Annotation of dietlibc/CHANGES, revision 1.69
1.65 fefe 1: 0.8.1:
2: fixed initgroups (oops)
1.66 fefe 3: Cleanup appending file open for stdio (thanks, James Antill)
4: imported vfprintf fix from James Antill.
1.67 fefe 5: add fnmatch
6: change index, rindex and bcmp as weak ELF symbols (were #defined before)
1.68 fefe 7: Olaf added character classes to *scanf.
1.69 ! fefe 8: Put object and library files into a separate directory (bin-$ARCH).
! 9: Add "cross" make target to make all architectures.
! 10: Add "diet" program and man page.
1.65 fefe 11:
1.64 fefe 12: 0.8:
1.34 fefe 13: vsnprintf(0,...) estimated the length of strings with a padding width
14: incorrectly.
1.35 fefe 15: new reboot() from Olaf.
1.36 fefe 16: added prototypes for getsockopt and setsockopt.
1.39 fefe 17: added experimental sem* and shm*
1.38 fefe 18: define EXIT_SUCCESS and EXIT_FAILURE
1.40 fefe 19: added setvbuf (only changes mode, not buffer)
1.41 fefe 20: added dietwarning.h (woohoo, linker warnings!) and used it to warn of
21: insecurity (system, sprintf) or bloat (printf and stdio).
1.42 fefe 22: added qsort() and bsearch(). Benchmark on 1000 random ints:
23: glibc: 679414 cycles qsort and 3352 cycles bsearch
24: diet libc: 300010 cycles qsort, 553 cycles bsearch
25: I have no idea why glibc is so slow.
1.43 fefe 26: added __isnan (references by libm.a from glibc).
27: added popen() and pclose(). Now the diet libc can link slsh (from
28: S-Lang!)
1.44 fefe 29: corrected usleep. It was more of an msleep until now.
1.45 fefe 30: corrected __dtostr (it looped when trying to format "0.0").
31: several fixes to stdio (thanks to S-Lang's excellent test suite!)
1.46 fefe 32: added ungetc (does it work?)
1.47 fefe 33: added inet_aton (yuck!)
1.50 fefe 34: added gethostbyname, gethostbyaddr, gethostbyname_r and gethostbyaddr_r
1.48 fefe 35: added h_errno and __h_errno_location
36: added inet_ntoa
1.49 fefe 37: added setpgrp and getpgrp, strsignal and execv (thanks, Olaf)
1.51 fefe 38: add bcopy, index, rindex macros
39: add isgraph and isxdigit
1.52 fefe 40: add strftime (ugh!!!)
1.53 olaf 41: Olaf fixed strftime, gmtime_r leap year bug.
1.54 fefe 42: Jeff Garzik contributed a spec file. I put it in contrib and hope
43: everbody knows what to do with it...
44: Olaf checked in a ton of libpthread stuff.
1.55 fefe 45: added mkstemp using /dev/urandom (harder to guess than getpid()).
1.56 fefe 46: add large file backwards compatibility support (so gzip will run on
47: kernel 2.2 and 2.0)
1.57 fefe 48: added shutdown
1.58 fefe 49: added nice
50: added pseudo-locale support for is* (they are now weak aliases,
51: isupper -> __isupper_ascii etc). So, for example, liblatin1 can
52: overwrite the default is* functions.
1.59 fefe 53: added liblatin1.a
1.60 fefe 54: added tzfile implementation (does it work for anyone else?)
1.61 fefe 55: added syslog(3) (thanks to Rene Müller!)
1.62 fefe 56: Jeff Garzik contributed getgr*, getpw* and getsp* in one tiny package.
57: Thanks!
1.63 fefe 58: merged duplicate code in execv and execvp into exec_lib.o
1.34 fefe 59:
1.21 fefe 60: 0.7.2:
61: added several prototypes to already implemented functions and
62: <sys/reboot.h>. Thanks for Abraham for pointing this out.
1.22 fefe 63: added prototypes for process group and dup syscalls. Thanks to
64: Abraham again. He also contributed an implementation for daemon(),
65: an ugly BSD hack. I decided to diversify the lib/ directory a
66: little and now created a libugly/ directory for stuff like system()
67: and daemon() ;-)
68: added libstdio and moved all that stdio crap from lib/ there. The
69: idea is to start separating now what we will separate for shared
70: libraries anyway.
