TODO 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374
  1. Busybox TODO
  2. Stuff that needs to be done. This is organized by who plans to get around to
  3. doing it eventually, but that doesn't mean they "own" the item. If you want to
  4. do one of these bounce an email off the person it's listed under to see if they
  5. have any suggestions how they plan to go about it, and to minimize conflicts
  6. between your work and theirs. But otherwise, all of these are fair game.
  7. Rob Landley suggested this:
  8. Implement bb_realpath() that can handle NULL on non-glibc.
  9. Remove obsolete _() wrapper crud for internationalization we don't do.
  10. Figure out where we need utf8 support, and add it.
  11. sh
  12. The command shell situation is a mess. We have two different
  13. shells that don't really share any code, and the "standalone shell" doesn't
  14. work all that well (especially not in a chroot environment), due to apps not
  15. being reentrant.
  16. Do a SUSv3 audit
  17. Look at the full Single Unix Specification version 3 (available online at
  18. "http://www.opengroup.org/onlinepubs/009695399/nfindex.html") and
  19. figure out which of our apps are compliant, and what we're missing that
  20. we might actually care about.
  21. Even better would be some kind of automated compliance test harness that
  22. exercises each command line option and the various corner cases.
  23. Internationalization
  24. How much internationalization should we do?
  25. The low hanging fruit is UTF-8 character set support. We should do this.
  26. (Vodz pointed out the shell's cmdedit as needing work here. What else?)
  27. We also have lots of hardwired english text messages. Consolidating this
  28. into some kind of message table not only makes translation easier, but
  29. also allows us to consolidate redundant (or close) strings.
  30. We probably don't want to be bloated with locale support. (Not unless we
  31. can cleanly export it from our underlying C library without having to
  32. concern ourselves with it directly. Perhaps a few specific things like a
  33. config option for "date" are low hanging fruit here?)
  34. What level should things happen at? How much do we care about
  35. internationalizing the text console when X11 and xterms are so much better
  36. at it? (There's some infrastructure here we don't implement: The
  37. "unicode_start" and "unicode_stop" shell scripts need "vt-is-UTF8" and a
  38. --unicode option to loadkeys. That implies a real loadkeys/dumpkeys
  39. implementation to replace loadkmap/dumpkmap. Plus messing with console font
  40. loading. Is it worth it, or do we just say "use X"?)
  41. Individual compilation of applets.
  42. It would be nice if busybox had the option to compile to individual applets,
  43. for people who want an alternate implementation less bloated than the gnu
  44. utils (or simply with less political baggage), but without it being one big
  45. executable.
  46. Turning libbb into a real dll is another possibility, especially if libbb
  47. could export some of the other library interfaces we've already more or less
  48. got the code for (like zlib).
  49. buildroot - Make a "dogfood" option
  50. Busybox 1.1 will be capable of replacing most gnu packages for real world
  51. use, such as developing software or in a live CD. It needs wider testing.
  52. Busybox should now be able to replace bzip2, coreutils, e2fsprogs, file,
  53. findutils, gawk, grep, inetutils, less, modutils, net-tools, patch, procps,
  54. sed, shadow, sysklogd, sysvinit, tar, util-linux, and vim. The resulting
  55. system should be self-hosting (I.E. able to rebuild itself from source
  56. code). This means it would need (at least) binutils, gcc, and make, or
  57. equivalents.
  58. It would be a good "eating our own dogfood" test if buildroot had the option
  59. of using a "make allyesconfig" busybox instead of the all of the above
  60. packages. Anything that's wrong with the resulting system, we can fix. (It
  61. would be nice to be able to upgrade busybox to be able to replace bash and
  62. diffutils as well, but we're not there yet.)
  63. One example of an existing system that does this already is Firmware Linux:
  64. http://www.landley.net/code/firmware
  65. initramfs
  66. Busybox should have a sample initramfs build script. This depends on
  67. bbsh, mdev, and switch_root.
