busybox/TODO
<<
>>
Prefs
   1Busybox TODO
   2
   3Harvest patches from
   4http://git.openembedded.org/cgit.cgi/openembedded/tree/recipes/busybox/
   5https://dev.openwrt.org/browser/trunk/package/busybox/patches/
   6
   7
   8Stuff that needs to be done.  This is organized by who plans to get around to
   9doing it eventually, but that doesn't mean they "own" the item.  If you want to
  10do one of these bounce an email off the person it's listed under to see if they
  11have any suggestions how they plan to go about it, and to minimize conflicts
  12between your work and theirs.  But otherwise, all of these are fair game.
  13
  14Rob Landley suggested this:
  15  Implement bb_realpath() that can handle NULL on non-glibc.
  16
  17  sh
  18    The command shell situation is a mess.  We have two different
  19    shells that don't really share any code, and the "standalone shell" doesn't
  20    work all that well (especially not in a chroot environment), due to apps not
  21    being reentrant.
  22
  23  Do a SUSv3 audit
  24    Look at the full Single Unix Specification version 3 (available online at
  25    "http://www.opengroup.org/onlinepubs/009695399/nfindex.html") and
  26    figure out which of our apps are compliant, and what we're missing that
  27    we might actually care about.
  28
  29    Even better would be some kind of automated compliance test harness that
  30    exercises each command line option and the various corner cases.
  31
  32  Internationalization
  33    How much internationalization should we do?
  34
  35    The low hanging fruit is UTF-8 character set support.  We should do this.
  36    See TODO_unicode file.
  37
  38    We also have lots of hardwired english text messages.  Consolidating this
  39    into some kind of message table not only makes translation easier, but
  40    also allows us to consolidate redundant (or close) strings.
  41
  42    We probably don't want to be bloated with locale support.  (Not unless we
  43    can cleanly export it from our underlying C library without having to
  44    concern ourselves with it directly.  Perhaps a few specific things like a
  45    config option for "date" are low hanging fruit here?)
  46
  47    What level should things happen at?  How much do we care about
  48    internationalizing the text console when X11 and xterms are so much better
  49    at it?  (There's some infrastructure here we don't implement: The
  50    "unicode_start" and "unicode_stop" shell scripts need "vt-is-UTF8" and a
  51    --unicode option to loadkeys.  That implies a real loadkeys/dumpkeys
  52    implementation to replace loadkmap/dumpkmap.  Plus messing with console font
  53    loading.  Is it worth it, or do we just say "use X"?)
  54
  55  Individual compilation of applets.
  56    It would be nice if busybox had the option to compile to individual applets,
  57    for people who want an alternate implementation less bloated than the gnu
  58    utils (or simply with less political baggage), but without it being one big
  59    executable.
  60
  61    Turning libbb into a real dll is another possibility, especially if libbb
  62    could export some of the other library interfaces we've already more or less
  63    got the code for (like zlib).
  64
  65  buildroot - Make a "dogfood" option
  66    Busybox 1.1 will be capable of replacing most gnu packages for real world
  67    use, such as developing software or in a live CD.  It needs wider testing.
  68
  69    Busybox should now be able to replace bzip2, coreutils, e2fsprogs, file,
  70    findutils, gawk, grep, inetutils, less, modutils, net-tools, patch, procps,
  71    sed, shadow, sysklogd, sysvinit, tar, util-linux, and vim.  The resulting
  72    system should be self-hosting (I.E. able to rebuild itself from source
  73    code).  This means it would need (at least) binutils, gcc, and make, or
  74    equivalents.
  75
  76    It would be a good "eating our own dogfood" test if buildroot had the option
  77    of using a "make allyesconfig" busybox instead of the all of the above
  78    packages.  Anything that's wrong with the resulting system, we can fix.  (It
  79    would be nice to be able to upgrade busybox to be able to replace bash and
  80    diffutils as well, but we're not there yet.)