1.26 fefe 71: optimized x86 unified-syscall by Olaf.
1.27 fefe 72: added limits.h, removed __ARCHBITS from endian.h (use __WORDSIZE from
73: limits.h instead)
74: added remove.c and enough cruft to limits.h to make it compile lzo,
75: lzop and bzip2.
1.28 fefe 76: added mktime, asctime_r, asctime and ctime (needed for gzip).
77: Someone will need to implement a reader for the glibc localtime file
1.31 fefe 78: format some day (man tzfile, ugh!) :-(
1.29 fefe 79: added unlink prototype to unistd.h.
80: added gmtime and gmtime_r.
1.31 fefe 81: fixed sparc unified syscall error handling (please upgrade!).
1.30 fefe 82: removed readdir glibc compatibility cruft.
1.31 fefe 83: removed stat glibc compatibility cruft.
1.32 fefe 84: oops, my fread was broken for pipes.
1.33 fefe 85: Olaf found a bug in pipe() for sparc, too.
1.21 fefe 86:
1.9 fefe 87: 0.7.1:
1.11 fefe 88: optimized away the static arrays from __ltostr and __lltostr.
89: optimized away the static arrays (and more!) from strtoul and strtoull.
90: added system (by Olaf).
1.12 fefe 91: added tolower and toupper
1.13 fefe 92: added getopt from Olaf (~600 bytes on x86!)
1.14 fefe 93: added clone for mips (by Olaf).
94: fixed longjmp and setjmp on sparc not to require glibc include files.
1.15 olaf 95: added pause for all (by Olaf).
1.17 fefe 96: added __attribute__ ((__const__)) to ctype.h
1.16 fefe 97: added isupper and islower.
98: added putc and putchar as macros.
1.17 fefe 99: added strcasecmp and strncasecmp (thanks to Abraham vd Merwe for
100: insightful discussions and patches I partially used for the last three)
1.18 fefe 101: added getpriority and setpriority (also by Abraham)
1.19 fefe 102: oops, I broke strtoul in a recent optimization attempt.
1.20 fefe 103: oi, setjmp and longjmp were broken on other platforms, too!
104: Thanks to Sébastien Côté for reporting this.
1.9 fefe 105:
106: 0.7:
1.6 olaf 107: clone support for i386,alpha,arm and sparc by Olaf.
1.1 cvs 108: Initial dynamic loading glue for i386, also by Olaf.
109: We don't have a dynamic loader yet, so this is not yet usable.
110: added readv, sigdelset, sigaltstack, sigfillset, sigismember, usleep,
111: vsprintf, writev, all by Olaf.
1.2 fefe 112: imported readdir fix from Guillaume. Mhh, can the kernel interface
113: really be this broken?
1.3 fefe 114: Found a kludgy way to detect whether the kernel includes define struct
115: stat64 (they also define STAT64_HAS_BROKEN_ST_INO). If this does
116: not work for you, please tell me!
1.4 fefe 117: Removed bogus __seek_types enum, it's a bunch of #defines now.
1.7 fefe 118: Fixed stdio line buffering. The code was already there but I forgot
119: to mark stdio and stdout as line buffered. This should really be
120: done dynamically using isatty, though.
1.8 fefe 121: Removed superfluous "if (1)" in execvp.c
1.1 cvs 122:
123: 0.6.14:
124: *scanf did not append the 0 byte for %s.
125: added creat and changed creat to call open directly instead of open64.
126: did for fread what 0.6.10 did for fwrite.
127: repaired buffered stdio a little (line buffering is still not
128: supported).
129: [All of these bugs have been reported by Guillaume. Thanks!]
130: added __pure__ to strchr and strrchr in <string.h>
131:
132: 0.6.13:
133: included COPYING to make clear that the diet libc is covered by the
134: GNU General Public License (and _not_ the LGPL). That means that
135: you need to obtain a license from me if you want to use the diet
136: libc in a proprietary program that you want to distribute.
