Annotation of dietlibc/CHANGES, revision 1.51
1.34 fefe 1: 0.7.3:
2: vsnprintf(0,...) estimated the length of strings with a padding width
3: incorrectly.
1.35 fefe 4: new reboot() from Olaf.
1.36 fefe 5: added prototypes for getsockopt and setsockopt.
1.39 fefe 6: added experimental sem* and shm*
1.38 fefe 7: define EXIT_SUCCESS and EXIT_FAILURE
1.40 fefe 8: added setvbuf (only changes mode, not buffer)
1.41 fefe 9: added dietwarning.h (woohoo, linker warnings!) and used it to warn of
10: insecurity (system, sprintf) or bloat (printf and stdio).
1.42 fefe 11: added qsort() and bsearch(). Benchmark on 1000 random ints:
12: glibc: 679414 cycles qsort and 3352 cycles bsearch
13: diet libc: 300010 cycles qsort, 553 cycles bsearch
14: I have no idea why glibc is so slow.
1.43 fefe 15: added __isnan (references by libm.a from glibc).
16: added popen() and pclose(). Now the diet libc can link slsh (from
17: S-Lang!)
1.44 fefe 18: corrected usleep. It was more of an msleep until now.
1.45 fefe 19: corrected __dtostr (it looped when trying to format "0.0").
20: several fixes to stdio (thanks to S-Lang's excellent test suite!)
1.46 fefe 21: added ungetc (does it work?)
1.47 fefe 22: added inet_aton (yuck!)
1.50 fefe 23: added gethostbyname, gethostbyaddr, gethostbyname_r and gethostbyaddr_r
1.48 fefe 24: added h_errno and __h_errno_location
25: added inet_ntoa
1.49 fefe 26: added setpgrp and getpgrp, strsignal and execv (thanks, Olaf)
1.51 ! fefe 27: add bcopy, index, rindex macros
! 28: add isgraph and isxdigit
1.34 fefe 29:
1.21 fefe 30: 0.7.2:
31: added several prototypes to already implemented functions and
32: <sys/reboot.h>. Thanks for Abraham for pointing this out.
1.22 fefe 33: added prototypes for process group and dup syscalls. Thanks to
34: Abraham again. He also contributed an implementation for daemon(),
35: an ugly BSD hack. I decided to diversify the lib/ directory a
36: little and now created a libugly/ directory for stuff like system()
37: and daemon() ;-)
38: added libstdio and moved all that stdio crap from lib/ there. The
39: idea is to start separating now what we will separate for shared
40: libraries anyway.
1.26 fefe 41: optimized x86 unified-syscall by Olaf.
1.27 fefe 42: added limits.h, removed __ARCHBITS from endian.h (use __WORDSIZE from
43: limits.h instead)
44: added remove.c and enough cruft to limits.h to make it compile lzo,
45: lzop and bzip2.
1.28 fefe 46: added mktime, asctime_r, asctime and ctime (needed for gzip).
47: Someone will need to implement a reader for the glibc localtime file
1.31 fefe 48: format some day (man tzfile, ugh!) :-(
1.29 fefe 49: added unlink prototype to unistd.h.
50: added gmtime and gmtime_r.
1.31 fefe 51: fixed sparc unified syscall error handling (please upgrade!).
1.30 fefe 52: removed readdir glibc compatibility cruft.
1.31 fefe 53: removed stat glibc compatibility cruft.
1.32 fefe 54: oops, my fread was broken for pipes.
1.33 fefe 55: Olaf found a bug in pipe() for sparc, too.
1.21 fefe 56:
1.9 fefe 57: 0.7.1:
1.11 fefe 58: optimized away the static arrays from __ltostr and __lltostr.
59: optimized away the static arrays (and more!) from strtoul and strtoull.
60: added system (by Olaf).
1.12 fefe 61: added tolower and toupper
1.13 fefe 62: added getopt from Olaf (~600 bytes on x86!)
1.14 fefe 63: added clone for mips (by Olaf).
64: fixed longjmp and setjmp on sparc not to require glibc include files.
1.15 olaf 65: added pause for all (by Olaf).
1.17 fefe 66: added __attribute__ ((__const__)) to ctype.h
1.16 fefe 67: added isupper and islower.
68: added putc and putchar as macros.
1.17 fefe 69: added strcasecmp and strncasecmp (thanks to Abraham vd Merwe for
70: insightful discussions and patches I partially used for the last three)
1.18 fefe 71: added getpriority and setpriority (also by Abraham)
1.19 fefe 72: oops, I broke strtoul in a recent optimization attempt.
1.20 fefe 73: oi, setjmp and longjmp were broken on other platforms, too!
74: Thanks to Sébastien Côté for reporting this.
