Annotation of dietlibc/CHANGES, revision 1.3
1.1 cvs 1: 0.6.15:
2: clone support for i386 and alpha by Olaf.
3: Initial dynamic loading glue for i386, also by Olaf.
4: We don't have a dynamic loader yet, so this is not yet usable.
5: added readv, sigdelset, sigaltstack, sigfillset, sigismember, usleep,
6: vsprintf, writev, all by Olaf.
1.2 fefe 7: imported readdir fix from Guillaume. Mhh, can the kernel interface
8: really be this broken?
1.3 ! fefe 9: Found a kludgy way to detect whether the kernel includes define struct
! 10: stat64 (they also define STAT64_HAS_BROKEN_ST_INO). If this does
! 11: not work for you, please tell me!
1.1 cvs 12:
13: 0.6.14:
14: *scanf did not append the 0 byte for %s.
15: added creat and changed creat to call open directly instead of open64.
16: did for fread what 0.6.10 did for fwrite.
17: repaired buffered stdio a little (line buffering is still not
18: supported).
19: [All of these bugs have been reported by Guillaume. Thanks!]
20: added __pure__ to strchr and strrchr in <string.h>
21:
22: 0.6.13:
23: included COPYING to make clear that the diet libc is covered by the
24: GNU General Public License (and _not_ the LGPL). That means that
25: you need to obtain a license from me if you want to use the diet
26: libc in a proprietary program that you want to distribute.
27: The x86 unified syscall swapped arguments #4 and #5 and noone
28: noticed until now! Thanks to Guillaume Cottenceau for reporting
29: this!
30: The malloc had a subtle bug with small allocations that could cause
31: segfault. Guillaume reported it, Olaf fixed it.
32:
33: 0.6.12:
34: added assert, statfs and fstatfs
35: added include/net/if.h so fget compiles again.
36:
37: 0.6.11:
38: added putenv.
39:
40: 0.6.10:
41: fixed fwrite. It returned the number of bytes written, not the number
42: of records. Thanks to Albert D. Cahalan for pointing this out.
43: fixed fgets. It returned EOF on empty lines. Thanks, Erik Frey.
44: added brk and sbrk. harold@nb.com.sg made me do it.
45:
46: 0.6.9:
47: fixed execvp not to return on ENOENT
48: did for {open|read|close|seek|tell}dir what I did for stat before.
49: The glibc compatibility part is currently non-functional.
50: fixed white space handling in sscanf " (" would not match " (".
51: fixed printf to accept 'l' flag.
52: fixed getpwuid (ignored last line)
53:
54: 0.6.8:
55: Olaf contributed initial thread-safe syscalls for i386, alpha, sparc,
56: mips and arm. I did the one for ppc. They can be enabled in dietfeatures.h.
57: added strlcpy and strlcat from OpenBSD.
58: added stat64, fstat64 and lstat64.
59: added endian.h and made ending of strlen.c endianness-aware
60: added string routines size tweak to dietfeatures.h. Most of the
61: string routines have been unrolled. This is up to three times as
62: fast but creates up to three times larger code. Now you can disable
63: the unrolling.
64: added sendfile.
65: I also added pread. It works on x86 and sparc but not on ppc and mips.
66: I have no idea what's going on. strace disagrees with the kernel.
67: Can anyone help? I asked the Linux kernel mailing list for help, too.
68: added a few aliases of the type __libc_open for open. nm on
69: libpthreads.so indicates that we will need them.
70:
71: 0.6.7:
72: the sources now compile without warnings with -Wall.
73: printf now prints "(null)" when %s is passed NULL. This can be
74: removed with WANT_NULL_PRINTF in dietfeatures.h
75: added vfprintf, execl, ttyname.
76: stat, lstat and fstat are now #defined to __dietstat, __dietlstat and
77: __dietfstat so they can use the normal kernel struct stat.
78: stat, lstat and fstat are now C wrappers that convert to the glibc
79: struct stat. So we can avoid that overhead for programs that use
80: dietlibc headers.
81:
82: 0.6.6:
83: changed the Makefiles so you can now set CFLAGS on the command line
84: added memccmp (analogous to memccpy) and strncmp.
85: Olaf fixed another bug in the sparc unified syscall.
86: Paul Clifford contributed a C version of his strlen.S that is much
87: more efficient than the previous strlen on all platforms! It also
88: looks like technology from Roswell. ;-}
89: added a "real" stdio. Well, almost. Please contribute!
90: added dietfeatures.h so you can remove features you don't need.
91: removed debug code from vsnprintf that changed \0 to ' ' (argh!).
92: stdio now works with simple programs and minigzip from libz.
