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