aboutsummaryrefslogtreecommitdiff
path: root/shell/ash.c (follow)
Commit message (Collapse)AuthorAgeFilesLines
* win32: use built-in applets for non-existent binaries with Unix-style pathsRon Yorston2020-08-131-9/+46
| | | | | | | | | | | | | | | | | | | | Shell scripts moved from Unix may contain hard-coded paths to binaries such as /bin/sh. A recent commit made it possible to execute such binaries reliably, but that does require them to be installed. As an alternative solution: if a binary with a standard Unix path prefix can't be found but is available as a built-in applet, run the applet. Add the function unix_path() to detect paths starting with /bin, /usr/bin, /sbin or /usr/sbin. Use this function in: - the 'which' applet - shellexec(), describe_command() and find_command() in ash - mingw_spawn_1() See GitHub issue #195.
* win32: handle Unix-style absolute paths for executablesRon Yorston2020-08-131-2/+28
| | | | | | | | | | | | | | | | | | | | | | As noted in commit 548ec7045 (win32: interpret absolute paths as relative to %SYSTEMDRIVE%) a path starting with a '/' in the Unix world is treated as relative to the current drive by Windows. To avoid ambiguity that commit considered certain such paths to be relative to %SYSTEMDRIVE%. Extend this to paths representing executables. Add the functions need_system_drive() and auto_add_system_drive() to detect the need for a system drive prefix and to add it if necessary. Use these functions in: - the 'which' applet - the find_executable() function - tab-completion code - PATH look-up, shellexec(), describe_command() and find_command() in ash - parse_interpreter() and mingw_spawn_1() With these changes executable paths starting with a slash are handled consistently, whatever the current drive.
* win32: use a static buffer in get_system_drive()Ron Yorston2020-08-131-2/+1
| | | | | | | Allocate static storage for the system drive string instead of making a new allocation on every call. This is easier to manage. Adds 16 bytes.
* win32: code shrinkRon Yorston2020-08-131-3/+6
| | | | | | | | | | Add a new function, has_path(), to detect that an executable name doesn't require a path look-up. Also, since is_absolute_path() is now only used in shell/ash.c move its definition there from include/mingw.h. Saves 128 bytes.
* Merge branch 'busybox' into mergeRon Yorston2020-07-091-19/+22
|\
| * nologin: make it possible to build it as single appletDenys Vlasenko2020-06-241-19/+22
| | | | | | | | Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
* | ash: tidy code, no functional changeRon Yorston2020-07-091-3/+2
| |
* | ash: fix shell optionsRon Yorston2020-07-071-7/+7
| | | | | | | | | | | | The Windows-specific shell options winxp, noconsole and nocaseglob have had the wrong indices in optlist since the merge with upstream BusyBox in commit 517cf74f6.
* | ash: improve handling of UNC pathsRon Yorston2020-06-301-37/+54
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Be more strict about identifying UNC paths in unc_root_len(). In updatepwd() in ash: - Skip duplicate leading slashes unless the directory is a UNC path. - Rewrite detection and handling of the five possible types of path. This improves cases like 'cd ///' and 'cd /xyz' when the current directory is a UNC path. See GitHub issue #192.
* | ash: avoid duplicated slashes in output of type/command builtinsRon Yorston2020-06-291-0/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Microsoft developers have a penchant for adding trailing slashes to entries in PATH: C:/Windows/System32/WindowsPowerShell/v1.0/ C:/Windows/System32/OpenSSH/ The 'type' and 'command -v' shell builtins return paths with duplicated slashes for executables in those directories. See GitHub issue #191. Bonus fixes: - handle backslashes as well as slashes in concat_path_file() - convert backslashes to slashes in the output of 'type', 'command -v'
* | ash: skip NULL argv during forkshellRon Yorston2020-06-211-67/+44
| | | | | | | | | | | | | | | | Alter argv_size()/argv_copy() to omit argv from the forkshell block if it's NULL. Improve forkshell debug code to properly account for parts of funcblock which aren't being used.
