Busybox uses FAST_FUNC macro to tweak with IA-32 calling conventions in
order to make the function call slightly smaller or slightly faster.
However, when I experiment with GCC's LTO (Link Time Optimization), I
discovered that FAST_FUNC could hinder LTO's optimization so that the
resulting executable become a few bytes larger (than what is compiled
without FAST_FUNC).
This change allows to specify e.g.
CONFIG_EXTRA_CFLAGS="-DFAST_FUNC= -flto"
and compile with LTO without a source code hack.
Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
* Fix a bug with a configuration in which the shell's kill builtin
would be mistreated as a killall command (i.e. '-q' works, and
'kill process_name' succeeds when it shouldn't):
CONFIG_ASH_JOB_CONTROL=y
CONFIG_HUSH_KILL=y
# CONFIG_KILL is not set
CONFIG_KILLALL=y
# CONFIG_KILLALL5 is not set
* Optimize out unneeded code when the relevant applets are not
selected.
* Move kbuild lines about shells' kill builtins from Kbuild.src to
kill.c, to accompany the new HAVE_SH_KILL macro. I hope this would
make maintanence a little bit easier.
Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
It's a bit overkill (who would want it off?) but ash already has it
configurable. Let's be symmetric.
Also tweak kbuild logic to use ASH_BUILTIN_ECHO to select echo.o,
not ASH.
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
These parts of the code essentially check whether
stepping back by rep0 goes negative or not.
LZMA SDK from lzma1604.7z has the following in the corresponding places:
... = dic[dicPos - rep0 + (dicPos < rep0 ? dicBufSize : 0)]
Clearly, not loop here.
Technically, "while" here works: if condition is false (because pos
underflowed), it iterates once, adds header.dict_size (a.k.a. dicBufSize),
this makes pos positive but smaller than header.dict_size, and loop exits.
Now we'll just check for negative result of subtraction, which is less code:
function old new delta
unpack_lzma_stream 2659 2641 -18
(I hope 2 Gbyte+ dictionaries won't be in use soon).
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Large nested indented code blocks made more sane with a few gotos.
function old new delta
unzip_main 2491 2519 +28
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Was throwing some build errors:
CONFIG_MODPROBE_SMALL=y
CONFIG_DEPMOD=y
CONFIG_LSMOD=y
CONFIG_MODINFO=y
CONFIG_FEATURE_MODPROBE_SMALL_CHECK_ALREADY_LOADED=y
error: unused variable 'exitcode'
modutils/modprobe-small.c: In function 'modprobe_main':
modutils/modprobe-small.c:1060: error: control reaches end of non-void function
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
(I'm requesting for a review first because I fear such an aggressive
change could lead to bugs. While I observe the sizes have reduced, I
haven't test the functionality of each applet after that. So please
test before merging.)
Aggressively cut off unneeded code when the relevant applets are not
built.
Correct dependencies of FEATURE_MODPROBE_SMALL_OPTIONS_ON_CMDLINE and
FEATURE_MODPROBE_SMALL_CHECK_ALREADY_LOADED.
Don't bother with the '-r' option check if only rmmod is built (assume
true then), or when neither rmmod or mobprobe is built (assume false
then).
Size comparison before and after the change (single applet
configuration):
text data bss dec hex filename
34778 946 112 35836 8bfc old/busybox_DEPMOD
34151 946 112 35209 8989 new/busybox_DEPMOD
34903 946 112 35961 8c79 old/busybox_INSMOD
28316 778 112 29206 7216 new/busybox_INSMOD
35228 962 112 36302 8dce old/busybox_LSMOD
5011 706 40 5757 167d new/busybox_LSMOD
34830 946 112 35888 8c30 old/busybox_MODPROBE
34795 946 112 35853 8c0d new/busybox_MODPROBE
34718 946 112 35776 8bc0 old/busybox_RMMOD
7502 714 104 8320 2080 new/busybox_RMMOD
Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Also made it and printf, type and wait builtins optional.
function old new delta
builtin_kill - 323 +323
bltins1 336 348 +12
builtin_type 114 116 +2
------------------------------------------------------------------------------
(add/remove: 1/0 grow/shrink: 2/0 up/down: 337/0) Total: 337 bytes
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
When httpd proxies a request to another server, it first creates
an AF_INET socket, then resolves the server name to a sockaddr,
then connects to it. This fails if the server name resolves to
an IPv6 address.
