Annotation of dietlibc/CHANGES, revision 1.7
1.1 cvs 1: 0.6.15:
1.6 olaf 2: clone support for i386,alpha,arm and sparc by Olaf.
1.1 cvs 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!
1.4 fefe 12: Removed bogus __seek_types enum, it's a bunch of #defines now.
1.7 ! fefe 13: Fixed stdio line buffering. The code was already there but I forgot
! 14: to mark stdio and stdout as line buffered. This should really be
! 15: done dynamically using isatty, though.
1.1 cvs 16:
17: 0.6.14:
18: *scanf did not append the 0 byte for %s.
19: added creat and changed creat to call open directly instead of open64.
20: did for fread what 0.6.10 did for fwrite.
21: repaired buffered stdio a little (line buffering is still not
22: supported).
23: [All of these bugs have been reported by Guillaume. Thanks!]
24: added __pure__ to strchr and strrchr in <string.h>
25:
26: 0.6.13:
27: included COPYING to make clear that the diet libc is covered by the
28: GNU General Public License (and _not_ the LGPL). That means that
29: you need to obtain a license from me if you want to use the diet
30: libc in a proprietary program that you want to distribute.
31: The x86 unified syscall swapped arguments #4 and #5 and noone
32: noticed until now! Thanks to Guillaume Cottenceau for reporting
33: this!
34: The malloc had a subtle bug with small allocations that could cause
35: segfault. Guillaume reported it, Olaf fixed it.
36:
37: 0.6.12:
38: added assert, statfs and fstatfs
39: added include/net/if.h so fget compiles again.
40:
41: 0.6.11:
42: added putenv.
43:
44: 0.6.10:
45: fixed fwrite. It returned the number of bytes written, not the number
46: of records. Thanks to Albert D. Cahalan for pointing this out.
47: fixed fgets. It returned EOF on empty lines. Thanks, Erik Frey.
48: added brk and sbrk. harold@nb.com.sg made me do it.
49:
50: 0.6.9:
51: fixed execvp not to return on ENOENT
52: did for {open|read|close|seek|tell}dir what I did for stat before.
53: The glibc compatibility part is currently non-functional.
54: fixed white space handling in sscanf " (" would not match " (".
55: fixed printf to accept 'l' flag.
56: fixed getpwuid (ignored last line)
57:
58: 0.6.8:
59: Olaf contributed initial thread-safe syscalls for i386, alpha, sparc,
60: mips and arm. I did the one for ppc. They can be enabled in dietfeatures.h.
61: added strlcpy and strlcat from OpenBSD.
62: added stat64, fstat64 and lstat64.
63: added endian.h and made ending of strlen.c endianness-aware
64: added string routines size tweak to dietfeatures.h. Most of the
65: string routines have been unrolled. This is up to three times as
66: fast but creates up to three times larger code. Now you can disable
67: the unrolling.
68: added sendfile.
69: I also added pread. It works on x86 and sparc but not on ppc and mips.
70: I have no idea what's going on. strace disagrees with the kernel.
71: Can anyone help? I asked the Linux kernel mailing list for help, too.
72: added a few aliases of the type __libc_open for open. nm on
73: libpthreads.so indicates that we will need them.
74:
75: 0.6.7:
76: the sources now compile without warnings with -Wall.
77: printf now prints "(null)" when %s is passed NULL. This can be
78: removed with WANT_NULL_PRINTF in dietfeatures.h
79: added vfprintf, execl, ttyname.
80: stat, lstat and fstat are now #defined to __dietstat, __dietlstat and
81: __dietfstat so they can use the normal kernel struct stat.
82: stat, lstat and fstat are now C wrappers that convert to the glibc
83: struct stat. So we can avoid that overhead for programs that use
84: dietlibc headers.
85:
86: 0.6.6:
87: changed the Makefiles so you can now set CFLAGS on the command line
88: added memccmp (analogous to memccpy) and strncmp.
89: Olaf fixed another bug in the sparc unified syscall.
90: Paul Clifford contributed a C version of his strlen.S that is much
91: more efficient than the previous strlen on all platforms! It also
92: looks like technology from Roswell. ;-}
93: added a "real" stdio. Well, almost. Please contribute!
94: added dietfeatures.h so you can remove features you don't need.
95: removed debug code from vsnprintf that changed \0 to ' ' (argh!).
96: stdio now works with simple programs and minigzip from libz.
97: stdio uses some major trickery to avoid linking the stdio, stderr and
98: stdout (and reserving space for them) when they are not used.
