3 Stuff that needs to be done. This is organized by who plans to get around to
4 doing it eventually, but that doesn't mean they "own" the item. If you want to
5 do one of these bounce an email off the person it's listed under to see if they
6 have any suggestions how they plan to go about it, and to minimize conflicts
7 between your work and theirs. But otherwise, all of these are fair game.
9 Rob Landley <rob@landley.net>:
10 Add BB_NOMMU to platform.h and migrate __uClinux__ tests to that.
11 #if defined __UCLIBC__ && !defined __ARCH_USE_MMU__
12 Add a libbb/platform.c
13 Implement fdprintf() for platforms that haven't got one.
14 Implement bb_realpath() that can handle NULL on non-glibc.
17 Migrate calloc() and bb_calloc() occurrences to bb_xzalloc().
18 Remove obsolete _() wrapper crud for internationalization we don't do.
19 Figure out where we need utf8 support, and add it.
22 The command shell situation is a big mess. We have three or four different
23 shells that don't really share any code, and the "standalone shell" doesn't
24 work all that well (especially not in a chroot environment), due to apps not
25 being reentrant. I'm writing a new shell (bbsh) to unify the various
26 shells and configurably add the minimal set of bash features people
27 actually use. The hardest part is it has to configure down as small as
28 lash while providing lash's features. The rest is easy in comparison.
30 Compression-side support.
32 General cleanup (should use ENABLE_FEATURE_INIT_SYSLOG and ENABLE_FEATURE_INIT_DEBUG).
34 busybox lacks a way to update module deps when running from firmware without the
35 use of the depmod.pl (perl is to bloated for most embedded setups) and or orig
36 modutils. The orig depmod is rather pointless to have to add to a firmware image
37 in when we already have a insmod/rmmod and friends.
38 Unify base64 handling.
39 There's base64 encoding and decoding going on in:
40 networking/wget.c:base64enc()
41 coreutils/uudecode.c:read_base64()
42 coreutils/uuencode.c:tbl_base64[]
43 networking/httpd.c:decodeBase64()
44 And probably elsewhere. That needs to be unified into libbb functions.
46 Look at the full Single Unix Specification version 3 (available online at
47 "http://www.opengroup.org/onlinepubs/009695399/nfindex.html") and
48 figure out which of our apps are compliant, and what we're missing that
49 we might actually care about.
51 Even better would be some kind of automated compliance test harness that
52 exercises each command line option and the various corner cases.
54 How much internationalization should we do?
56 The low hanging fruit is UTF-8 character set support. We should do this.
57 (Vodz pointed out the shell's cmdedit as needing work here. What else?)
59 We also have lots of hardwired english text messages. Consolidating this
60 into some kind of message table not only makes translation easier, but
61 also allows us to consolidate redundant (or close) strings.
63 We probably don't want to be bloated with locale support. (Not unless we
64 can cleanly export it from our underlying C library without having to
65 concern ourselves with it directly. Perhaps a few specific things like a
66 config option for "date" are low hanging fruit here?)
68 What level should things happen at? How much do we care about
69 internationalizing the text console when X11 and xterms are so much better
70 at it? (There's some infrastructure here we don't implement: The
71 "unicode_start" and "unicode_stop" shell scripts need "vt-is-UTF8" and a
72 --unicode option to loadkeys. That implies a real loadkeys/dumpkeys
73 implementation to replace loadkmap/dumpkmap. Plus messing with console font
74 loading. Is it worth it, or do we just say "use X"?)
76 Individual compilation of applets.
77 It would be nice if busybox had the option to compile to individual applets,
78 for people who want an alternate implementation less bloated than the gnu
79 utils (or simply with less political baggage), but without it being one big
82 Turning libbb into a real dll is another possibility, especially if libbb
83 could export some of the other library interfaces we've already more or less
84 got the code for (like zlib).
85 buildroot - Make a "dogfood" option
86 Busybox 1.1 will be capable of replacing most gnu packages for real world
87 use, such as developing software or in a live CD. It needs wider testing.
