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