Annotation of dietlibc/CHANGES, revision 1.37
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.
1.37 ! fefe 6: added experimental sem*
1.34 fefe 7:
1.21 fefe 8: 0.7.2:
9: added several prototypes to already implemented functions and
10: <sys/reboot.h>. Thanks for Abraham for pointing this out.
1.22 fefe 11: added prototypes for process group and dup syscalls. Thanks to
12: Abraham again. He also contributed an implementation for daemon(),
13: an ugly BSD hack. I decided to diversify the lib/ directory a
14: little and now created a libugly/ directory for stuff like system()
15: and daemon() ;-)
16: added libstdio and moved all that stdio crap from lib/ there. The
17: idea is to start separating now what we will separate for shared
18: libraries anyway.
1.26 fefe 19: optimized x86 unified-syscall by Olaf.
1.27 fefe 20: added limits.h, removed __ARCHBITS from endian.h (use __WORDSIZE from
21: limits.h instead)
22: added remove.c and enough cruft to limits.h to make it compile lzo,
23: lzop and bzip2.
1.28 fefe 24: added mktime, asctime_r, asctime and ctime (needed for gzip).
25: Someone will need to implement a reader for the glibc localtime file
1.31 fefe 26: format some day (man tzfile, ugh!) :-(
1.29 fefe 27: added unlink prototype to unistd.h.
28: added gmtime and gmtime_r.
1.31 fefe 29: fixed sparc unified syscall error handling (please upgrade!).
1.30 fefe 30: removed readdir glibc compatibility cruft.
1.31 fefe 31: removed stat glibc compatibility cruft.
1.32 fefe 32: oops, my fread was broken for pipes.
1.33 fefe 33: Olaf found a bug in pipe() for sparc, too.
1.21 fefe 34:
1.9 fefe 35: 0.7.1:
1.11 fefe 36: optimized away the static arrays from __ltostr and __lltostr.
37: optimized away the static arrays (and more!) from strtoul and strtoull.
38: added system (by Olaf).
1.12 fefe 39: added tolower and toupper
1.13 fefe 40: added getopt from Olaf (~600 bytes on x86!)
1.14 fefe 41: added clone for mips (by Olaf).
42: fixed longjmp and setjmp on sparc not to require glibc include files.
1.15 olaf 43: added pause for all (by Olaf).
1.17 fefe 44: added __attribute__ ((__const__)) to ctype.h
1.16 fefe 45: added isupper and islower.
46: added putc and putchar as macros.
1.17 fefe 47: added strcasecmp and strncasecmp (thanks to Abraham vd Merwe for
48: insightful discussions and patches I partially used for the last three)
1.18 fefe 49: added getpriority and setpriority (also by Abraham)
1.19 fefe 50: oops, I broke strtoul in a recent optimization attempt.
1.20 fefe 51: oi, setjmp and longjmp were broken on other platforms, too!
52: Thanks to Sébastien Côté for reporting this.
1.9 fefe 53:
54: 0.7:
1.6 olaf 55: clone support for i386,alpha,arm and sparc by Olaf.
1.1 cvs 56: Initial dynamic loading glue for i386, also by Olaf.
57: We don't have a dynamic loader yet, so this is not yet usable.
58: added readv, sigdelset, sigaltstack, sigfillset, sigismember, usleep,
59: vsprintf, writev, all by Olaf.
1.2 fefe 60: imported readdir fix from Guillaume. Mhh, can the kernel interface
61: really be this broken?
1.3 fefe 62: Found a kludgy way to detect whether the kernel includes define struct
63: stat64 (they also define STAT64_HAS_BROKEN_ST_INO). If this does
64: not work for you, please tell me!
1.4 fefe 65: Removed bogus __seek_types enum, it's a bunch of #defines now.
1.7 fefe 66: Fixed stdio line buffering. The code was already there but I forgot
67: to mark stdio and stdout as line buffered. This should really be
68: done dynamically using isatty, though.
1.8 fefe 69: Removed superfluous "if (1)" in execvp.c
1.1 cvs 70:
71: 0.6.14:
72: *scanf did not append the 0 byte for %s.
73: added creat and changed creat to call open directly instead of open64.
74: did for fread what 0.6.10 did for fwrite.
75: repaired buffered stdio a little (line buffering is still not
76: supported).