99:
100: 0.6.5:
101: Red Plait <redplait@ixcelerator.com> found several bugs in diet libc,
102: most of them bugs in the new header files, but also missing sigset
103: functions.
104: Paul Clifford contributed an assembly strlen.S for ARM and fixed
105: several bugs.
106: Fixed ppc/setjmp and mips/pipe.
107: Olaf contributed a new sparc unified syscall.
108:
109: 0.6.4:
110: declared is* static inline in getservent.
111: added assert_fail, strtoul, isalpha, isdigit, isalnum, isascii.
112: changed strlen to return 0 when passed NULL.
113: new, much smaller unified syscall for MIPS. Thanks to Olaf the Mad
114: Scientist who actually implemented this without access to a MIPS
115: box, just from reading the architecture manual. And it worked out
116: of the box.
117: started a set of system includes, mainly so that I can use lcc and
118: my alpha-linux cross compiler (which is unable to cross-compile
119: glibc). diet libc can now be compiled without any include files
120: from a normal libc. The includes do declare more than diet libc
121: currently delivers and they are still far from complete for real
122: applications.
123: I will only add assembly versions that are smaller _and_ faster than
124: the C version. Larger routines are only accepted if they are called
125: very often and are substantially faster.
126: added i386 assembly strchr, which is smaller and faster than the old
127: version. It is, however, larger than the version contributed by
128: proton (thanks, anyway).
129: added i386 strlen (31% faster, 14% smaller)
130: "make t" will now create a map file called "mapfile".
131:
132: 0.6.3:
133: added sys_errlist, strerror and perror
134: added isblank
135: added atol
136:
137: 0.6.2:
138: mmap for ARM didn't compile. Thanks, Paul!
139:
140: 0.6.1:
141: split mmap into the architecture specific subdirectories.
142: getenv now copes with environ==NULL, thanks Paul Clifford.
143: Paul also contributed a smaller ARM startup code.
144:
145: 0.6:
146: strcat returned the wrong result. Thanks, Dietz Pröpper.
147: strtod now understands a negative exponent (oops, thanks Bertram Barth)
148: Port to arm-linux-gnu, but on the Netwinder I use for testing the
149: __dtostr does not work (I have no idea why!)
150: The MIPS port now uses (much smaller) non-PIC code. That means that
151: the applications you link against diet libc also have to be compiled
152: non-PIC. I suggest copying the CFLAGS from the diet libc Makefile.
153: Thanks to Ralf Bächle for helping me with this!
154: I made subdirectories for the architectures and use VPATH to override
155: VPATH so that make finds the source file automatically. That should
156: simplify the sources greatly.
157: Olaf Dreesen contributed Alpha support including setjmp and longjmp!
158: ARM and MIPS now also have setjmp and longjmp
159: "compile" and "load" are now make targets. Use them for djb programs.
160: added contrib/elftrunc.c which will remove unnecessary ELF headers.
161: Again, contributed from Olaf. Great work!
162:
163: 0.5.12:
164: printf also does signed numbers.
165: If you don't use atexit, dietlibc now does not link exit, only _exit.
166: *printf now correctly returns the number of bytes written.
167: Olaf contributed experimental sscanf and vsscanf implementations.
168: If passed NULL as buffer, snprintf will not write anything but still
169: return the number of bytes it would have written.
170: Initial MIPS port! (Oh, the agony!)
171: No setjmp and longjmp support yet! Please contribute!
172: I even implemented unified syscalls for MIPS. Still, MIPS code is
173: almost twice the size of SPARC code. If anyone knows why: please
174: tell me!
175:
176: 0.5.11:
177: I implemented new unified syscalls for x86, sparc and ppc, this time based
178: on .s files and not .c files, and I moved the syscalls into
179: subdirectores to clean the dietlibc sources up a little.
180: A binary consisting of printf("%s is %d\n","olaf",23) is now
181: 2864 bytes on sparc
182: 2488 bytes on intel
183: I kludgily implemented fprintf(stdout,... and fprintf(stderr,...
184: to make a few more applications work.
185: Olaf Dreesen also implemented some unified syscall stuff for x86.
186: He discovered that with our unified syscall interface it actually
187: costs <10 bytes total to make all system calls thread safe!
188: I implemented atexit() (can register up to 4 callbacks).
189: I implemented strtod and __dtostr (the opposite). Now we can add
190: floating point support to vsnprintf and sscanf!
191:
192: 0.5.10:
193: I actually saw that I can not only merge the errno handling code of
194: the system calls on x86, I can also merge the rest (including the
195: arguments) except for setting the system call number. All those
196: system calls are now just a jump to a unified system call handler.