89 Busybox should now be able to replace bzip2, coreutils, e2fsprogs, file,
90 findutils, gawk, grep, inetutils, less, modutils, net-tools, patch, procps,
91 sed, shadow, sysklogd, sysvinit, tar, util-linux, and vim. The resulting
92 system should be self-hosting (I.E. able to rebuild itself from source
93 code). This means it would need (at least) binutils, gcc, and make, or
96 It would be a good "eating our own dogfood" test if buildroot had the option
97 of using a "make allyesconfig" busybox instead of the all of the above
98 packages. Anything that's wrong with the resulting system, we can fix. (It
99 would be nice to be able to upgrade busybox to be able to replace bash and
100 diffutils as well, but we're not there yet.)
102 One example of an existing system that does this already is Firmware Linux:
103 http://www.landley.net/code/firmware
105 Busybox should have a sample initramfs build script. This depends on
106 bbsh, mdev, and switch_root.
108 Write a mkdep that doesn't segfault if there's a directory it doesn't
109 have permission to read, isn't based on manually editing the output of
110 lexx and yacc, doesn't make such a mess under include/config, etc.
111 Group globals into unions of structures.
112 Go through and turn all the global and static variables into structures,
113 and have all those structures be in a big union shared between processes,
114 so busybox uses less bss. (This is a big win on nommu machines.) See
115 sed.c and mdev.c for examples.
116 Go through bugs.busybox.net and close out all of that somehow.
117 This one's open to everybody, but I'll wind up doing it...
120 Bernhard Fischer <busybox@busybox.net> suggests to look at these:
123 Cleanup any big users
125 Facilitate applet PROTOTYPES to provide means for having applets that
126 do a) not take any arguments b) need only one of argc or argv c) need
127 both argc and argv. All of these three options should go for the most
128 feature complete denominator.
129 Collate BUFSIZ IOBUF_SIZE MY_BUF_SIZE PIPE_PROGRESS_SIZE BUFSIZE PIPESIZE
130 make bb_common_bufsiz1 configurable, size wise.
131 make pipesize configurable, size wise.
132 Use bb_common_bufsiz1 throughout applets!
133 Add chrt applet. Please CC Bernhard if you suggest a patch.
139 doesn't understand (), lots of susv3 stuff.
142 Make sure we handle empty files properly:
143 From the patch man page:
145 you can remove a file by sending out a context diff that compares
146 the file to be deleted with an empty file dated the Epoch. The
147 file will be removed unless patch is conforming to POSIX and the
148 -E or --remove-empty-files option is not given.
151 Should have simple fuzz factor support to apply patches at an offset which
152 shouldn't take up too much space.
154 And while we're at it, a new patch filename quoting format is apparently
155 coming soon: http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
158 It would be nice to have a man command. Not one that handles troff or
159 anything, just one that can handle preformatted ascii man pages, possibly
160 compressed. This could probably be a script in the extras directory that
161 calls cat/zcat/bzcat | less
163 (How doclifter might work into this is anybody's guess.)
169 Architectural issues:
171 bb_close() with fsync()
172 We should have a bb_close() in place of normal close, with a CONFIG_ option
173 to not just check the return value of close() for an error, but fsync().
174 Close can't reliably report anything useful because if write() accepted the
175 data then it either went out to the network or it's in cache or a pipe
176 buffer. Either way, there's no guarantee it'll make it to its final
177 destination before close() gets called, so there's no guarantee that any
178 error will be reported.
180 You need to call fsync() if you care about errors that occur after write(),
181 but that can have a big performance impact. So make it a config option.
184 Lots of archivers have the same general infrastructure. The directory
185 traversal code should be factored out, and the guts of each archiver could
186 be some setup code and a series of callbacks for "add this file",
187 "add this directory", "add this symlink" and so on.
189 This could clean up tar and zip, and make it cheaper to add cpio and ar
190 write support, and possibly even cheaply add things like mkisofs or
191 mksquashfs someday, if they become relevant.
194 Several existing applets (sort, vi, less...) read
195 a whole file into memory and act on it. There might be an opportunity
196 for shared code in there that could be moved into libbb...
199 We have a CONFIG_BUFFER mechanism that lets us select whether to do memory
200 allocation on the stack or the heap. Unfortunately, we're not using it much.
201 We need to audit our memory allocations and turn a lot of malloc/free calls
202 into RESERVE_CONFIG_BUFFER/RELEASE_CONFIG_BUFFER.