93: stdio uses some major trickery to avoid linking the stdio, stderr and
94: stdout (and reserving space for them) when they are not used.
95:
96: 0.6.5:
97: Red Plait <redplait@ixcelerator.com> found several bugs in diet libc,
98: most of them bugs in the new header files, but also missing sigset
99: functions.
100: Paul Clifford contributed an assembly strlen.S for ARM and fixed
101: several bugs.
102: Fixed ppc/setjmp and mips/pipe.
103: Olaf contributed a new sparc unified syscall.
104:
105: 0.6.4:
106: declared is* static inline in getservent.
107: added assert_fail, strtoul, isalpha, isdigit, isalnum, isascii.
108: changed strlen to return 0 when passed NULL.
109: new, much smaller unified syscall for MIPS. Thanks to Olaf the Mad
110: Scientist who actually implemented this without access to a MIPS
111: box, just from reading the architecture manual. And it worked out
112: of the box.
113: started a set of system includes, mainly so that I can use lcc and
114: my alpha-linux cross compiler (which is unable to cross-compile
115: glibc). diet libc can now be compiled without any include files
116: from a normal libc. The includes do declare more than diet libc
117: currently delivers and they are still far from complete for real
118: applications.
119: I will only add assembly versions that are smaller _and_ faster than
120: the C version. Larger routines are only accepted if they are called
121: very often and are substantially faster.
122: added i386 assembly strchr, which is smaller and faster than the old
123: version. It is, however, larger than the version contributed by
124: proton (thanks, anyway).
125: added i386 strlen (31% faster, 14% smaller)
126: "make t" will now create a map file called "mapfile".
127:
128: 0.6.3:
129: added sys_errlist, strerror and perror
130: added isblank
131: added atol
132:
133: 0.6.2:
134: mmap for ARM didn't compile. Thanks, Paul!
135:
136: 0.6.1:
137: split mmap into the architecture specific subdirectories.
138: getenv now copes with environ==NULL, thanks Paul Clifford.
139: Paul also contributed a smaller ARM startup code.
140:
141: 0.6:
142: strcat returned the wrong result. Thanks, Dietz Pröpper.
143: strtod now understands a negative exponent (oops, thanks Bertram Barth)
144: Port to arm-linux-gnu, but on the Netwinder I use for testing the
145: __dtostr does not work (I have no idea why!)
146: The MIPS port now uses (much smaller) non-PIC code. That means that
147: the applications you link against diet libc also have to be compiled
148: non-PIC. I suggest copying the CFLAGS from the diet libc Makefile.
149: Thanks to Ralf Bächle for helping me with this!
150: I made subdirectories for the architectures and use VPATH to override
151: VPATH so that make finds the source file automatically. That should
152: simplify the sources greatly.
153: Olaf Dreesen contributed Alpha support including setjmp and longjmp!
154: ARM and MIPS now also have setjmp and longjmp
155: "compile" and "load" are now make targets. Use them for djb programs.
156: added contrib/elftrunc.c which will remove unnecessary ELF headers.
157: Again, contributed from Olaf. Great work!
158:
159: 0.5.12:
160: printf also does signed numbers.
161: If you don't use atexit, dietlibc now does not link exit, only _exit.
162: *printf now correctly returns the number of bytes written.
163: Olaf contributed experimental sscanf and vsscanf implementations.
164: If passed NULL as buffer, snprintf will not write anything but still
165: return the number of bytes it would have written.
166: Initial MIPS port! (Oh, the agony!)
167: No setjmp and longjmp support yet! Please contribute!
168: I even implemented unified syscalls for MIPS. Still, MIPS code is
169: almost twice the size of SPARC code. If anyone knows why: please
170: tell me!
171:
172: 0.5.11:
173: I implemented new unified syscalls for x86, sparc and ppc, this time based
174: on .s files and not .c files, and I moved the syscalls into
175: subdirectores to clean the dietlibc sources up a little.
176: A binary consisting of printf("%s is %d\n","olaf",23) is now
177: 2864 bytes on sparc
178: 2488 bytes on intel
179: I kludgily implemented fprintf(stdout,... and fprintf(stderr,...
180: to make a few more applications work.
181: Olaf Dreesen also implemented some unified syscall stuff for x86.
182: He discovered that with our unified syscall interface it actually
183: costs <10 bytes total to make all system calls thread safe!
184: I implemented atexit() (can register up to 4 callbacks).
185: I implemented strtod and __dtostr (the opposite). Now we can add
186: floating point support to vsnprintf and sscanf!
187:
188: 0.5.10:
189: I actually saw that I can not only merge the errno handling code of
190: the system calls on x86, I can also merge the rest (including the
191: arguments) except for setting the system call number. All those
192: system calls are now just a jump to a unified system call handler.