  68. mkdep
  69. Write a mkdep that doesn't segfault if there's a directory it doesn't
  70. have permission to read, isn't based on manually editing the output of
  71. lexx and yacc, doesn't make such a mess under include/config, etc.
  72. Group globals into unions of structures.
  73. Go through and turn all the global and static variables into structures,
  74. and have all those structures be in a big union shared between processes,
  75. so busybox uses less bss. (This is a big win on nommu machines.) See
  76. sed.c and mdev.c for examples.
  77. Go through bugs.busybox.net and close out all of that somehow.
  78. This one's open to everybody, but I'll wind up doing it...
  79. Bernhard Reutner-Fischer <busybox@busybox.net> suggests to look at these:
  80. New debug options:
  81. -Wlarger-than-127
  82. Cleanup any big users
  83. Collate BUFSIZ IOBUF_SIZE MY_BUF_SIZE PIPE_PROGRESS_SIZE BUFSIZE PIPESIZE
  84. make bb_common_bufsiz1 configurable, size wise.
  85. make pipesize configurable, size wise.
  86. Use bb_common_bufsiz1 throughout applets!
  87. As yet unclaimed:
  88. ----
  89. diff
  90. Make sure we handle empty files properly:
  91. From the patch man page:
  92. you can remove a file by sending out a context diff that compares
  93. the file to be deleted with an empty file dated the Epoch. The
  94. file will be removed unless patch is conforming to POSIX and the
  95. -E or --remove-empty-files option is not given.
  96. ---
  97. patch
  98. Should have simple fuzz factor support to apply patches at an offset which
  99. shouldn't take up too much space.
  100. And while we're at it, a new patch filename quoting format is apparently
  101. coming soon: http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
  102. ---
  103. stty / catv
  104. stty's visible() function and catv's guts are identical. Merge them into
  105. an appropriate libbb function.
  106. ---
  107. struct suffix_mult
  108. Several duplicate users of: grep -r "1024\*1024" * -B2 -A1
  109. Merge to a single size_suffixes[] in libbb.
  110. Users: head tail od_bloaty hexdump and (partially as it wouldn't hurt) svlogd
  111. ---
  112. tail
  113. ./busybox tail -f foo.c~ TODO
  114. should not print fmt=header_fmt for subsequent date >> TODO; i.e. only
  115. fmt+ if another (not the current) file did change
  116. Architectural issues:
  117. bb_close() with fsync()
  118. We should have a bb_close() in place of normal close, with a CONFIG_ option
  119. to not just check the return value of close() for an error, but fsync().
  120. Close can't reliably report anything useful because if write() accepted the
  121. data then it either went out to the network or it's in cache or a pipe
  122. buffer. Either way, there's no guarantee it'll make it to its final
  123. destination before close() gets called, so there's no guarantee that any
  124. error will be reported.
  125. You need to call fsync() if you care about errors that occur after write(),
  126. but that can have a big performance impact. So make it a config option.
  127. ---
  128. Unify archivers
  129. Lots of archivers have the same general infrastructure. The directory
  130. traversal code should be factored out, and the guts of each archiver could
  131. be some setup code and a series of callbacks for "add this file",
  132. "add this directory", "add this symlink" and so on.
  133. This could clean up tar and zip, and make it cheaper to add cpio and ar
  134. write support, and possibly even cheaply add things like mkisofs or
  135. mksquashfs someday, if they become relevant.
  136. ---
  137. Text buffer support.
  138. Several existing applets (sort, vi, less...) read
  139. a whole file into memory and act on it. Use open_read_close().
  140. ---
  141. Memory Allocation
  142. We have a CONFIG_BUFFER mechanism that lets us select whether to do memory
  143. allocation on the stack or the heap. Unfortunately, we're not using it much.
  144. We need to audit our memory allocations and turn a lot of malloc/free calls
  145. into RESERVE_CONFIG_BUFFER/RELEASE_CONFIG_BUFFER.