  81
  82    One example of an existing system that does this already is Firmware Linux:
  83      http://www.landley.net/code/firmware
  84
  85  initramfs
  86    Busybox should have a sample initramfs build script.  This depends on
  87    shell, mdev, and switch_root.
  88
  89  mkdep
  90    Write a mkdep that doesn't segfault if there's a directory it doesn't
  91    have permission to read, isn't based on manually editing the output of
  92    lexx and yacc, doesn't make such a mess under include/config, etc.
  93
  94  Group globals into unions of structures.
  95    Go through and turn all the global and static variables into structures,
  96    and have all those structures be in a big union shared between processes,
  97    so busybox uses less bss.  (This is a big win on nommu machines.)  See
  98    sed.c and mdev.c for examples.
  99
 100  Go through bugs.busybox.net and close out all of that somehow.
 101    This one's open to everybody, but I'll wind up doing it...
 102
 103Bernhard Reutner-Fischer <busybox@busybox.net> suggests to look at these:
 104  New debug options:
 105    -Wlarger-than-127
 106    Cleanup any big users
 107  Collate BUFSIZ IOBUF_SIZE MY_BUF_SIZE PIPE_PROGRESS_SIZE BUFSIZE PIPESIZE
 108    make bb_common_bufsiz1 configurable, size wise.
 109    make pipesize configurable, size wise.
 110    Use bb_common_bufsiz1 throughout applets!
 111
 112As yet unclaimed:
 113
 114----
 115diff
 116  Make sure we handle empty files properly:
 117    From the patch man page:
 118
 119    you can remove a file by sending out a context diff that compares
 120    the file to be deleted with an empty file dated the Epoch.  The
 121    file will be removed unless patch is conforming to POSIX and the
 122    -E or --remove-empty-files option is not given.
 123---
 124patch
 125  Should have simple fuzz factor support to apply patches at an offset which
 126  shouldn't take up too much space.
 127
 128  And while we're at it, a new patch filename quoting format is apparently
 129  coming soon:  http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
 130
 131Architectural issues:
 132
 133bb_close() with fsync()
 134  We should have a bb_close() in place of normal close, with a CONFIG_ option
 135  to not just check the return value of close() for an error, but fsync().
 136  Close can't reliably report anything useful because if write() accepted the
 137  data then it either went out to the network or it's in cache or a pipe
 138  buffer.  Either way, there's no guarantee it'll make it to its final
 139  destination before close() gets called, so there's no guarantee that any
 140  error will be reported.
 141
 142  You need to call fsync() if you care about errors that occur after write(),
 143  but that can have a big performance impact.  So make it a config option.
 144---
 145Unify archivers
 146  Lots of archivers have the same general infrastructure.  The directory
 147  traversal code should be factored out, and the guts of each archiver could
 148  be some setup code and a series of callbacks for "add this file",
 149  "add this directory", "add this symlink" and so on.
 150
 151  This could clean up tar and zip, and make it cheaper to add cpio and ar
 152  write support, and possibly even cheaply add things like mkisofs or
 153  mksquashfs someday, if they become relevant.
 154---
 155Text buffer support.
 156  Several existing applets (sort, vi, less...) read
 157  a whole file into memory and act on it.  Use open_read_close().
 158---
 159Memory Allocation
 160  We have a CONFIG_BUFFER mechanism that lets us select whether to do memory
 161  allocation on the stack or the heap.  Unfortunately, we're not using it much.
 162  We need to audit our memory allocations and turn a lot of malloc/free calls
 163  into RESERVE_CONFIG_BUFFER/RELEASE_CONFIG_BUFFER.
 164  For a start, see e.g. make EXTRA_CFLAGS=-Wlarger-than-64
 165
 166  And while we're at it, many of the CONFIG_FEATURE_CLEAN_UP #ifdefs will be
 167  optimized out by the compiler in the stack allocation case (since there's no
 168  free for an alloca()), and this means that various cleanup loops that just
 169  call free might also be optimized out by the compiler if written right, so
 170  we can yank those #ifdefs too, and generally clean up the code.