77: [All of these bugs have been reported by Guillaume. Thanks!]
78: added __pure__ to strchr and strrchr in <string.h>
79:
80: 0.6.13:
81: included COPYING to make clear that the diet libc is covered by the
82: GNU General Public License (and _not_ the LGPL). That means that
83: you need to obtain a license from me if you want to use the diet
84: libc in a proprietary program that you want to distribute.
85: The x86 unified syscall swapped arguments #4 and #5 and noone
86: noticed until now! Thanks to Guillaume Cottenceau for reporting
87: this!
88: The malloc had a subtle bug with small allocations that could cause
89: segfault. Guillaume reported it, Olaf fixed it.
90:
91: 0.6.12:
92: added assert, statfs and fstatfs
93: added include/net/if.h so fget compiles again.
94:
95: 0.6.11:
96: added putenv.
97:
98: 0.6.10:
99: fixed fwrite. It returned the number of bytes written, not the number
100: of records. Thanks to Albert D. Cahalan for pointing this out.
101: fixed fgets. It returned EOF on empty lines. Thanks, Erik Frey.
102: added brk and sbrk. harold@nb.com.sg made me do it.
103:
104: 0.6.9:
105: fixed execvp not to return on ENOENT
106: did for {open|read|close|seek|tell}dir what I did for stat before.
107: The glibc compatibility part is currently non-functional.
108: fixed white space handling in sscanf " (" would not match " (".
109: fixed printf to accept 'l' flag.
110: fixed getpwuid (ignored last line)
111:
112: 0.6.8:
113: Olaf contributed initial thread-safe syscalls for i386, alpha, sparc,
114: mips and arm. I did the one for ppc. They can be enabled in dietfeatures.h.
115: added strlcpy and strlcat from OpenBSD.
116: added stat64, fstat64 and lstat64.
117: added endian.h and made ending of strlen.c endianness-aware
118: added string routines size tweak to dietfeatures.h. Most of the
119: string routines have been unrolled. This is up to three times as
120: fast but creates up to three times larger code. Now you can disable
121: the unrolling.
122: added sendfile.
123: I also added pread. It works on x86 and sparc but not on ppc and mips.
124: I have no idea what's going on. strace disagrees with the kernel.
125: Can anyone help? I asked the Linux kernel mailing list for help, too.
126: added a few aliases of the type __libc_open for open. nm on
127: libpthreads.so indicates that we will need them.
128:
129: 0.6.7:
130: the sources now compile without warnings with -Wall.
131: printf now prints "(null)" when %s is passed NULL. This can be
132: removed with WANT_NULL_PRINTF in dietfeatures.h
133: added vfprintf, execl, ttyname.
134: stat, lstat and fstat are now #defined to __dietstat, __dietlstat and
135: __dietfstat so they can use the normal kernel struct stat.
136: stat, lstat and fstat are now C wrappers that convert to the glibc
137: struct stat. So we can avoid that overhead for programs that use
138: dietlibc headers.
139:
140: 0.6.6:
141: changed the Makefiles so you can now set CFLAGS on the command line
142: added memccmp (analogous to memccpy) and strncmp.
143: Olaf fixed another bug in the sparc unified syscall.
144: Paul Clifford contributed a C version of his strlen.S that is much
145: more efficient than the previous strlen on all platforms! It also
146: looks like technology from Roswell. ;-}
147: added a "real" stdio. Well, almost. Please contribute!
148: added dietfeatures.h so you can remove features you don't need.
149: removed debug code from vsnprintf that changed \0 to ' ' (argh!).
150: stdio now works with simple programs and minigzip from libz.
151: stdio uses some major trickery to avoid linking the stdio, stderr and
152: stdout (and reserving space for them) when they are not used.
153:
154: 0.6.5:
155: Red Plait <redplait@ixcelerator.com> found several bugs in diet libc,
156: most of them bugs in the new header files, but also missing sigset
157: functions.
158: Paul Clifford contributed an assembly strlen.S for ARM and fixed
159: several bugs.
160: Fixed ppc/setjmp and mips/pipe.
161: Olaf contributed a new sparc unified syscall.
162:
163: 0.6.4:
164: declared is* static inline in getservent.
165: added assert_fail, strtoul, isalpha, isdigit, isalnum, isascii.
166: changed strlen to return 0 when passed NULL.