This patch ensures that the socket is created with the correct
address family (AF_INET6 if the server resolves to an IPv6 address
and AF_INET otherwise).
Signed-off-by: Laurent Bercot <ska-dietlibc@skarnet.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
With the new "select PLATFORM_LINUX" mechanism
(commit e3b1a1fd28), the PLATFORM_LINUX
option alone no longer has any purpose of changing program behavior or
affecting compiled code. So there is no longer need to prompt user of
this config question.
Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
This example single-applet configuration would trigger the bloat:
CONFIG_FEATURE_SEAMLESS_XZ=y
CONFIG_FEATURE_SEAMLESS_LZMA=y
CONFIG_FEATURE_SEAMLESS_BZ2=y
CONFIG_FEATURE_SEAMLESS_GZ=y
CONFIG_BUNZIP2=y
# CONFIG_ZCAT is not set
# All other applets disabled
Here, the resulting "busybox-bunzip2" binary would contain
unpack_gz_stream, unpack_lzma_stream and unpack_xz_stream functions
code. In other words, the gzip, lzma and xz decompressors' code are
linked into the binary unnecessarily.
This happens because SEAMLESS_MAGIC != 0 and compiler is unable
to figure out that SEAMLESS_MAGIC bit is never set.
Fix this by disabling SEAMLESS_MAGIC option flag (setting its value
to 0) when zcat is disabled. This will help the compiler optimize out
bbunpack() and no longer generate open_zipped() function call.
Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
ix ash "kill %1" not working if CONFIG_ASH is disabled but
ash is launched by 'sh' or 'bash' name
Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
While at it, shorten many field and variable names.
function old new delta
unzip_main 2334 2376 +42
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
commit e19923f665 deleted clearredir()
call in shellexec():
ash: [REDIR] Remove redundant CLOEXEC calls
Upstream commit:
Now that we're marking file descriptors as CLOEXEC in savefd, we no longer
need to close them on exec or in setinputfd.
but it missed one place where we don't set CLOEXEC. Fixing this.
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
This would makes all ash options indented inside "ash" in menuconfig.
It appears that menuconfig has a limit at tracking multiple dependency
lines like this (it looks like a "diamond problem" but I'm not sure if
it is):
---ASH <----------
/ \ ASH_OPTIMIZE_FOR_SIZE
!NOMMU <-*----SH_IS_ASH <----[OR] <--ASH_INTERNAL_GLOB
\ / ASH_RANDOM_SUPPORT
---BASH_IS_ASH <-- [...]
The kconfig-language document [1] states that:
> If a menu entry somehow depends on the previous entry, it can be
> made a submenu of it. First, the previous (parent) symbol must be
> part of the dependency list and then one of these two conditions
> must be true:
> - the child entry must become invisible, if the parent is set to 'n'
[BusyBox ash used to satisfy this, but no longer does]
> - the child entry must only be visible, if the parent is visible
[BusyBox ash configs actually satisfy this, but because of
"diamond" above this might not be easily detected]
So I found out a direct workaround: by making ash options explicitly
depend on !NOMMU, we can tell menuconfig that rule 2 above is satisfied
without any more tracking.
---------------------
/ \
!NOMMU <-*-----ASH <-------- \
\ \ \ ASH_OPTIMIZE_FOR_SIZE
*---SH_IS_ASH <---[OR]-[AND] <--ASH_INTERNAL_GLOB
\ / ASH_RANDOM_SUPPORT
--BASH_IS_ASH <- [...]
So all ash options would now be indented under "ash".
[1] "Documentation/kbuild/kconfig-language.txt" in Linux kernel source
Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Run the namelookup from the main loop so a misspelled first ntp server
name does not block everything forever.
This fixes the following situation which would block forever:
$ sudo ./busybox ntpd -dn -p foobar -p pool.ntp.org
ntpd: bad address 'foobar'
ntpd: bad address 'foobar'
ntpd: bad address 'foobar'
...