  146. For a start, see e.g. make EXTRA_CFLAGS=-Wlarger-than-64
  147. And while we're at it, many of the CONFIG_FEATURE_CLEAN_UP #ifdefs will be
  148. optimized out by the compiler in the stack allocation case (since there's no
  149. free for an alloca()), and this means that various cleanup loops that just
  150. call free might also be optimized out by the compiler if written right, so
  151. we can yank those #ifdefs too, and generally clean up the code.
  152. ---
  153. Switch CONFIG_SYMBOLS to ENABLE_SYMBOLS
  154. In busybox 1.0 and earlier, configuration was done by CONFIG_SYMBOLS
  155. that were either defined or undefined to indicate whether the symbol was
  156. selected in the .config file. They were used with #ifdefs, ala:
  157. #ifdef CONFIG_SYMBOL
  158. if (other_test) {
  159. do_code();
  160. }
  161. #endif
  162. In 1.1, we have new ENABLE_SYMBOLS which are always defined (as 0 or 1),
  163. meaning you can still use them for preprocessor tests by replacing
  164. "#ifdef CONFIG_SYMBOL" with "#if ENABLE_SYMBOL". But more importantly, we
  165. can use them as a true or false test in normal C code:
  166. if (ENABLE_SYMBOL && other_test) {
  167. do_code();
  168. }
  169. (Optimizing away if() statements that resolve to a constant value
  170. is known as "dead code elimination", an optimization so old and simple that
  171. Turbo Pascal for DOS did it twenty years ago. Even modern mini-compilers
  172. like the Tiny C Compiler (tcc) and the Small Device C Compiler (SDCC)
  173. perform dead code elimination.)
  174. Right now, busybox.h is #including both "config.h" (defining the
  175. CONFIG_SYMBOLS) and "bb_config.h" (defining the ENABLE_SYMBOLS). At some
  176. point in the future, it would be nice to wean ourselves off of the
  177. CONFIG versions. (Among other things, some defective build environments
  178. leak the Linux kernel's CONFIG_SYMBOLS into the system's standard #include
  179. files. We've experienced collisions before.)
  180. ---
  181. FEATURE_CLEAN_UP
  182. This is more an unresolved issue than a to-do item. More thought is needed.
  183. Normally we rely on exit() to free memory, close files and unmap segments
  184. for us. This makes most calls to free(), close(), and unmap() optional in
  185. busybox applets that don't intend to run for very long, and optional stuff
  186. can be omitted to save size.
  187. The idea was raised that we could simulate fork/exit with setjmp/longjmp
  188. for _really_ brainless embedded systems, or speed up the standalone shell
  189. by not forking. Doing so would require a reliable FEATURE_CLEAN_UP.
  190. Unfortunately, this isn't as easy as it sounds.
  191. The problem is, lots of things exit(), sometimes unexpectedly (xmalloc())
  192. and sometimes reliably (bb_perror_msg_and_die() or show_usage()). This
  193. jumps out of the normal flow control and bypasses any cleanup code we
  194. put at the end of our applets.
  195. It's possible to add hooks to libbb functions like xmalloc() and xopen()
  196. to add their entries to a linked list, which could be traversed and
  197. freed/closed automatically. (This would need to be able to free just the
  198. entries after a checkpoint to be usable for a forkless standalone shell.
  199. You don't want to free the shell's own resources.)
  200. Right now, FEATURE_CLEAN_UP is more or less a debugging aid, to make things
  201. like valgrind happy. It's also documentation of _what_ we're trusting
  202. exit() to clean up for us. But new infrastructure to auto-free stuff would
  203. render the existing FEATURE_CLEAN_UP code redundant.
  204. For right now, exit() handles it just fine.
  205. Minor stuff:
  206. watchdog.c could autodetect the timer duration via:
  207. if(!ioctl (fd, WDIOC_GETTIMEOUT, &tmo)) timer_duration = 1 + (tmo / 2);
  208. Unfortunately, that needs linux/watchdog.h and that contains unfiltered
  209. kernel types on some distros, which breaks the build.