* | ash: changes to ctrl-c handlingRon Yorston2020-06-201-57/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Make ctrl-c handling more like Unix. Remove the hSIGINT event and the code in waitpid_child() to catch it. Add a call to raise_interrupt() in ctrl_handler(). Prior to these changes interrupts in a child shell weren't properly handled. In release FRP-3466-g53c09d0e1 interrupting a sleep in a child shell results in both shells competing for input: ~ $ sh ~ $ sleep 10 ^C^C ~ $ ~ $ pwd sh: w: not found ~ $ sh: pd: not found ~ $ With recent changes: ~ $ sh ~ $ sleep 10 ^C ~ $ echo $? 130 ~ $ exit ^C ~ $ echo $? 130 ~ $ The error return from the parent shell is due to the lack of job control. Upstream BusyBox ash and dash both do the same when job control is disabled.
* | ash: run ash_main() directly from a FS_SHELLEXEC shellRon Yorston2020-06-201-3/+6
| |
* | ash: reduce forkshell block size for FS_SHELLEXECRon Yorston2020-06-201-25/+46
| | | | | | | | | | It isn't necessary to pass the alias table or line editing history to a forkshell shell that's handling FS_SHELLEXEC.
* | ash: reduce forkshell block size for FS_OPENHERERon Yorston2020-06-201-33/+50
| | | | | | | | | | | | | | | | When handling FS_OPENHERE the forkshell data block only needs to contain the forkshell structure and the here document. Omit everything else. Update forkshell_print() for this case.
* | ash: forkshell code shrinkRon Yorston2020-06-201-18/+8
| | | | | | | | | | Don't pass global pointers to forkshell_size()/forkshell_copy() via the forkshell structure, just reference them directly.
* | ash: treat all applets as NOEXECRon Yorston2020-06-091-0/+5
| | | | | | | | | | | | | | | | | | In standalone shell mode tryexec() treats all applets as NOEXEC. Except for the shell: calling ash_main() recursively seems unwise. This avoids creating a new process for the applet (which is a Good Thing on Microsoft Windows where creating processes is expensive) but leaves the resources used by the invoking shell in the current process.
* | ash: minimise work when a forkshell process is startedRon Yorston2020-06-091-19/+27
| | | | | | | | | | | | | | | | | | | | | | There's no need to create or initialise the global data structures when a forkshell is being started: they'll soon be replaced by data from the forkshell data block. Move the call to forkshell_init() to the top of ash_main() along with some initialisation that's common to forkshell shells and ordinary shells. Add the minimum necessary initialisation to forkshell_init().
* | ash: don't initialise shell variables in forkshell childRon Yorston2020-06-011-2/+9
| | | | | | | | | | | | | | | | | | | | The forkshell child process calls init(), which mostly just initialises shell variables. Since these variables are overwritten with values from the parent process there's no point in doing that. In particular, init() initialises HOSTNAME by calling gethostname() which involves network access and is slow. Avoid this cost when we know the shell is a forkshell child.
* | ash: simplify spawning during forkshellRon Yorston2020-06-011-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | spawn_forkshell() uses mingw_spawn_proc() to start the child shell. mingw_spawn_proc() then calls mingw_spawn_1() which determines that "sh" is an applet, thus calling mingw_spawn_applet() which finally calls spawnveq(). Not only is this convoluted it also won't work if PREFER_APPLETS and SH_STANDALONE aren't enabled. Simplify matters by adding a new function, mingw_spawn_forkshell(), which is tailored for just this case.
* | ash: minor changes to forkshell error handlingRon Yorston2020-06-011-3/+9
| | | | | | | | | | | | | | | | forkshell_prepare() now returns NULL if it fails to map its view of the data block. Detect and report this error in spawn_forkshell(). If forkshell_init() fails it now returns to ash_main() where a generic error is reported.
* | ash: improve forkshell debuggingRon Yorston2020-06-011-4/+14
| | | | | | | | | | | | | | | | | | The name of the output file depended only on the pid of the shell. Since a single shell can execute many forkshells it's useful to include a serial number in the filename so it isn't constantly overwritten. Also, output the type of forkshell.