203 For a start, see e.g. make EXTRA_CFLAGS=-Wlarger-than-64
205 And while we're at it, many of the CONFIG_FEATURE_CLEAN_UP #ifdefs will be
206 optimized out by the compiler in the stack allocation case (since there's no
207 free for an alloca()), and this means that various cleanup loops that just
208 call free might also be optimized out by the compiler if written right, so
209 we can yank those #ifdefs too, and generally clean up the code.
211 Switch CONFIG_SYMBOLS to ENABLE_SYMBOLS
213 In busybox 1.0 and earlier, configuration was done by CONFIG_SYMBOLS
214 that were either defined or undefined to indicate whether the symbol was
215 selected in the .config file. They were used with #ifdefs, ala:
223 In 1.1, we have new ENABLE_SYMBOLS which are always defined (as 0 or 1),
224 meaning you can still use them for preprocessor tests by replacing
225 "#ifdef CONFIG_SYMBOL" with "#if ENABLE_SYMBOL". But more importantly, we
226 can use them as a true or false test in normal C code:
228 if (ENABLE_SYMBOL && other_test) {
232 (Optimizing away if() statements that resolve to a constant value
233 is known as "dead code elimination", an optimization so old and simple that
234 Turbo Pascal for DOS did it twenty years ago. Even modern mini-compilers
235 like the Tiny C Compiler (tcc) and the Small Device C Compiler (SDCC)
236 perform dead code elimination.)
238 Right now, busybox.h is #including both "config.h" (defining the
239 CONFIG_SYMBOLS) and "bb_config.h" (defining the ENABLE_SYMBOLS). At some
240 point in the future, it would be nice to wean ourselves off of the
241 CONFIG versions. (Among other things, some defective build environments
242 leak the Linux kernel's CONFIG_SYMBOLS into the system's standard #include
243 files. We've experienced collisions before.)
246 This is more an unresolved issue than a to-do item. More thought is needed.
248 Normally we rely on exit() to free memory, close files, and unmap segments
249 for us. This makes most calls to free(), close(), and unmap() optional in
250 busybox applets that don't intend to run for very long, and optional stuff
251 can be omitted to save size.
253 The idea was raised that we could simulate fork/exit with setjmp/longjmp
254 for _really_ brainless embedded systems, or speed up the standalone shell
255 by not forking. Doing so would require a reliable FEATURE_CLEAN_UP.
256 Unfortunately, this isn't as easy as it sounds.
258 The problem is, lots of things exit(), sometimes unexpectedly (xmalloc())
259 and sometimes reliably (bb_perror_msg_and_die() or show_usage()). This
260 jumps out of the normal flow control and bypasses any cleanup code we
261 put at the end of our applets.
263 It's possible to add hooks to libbb functions like xmalloc() and xopen()
264 to add their entries to a linked list, which could be traversed and
265 freed/closed automatically. (This would need to be able to free just the
266 entries after a checkpoint to be usable for a forkless standalone shell.
267 You don't want to free the shell's own resources.)
269 Right now, FEATURE_CLEAN_UP is more or less a debugging aid, to make things
270 like valgrind happy. It's also documentation of _what_ we're trusting
271 exit() to clean up for us. But new infrastructure to auto-free stuff would
272 render the existing FEATURE_CLEAN_UP code redundant.
274 For right now, exit() handles it just fine.
279 watchdog.c could autodetect the timer duration via:
280 if(!ioctl (fd, WDIOC_GETTIMEOUT, &tmo)) timer_duration = 1 + (tmo / 2);
281 Unfortunately, that needs linux/watchdog.h and that contains unfiltered
282 kernel types on some distros, which breaks the build.
284 use bb_error_msg where appropriate: See
285 egrep "(printf.*\([[:space:]]*(stderr|2)|[^_]write.*\([[:space:]]*(stderr|2))"
287 use bb_perror_msg where appropriate: See
290 Remove superfluous fmt occurances: e.g.
291 fprintf(stderr, "%s: %s not found\n", "unalias", *argptr);
292 -> fprintf(stderr, "unalias: %s not found\n", *argptr);
294 possible code duplication ingroup() and is_a_group_member()
296 Move __get_hz() to a better place and (re)use it in route.c, ash.c, msh.c
302 Replace deprecated functions.
306 sigblock(), siggetmask(), sigsetmask(), sigmask() -> sigprocmask et al
308 vdprintf() -> similar sized functionality