1.9 fefe 75:
76: 0.7:
1.6 olaf 77: clone support for i386,alpha,arm and sparc by Olaf.
1.1 cvs 78: Initial dynamic loading glue for i386, also by Olaf.
79: We don't have a dynamic loader yet, so this is not yet usable.
80: added readv, sigdelset, sigaltstack, sigfillset, sigismember, usleep,
81: vsprintf, writev, all by Olaf.
1.2 fefe 82: imported readdir fix from Guillaume. Mhh, can the kernel interface
83: really be this broken?
1.3 fefe 84: Found a kludgy way to detect whether the kernel includes define struct
85: stat64 (they also define STAT64_HAS_BROKEN_ST_INO). If this does
86: not work for you, please tell me!
1.4 fefe 87: Removed bogus __seek_types enum, it's a bunch of #defines now.
1.7 fefe 88: Fixed stdio line buffering. The code was already there but I forgot
89: to mark stdio and stdout as line buffered. This should really be
90: done dynamically using isatty, though.
1.8 fefe 91: Removed superfluous "if (1)" in execvp.c
1.1 cvs 92:
93: 0.6.14:
94: *scanf did not append the 0 byte for %s.
95: added creat and changed creat to call open directly instead of open64.
96: did for fread what 0.6.10 did for fwrite.
97: repaired buffered stdio a little (line buffering is still not
98: supported).
99: [All of these bugs have been reported by Guillaume. Thanks!]
100: added __pure__ to strchr and strrchr in <string.h>
101:
102: 0.6.13:
103: included COPYING to make clear that the diet libc is covered by the
104: GNU General Public License (and _not_ the LGPL). That means that
105: you need to obtain a license from me if you want to use the diet
106: libc in a proprietary program that you want to distribute.
107: The x86 unified syscall swapped arguments #4 and #5 and noone
108: noticed until now! Thanks to Guillaume Cottenceau for reporting
109: this!
110: The malloc had a subtle bug with small allocations that could cause
111: segfault. Guillaume reported it, Olaf fixed it.
112:
113: 0.6.12:
114: added assert, statfs and fstatfs
115: added include/net/if.h so fget compiles again.
116:
117: 0.6.11:
118: added putenv.
119:
120: 0.6.10:
121: fixed fwrite. It returned the number of bytes written, not the number
122: of records. Thanks to Albert D. Cahalan for pointing this out.
123: fixed fgets. It returned EOF on empty lines. Thanks, Erik Frey.
124: added brk and sbrk. harold@nb.com.sg made me do it.
125:
126: 0.6.9:
127: fixed execvp not to return on ENOENT
128: did for {open|read|close|seek|tell}dir what I did for stat before.
129: The glibc compatibility part is currently non-functional.
130: fixed white space handling in sscanf " (" would not match " (".
131: fixed printf to accept 'l' flag.
132: fixed getpwuid (ignored last line)
133:
134: 0.6.8:
135: Olaf contributed initial thread-safe syscalls for i386, alpha, sparc,
136: mips and arm. I did the one for ppc. They can be enabled in dietfeatures.h.
137: added strlcpy and strlcat from OpenBSD.
138: added stat64, fstat64 and lstat64.
139: added endian.h and made ending of strlen.c endianness-aware
140: added string routines size tweak to dietfeatures.h. Most of the
141: string routines have been unrolled. This is up to three times as
142: fast but creates up to three times larger code. Now you can disable
143: the unrolling.
144: added sendfile.
145: I also added pread. It works on x86 and sparc but not on ppc and mips.
146: I have no idea what's going on. strace disagrees with the kernel.
147: Can anyone help? I asked the Linux kernel mailing list for help, too.
148: added a few aliases of the type __libc_open for open. nm on
149: libpthreads.so indicates that we will need them.
150:
151: 0.6.7:
152: the sources now compile without warnings with -Wall.
153: printf now prints "(null)" when %s is passed NULL. This can be
154: removed with WANT_NULL_PRINTF in dietfeatures.h
155: added vfprintf, execl, ttyname.
156: stat, lstat and fstat are now #defined to __dietstat, __dietlstat and
157: __dietfstat so they can use the normal kernel struct stat.
158: stat, lstat and fstat are now C wrappers that convert to the glibc
159: struct stat. So we can avoid that overhead for programs that use
160: dietlibc headers.
161:
162: 0.6.6:
163: changed the Makefiles so you can now set CFLAGS on the command line
164: added memccmp (analogous to memccpy) and strncmp.
165: Olaf fixed another bug in the sparc unified syscall.
166: Paul Clifford contributed a C version of his strlen.S that is much
167: more efficient than the previous strlen on all platforms! It also
168: looks like technology from Roswell. ;-}
169: added a "real" stdio. Well, almost. Please contribute!
