TODO 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260
  1. Busybox TODO
  2. Harvest patches from
  3. http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/busybox/
  4. https://dev.openwrt.org/browser/trunk/package/busybox/patches/
  5. Stuff that needs to be done. This is organized by who plans to get around to
  6. doing it eventually, but that doesn't mean they "own" the item. If you want to
  7. do one of these bounce an email off the person it's listed under to see if they
  8. have any suggestions how they plan to go about it, and to minimize conflicts
  9. between your work and theirs. But otherwise, all of these are fair game.
  10. Rob Landley suggested this:
  11. Implement bb_realpath() that can handle NULL on non-glibc.
  12. sh
  13. The command shell situation is a mess. We have two different
  14. shells that don't really share any code, and the "standalone shell" doesn't
  15. work all that well (especially not in a chroot environment), due to apps not
  16. being reentrant.
  17. Do a SUSv3 audit
  18. Look at the full Single Unix Specification version 3 (available online at
  19. "http://www.opengroup.org/onlinepubs/009695399/nfindex.html") and
  20. figure out which of our apps are compliant, and what we're missing that
  21. we might actually care about.
  22. Even better would be some kind of automated compliance test harness that
  23. exercises each command line option and the various corner cases.
  24. Internationalization
  25. How much internationalization should we do?
  26. The low hanging fruit is UTF-8 character set support. We should do this.
  27. See TODO_unicode file.
  28. We also have lots of hardwired english text messages. Consolidating this
  29. into some kind of message table not only makes translation easier, but
  30. also allows us to consolidate redundant (or close) strings.
  31. We probably don't want to be bloated with locale support. (Not unless we
  32. can cleanly export it from our underlying C library without having to
  33. concern ourselves with it directly. Perhaps a few specific things like a
  34. config option for "date" are low hanging fruit here?)
  35. What level should things happen at? How much do we care about
  36. internationalizing the text console when X11 and xterms are so much better
  37. at it? (There's some infrastructure here we don't implement: The
  38. "unicode_start" and "unicode_stop" shell scripts need "vt-is-UTF8" and a
  39. --unicode option to loadkeys. That implies a real loadkeys/dumpkeys
  40. implementation to replace loadkmap/dumpkmap. Plus messing with console font
  41. loading. Is it worth it, or do we just say "use X"?)
  42. Individual compilation of applets.
  43. It would be nice if busybox had the option to compile to individual applets,
  44. for people who want an alternate implementation less bloated than the gnu
  45. utils (or simply with less political baggage), but without it being one big
  46. executable.
  47. Turning libbb into a real dll is another possibility, especially if libbb
  48. could export some of the other library interfaces we've already more or less
  49. got the code for (like zlib).
  50. buildroot - Make a "dogfood" option
  51. Busybox 1.1 will be capable of replacing most gnu packages for real world
  52. use, such as developing software or in a live CD. It needs wider testing.
  53. Busybox should now be able to replace bzip2, coreutils, e2fsprogs, file,
  54. findutils, gawk, grep, inetutils, less, modutils, net-tools, patch, procps,
  55. sed, shadow, sysklogd, sysvinit, tar, util-linux, and vim. The resulting
  56. system should be self-hosting (I.E. able to rebuild itself from source
  57. code). This means it would need (at least) binutils, gcc, and make, or
  58. equivalents.
  59. It would be a good "eating our own dogfood" test if buildroot had the option
  60. of using a "make allyesconfig" busybox instead of the all of the above
  61. packages. Anything that's wrong with the resulting system, we can fix. (It
  62. would be nice to be able to upgrade busybox to be able to replace bash and
  63. diffutils as well, but we're not there yet.)
  64. One example of an existing system that does this already is Firmware Linux:
  65. http://www.landley.net/code/firmware
  66. initramfs
  67. Busybox should have a sample initramfs build script. This depends on
  68. shell, mdev, and switch_root.
