, statfs and fstatfs
156: added include/net/if.h so fget compiles again.
157:
158: 0.6.11:
159: added putenv.
160:
161: 0.6.10:
162: fixed fwrite. It returned the number of bytes written, not the number
163: of records. Thanks to Albert D. Cahalan for pointing this out.
164: fixed fgets. It returned EOF on empty lines. Thanks, Erik Frey.
165: added brk and sbrk. harold@nb.com.sg made me do it.
166:
167: 0.6.9:
168: fixed execvp not to return on ENOENT
169: did for {open|read|close|seek|tell}dir what I did for stat before.
170: The glibc compatibility part is currently non-functional.
171: fixed white space handling in sscanf " (" would not match " (".
172: fixed printf to accept 'l' flag.
173: fixed getpwuid (ignored last line)
174:
175: 0.6.8:
176: Olaf contributed initial thread-safe syscalls for i386, alpha, sparc,
177: mips and arm. I did the one for ppc. They can be enabled in dietfeatures.h.
178: added strlcpy and strlcat from OpenBSD.
179: added stat64, fstat64 and lstat64.
180: added endian.h and made ending of strlen.c endianness-aware
181: added string routines size tweak to dietfeatures.h. Most of the
182: string routines have been unrolled. This is up to three times as
183: fast but creates up to three times larger code. Now you can disable
184: the unrolling.
185: added sendfile.
186: I also added pread. It works on x86 and sparc but not on ppc and mips.
187: I have no idea what's going on. strace disagrees with the kernel.
188: Can anyone help? I asked the Linux kernel mailing list for help, too.
189: added a few aliases of the type __libc_open for open. nm on
190: libpthreads.so indicates that we will need them.
191:
192: 0.6.7:
193: the sources now compile without warnings with -Wall.
194: printf now prints "(null)" when %s is passed NULL. This can be
195: removed with WANT_NULL_PRINTF in dietfeatures.h
196: added vfprintf, execl, ttyname.
197: stat, lstat and fstat are now #defined to __dietstat, __dietlstat and
198: __dietfstat so they can use the normal kernel struct stat.
199: stat, lstat and fstat are now C wrappers that convert to the glibc
200: struct stat. So we can avoid that overhead for programs that use
201: dietlibc headers.
202:
203: 0.6.6:
204: changed the Makefiles so you can now set CFLAGS on the command line
205: added memccmp (analogous to memccpy) and strncmp.
206: Olaf fixed another bug in the sparc unified syscall.
207: Paul Clifford contributed a C version of his strlen.S that is much
208: more efficient than the previous strlen on all platforms! It also
209: looks like technology from Roswell. ;-}
210: added a "real" stdio. Well, almost. Please contribute!
211: added dietfeatures.h so you can remove features you don't need.
212: removed debug code from vsnprintf that changed \0 to ' ' (argh!).
213: stdio now works with simple programs and minigzip from libz.
214: stdio uses some major trickery to avoid linking the stdio, stderr and
215: stdout (and reserving space for them) when they are not used.
216:
217: 0.6.5:
218: Red Plait <redplait@ixcelerator.com> found several bugs in diet libc,
219: most of them bugs in the new header files, but also missing sigset
220: functions.
221: Paul Clifford contributed an assembly strlen.S for ARM and fixed
222: several bugs.
223: Fixed ppc/setjmp and mips/pipe.
224: Olaf contributed a new sparc unified syscall.
225:
226: 0.6.4:
227: declared is* static inline in getservent.
228: added assert_fail, strtoul, isalpha, isdigit, isalnum, isascii.
229: changed strlen to return 0 when passed NULL.
230: new, much smaller unified syscall for MIPS. Thanks to Olaf the Mad
231: Scientist who actually implemented this without access to a MIPS
232: box, just from reading the architecture manual. And it worked out
233: of the box.
234: started a set of system includes, mainly so that I can use lcc and
235: my alpha-linux cross compiler (which is unable to cross-compile
236: glibc). diet libc can now be compiled without any include files
237: from a normal libc. The includes do declare more than diet libc
238: currently delivers and they are still far from complete for real
239: applications.