193: I got rid of x86openclose again. The savings are substantial:
194: chown from embutils went from 7664 to 7184 bytes!
195: If I'd move the system call wrappers to assembly language, I could
196: even reduce the alignment (does not matter since it's just a jump
197: anyway) and get rid of the "ret" behind each jump (one byte per
198: system call!)
199:
200: 0.5.9:
201: added memchr, strpbrk, strstr, strtol, isspace (hehe)
202: fixed strdup
203: fixed return values for strcpy and strcat (thanks to Norbert Berzen)
204: Olaf Dreesen contributed a strtol and initial {sn|vsn|}printf implementation
205: (no signed integers, only strings and unsigned integers (but
206: supporting octal, hex and decimal). Thanks, Olaf! (by the way:
207: that code must be wonderful, I don't understand it at all *bg*)
208: Olaf's printf does understand padding like in "%08d" and "%8d" and
209: automatically pads pointers with '0'.
210: Rewrote the x86 start code, old: 22 instructions, new: 12 instructions.
211: Thanks to proton for inspiration on this.
212: The new x86 start code also does not reference exit any more, thus saving
213: 64 bytes for executables that don't call exit explicitly.
214: [insert maniacal laughter] EVERY BYTE COUNTS!1!! ;-)
215: Moved errno and environ to start.S, saving no byte binary size but
216: speeding up compilation and slightly shortening dietlibc.a ;-)
217: For x86: joined open and close into one assembler file, sharing the
218: errno handling. This is a feasibility test and it actually saves a
219: few bytes. I think I will reimplement the _syscall[1-6] macros on
220: all platforms now to share the errno handling code for them.
221:
222: 0.5.8:
223: fixed strchr to be able to look for 0.
224:
225: added _llseek, ftruncate, getpgid, getresgid, getresuid, getsid,
226: memccpy, memmove, mprotect, setregid, setresgid, setresuid, setreuid,
227: strncpy, swapon, truncate, strtok, strtok_r, strspn, strcspn (all
228: contributed by Olaf Dreesen)
229:
230: added execvp, getcwd
231:
232: fixed __xmknod
233:
234: Note: can it be that the ftw interface really is so broken that I
235: cannot implement it without having to implement some searching data
236: structure? I included an experimental ftw implementation that will
237: not follow symlinks.
238:
239: 0.5.7:
240: oops, if_nametoindex was broken!
241:
242: 0.5.6:
243: ported to ppc-linux.
244: fixed i386 sigsetjmp (I mistyped the function name)
245: included sigjmp.c
246:
247: 0.5.5:
248: added wait, sys_siglist, longjmp/setjmp/sigsetjmp for i386 and sparc
249: actually, it wasn't fork that wasn't working for sparc, it was pipe.
250: I fixed it now.
251: dietlibc/SPARC assumes -msupersparc (does not provide div, mul, etc)
252: fixed readdir to use getdents and not the intel inline asm
253:
254: 0.5.4:
255: ported to sparclinux (sparc32 only). Beware: does not work yet.
256: added raise, abort, readlink, strcat, geteuid, geteuid, wait3, access
257:
258: 0.5.3:
259: added tcsetattr and getenv to compile e3
260:
261: 0.5.2:
262: added vhangup, tcgetattr, isatty and memcmp for fgetty
263: (http://www.fefe.de/fgetty/)
264: added localtime from uC-libc.
265:
266: 0.5.1:
267: stat and friends actually work now. Yuck! Another case of
268: translation between kernel and userland. Why can't the kernel
269: people and the libc people simply agree on a standard?
270: removed many unnecessary includes to speed up compilation.
271: added -fomit-frame-pointer and i386 compilation target to reduce code size.
272: inlined socketcall to reduce code size.
273: "load" and "compile" are examples for djb code.
274:
275: 0.5:
276: Olaf Dreesen contributed a much smaller implementation of malloc and friends.
277: each object file is now treated with "strip -x -R .note -R .comment"
278: split each system call into a separate object file
279: added ntohs, htons, alarm, if_indextoname and if_nametoindex
280:
281: 0.4.1:
282: oops, the strchr implementation was wrong. Thanks Jens Laas!
283:
284: 0.4:
285: copied opendir and friends from uC-libc.
286: added getservent and getservby* in gerservent.c
287:
288: 0.3:
289: remove readdir system call
290: added getdents system call
291: added getpwnam/getpwuid implementation in getpwnam.c
292: added a few string functions (in str*.c)
293:
LinuxTV legacy CVS <linuxtv.org/cvs>