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