New behavior:
ntpd: bad address 'foobar'
ntpd: sending query to 137.190.2.4
ntpd: reply from 137.190.2.4: offset:-1.009775 delay:0.175550 status:0x24 strat:1 refid:0x00535047 rootdelay:0.000000 reach:0x01
ntpd: sending query to 137.190.2.4
ntpd: reply from 137.190.2.4: offset:-1.009605 delay:0.175461 status:0x24 strat:1 refid:0x00535047 rootdelay:0.000000 reach:0x03
ntpd: sending query to 137.190.2.4
ntpd: reply from 137.190.2.4: offset:-1.005327 delay:0.167027 status:0x24 strat:1 refid:0x00535047 rootdelay:0.000000 reach:0x07
ntpd: sending query to 137.190.2.4
ntpd: bad address 'foobar'
ntpd: reply from 137.190.2.4: offset:-1.046349 delay:0.248705 status:0x24 strat:1 refid:0x00535047 rootdelay:0.000000 reach:0x0f
This patch is based on Kaarle Ritvanens work.
http://lists.busybox.net/pipermail/busybox/2016-May/084197.html
function old new delta
ntpd_main 1061 1079 +18
ntp_init 556 560 +4
resolve_peer_hostname 81 75 -6
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 2/1 up/down: 22/-6) Total: 16 bytes
Signed-off-by: Kaarle Ritvanen <kaarle.ritvanen@datakunkku.fi>
Signed-off-by: Natanael Copa <ncopa@alpinelinux.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
archival/unzip.c: In function 'read_next_cdf':
archival/unzip.c:271:8: warning: variable 'org' set but
not used [-Wunused-but-set-variable]
off_t org;
^~~
Signed-off-by: Cristian Ionescu-Idbohrn <cristian.ionescu-idbohrn@axis.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Commit 3a45b87ac3 (modutils: support finit_module syscall) introduced
macro finit_module. But it is not defined for uClibc.
The compilation for busybox fails for MIPS with:
With uClibc, we get following build errors:
modutils/lib.a(modutils.o): In function `bb_init_module':
modutils.c:(.text.bb_init_module+0x94): undefined reference to `finit_module'
modutils.c:(.text.bb_init_module+0xa0): undefined reference to `finit_module'
We can just use syscall() without any need for the
uClibc wrappers.
Newer versions of uClibc-ng (>1.0.20) will remove the
module syscall wrappers.
Found via Buildroot autobuilders:
http://autobuild.buildroot.net/results/556/55655daef23788fb3967f801ec8b79e9bed7122b/build-end.log
function old new delta
bb_delete_module 26 32 +6
bb_init_module 90 95 +5
delete_module 37 - -37
init_module 53 - -53
------------------------------------------------------------------------------
(add/remove: 0/4 grow/shrink: 2/0 up/down: 11/-90) Total: -79 bytes
Reported-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Mention the behavior if user selects CONFIG_SH_IS_ASH but not
CONFIG_ASH. We will be explicit that invocations like "busybox ash"
will not work for such configuration.
Also clarify help text of CONFIG_BASH_IS_* that bash compatibility in
ash is not complete. (It shouldn't be anyway - ash can't support every
bash quirk out there.)
Signed-off-by: Kang-Che Sung <explorer09@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
All other applets are listed simply by their name, no reason why
dumpleases doesn't do that.
Group all udhcpd feature options directly after it.
Put "NOT READY" into udhcpc6 item (some users actually tried to use it,
and complained).
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Commit a8c696bf09 makes ifup and ifdown
individually selectable, but forgets to update the dependency to
IFUPDOWN_UDHCPC_CMD_OPTIONS, so it is not selectable anymore.
This patch fixes the dependency by checking for IFUP or IFDOWN, instead
of the obsolete IFUPDOWN.
Also, it drops dependency on UDHCPC: udhcpc on the target system
does not have to come from the _same_ binary.
Signed-off-by: Jörg Krause <joerg.krause@embedded.rocks>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
libbb/appletlib.c:558:12: warning: 'ingroup' defined but
not used [-Wunused-function]
static int ingroup(uid_t u, gid_t g)
^~~~~~~
That function is used only if FEATURE_SUID_CONFIG is also enabled.
Signed-off-by: Cristian Ionescu-Idbohrn <cristian.ionescu-idbohrn@axis.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
The commit 'ash,hush: set exit code 127 in "sh /does/not/exist" case'
only partly implemented the dash commit '[ERROR] Allow the originator
of EXERROR to set the exit status'. This resulted in incorrect error
codes for a syntax error:
$ )
$ echo $?
0
or a redirection error for a special builtin:
$ rm -f xxx
$ eval cat <xxx
$ echo $?
0
Signed-off-by: Ron Yorston <rmy@pobox.com>
Reported-by: Martijn Dekker <martijn@inlv.org>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
musl does not like including linux/netfilter_ipv4.h
(enum / #define collision in two headers, resulting in "3 = 3"
type situation in enum definition).
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