137: The x86 unified syscall swapped arguments #4 and #5 and noone
138: noticed until now! Thanks to Guillaume Cottenceau for reporting
139: this!
140: The malloc had a subtle bug with small allocations that could cause
141: segfault. Guillaume reported it, Olaf fixed it.
142:
143: 0.6.12:
144: added assert, statfs and fstatfs
145: added include/net/if.h so fget compiles again.
146:
147: 0.6.11:
148: added putenv.
149:
150: 0.6.10:
151: fixed fwrite. It returned the number of bytes written, not the number
152: of records. Thanks to Albert D. Cahalan for pointing this out.
153: fixed fgets. It returned EOF on empty lines. Thanks, Erik Frey.
154: added brk and sbrk. harold@nb.com.sg made me do it.
155:
156: 0.6.9:
157: fixed execvp not to return on ENOENT
158: did for {open|read|close|seek|tell}dir what I did for stat before.
159: The glibc compatibility part is currently non-functional.
160: fixed white space handling in sscanf " (" would not match " (".
161: fixed printf to accept 'l' flag.
162: fixed getpwuid (ignored last line)
163:
164: 0.6.8:
165: Olaf contributed initial thread-safe syscalls for i386, alpha, sparc,
166: mips and arm. I did the one for ppc. They can be enabled in dietfeatures.h.
167: added strlcpy and strlcat from OpenBSD.
168: added stat64, fstat64 and lstat64.
169: added endian.h and made ending of strlen.c endianness-aware
170: added string routines size tweak to dietfeatures.h. Most of the
171: string routines have been unrolled. This is up to three times as
172: fast but creates up to three times larger code. Now you can disable
173: the unrolling.
174: added sendfile.
175: I also added pread. It works on x86 and sparc but not on ppc and mips.
176: I have no idea what's going on. strace disagrees with the kernel.
177: Can anyone help? I asked the Linux kernel mailing list for help, too.
178: added a few aliases of the type __libc_open for open. nm on
179: libpthreads.so indicates that we will need them.
180:
181: 0.6.7:
182: the sources now compile without warnings with -Wall.
183: printf now prints "(null)" when %s is passed NULL. This can be
184: removed with WANT_NULL_PRINTF in dietfeatures.h
185: added vfprintf, execl, ttyname.
186: stat, lstat and fstat are now #defined to __dietstat, __dietlstat and
187: __dietfstat so they can use the normal kernel struct stat.
188: stat, lstat and fstat are now C wrappers that convert to the glibc
189: struct stat. So we can avoid that overhead for programs that use
190: dietlibc headers.
191:
192: 0.6.6:
193: changed the Makefiles so you can now set CFLAGS on the command line
194: added memccmp (analogous to memccpy) and strncmp.
195: Olaf fixed another bug in the sparc unified syscall.
196: Paul Clifford contributed a C version of his strlen.S that is much
197: more efficient than the previous strlen on all platforms! It also
198: looks like technology from Roswell. ;-}
199: added a "real" stdio. Well, almost. Please contribute!
200: added dietfeatures.h so you can remove features you don't need.
201: removed debug code from vsnprintf that changed \0 to ' ' (argh!).
202: stdio now works with simple programs and minigzip from libz.
203: stdio uses some major trickery to avoid linking the stdio, stderr and
204: stdout (and reserving space for them) when they are not used.
205:
206: 0.6.5:
207: Red Plait <redplait@ixcelerator.com> found several bugs in diet libc,
208: most of them bugs in the new header files, but also missing sigset
209: functions.
210: Paul Clifford contributed an assembly strlen.S for ARM and fixed
211: several bugs.
212: Fixed ppc/setjmp and mips/pipe.
213: Olaf contributed a new sparc unified syscall.
214:
215: 0.6.4:
216: declared is* static inline in getservent.
217: added assert_fail, strtoul, isalpha, isdigit, isalnum, isascii.
218: changed strlen to return 0 when passed NULL.
219: new, much smaller unified syscall for MIPS. Thanks to Olaf the Mad
220: Scientist who actually implemented this without access to a MIPS
221: box, just from reading the architecture manual. And it worked out
222: of the box.
