source: branches/3.2/mindi-busybox/TODO @ 3232

Last change on this file since 3232 was 3232, checked in by Bruno Cornec, 7 years ago
  • Update mindi-busybox to 1.21.1
File size: 11.9 KB
1Busybox TODO
3Harvest patches from
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.
14Rob Landley suggested this:
15  Implement bb_realpath() that can handle NULL on non-glibc.
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.
23  Do a SUSv3 audit
24    Look at the full Single Unix Specification version 3 (available online at
25    "") and
26    figure out which of our apps are compliant, and what we're missing that
27    we might actually care about.
29    Even better would be some kind of automated compliance test harness that
30    exercises each command line option and the various corner cases.
32  Internationalization
33    How much internationalization should we do?
35    The low hanging fruit is UTF-8 character set support.  We should do this.
36    See TODO_unicode file.
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.
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?)
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"?)
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.
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).
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.
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.
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.)
82    One example of an existing system that does this already is Firmware Linux:
85  initramfs
86    Busybox should have a sample initramfs build script.  This depends on
87    shell, mdev, and switch_root.
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.
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.
100  Go through and close out all of that somehow.
101    This one's open to everybody, but I'll wind up doing it...
103Bernhard Reutner-Fischer <> suggests to look at these:
104  New debug options:
105    -Wlarger-than-127
106    Cleanup any big users
108    make bb_common_bufsiz1 configurable, size wise.
109    make pipesize configurable, size wise.
110    Use bb_common_bufsiz1 throughout applets!
112As yet unclaimed:
116  Make sure we handle empty files properly:
117    From the patch man page:
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.
125  Should have simple fuzz factor support to apply patches at an offset which
126  shouldn't take up too much space.
128  And while we're at it, a new patch filename quoting format is apparently
129  coming soon:
131stty / catv
132  stty's visible() function and catv's guts are identical. Merge them into
133  an appropriate libbb function.
135struct suffix_mult
136  Several duplicate users of: grep -r "1024\*1024" * -B2 -A1
137  Merge to a single size_suffixes[] in libbb.
138  Users: head tail od_bloaty hexdump and (partially as it wouldn't hurt) svlogd
141  ./busybox tail -f foo.c~ TODO
142  should not print fmt=header_fmt for subsequent date >> TODO; i.e. only
143  fmt+ if another (not the current) file did change
145Architectural issues:
147bb_close() with fsync()
148  We should have a bb_close() in place of normal close, with a CONFIG_ option
149  to not just check the return value of close() for an error, but fsync().
150  Close can't reliably report anything useful because if write() accepted the
151  data then it either went out to the network or it's in cache or a pipe
152  buffer.  Either way, there's no guarantee it'll make it to its final
153  destination before close() gets called, so there's no guarantee that any
154  error will be reported.
156  You need to call fsync() if you care about errors that occur after write(),
157  but that can have a big performance impact.  So make it a config option.
159Unify archivers
160  Lots of archivers have the same general infrastructure.  The directory
161  traversal code should be factored out, and the guts of each archiver could
162  be some setup code and a series of callbacks for "add this file",
163  "add this directory", "add this symlink" and so on.
165  This could clean up tar and zip, and make it cheaper to add cpio and ar
166  write support, and possibly even cheaply add things like mkisofs or
167  mksquashfs someday, if they become relevant.
169Text buffer support.
170  Several existing applets (sort, vi, less...) read
171  a whole file into memory and act on it.  Use open_read_close().
173Memory Allocation
174  We have a CONFIG_BUFFER mechanism that lets us select whether to do memory
175  allocation on the stack or the heap.  Unfortunately, we're not using it much.
176  We need to audit our memory allocations and turn a lot of malloc/free calls
178  For a start, see e.g. make EXTRA_CFLAGS=-Wlarger-than-64
180  And while we're at it, many of the CONFIG_FEATURE_CLEAN_UP #ifdefs will be
181  optimized out by the compiler in the stack allocation case (since there's no
182  free for an alloca()), and this means that various cleanup loops that just
183  call free might also be optimized out by the compiler if written right, so
184  we can yank those #ifdefs too, and generally clean up the code.