* | Merge branch 'busybox' into mergeFRP-3466-g53c09d0e1Ron Yorston2020-05-291-3/+4
|\|
| * ash,hush: drop pointer check before calls to show_historyRon Yorston2020-04-301-2/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | show_history() checks that its argument in non-null so there's no need to repeat the test at call sites. function old new delta historycmd 25 17 -8 builtin_history 29 21 -8 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 0/2 up/down: 0/-16) Total: -16 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: fix build failure when command built-in is disabledRon Yorston2020-04-301-1/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Since commit 7eb8eecbb (ash: eval: Add assignment built-in support again) building BusyBox with the 'command' built-in disabled fails. parse_command_args() only needs to be called when the 'command' built-in is run. Which it won't be if it's disabled. v2: Avoiding infinite loops is good, too. Thanks, Harald van Dijk. Reported-by: Deweloper <deweloper@wp.pl> Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
* | win32: changes to '#!' supportRon Yorston2020-05-241-2/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | Since the earliest days of busybox-w32 '#!' has searched PATH for the interpreter. This doesn't seem to be supported by precedent. In testing I also found that additional code was needed for the case where PATH has been altered in the current shell or is modified on the command line: PATH has to be extracted from envp rather than getenv("PATH"). Drop this non-standard feature. *NIX implementations disagree on whether the interpreter can itself be a script. Follow Linux and allow this with a limit of four levels of nesting.
* | ash: reset ANSI emulation when BB_SKIP_ANSI_EMULATION changesRon Yorston2020-04-131-0/+11
| | | | | | | | | | Update the skip status on the first call to skip_ansi_emulation() and whenever the BB_SKIP_ANSI_EMULATION variable changes.
* | Fix POSIX buildFRP-3445-g10e14d5ebRon Yorston2020-04-111-1/+1
| |
* | win32: improve handling of 'c:file' paths for executablesRon Yorston2020-03-181-2/+10
| | | | | | | | | | | | | | | | bb_basename() didn't properly handle paths of the form 'c:file'. Attempting to run 'c:busybox' resulted in 'c:busybox: applet not found'. The shell had a similar problem: trying to run 'c:busybox' reported 'sh: c:busybox: not found'.
* | Merge branch 'busybox' into mergeRon Yorston2020-02-251-207/+183
|\|
| * ash: expand: Fix multiple issues with EXP_DISCARD in evalvarDenys Vlasenko2020-02-241-14/+14
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Wed, 12 Sep 2018 14:27:16 +0800 expand: Fix multiple issues with EXP_DISCARD in evalvar The commit 3cd538634f71538370f5af239f342aec48b7470b broke parameter expansion in multiple ways because the EXP_DISCARD flag wasn't set or tested for various cases: $ src/dash -c 'var=; echo ${var:+nonempty}' nonempty $ src/dash -u -c 'unset foo bar; echo ${foo+${bar}}' dash: 1: bar: parameter not set $ src/dash -c 'foo=bar; echo ${foo=BUG}; echo $foo' barBUG bar $ This patch fixes them by introducing a new discard variable that tracks whether the extra word should be discarded or not when it is parsed. Reported-by: Martijn Dekker <martijn@inlv.org> Fixes: 3cd538634f71 ("expand: Do not reprocess data when...") Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Reported-by: Martijn Dekker <martijn@inlv.org> Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: expand: Do not reprocess data when expanding wordsDenys Vlasenko2020-02-241-178/+154
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream patch: Date: Wed, 30 May 2018 02:06:03 +0800 expand: Do not reprocess data when expanding words Currently various paths will reprocess data when performing word expansion. For example, expari will skip backwards looking for the start of the arithmetic expansion, while evalvar will skip unexpanded words manually. This is cumbersome and error-prone. This patch fixes this by making word expansions proceed in a linear fashion. This means changing argstr and the various expansion functions such as expari and subevalvar to return the next character to be expanded. This is inspired by similar code from FreeBSD. However, we take things one step further and completely remove the manual word skipping in evalvar. This is accomplished by introducing a new EXP_DISCARD flag that tells argstr to only parse and not produce any actual expansions. Incidentally, argstr will now always NUL-terminate the expansion unless the EXP_WORD flag is set. This is because all but one caller of argstr wants the result to be NUL-termianted. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Also includes two one-line follow-up fixes: expand: Eat closing brace for length parameter if (subtype == VSLENGTH) { + p++; if (flag & EXP_DISCARD) expand: Fix double-decrement in argstr - newloc = expdest - (char *)stackblock() - end; + newloc = q - (char *)stackblock() - end; and changes in code for bash substring extensions. Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: move TRACE statement in evalcommand()Ron Yorston2020-02-241-1/+2