  69. mkdep
  70. Write a mkdep that doesn't segfault if there's a directory it doesn't
  71. have permission to read, isn't based on manually editing the output of
  72. lexx and yacc, doesn't make such a mess under include/config, etc.
  73. Group globals into unions of structures.
  74. Go through and turn all the global and static variables into structures,
  75. and have all those structures be in a big union shared between processes,
  76. so busybox uses less bss. (This is a big win on nommu machines.) See
  77. sed.c and mdev.c for examples.
  78. Go through bugs.busybox.net and close out all of that somehow.
  79. This one's open to everybody, but I'll wind up doing it...
  80. Bernhard Reutner-Fischer <busybox@busybox.net> suggests to look at these:
  81. New debug options:
  82. -Wlarger-than-127
  83. Cleanup any big users
  84. Collate BUFSIZ IOBUF_SIZE MY_BUF_SIZE PIPE_PROGRESS_SIZE BUFSIZE PIPESIZE
  85. make bb_common_bufsiz1 configurable, size wise.
  86. make pipesize configurable, size wise.
  87. Use bb_common_bufsiz1 throughout applets!
  88. As yet unclaimed:
  89. ----
  90. diff
  91. Make sure we handle empty files properly:
  92. From the patch man page:
  93. you can remove a file by sending out a context diff that compares
  94. the file to be deleted with an empty file dated the Epoch. The
  95. file will be removed unless patch is conforming to POSIX and the
  96. -E or --remove-empty-files option is not given.
  97. ---
  98. patch
  99. Should have simple fuzz factor support to apply patches at an offset which
  100. shouldn't take up too much space.
  101. And while we're at it, a new patch filename quoting format is apparently
  102. coming soon: http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
  103. Architectural issues:
  104. bb_close() with fsync()
  105. We should have a bb_close() in place of normal close, with a CONFIG_ option
  106. to not just check the return value of close() for an error, but fsync().
  107. Close can't reliably report anything useful because if write() accepted the
  108. data then it either went out to the network or it's in cache or a pipe
  109. buffer. Either way, there's no guarantee it'll make it to its final
  110. destination before close() gets called, so there's no guarantee that any
  111. error will be reported.
  112. You need to call fsync() if you care about errors that occur after write(),
  113. but that can have a big performance impact. So make it a config option.
  114. ---
  115. Unify archivers
  116. Lots of archivers have the same general infrastructure. The directory
  117. traversal code should be factored out, and the guts of each archiver could
  118. be some setup code and a series of callbacks for "add this file",
  119. "add this directory", "add this symlink" and so on.
  120. This could clean up tar and zip, and make it cheaper to add cpio and ar
  121. write support, and possibly even cheaply add things like mkisofs or
  122. mksquashfs someday, if they become relevant.
  123. ---
  124. Text buffer support.
  125. Several existing applets (sort, vi, less...) read
  126. a whole file into memory and act on it. Use open_read_close().
  127. ---
  128. Memory Allocation
  129. We have a CONFIG_BUFFER mechanism that lets us select whether to do memory
  130. allocation on the stack or the heap. Unfortunately, we're not using it much.
  131. We need to audit our memory allocations and turn a lot of malloc/free calls
  132. into RESERVE_CONFIG_BUFFER/RELEASE_CONFIG_BUFFER.
  133. For a start, see e.g. make EXTRA_CFLAGS=-Wlarger-than-64
  134. And while we're at it, many of the CONFIG_FEATURE_CLEAN_UP #ifdefs will be
  135. optimized out by the compiler in the stack allocation case (since there's no
  136. free for an alloca()), and this means that various cleanup loops that just
  137. call free might also be optimized out by the compiler if written right, so
  138. we can yank those #ifdefs too, and generally clean up the code.
