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