167: new, much smaller unified syscall for MIPS. Thanks to Olaf the Mad
168: Scientist who actually implemented this without access to a MIPS
169: box, just from reading the architecture manual. And it worked out
170: of the box.
171: started a set of system includes, mainly so that I can use lcc and
172: my alpha-linux cross compiler (which is unable to cross-compile
173: glibc). diet libc can now be compiled without any include files
174: from a normal libc. The includes do declare more than diet libc
175: currently delivers and they are still far from complete for real
176: applications.
177: I will only add assembly versions that are smaller _and_ faster than
178: the C version. Larger routines are only accepted if they are called
179: very often and are substantially faster.
180: added i386 assembly strchr, which is smaller and faster than the old
181: version. It is, however, larger than the version contributed by
182: proton (thanks, anyway).
183: added i386 strlen (31% faster, 14% smaller)
184: "make t" will now create a map file called "mapfile".
185:
186: 0.6.3:
187: added sys_errlist, strerror and perror
188: added isblank
189: added atol
190:
191: 0.6.2:
192: mmap for ARM didn't compile. Thanks, Paul!
193:
194: 0.6.1:
195: split mmap into the architecture specific subdirectories.
196: getenv now copes with environ==NULL, thanks Paul Clifford.
197: Paul also contributed a smaller ARM startup code.
198:
199: 0.6:
200: strcat returned the wrong result. Thanks, Dietz Pröpper.
201: strtod now understands a negative exponent (oops, thanks Bertram Barth)
202: Port to arm-linux-gnu, but on the Netwinder I use for testing the
203: __dtostr does not work (I have no idea why!)
204: The MIPS port now uses (much smaller) non-PIC code. That means that
205: the applications you link against diet libc also have to be compiled
206: non-PIC. I suggest copying the CFLAGS from the diet libc Makefile.
207: Thanks to Ralf Bächle for helping me with this!
208: I made subdirectories for the architectures and use VPATH to override
209: VPATH so that make finds the source file automatically. That should
210: simplify the sources greatly.
211: Olaf Dreesen contributed Alpha support including setjmp and longjmp!
212: ARM and MIPS now also have setjmp and longjmp
213: "compile" and "load" are now make targets. Use them for djb programs.
214: added contrib/elftrunc.c which will remove unnecessary ELF headers.
215: Again, contributed from Olaf. Great work!
216:
217: 0.5.12:
218: printf also does signed numbers.
219: If you don't use atexit, dietlibc now does not link exit, only _exit.
220: *printf now correctly returns the number of bytes written.
221: Olaf contributed experimental sscanf and vsscanf implementations.
222: If passed NULL as buffer, snprintf will not write anything but still
223: return the number of bytes it would have written.
224: Initial MIPS port! (Oh, the agony!)
225: No setjmp and longjmp support yet! Please contribute!
226: I even implemented unified syscalls for MIPS. Still, MIPS code is
227: almost twice the size of SPARC code. If anyone knows why: please
228: tell me!
229:
230: 0.5.11:
231: I implemented new unified syscalls for x86, sparc and ppc, this time based
232: on .s files and not .c files, and I moved the syscalls into
233: subdirectores to clean the dietlibc sources up a little.
234: A binary consisting of printf("%s is %d\n","olaf",23) is now
235: 2864 bytes on sparc
236: 2488 bytes on intel
237: I kludgily implemented fprintf(stdout,... and fprintf(stderr,...
238: to make a few more applications work.
239: Olaf Dreesen also implemented some unified syscall stuff for x86.
240: He discovered that with our unified syscall interface it actually
241: costs <10 bytes total to make all system calls thread safe!
242: I implemented atexit() (can register up to 4 callbacks).
243: I implemented strtod and __dtostr (the opposite). Now we can add
244: floating point support to vsnprintf and sscanf!
245:
246: 0.5.10:
247: I actually saw that I can not only merge the errno handling code of
248: the system calls on x86, I can also merge the rest (including the
249: arguments) except for setting the system call number. All those
250: system calls are now just a jump to a unified system call handler.
251: I got rid of x86openclose again. The savings are substantial:
252: chown from embutils went from 7664 to 7184 bytes!
253: If I'd move the system call wrappers to assembly language, I could
254: even reduce the alignment (does not matter since it's just a jump
255: anyway) and get rid of the "ret" behind each jump (one byte per
256: system call!)