  210. ---
  211. use bb_error_msg where appropriate: See
  212. egrep "(printf.*\([[:space:]]*(stderr|2)|[^_]write.*\([[:space:]]*(stderr|2))"
  213. ---
  214. use bb_perror_msg where appropriate: See
  215. egrep "[^_]perror"
  216. ---
  217. possible code duplication ingroup() and is_a_group_member()
  218. ---
  219. Move __get_hz() to a better place and (re)use it in route.c, ash.c
  220. ---
  221. See grep -r strtod
  222. Alot of duplication that wants cleanup.
  223. ---
  224. in_ether duplicated in network/{interface,ifconfig}.c
  225. ---
  226. unify progress_meter. wget, flash_eraseall, pipe_progress, fbsplash, setfiles.
  227. ---
  228. support start-stop-daemon -d <chdir-path>
  229. Code cleanup:
  230. Replace deprecated functions.
  231. ---
  232. vdprintf() -> similar sized functionality
  233. ---
  234. (TODO list after discussion 11.05.2009)
  235. * shrink tc/brctl/ip
  236. tc/brctl seem like fairly large things to try and tackle in your timeframe,
  237. and i think people have posted attempts in the past. Adding additional
  238. options to ip though seems reasonable.
  239. * add tests for some applets
  240. * implement POSIX utilities and audit them for POSIX conformance. then
  241. audit them for GNU conformance. then document all your findings in a new
  242. doc/conformance.txt file while perhaps implementing some of the missing
  243. features.
  244. you can find the latest POSIX documentation (1003.1-2008) here:
  245. http://www.opengroup.org/onlinepubs/9699919799/
  246. and the complete list of all utilities that POSIX covers:
  247. http://www.opengroup.org/onlinepubs/9699919799/idx/utilities.html
  248. The first step would to generate a file/matrix what is already archived
  249. (also IPV6)
  250. * ntpdate/ntpd (see ntpclient and openntp for examples)
  251. * implement 'at'
  252. * rpcbind (former portmap) or equivalent
  253. so that we don't have to use -o nolock on nfs mounts
  254. * check IPV6 compliance
  255. * generate a mini example using kernel+busybox only (+libc) for example
  256. * more support for advanced linux 2.6.x features, see: iotop
  257. most likely there is more
  258. * even more support for statistics: mpstat, iostat, powertop....
  259. Unicode work needed:
  260. Unicode support uses libc multibyte functions if LOCALE_SUPPORT is on
  261. (in this case, the code will also support many more encodings),
  262. or uses a limited subset of re-implemented multibyte functions
  263. which only understand "one byte == one char" and unicode.
  264. This is useful if you build against uclibc with locale support disabled.
  265. Unicode-dependent applets must call check_unicode_in_env() when they
  266. begin executing.
  267. Applet code may conditionalize on UNICODE_SUPPORT in order to use
  268. more efficient code if unicode support is not requested.
  269. Available functions (if you need more, implement them in libbb/unicode.c
  270. so that they work without LOCALE_SUPPORT too):
  271. int bb_mbstrlen(str) - multibyte-aware strlen
  272. size_t mbstowcs(wdest, src, n)
  273. size_t wcstombs(dest, wsrc, n)
  274. size_t wcrtomb(str, wc, wstate)
  275. int iswspace(wc)
  276. int iswalnum(wc)
  277. int iswpunct(wc)
  278. Applets which only need to align columns on screen correctly:
  279. ls - already done, use source as an example
  280. df
  281. dumpleases
  282. lsmod
  283. Applets which need to account for Unicode chars
  284. while processing the output:
  285. [un]expand
  286. fold
  287. man
  288. watch
  289. cut (-b and -c are currently the same, needs fixing)
  290. These applets need to ensure that unicode input
  291. is handled correctly (say, <unicode><backspace> sequence):
  292. getty, login
  293. rm -i
  294. unzip (overwrite prompt)
  295. Viewers/editors are more difficult (many cases to get right).
  296. libbb/lineedit.c is an example how to do it:
  297. less, most, ed, vi
  298. awk
  299. [ef]grep
  300. sed
  301. Probably needs some specialized work:
  302. loadkeys