170: added dietfeatures.h so you can remove features you don't need.
171: removed debug code from vsnprintf that changed \0 to ' ' (argh!).
172: stdio now works with simple programs and minigzip from libz.
173: stdio uses some major trickery to avoid linking the stdio, stderr and
174: stdout (and reserving space for them) when they are not used.
175:
176: 0.6.5:
177: Red Plait <redplait@ixcelerator.com> found several bugs in diet libc,
178: most of them bugs in the new header files, but also missing sigset
179: functions.
180: Paul Clifford contributed an assembly strlen.S for ARM and fixed
181: several bugs.
182: Fixed ppc/setjmp and mips/pipe.
183: Olaf contributed a new sparc unified syscall.
184:
185: 0.6.4:
186: declared is* static inline in getservent.
187: added assert_fail, strtoul, isalpha, isdigit, isalnum, isascii.
188: changed strlen to return 0 when passed NULL.
189: new, much smaller unified syscall for MIPS. Thanks to Olaf the Mad
190: Scientist who actually implemented this without access to a MIPS
191: box, just from reading the architecture manual. And it worked out
192: of the box.
193: started a set of system includes, mainly so that I can use lcc and
194: my alpha-linux cross compiler (which is unable to cross-compile
195: glibc). diet libc can now be compiled without any include files
196: from a normal libc. The includes do declare more than diet libc
197: currently delivers and they are still far from complete for real
198: applications.
199: I will only add assembly versions that are smaller _and_ faster than
200: the C version. Larger routines are only accepted if they are called
201: very often and are substantially faster.
202: added i386 assembly strchr, which is smaller and faster than the old
203: version. It is, however, larger than the version contributed by
204: proton (thanks, anyway).
205: added i386 strlen (31% faster, 14% smaller)
206: "make t" will now create a map file called "mapfile".
207:
208: 0.6.3:
209: added sys_errlist, strerror and perror
210: added isblank
211: added atol
212:
213: 0.6.2:
214: mmap for ARM didn't compile. Thanks, Paul!
215:
216: 0.6.1:
217: split mmap into the architecture specific subdirectories.
218: getenv now copes with environ==NULL, thanks Paul Clifford.
219: Paul also contributed a smaller ARM startup code.
220:
221: 0.6:
222: strcat returned the wrong result. Thanks, Dietz Pröpper.
223: strtod now understands a negative exponent (oops, thanks Bertram Barth)
224: Port to arm-linux-gnu, but on the Netwinder I use for testing the
225: __dtostr does not work (I have no idea why!)
226: The MIPS port now uses (much smaller) non-PIC code. That means that
227: the applications you link against diet libc also have to be compiled
228: non-PIC. I suggest copying the CFLAGS from the diet libc Makefile.
229: Thanks to Ralf Bächle for helping me with this!
230: I made subdirectories for the architectures and use VPATH to override
231: VPATH so that make finds the source file automatically. That should
232: simplify the sources greatly.
233: Olaf Dreesen contributed Alpha support including setjmp and longjmp!
234: ARM and MIPS now also have setjmp and longjmp
235: "compile" and "load" are now make targets. Use them for djb programs.
236: added contrib/elftrunc.c which will remove unnecessary ELF headers.
237: Again, contributed from Olaf. Great work!
238:
239: 0.5.12:
240: printf also does signed numbers.
241: If you don't use atexit, dietlibc now does not link exit, only _exit.
242: *printf now correctly returns the number of bytes written.
243: Olaf contributed experimental sscanf and vsscanf implementations.
244: If passed NULL as buffer, snprintf will not write anything but still
245: return the number of bytes it would have written.
246: Initial MIPS port! (Oh, the agony!)
247: No setjmp and longjmp support yet! Please contribute!
248: I even implemented unified syscalls for MIPS. Still, MIPS code is
249: almost twice the size of SPARC code. If anyone knows why: please
250: tell me!
251:
252: 0.5.11:
253: I implemented new unified syscalls for x86, sparc and ppc, this time based
254: on .s files and not .c files, and I moved the syscalls into
255: subdirectores to clean the dietlibc sources up a little.
256: A binary consisting of printf("%s is %d\n","olaf",23) is now
257: 2864 bytes on sparc
258: 2488 bytes on intel
259: I kludgily implemented fprintf(stdout,... and fprintf(stderr,...
260: to make a few more applications work.
261: Olaf Dreesen also implemented some unified syscall stuff for x86.
262: He discovered that with our unified syscall interface it actually
263: costs <10 bytes total to make all system calls thread safe!
264: I implemented atexit() (can register up to 4 callbacks).
265: I implemented strtod and __dtostr (the opposite). Now we can add
266: floating point support to vsnprintf and sscanf!