| | | | | | | | | | | | | | | | Following recent work on evalcommand() a TRACE statement to report the status of a forked command was left in the wrong place. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: rename some function parameters to match dashDenys Vlasenko2020-02-241-29/+29
| | | | | | | | Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: rename stack_nputstr() back to stnputs() to match dashDenys Vlasenko2020-02-241-3/+3
| | | | | | | | Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
* | Merge branch 'busybox' into mergeRon Yorston2020-02-251-54/+56
|\|
| * ash: Expand here-documents in the current shell environmentDenys Vlasenko2020-02-221-13/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Sun, 11 Nov 2007 15:27:00 +0800 Expand here-documents in the current shell environment Previously we always expanded here-documents in a subshell. This is contrary to the POSIX specification and how other shells behave. What's more this slows down many expansions due to the extra fork (however, it must be said that it is possible for it speed up certain expansions by running it simultaneously with the command on two CPUs). This patch move the expansion into the current shell environment. Test case: unset a cat <<- EOF > /dev/null ${a=NOT} EOF echo ${a}BAD Old result: BAD New result: NOTBAD Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: parser: Do not push token back before parseheredocDenys Vlasenko2020-02-221-1/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Mon, 19 Nov 2018 18:43:58 +0800 parser: Do not push token back before parseheredoc When we read the first token in list() we use peektoken instead of readtoken as the following code needs to use the same token again. However, this is wrong when we're in a here-document as it will clobber the saved token without resetting the tokpushback flag. This patch fixes it by doing the tokpushback after parseheredoc and setting lasttoken again if parseheredoc was called. Reported-by: Ron Yorston <rmy@frippery.org> Fixes: 7c245aa8ed33 ("[PARSER] Simplify EOF/newline handling in...") Fixes: ee5cbe9fd6bc ("[SHELL] Optimize dash -c "command" to avoid a fork") Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Tested-by: Simon Ser <contact@emersion.fr> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: parser: Fix old-style command substitution here-document crashDenys Vlasenko2020-02-221-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Fri, 29 Mar 2019 13:49:59 +0800 parser: Fix old-style command substitution here-document crash ... This is caused by the recent change to save/restore here-docment list around command substitutions. In doing so we must finish existing here-documents prior to restoring the old here-document list. This is done for new-style command substitutions but not for old-style. This patch fixes it by doing it for both. Fixes: 51e2d88d6e51 ("parser: Save/restore here-documents in...") Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: expand: Ensure result is escaped in cvtnumDenys Vlasenko2020-02-211-25/+24
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Fri, 1 Jun 2018 18:25:29 +0800 expand: Ensure result is escaped in cvtnum The minus sign generated from arithmetic expansion is currently unquoted which causes anomalies when the result is used in where the quoting matters. This patch fixes it by explicitly calling memtodest on the result in cvtnum. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: memalloc: Avoid looping in growstacktoDenys Vlasenko2020-02-211-7/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Thu, 31 May 2018 01:51:48 +0800 memalloc: Avoid looping in growstackto Currently growstackto will repeatedly call growstackblock until the requisite size is obtained. This is wasteful. This patch changes growstackblock to take a minimum size instead. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
* | Merge branch 'busybox' into mergeRon Yorston2020-02-211-74/+88
|\|
| * ash: eval: Only restore exit status on exit/returnDenys Vlasenko2020-02-201-48/+51