  139. ---
  140. FEATURE_CLEAN_UP
  141. This is more an unresolved issue than a to-do item. More thought is needed.
  142. Normally we rely on exit() to free memory, close files and unmap segments
  143. for us. This makes most calls to free(), close(), and unmap() optional in
  144. busybox applets that don't intend to run for very long, and optional stuff
  145. can be omitted to save size.
  146. The idea was raised that we could simulate fork/exit with setjmp/longjmp
  147. for _really_ brainless embedded systems, or speed up the standalone shell
  148. by not forking. Doing so would require a reliable FEATURE_CLEAN_UP.
  149. Unfortunately, this isn't as easy as it sounds.
  150. The problem is, lots of things exit(), sometimes unexpectedly (xmalloc())
  151. and sometimes reliably (bb_perror_msg_and_die() or show_usage()). This
  152. jumps out of the normal flow control and bypasses any cleanup code we
  153. put at the end of our applets.
  154. It's possible to add hooks to libbb functions like xmalloc() and xopen()
  155. to add their entries to a linked list, which could be traversed and
  156. freed/closed automatically. (This would need to be able to free just the
  157. entries after a checkpoint to be usable for a forkless standalone shell.
  158. You don't want to free the shell's own resources.)
  159. Right now, FEATURE_CLEAN_UP is more or less a debugging aid, to make things
  160. like valgrind happy. It's also documentation of _what_ we're trusting
  161. exit() to clean up for us. But new infrastructure to auto-free stuff would
  162. render the existing FEATURE_CLEAN_UP code redundant.
  163. For right now, exit() handles it just fine.
  164. Minor stuff:
  165. watchdog.c could autodetect the timer duration via:
  166. if(!ioctl (fd, WDIOC_GETTIMEOUT, &tmo)) timer_duration = 1 + (tmo / 2);
  167. Unfortunately, that needs linux/watchdog.h and that contains unfiltered
  168. kernel types on some distros, which breaks the build.
  169. ---
  170. use bb_error_msg where appropriate: See
  171. egrep "(printf.*\([[:space:]]*(stderr|2)|[^_]write.*\([[:space:]]*(stderr|2))"
  172. ---
  173. use bb_perror_msg where appropriate: See
  174. egrep "[^_]perror"
  175. ---
  176. possible code duplication ingroup() and is_a_group_member()
  177. ---
  178. Move __get_hz() to a better place and (re)use it in route.c, ash.c
  179. ---
  180. See grep -r strtod
  181. Alot of duplication that wants cleanup.
  182. ---
  183. unify progress_meter. wget, flash_eraseall, pipe_progress, fbsplash, setfiles.
  184. ---
  185. support start-stop-daemon -d <chdir-path>
  186. ---
  187. vdprintf() -> similar sized functionality
  188. ---
  189. (TODO list after discussion 11.05.2009)
  190. * shrink tc/brctl/ip
  191. tc/brctl seem like fairly large things to try and tackle in your timeframe,
  192. and i think people have posted attempts in the past. Adding additional
  193. options to ip though seems reasonable.
  194. * add tests for some applets
  195. * implement POSIX utilities and audit them for POSIX conformance. then
  196. audit them for GNU conformance. then document all your findings in a new
  197. doc/conformance.txt file while perhaps implementing some of the missing
  198. features.
  199. you can find the latest POSIX documentation (1003.1-2008) here:
  200. http://www.opengroup.org/onlinepubs/9699919799/
  201. and the complete list of all utilities that POSIX covers:
  202. http://www.opengroup.org/onlinepubs/9699919799/idx/utilities.html
  203. The first step would to generate a file/matrix what is already archived
  204. (also IPV6)
  205. * implement 'at'
  206. * rpcbind (former portmap) or equivalent
  207. so that we don't have to use -o nolock on nfs mounts
  208. * check IPV6 compliance
  209. * generate a mini example using kernel+busybox only (+libc) for example
  210. * more support for advanced linux 2.6.x features, see: iotop
  211. most likely there is more