197: I got rid of x86openclose again. The savings are substantial:
198: chown from embutils went from 7664 to 7184 bytes!
199: If I'd move the system call wrappers to assembly language, I could
200: even reduce the alignment (does not matter since it's just a jump
201: anyway) and get rid of the "ret" behind each jump (one byte per
202: system call!)
203:
204: 0.5.9:
205: added memchr, strpbrk, strstr, strtol, isspace (hehe)
206: fixed strdup
207: fixed return values for strcpy and strcat (thanks to Norbert Berzen)
208: Olaf Dreesen contributed a strtol and initial {sn|vsn|}printf implementation
209: (no signed integers, only strings and unsigned integers (but
210: supporting octal, hex and decimal). Thanks, Olaf! (by the way:
211: that code must be wonderful, I don't understand it at all *bg*)
212: Olaf's printf does understand padding like in "%08d" and "%8d" and
213: automatically pads pointers with '0'.
214: Rewrote the x86 start code, old: 22 instructions, new: 12 instructions.
215: Thanks to proton for inspiration on this.
216: The new x86 start code also does not reference exit any more, thus saving
217: 64 bytes for executables that don't call exit explicitly.
218: [insert maniacal laughter] EVERY BYTE COUNTS!1!! ;-)
219: Moved errno and environ to start.S, saving no byte binary size but
220: speeding up compilation and slightly shortening dietlibc.a ;-)
221: For x86: joined open and close into one assembler file, sharing the
222: errno handling. This is a feasibility test and it actually saves a
223: few bytes. I think I will reimplement the _syscall[1-6] macros on
224: all platforms now to share the errno handling code for them.
225:
226: 0.5.8:
227: fixed strchr to be able to look for 0.
228:
229: added _llseek, ftruncate, getpgid, getresgid, getresuid, getsid,
230: memccpy, memmove, mprotect, setregid, setresgid, setresuid, setreuid,
231: strncpy, swapon, truncate, strtok, strtok_r, strspn, strcspn (all
232: contributed by Olaf Dreesen)
233:
234: added execvp, getcwd
235:
236: fixed __xmknod
237:
238: Note: can it be that the ftw interface really is so broken that I
239: cannot implement it without having to implement some searching data
240: structure? I included an experimental ftw implementation that will
241: not follow symlinks.
242:
243: 0.5.7:
244: oops, if_nametoindex was broken!
245:
246: 0.5.6:
247: ported to ppc-linux.
248: fixed i386 sigsetjmp (I mistyped the function name)
249: included sigjmp.c
250:
251: 0.5.5:
252: added wait, sys_siglist, longjmp/setjmp/sigsetjmp for i386 and sparc
253: actually, it wasn't fork that wasn't working for sparc, it was pipe.
254: I fixed it now.
255: dietlibc/SPARC assumes -msupersparc (does not provide div, mul, etc)
256: fixed readdir to use getdents and not the intel inline asm
257:
258: 0.5.4:
259: ported to sparclinux (sparc32 only). Beware: does not work yet.
260: added raise, abort, readlink, strcat, geteuid, geteuid, wait3, access
261:
262: 0.5.3:
263: added tcsetattr and getenv to compile e3
264:
265: 0.5.2:
266: added vhangup, tcgetattr, isatty and memcmp for fgetty
267: (http://www.fefe.de/fgetty/)
268: added localtime from uC-libc.
269:
270: 0.5.1:
271: stat and friends actually work now. Yuck! Another case of
272: translation between kernel and userland. Why can't the kernel
273: people and the libc people simply agree on a standard?
274: removed many unnecessary includes to speed up compilation.
275: added -fomit-frame-pointer and i386 compilation target to reduce code size.
276: inlined socketcall to reduce code size.
277: "load" and "compile" are examples for djb code.
278:
279: 0.5:
280: Olaf Dreesen contributed a much smaller implementation of malloc and friends.
281: each object file is now treated with "strip -x -R .note -R .comment"
282: split each system call into a separate object file
283: added ntohs, htons, alarm, if_indextoname and if_nametoindex
284:
285: 0.4.1:
286: oops, the strchr implementation was wrong. Thanks Jens Laas!
287:
288: 0.4:
289: copied opendir and friends from uC-libc.
290: added getservent and getservby* in gerservent.c
291:
292: 0.3:
293: remove readdir system call
294: added getdents system call
295: added getpwnam/getpwuid implementation in getpwnam.c
296: added a few string functions (in str*.c)
297:
LinuxTV legacy CVS <linuxtv.org/cvs>