 171---
 172FEATURE_CLEAN_UP
 173  This is more an unresolved issue than a to-do item.  More thought is needed.
 174
 175  Normally we rely on exit() to free memory, close files and unmap segments
 176  for us.  This makes most calls to free(), close(), and unmap() optional in
 177  busybox applets that don't intend to run for very long, and optional stuff
 178  can be omitted to save size.
 179
 180  The idea was raised that we could simulate fork/exit with setjmp/longjmp
 181  for _really_ brainless embedded systems, or speed up the standalone shell
 182  by not forking.  Doing so would require a reliable FEATURE_CLEAN_UP.
 183  Unfortunately, this isn't as easy as it sounds.
 184
 185  The problem is, lots of things exit(), sometimes unexpectedly (xmalloc())
 186  and sometimes reliably (bb_perror_msg_and_die() or show_usage()).  This
 187  jumps out of the normal flow control and bypasses any cleanup code we
 188  put at the end of our applets.
 189
 190  It's possible to add hooks to libbb functions like xmalloc() and xopen()
 191  to add their entries to a linked list, which could be traversed and
 192  freed/closed automatically.  (This would need to be able to free just the
 193  entries after a checkpoint to be usable for a forkless standalone shell.
 194  You don't want to free the shell's own resources.)
 195
 196  Right now, FEATURE_CLEAN_UP is more or less a debugging aid, to make things
 197  like valgrind happy.  It's also documentation of _what_ we're trusting
 198  exit() to clean up for us.  But new infrastructure to auto-free stuff would
 199  render the existing FEATURE_CLEAN_UP code redundant.
 200
 201  For right now, exit() handles it just fine.
 202
 203
 204Minor stuff:
 205  watchdog.c could autodetect the timer duration via:
 206    if(!ioctl (fd, WDIOC_GETTIMEOUT, &tmo)) timer_duration = 1 + (tmo / 2);
 207  Unfortunately, that needs linux/watchdog.h and that contains unfiltered
 208  kernel types on some distros, which breaks the build.
 209---
 210  use bb_error_msg where appropriate: See
 211  egrep "(printf.*\([[:space:]]*(stderr|2)|[^_]write.*\([[:space:]]*(stderr|2))"
 212---
 213  use bb_perror_msg where appropriate: See
 214  egrep "[^_]perror"
 215---
 216  possible code duplication ingroup() and is_a_group_member()
 217---
 218  Move __get_hz() to a better place and (re)use it in route.c, ash.c
 219---
 220  See grep -r strtod
 221  Alot of duplication that wants cleanup.
 222---
 223  unify progress_meter. wget, flash_eraseall, pipe_progress, fbsplash, setfiles.
 224---
 225  support start-stop-daemon -d <chdir-path>
 226---
 227
 228(TODO list after discussion 11.05.2009)
 229
 230* shrink tc/brctl/ip
 231  tc/brctl seem like fairly large things to try and tackle in your timeframe,
 232  and i think people have posted attempts in the past. Adding additional
 233  options to ip though seems reasonable.
 234
 235* add tests for some applets
 236
 237* implement POSIX utilities and audit them for POSIX conformance. then
 238  audit them for GNU conformance. then document all your findings in a new
 239  doc/conformance.txt file while perhaps implementing some of the missing
 240  features.
 241  you can find the latest POSIX documentation (1003.1-2008) here:
 242  http://www.opengroup.org/onlinepubs/9699919799/
 243  and the complete list of all utilities that POSIX covers:
 244  http://www.opengroup.org/onlinepubs/9699919799/idx/utilities.html
 245  The first step would to generate a file/matrix what is already archived
 246  (also IPV6)
 247
 248* implement 'at'
 249
 250* rpcbind (former portmap) or equivalent
 251  so that we don't have to use -o nolock on nfs mounts
 252
 253* check IPV6 compliance
 254
 255* generate a mini example using kernel+busybox only (+libc) for example
 256
 257* more support for advanced linux 2.6.x features, see: iotop
 258  most likely there is more
 259