223: started a set of system includes, mainly so that I can use lcc and
224: my alpha-linux cross compiler (which is unable to cross-compile
225: glibc). diet libc can now be compiled without any include files
226: from a normal libc. The includes do declare more than diet libc
227: currently delivers and they are still far from complete for real
228: applications.
229: I will only add assembly versions that are smaller _and_ faster than
230: the C version. Larger routines are only accepted if they are called
231: very often and are substantially faster.
232: added i386 assembly strchr, which is smaller and faster than the old
233: version. It is, however, larger than the version contributed by
234: proton (thanks, anyway).
235: added i386 strlen (31% faster, 14% smaller)
236: "make t" will now create a map file called "mapfile".
237:
238: 0.6.3:
239: added sys_errlist, strerror and perror
240: added isblank
241: added atol
242:
243: 0.6.2:
244: mmap for ARM didn't compile. Thanks, Paul!
245:
246: 0.6.1:
247: split mmap into the architecture specific subdirectories.
248: getenv now copes with environ==NULL, thanks Paul Clifford.
249: Paul also contributed a smaller ARM startup code.
250:
251: 0.6:
252: strcat returned the wrong result. Thanks, Dietz Pröpper.
253: strtod now understands a negative exponent (oops, thanks Bertram Barth)
254: Port to arm-linux-gnu, but on the Netwinder I use for testing the
255: __dtostr does not work (I have no idea why!)
256: The MIPS port now uses (much smaller) non-PIC code. That means that
257: the applications you link against diet libc also have to be compiled
258: non-PIC. I suggest copying the CFLAGS from the diet libc Makefile.
259: Thanks to Ralf Bächle for helping me with this!
260: I made subdirectories for the architectures and use VPATH to override
261: VPATH so that make finds the source file automatically. That should
262: simplify the sources greatly.
263: Olaf Dreesen contributed Alpha support including setjmp and longjmp!
264: ARM and MIPS now also have setjmp and longjmp
265: "compile" and "load" are now make targets. Use them for djb programs.
266: added contrib/elftrunc.c which will remove unnecessary ELF headers.
267: Again, contributed from Olaf. Great work!
268:
269: 0.5.12:
270: printf also does signed numbers.
271: If you don't use atexit, dietlibc now does not link exit, only _exit.
272: *printf now correctly returns the number of bytes written.
273: Olaf contributed experimental sscanf and vsscanf implementations.
274: If passed NULL as buffer, snprintf will not write anything but still
275: return the number of bytes it would have written.
276: Initial MIPS port! (Oh, the agony!)
277: No setjmp and longjmp support yet! Please contribute!
278: I even implemented unified syscalls for MIPS. Still, MIPS code is
279: almost twice the size of SPARC code. If anyone knows why: please
280: tell me!
281:
282: 0.5.11:
283: I implemented new unified syscalls for x86, sparc and ppc, this time based
284: on .s files and not .c files, and I moved the syscalls into
285: subdirectores to clean the dietlibc sources up a little.
286: A binary consisting of printf("%s is %d\n","olaf",23) is now
287: 2864 bytes on sparc
288: 2488 bytes on intel
289: I kludgily implemented fprintf(stdout,... and fprintf(stderr,...
290: to make a few more applications work.
291: Olaf Dreesen also implemented some unified syscall stuff for x86.
292: He discovered that with our unified syscall interface it actually
293: costs <10 bytes total to make all system calls thread safe!
294: I implemented atexit() (can register up to 4 callbacks).
295: I implemented strtod and __dtostr (the opposite). Now we can add
296: floating point support to vsnprintf and sscanf!
297:
298: 0.5.10:
299: I actually saw that I can not only merge the errno handling code of
300: the system calls on x86, I can also merge the rest (including the
301: arguments) except for setting the system call number. All those
302: system calls are now just a jump to a unified system call handler.
303: I got rid of x86openclose again. The savings are substantial:
304: chown from embutils went from 7664 to 7184 bytes!
305: If I'd move the system call wrappers to assembly language, I could
306: even reduce the alignment (does not matter since it's just a jump
307: anyway) and get rid of the "ret" behind each jump (one byte per
308: system call!)