240: I will only add assembly versions that are smaller _and_ faster than
241: the C version. Larger routines are only accepted if they are called
242: very often and are substantially faster.
243: added i386 assembly strchr, which is smaller and faster than the old
244: version. It is, however, larger than the version contributed by
245: proton (thanks, anyway).
246: added i386 strlen (31% faster, 14% smaller)
247: "make t" will now create a map file called "mapfile".
248:
249: 0.6.3:
250: added sys_errlist, strerror and perror
251: added isblank
252: added atol
253:
254: 0.6.2:
255: mmap for ARM didn't compile. Thanks, Paul!
256:
257: 0.6.1:
258: split mmap into the architecture specific subdirectories.
259: getenv now copes with environ==NULL, thanks Paul Clifford.
260: Paul also contributed a smaller ARM startup code.
261:
262: 0.6:
263: strcat returned the wrong result. Thanks, Dietz Pröpper.
264: strtod now understands a negative exponent (oops, thanks Bertram Barth)
265: Port to arm-linux-gnu, but on the Netwinder I use for testing the
266: __dtostr does not work (I have no idea why!)
267: The MIPS port now uses (much smaller) non-PIC code. That means that
268: the applications you link against diet libc also have to be compiled
269: non-PIC. I suggest copying the CFLAGS from the diet libc Makefile.
270: Thanks to Ralf Bächle for helping me with this!
271: I made subdirectories for the architectures and use VPATH to override
272: VPATH so that make finds the source file automatically. That should
273: simplify the sources greatly.
274: Olaf Dreesen contributed Alpha support including setjmp and longjmp!
275: ARM and MIPS now also have setjmp and longjmp
276: "compile" and "load" are now make targets. Use them for djb programs.
277: added contrib/elftrunc.c which will remove unnecessary ELF headers.
278: Again, contributed from Olaf. Great work!
279:
280: 0.5.12:
281: printf also does signed numbers.
282: If you don't use atexit, dietlibc now does not link exit, only _exit.
283: *printf now correctly returns the number of bytes written.
284: Olaf contributed experimental sscanf and vsscanf implementations.
285: If passed NULL as buffer, snprintf will not write anything but still
286: return the number of bytes it would have written.
287: Initial MIPS port! (Oh, the agony!)
288: No setjmp and longjmp support yet! Please contribute!
289: I even implemented unified syscalls for MIPS. Still, MIPS code is
290: almost twice the size of SPARC code. If anyone knows why: please
291: tell me!
292:
293: 0.5.11:
294: I implemented new unified syscalls for x86, sparc and ppc, this time based
295: on .s files and not .c files, and I moved the syscalls into
296: subdirectores to clean the dietlibc sources up a little.
297: A binary consisting of printf("%s is %d\n","olaf",23) is now
298: 2864 bytes on sparc
299: 2488 bytes on intel
300: I kludgily implemented fprintf(stdout,... and fprintf(stderr,...
301: to make a few more applications work.
302: Olaf Dreesen also implemented some unified syscall stuff for x86.
303: He discovered that with our unified syscall interface it actually
304: costs <10 bytes total to make all system calls thread safe!
305: I implemented atexit() (can register up to 4 callbacks).
306: I implemented strtod and __dtostr (the opposite). Now we can add
307: floating point support to vsnprintf and sscanf!
308:
309: 0.5.10:
310: I actually saw that I can not only merge the errno handling code of
311: the system calls on x86, I can also merge the rest (including the
312: arguments) except for setting the system call number. All those
313: system calls are now just a jump to a unified system call handler.
314: I got rid of x86openclose again. The savings are substantial:
315: chown from embutils went from 7664 to 7184 bytes!
316: If I'd move the system call wrappers to assembly language, I could
317: even reduce the alignment (does not matter since it's just a jump
318: anyway) and get rid of the "ret" behind each jump (one byte per
319: system call!)