257:
258: 0.5.9:
259: added memchr, strpbrk, strstr, strtol, isspace (hehe)
260: fixed strdup
261: fixed return values for strcpy and strcat (thanks to Norbert Berzen)
262: Olaf Dreesen contributed a strtol and initial {sn|vsn|}printf implementation
263: (no signed integers, only strings and unsigned integers (but
264: supporting octal, hex and decimal). Thanks, Olaf! (by the way:
265: that code must be wonderful, I don't understand it at all *bg*)
266: Olaf's printf does understand padding like in "%08d" and "%8d" and
267: automatically pads pointers with '0'.
268: Rewrote the x86 start code, old: 22 instructions, new: 12 instructions.
269: Thanks to proton for inspiration on this.
270: The new x86 start code also does not reference exit any more, thus saving
271: 64 bytes for executables that don't call exit explicitly.
272: [insert maniacal laughter] EVERY BYTE COUNTS!1!! ;-)
273: Moved errno and environ to start.S, saving no byte binary size but
274: speeding up compilation and slightly shortening dietlibc.a ;-)
275: For x86: joined open and close into one assembler file, sharing the
276: errno handling. This is a feasibility test and it actually saves a
277: few bytes. I think I will reimplement the _syscall[1-6] macros on
278: all platforms now to share the errno handling code for them.
279:
280: 0.5.8:
281: fixed strchr to be able to look for 0.
282:
283: added _llseek, ftruncate, getpgid, getresgid, getresuid, getsid,
284: memccpy, memmove, mprotect, setregid, setresgid, setresuid, setreuid,
285: strncpy, swapon, truncate, strtok, strtok_r, strspn, strcspn (all
286: contributed by Olaf Dreesen)
287:
288: added execvp, getcwd
289:
290: fixed __xmknod
291:
292: Note: can it be that the ftw interface really is so broken that I
293: cannot implement it without having to implement some searching data
294: structure? I included an experimental ftw implementation that will
295: not follow symlinks.
296:
297: 0.5.7:
298: oops, if_nametoindex was broken!
299:
300: 0.5.6:
301: ported to ppc-linux.
302: fixed i386 sigsetjmp (I mistyped the function name)
303: included sigjmp.c
304:
305: 0.5.5:
306: added wait, sys_siglist, longjmp/setjmp/sigsetjmp for i386 and sparc
307: actually, it wasn't fork that wasn't working for sparc, it was pipe.
308: I fixed it now.
309: dietlibc/SPARC assumes -msupersparc (does not provide div, mul, etc)
310: fixed readdir to use getdents and not the intel inline asm
311:
312: 0.5.4:
313: ported to sparclinux (sparc32 only). Beware: does not work yet.
314: added raise, abort, readlink, strcat, geteuid, geteuid, wait3, access
315:
316: 0.5.3:
317: added tcsetattr and getenv to compile e3
318:
319: 0.5.2:
320: added vhangup, tcgetattr, isatty and memcmp for fgetty
321: (http://www.fefe.de/fgetty/)
322: added localtime from uC-libc.
323:
324: 0.5.1:
325: stat and friends actually work now. Yuck! Another case of
326: translation between kernel and userland. Why can't the kernel
327: people and the libc people simply agree on a standard?
328: removed many unnecessary includes to speed up compilation.
329: added -fomit-frame-pointer and i386 compilation target to reduce code size.
330: inlined socketcall to reduce code size.
331: "load" and "compile" are examples for djb code.
332:
333: 0.5:
334: Olaf Dreesen contributed a much smaller implementation of malloc and friends.
335: each object file is now treated with "strip -x -R .note -R .comment"
336: split each system call into a separate object file
337: added ntohs, htons, alarm, if_indextoname and if_nametoindex
338:
339: 0.4.1:
340: oops, the strchr implementation was wrong. Thanks Jens Laas!
341:
342: 0.4:
343: copied opendir and friends from uC-libc.
344: added getservent and getservby* in gerservent.c
345:
346: 0.3:
347: remove readdir system call
348: added getdents system call
349: added getpwnam/getpwuid implementation in getpwnam.c
350: added a few string functions (in str*.c)
351:
LinuxTV legacy CVS <linuxtv.org/cvs>