267:
268: 0.5.10:
269: I actually saw that I can not only merge the errno handling code of
270: the system calls on x86, I can also merge the rest (including the
271: arguments) except for setting the system call number. All those
272: system calls are now just a jump to a unified system call handler.
273: I got rid of x86openclose again. The savings are substantial:
274: chown from embutils went from 7664 to 7184 bytes!
275: If I'd move the system call wrappers to assembly language, I could
276: even reduce the alignment (does not matter since it's just a jump
277: anyway) and get rid of the "ret" behind each jump (one byte per
278: system call!)
279:
280: 0.5.9:
281: added memchr, strpbrk, strstr, strtol, isspace (hehe)
282: fixed strdup
283: fixed return values for strcpy and strcat (thanks to Norbert Berzen)
284: Olaf Dreesen contributed a strtol and initial {sn|vsn|}printf implementation
285: (no signed integers, only strings and unsigned integers (but
286: supporting octal, hex and decimal). Thanks, Olaf! (by the way:
287: that code must be wonderful, I don't understand it at all *bg*)
288: Olaf's printf does understand padding like in "%08d" and "%8d" and
289: automatically pads pointers with '0'.
290: Rewrote the x86 start code, old: 22 instructions, new: 12 instructions.
291: Thanks to proton for inspiration on this.
292: The new x86 start code also does not reference exit any more, thus saving
293: 64 bytes for executables that don't call exit explicitly.
294: [insert maniacal laughter] EVERY BYTE COUNTS!1!! ;-)
295: Moved errno and environ to start.S, saving no byte binary size but
296: speeding up compilation and slightly shortening dietlibc.a ;-)
297: For x86: joined open and close into one assembler file, sharing the
298: errno handling. This is a feasibility test and it actually saves a
299: few bytes. I think I will reimplement the _syscall[1-6] macros on
300: all platforms now to share the errno handling code for them.
301:
302: 0.5.8:
303: fixed strchr to be able to look for 0.
304:
305: added _llseek, ftruncate, getpgid, getresgid, getresuid, getsid,
306: memccpy, memmove, mprotect, setregid, setresgid, setresuid, setreuid,
307: strncpy, swapon, truncate, strtok, strtok_r, strspn, strcspn (all
308: contributed by Olaf Dreesen)
309:
310: added execvp, getcwd
311:
312: fixed __xmknod
313:
314: Note: can it be that the ftw interface really is so broken that I
315: cannot implement it without having to implement some searching data
316: structure? I included an experimental ftw implementation that will
317: not follow symlinks.
318:
319: 0.5.7:
320: oops, if_nametoindex was broken!
321:
322: 0.5.6:
323: ported to ppc-linux.
324: fixed i386 sigsetjmp (I mistyped the function name)
325: included sigjmp.c
326:
327: 0.5.5:
328: added wait, sys_siglist, longjmp/setjmp/sigsetjmp for i386 and sparc
329: actually, it wasn't fork that wasn't working for sparc, it was pipe.
330: I fixed it now.
331: dietlibc/SPARC assumes -msupersparc (does not provide div, mul, etc)
332: fixed readdir to use getdents and not the intel inline asm
333:
334: 0.5.4:
335: ported to sparclinux (sparc32 only). Beware: does not work yet.
336: added raise, abort, readlink, strcat, geteuid, geteuid, wait3, access
337:
338: 0.5.3:
339: added tcsetattr and getenv to compile e3
340:
341: 0.5.2:
342: added vhangup, tcgetattr, isatty and memcmp for fgetty
343: (http://www.fefe.de/fgetty/)
344: added localtime from uC-libc.
345:
346: 0.5.1:
347: stat and friends actually work now. Yuck! Another case of
348: translation between kernel and userland. Why can't the kernel
349: people and the libc people simply agree on a standard?
350: removed many unnecessary includes to speed up compilation.
351: added -fomit-frame-pointer and i386 compilation target to reduce code size.
352: inlined socketcall to reduce code size.
353: "load" and "compile" are examples for djb code.
354:
355: 0.5:
356: Olaf Dreesen contributed a much smaller implementation of malloc and friends.
357: each object file is now treated with "strip -x -R .note -R .comment"
358: split each system call into a separate object file
359: added ntohs, htons, alarm, if_indextoname and if_nametoindex
360:
361: 0.4.1:
362: oops, the strchr implementation was wrong. Thanks Jens Laas!
363:
364: 0.4:
365: copied opendir and friends from uC-libc.
366: added getservent and getservby* in gerservent.c
367:
368: 0.3:
369: remove readdir system call
370: added getdents system call
371: added getpwnam/getpwuid implementation in getpwnam.c
372: added a few string functions (in str*.c)
373:
LinuxTV legacy CVS <linuxtv.org/cvs>