320:
321: 0.5.9:
322: added memchr, strpbrk, strstr, strtol, isspace (hehe)
323: fixed strdup
324: fixed return values for strcpy and strcat (thanks to Norbert Berzen)
325: Olaf Dreesen contributed a strtol and initial {sn|vsn|}printf implementation
326: (no signed integers, only strings and unsigned integers (but
327: supporting octal, hex and decimal). Thanks, Olaf! (by the way:
328: that code must be wonderful, I don't understand it at all *bg*)
329: Olaf's printf does understand padding like in "%08d" and "%8d" and
330: automatically pads pointers with '0'.
331: Rewrote the x86 start code, old: 22 instructions, new: 12 instructions.
332: Thanks to proton for inspiration on this.
333: The new x86 start code also does not reference exit any more, thus saving
334: 64 bytes for executables that don't call exit explicitly.
335: [insert maniacal laughter] EVERY BYTE COUNTS!1!! ;-)
336: Moved errno and environ to start.S, saving no byte binary size but
337: speeding up compilation and slightly shortening dietlibc.a ;-)
338: For x86: joined open and close into one assembler file, sharing the
339: errno handling. This is a feasibility test and it actually saves a
340: few bytes. I think I will reimplement the _syscall[1-6] macros on
341: all platforms now to share the errno handling code for them.
342:
343: 0.5.8:
344: fixed strchr to be able to look for 0.
345:
346: added _llseek, ftruncate, getpgid, getresgid, getresuid, getsid,
347: memccpy, memmove, mprotect, setregid, setresgid, setresuid, setreuid,
348: strncpy, swapon, truncate, strtok, strtok_r, strspn, strcspn (all
349: contributed by Olaf Dreesen)
350:
351: added execvp, getcwd
352:
353: fixed __xmknod
354:
355: Note: can it be that the ftw interface really is so broken that I
356: cannot implement it without having to implement some searching data
357: structure? I included an experimental ftw implementation that will
358: not follow symlinks.
359:
360: 0.5.7:
361: oops, if_nametoindex was broken!
362:
363: 0.5.6:
364: ported to ppc-linux.
365: fixed i386 sigsetjmp (I mistyped the function name)
366: included sigjmp.c
367:
368: 0.5.5:
369: added wait, sys_siglist, longjmp/setjmp/sigsetjmp for i386 and sparc
370: actually, it wasn't fork that wasn't working for sparc, it was pipe.
371: I fixed it now.
372: dietlibc/SPARC assumes -msupersparc (does not provide div, mul, etc)
373: fixed readdir to use getdents and not the intel inline asm
374:
375: 0.5.4:
376: ported to sparclinux (sparc32 only). Beware: does not work yet.
377: added raise, abort, readlink, strcat, geteuid, geteuid, wait3, access
378:
379: 0.5.3:
380: added tcsetattr and getenv to compile e3
381:
382: 0.5.2:
383: added vhangup, tcgetattr, isatty and memcmp for fgetty
384: (http://www.fefe.de/fgetty/)
385: added localtime from uC-libc.
386:
387: 0.5.1:
388: stat and friends actually work now. Yuck! Another case of
389: translation between kernel and userland. Why can't the kernel
390: people and the libc people simply agree on a standard?
391: removed many unnecessary includes to speed up compilation.
392: added -fomit-frame-pointer and i386 compilation target to reduce code size.
393: inlined socketcall to reduce code size.
394: "load" and "compile" are examples for djb code.
395:
396: 0.5:
397: Olaf Dreesen contributed a much smaller implementation of malloc and friends.
398: each object file is now treated with "strip -x -R .note -R .comment"
399: split each system call into a separate object file
400: added ntohs, htons, alarm, if_indextoname and if_nametoindex
401:
402: 0.4.1:
403: oops, the strchr implementation was wrong. Thanks Jens Laas!
404:
405: 0.4:
406: copied opendir and friends from uC-libc.
407: added getservent and getservby* in gerservent.c
408:
409: 0.3:
410: remove readdir system call
411: added getdents system call
412: added getpwnam/getpwuid implementation in getpwnam.c
413: added a few string functions (in str*.c)
414:
LinuxTV legacy CVS <linuxtv.org/cvs>