187  This is more an unresolved issue than a to-do item.  More thought is needed.
189  Normally we rely on exit() to free memory, close files and unmap segments
190  for us.  This makes most calls to free(), close(), and unmap() optional in
191  busybox applets that don't intend to run for very long, and optional stuff
192  can be omitted to save size.
194  The idea was raised that we could simulate fork/exit with setjmp/longjmp
195  for _really_ brainless embedded systems, or speed up the standalone shell
196  by not forking.  Doing so would require a reliable FEATURE_CLEAN_UP.
197  Unfortunately, this isn't as easy as it sounds.
199  The problem is, lots of things exit(), sometimes unexpectedly (xmalloc())
200  and sometimes reliably (bb_perror_msg_and_die() or show_usage()).  This
201  jumps out of the normal flow control and bypasses any cleanup code we
202  put at the end of our applets.
204  It's possible to add hooks to libbb functions like xmalloc() and xopen()
205  to add their entries to a linked list, which could be traversed and
206  freed/closed automatically.  (This would need to be able to free just the
207  entries after a checkpoint to be usable for a forkless standalone shell.
208  You don't want to free the shell's own resources.)
210  Right now, FEATURE_CLEAN_UP is more or less a debugging aid, to make things
211  like valgrind happy.  It's also documentation of _what_ we're trusting
212  exit() to clean up for us.  But new infrastructure to auto-free stuff would
213  render the existing FEATURE_CLEAN_UP code redundant.
215  For right now, exit() handles it just fine.
218Minor stuff:
219  watchdog.c could autodetect the timer duration via:
220    if(!ioctl (fd, WDIOC_GETTIMEOUT, &tmo)) timer_duration = 1 + (tmo / 2);
221  Unfortunately, that needs linux/watchdog.h and that contains unfiltered
222  kernel types on some distros, which breaks the build.
224  use bb_error_msg where appropriate: See
225  egrep "(printf.*\([[:space:]]*(stderr|2)|[^_]write.*\([[:space:]]*(stderr|2))"
227  use bb_perror_msg where appropriate: See
228  egrep "[^_]perror"
230  possible code duplication ingroup() and is_a_group_member()
232  Move __get_hz() to a better place and (re)use it in route.c, ash.c
234  See grep -r strtod
235  Alot of duplication that wants cleanup.
237  in_ether duplicated in network/{interface,ifconfig}.c
239  unify progress_meter. wget, flash_eraseall, pipe_progress, fbsplash, setfiles.
241  support start-stop-daemon -d <chdir-path>
243vdprintf() -> similar sized functionality
246(TODO list after discussion 11.05.2009)
248* shrink tc/brctl/ip
249  tc/brctl seem like fairly large things to try and tackle in your timeframe,
250  and i think people have posted attempts in the past. Adding additional
251  options to ip though seems reasonable.
253* add tests for some applets
255* implement POSIX utilities and audit them for POSIX conformance. then
256  audit them for GNU conformance. then document all your findings in a new
257  doc/conformance.txt file while perhaps implementing some of the missing
258  features.
259  you can find the latest POSIX documentation (1003.1-2008) here:
261  and the complete list of all utilities that POSIX covers:
263  The first step would to generate a file/matrix what is already archived
264  (also IPV6)
266* implement 'at'
268* rpcbind (former portmap) or equivalent
269  so that we don't have to use -o nolock on nfs mounts
271* check IPV6 compliance
273* generate a mini example using kernel+busybox only (+libc) for example
275* more support for advanced linux 2.6.x features, see: iotop
276  most likely there is more
278* even more support for statistics: mpstat, iostat, powertop....
Note: See TracBrowser for help on using the repository browser.