309:
310: 0.5.9:
311: added memchr, strpbrk, strstr, strtol, isspace (hehe)
312: fixed strdup
313: fixed return values for strcpy and strcat (thanks to Norbert Berzen)
314: Olaf Dreesen contributed a strtol and initial {sn|vsn|}printf implementation
315: (no signed integers, only strings and unsigned integers (but
316: supporting octal, hex and decimal). Thanks, Olaf! (by the way:
317: that code must be wonderful, I don't understand it at all *bg*)
318: Olaf's printf does understand padding like in "%08d" and "%8d" and
319: automatically pads pointers with '0'.
320: Rewrote the x86 start code, old: 22 instructions, new: 12 instructions.
321: Thanks to proton for inspiration on this.
322: The new x86 start code also does not reference exit any more, thus saving
323: 64 bytes for executables that don't call exit explicitly.
324: [insert maniacal laughter] EVERY BYTE COUNTS!1!! ;-)
325: Moved errno and environ to start.S, saving no byte binary size but
326: speeding up compilation and slightly shortening dietlibc.a ;-)
327: For x86: joined open and close into one assembler file, sharing the
328: errno handling. This is a feasibility test and it actually saves a
329: few bytes. I think I will reimplement the _syscall[1-6] macros on
330: all platforms now to share the errno handling code for them.
331:
332: 0.5.8:
333: fixed strchr to be able to look for 0.
334:
335: added _llseek, ftruncate, getpgid, getresgid, getresuid, getsid,
336: memccpy, memmove, mprotect, setregid, setresgid, setresuid, setreuid,
337: strncpy, swapon, truncate, strtok, strtok_r, strspn, strcspn (all
338: contributed by Olaf Dreesen)
339:
340: added execvp, getcwd
341:
342: fixed __xmknod
343:
344: Note: can it be that the ftw interface really is so broken that I
345: cannot implement it without having to implement some searching data
346: structure? I included an experimental ftw implementation that will
347: not follow symlinks.
348:
349: 0.5.7:
350: oops, if_nametoindex was broken!
351:
352: 0.5.6:
353: ported to ppc-linux.
354: fixed i386 sigsetjmp (I mistyped the function name)
355: included sigjmp.c
356:
357: 0.5.5:
358: added wait, sys_siglist, longjmp/setjmp/sigsetjmp for i386 and sparc
359: actually, it wasn't fork that wasn't working for sparc, it was pipe.
360: I fixed it now.
361: dietlibc/SPARC assumes -msupersparc (does not provide div, mul, etc)
362: fixed readdir to use getdents and not the intel inline asm
363:
364: 0.5.4:
365: ported to sparclinux (sparc32 only). Beware: does not work yet.
366: added raise, abort, readlink, strcat, geteuid, geteuid, wait3, access
367:
368: 0.5.3:
369: added tcsetattr and getenv to compile e3
370:
371: 0.5.2:
372: added vhangup, tcgetattr, isatty and memcmp for fgetty
373: (http://www.fefe.de/fgetty/)
374: added localtime from uC-libc.
375:
376: 0.5.1:
377: stat and friends actually work now. Yuck! Another case of
378: translation between kernel and userland. Why can't the kernel
379: people and the libc people simply agree on a standard?
380: removed many unnecessary includes to speed up compilation.
381: added -fomit-frame-pointer and i386 compilation target to reduce code size.
382: inlined socketcall to reduce code size.
383: "load" and "compile" are examples for djb code.
384:
385: 0.5:
386: Olaf Dreesen contributed a much smaller implementation of malloc and friends.
387: each object file is now treated with "strip -x -R .note -R .comment"
388: split each system call into a separate object file
389: added ntohs, htons, alarm, if_indextoname and if_nametoindex
390:
391: 0.4.1:
392: oops, the strchr implementation was wrong. Thanks Jens Laas!
393:
394: 0.4:
395: copied opendir and friends from uC-libc.
396: added getservent and getservby* in gerservent.c
397:
398: 0.3:
399: remove readdir system call
400: added getdents system call
401: added getpwnam/getpwuid implementation in getpwnam.c
402: added a few string functions (in str*.c)
403:
LinuxTV legacy CVS <linuxtv.org/cvs>