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Fri, 14 Dec 2018 13:52:02 +0800 eval: Only restore exit status on exit/return We unconditionally restore the saved status in exitreset, which is incorrect as we only want to do it for exitcmd and returncmd. This patch fixes the problem by introducing EXEND. Reported-by: Martijn Dekker <martijn@inlv.org> Fixes: da30b4b78769 ("[BUILTIN] Exit without arguments in a trap...") Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: Return without arguments in a trap should use status outside trapsDenys Vlasenko2020-02-201-5/+18
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Fixes exitcode_trap4.tests. Upstream commit: Date: Mon, 6 Oct 2014 21:51:26 +0800 Return without arguments in a trap should use status outside traps POSIX now requires that return without arguments in a trap should return the last command status prior to executing traps. This patch implements this behaviour. Incidentally this also changes the behaviour of return without arguments in a loop conditional to use the last exit status in the body as opposed to the last command in the conditional when there is one. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: return exit status of nofork applets (again)Ron Yorston2020-02-201-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Since commit d81af7216 (ash: eval: Reap zombies after built-in commands and functions) if the shell is compiled with SH_STANDALONE and SH_NOFORK enabled nofork applets potentially return the incorrect status. The status value returned by evalcommand() in this case is obtained from exitstatus in a call to waitforjob(NULL). This overwrites the status set for nonfork applets. If this commit seems familiar it's essentially a reversion of commit 5ccb0e92fa (ash: return exit status of nofork applets). What was correct in 2016 is wrong in 2020 and vice versa. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: expand: Merge syntax/quotes in memtodest with flagsDenys Vlasenko2020-02-201-21/+19
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Mon, 28 May 2018 00:17:39 +0800 expand: Merge syntax/quotes in memtodest with flags The function arguments syntax and quotes are both derived from the expansion flags. As syntax is only used by memtodest we do not need to maintain it outside of the function at all. The only place that uses something other than BASESYNTAX or DQSYNTAX is exptilde. However in that case DQSYNTAX has exactly the same effect as SQSYNTAX. This patch merges these two arguments into a single flags. The macro QUOTES_KEEPNUL has been renamed to EXP_KEEPNUL in order to keep the namespace separate. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
* | ash: merge changes to local variables from upstreamRon Yorston2020-02-211-175/+136
|\| | | | | | | | | | | | | | | | | This brings in dash commit cbb71a8 (eval: Add assignment built-in support again) which finally makes it possible to sort out local commit 65189dacb (ash: fix local PATH assignments). These changes also simplify handling of shellexec and remove the need for varlist in the forkshell structure.
| * ash: eval: Always set localvar_stopDenys Vlasenko2020-02-201-11/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Thu, 31 May 2018 01:15:34 +0800 eval: Always set localvar_stop The variable localvar_stop is set iff vlocal is true. gcc doesn't get this so we get a spurious warning. This patch fixes this by always calling pushlocalvars with vlocal and making it only actually do the push if vlocal is non-zero. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: eval: Replace with listsetvar with mklocal/setvareqDenys Vlasenko2020-02-201-35/+28
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Sat, 19 May 2018 02:39:55 +0800 eval: Replace with listsetvar with mklocal/setvareq This patch replaces listsetvar with mklocal/setvareq. As we now determine special built-in status prior to variable assignment, we no longer have to do a second pass listsetvar. Instead we will call setvareq directly instead of mklocal when necessary. In order to do this mklocal can now take a flag in order to mark a variable for export. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
| * ash: eval: Fail immediately with redirections errors for simple commandDenys Vlasenko2020-02-201-16/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Upstream commit: Date: Sat, 19 May 2018 02:39:54 +0800 eval: Fail immediately with redirections errors for simple command Previously, dash would continue to perform variable expansions even if a redirection error occured. This patch changes it so that it fails immediately. Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-03-22 22:23:46 +0000