| Age | Commit message (Collapse) | Author |
|---|
|
The following three cases behave identical in bash(1), but our ksh
(ksh93 also) fails to run the trap in the last case:
(non-zero exit code is trigger, no redirection)
$ ksh -c 'trap "echo ERR" ERR ; false'
ERR
(failed redirection is trigger, 'echo' was not executed)
$ ksh -c 'trap "echo ERR" ERR ; echo >/'
ksh: cannot create /: Is a directory
ERR
(failed redirection, no execution, trap was NOT triggered)
$ ksh -c 'trap "echo ERR" ERR ; exec >/'
ksh: cannot create /: Is a directory
bash(1) prints "ERR" in all three cases, as expected.
ksh93 behaves like our ksh(1).
In ksh `exec' is a builtin (CSHELL), but also special (SPEC_BI):
$ type alias
alias is a shell builtin
$ type exec
exec is a special shell builtin
Without command and redirection alone, `exec' permanently redirects I/O for
the shell itself, not executing anything; it is the only (special) builtin
with such a special use-case, implemented as c_sh.c:c_exec().
This corner-case is overlooked in exec.c:execute() which handles iosetup()
failure for all commands, incl. builtins.
Exclude c_exec() from the rest of special builtins to ensure it runs the
ERR trap as expected:
$ ./obj/ksh -c 'trap "echo ERR" ERR ; exec >/'
ksh: cannot create /: Is a directory
ERR
Also add three new regress cases covering this; rest keep passing.
OK millert
|
|
OK millert
|
|
nudge from luka krmpotic
|
|
jmc@ dislikes a comma before "then" in a conditional, so leave those
untouched.
ok jmc@
|
|
not print the shell script line number where this occured. Doing so is
pointless, or an information leak.
This change does not affect any other error reporting.
ok millert
|
|
is not unneccesary. Different buffer sizes are actually dangerous, though
major problems are strangely rare.
ok millert
|
|
3rd (variadic) mode_t parameter is irrelevant. Many developers in the past
have passed mode_t (0, 044, 0644, or such), which might lead future people
to copy this broken idiom, and perhaps even believe this parameter has some
meaning or implication or application. Delete them all.
This comes out of a conversation where tb@ noticed that a strange (but
intentional) pledge behaviour is to always knock-out high-bits from
mode_t on a number of system calls as a safety factor, and his bewilderment
that this appeared to be happening against valid modes (at least visually),
but no sorry, they are all irrelevant junk. They could all be 0xdeafbeef.
ok millert
|
|
invalid matches and unexpected behaviour.
Fix this by instead making a NUL character abort the search-history mode,
leaving the handling of said input to the "ordinary" command editing.
ok tb@
|
|
This avoids a cpu loop for "while do done" and is consistent with
the behavior of AT&T ksh and most other shells. OK jca@ halex@
|
|
4-byte UTF-8 sequences and not just some of them, to keep them together
and avoid passing them on byte by byte, helping tools like tmux(1).
While here, also do all the range tests with < and > rather than &
for uniformity and readability, and add some comments.
Input and OK jca@ and nicm@.
Soeren at Soeren dash Tempel dot net originally reported the bug
and provided an incomplete patch that was used as a starting point,
and he also tested this final patch.
|
|
Add a prominent deprecation notice to getopt.1.
Add examples of the getopts idiom to sh.1 and ksh.1.
Requested by and ok espie@, ok jmc@
|
|
Fixes a portability issue. From Benjamin Baier
|
|
so escape it;
|
|
From gotroyb127, OK tb@
|
|
From gotroyb127 OK tb@
|
|
|
|
and installing USD/SMM/PSD docs.
jmc@ agrees with the direction, ok millert@ on an earlier diff
|
|
This results, e.g., in allowing the first item of a read-only array to
be overwritten, as found by Jordan Geoghegan.
okay tb@
|
|
Mark the pointer used to walk the array in ksh const as well.
From Matthew Martin
ok guenther
|
|
This is similar to the emacs mode clear-screen command. Unlike
bash, but like zsh, ^L also works in input mode, not just command
mode. OK kn@ tb@
|
|
One uncommon but useful way of writing shell scripts is to start off by
disabling field/word splitting (IFS='') and pathname expansion/globbing
(set -f), re-enabling either or both only for the commands that need
them, e.g. within a subshell. This helps avoid a lot of snags with field
splitting and globbing if you forget to quote a variable somewhere,
adding to the general robustness of a script. (In fact it eliminates
much of the need to quote variable/parameter expansions, with empty
removal remaining as the only issue.)
Unfortunately OpenBSD ksh (like all pdksh variants except mksh) has a
POSIX compliance bug that is a show stopper for this approach: "$@" does
not generate words (arguments) if IFS is empty. As a result, the
separate command arguments represented by "$@" become a single argument.
So passing on an intact set of positional parameters to a command or
function is impossible with field splitting disabled.
Of course this is illogical: the quoted special parameter "$@" generates
zero or more words, it doesn't split any words, so the contents of IFS
(or lack thereof) should be neither here nor there. It's old ksh88
behaviour copied by the original pdksh, but it violates POSIX and it has
been fixed many years ago in ksh93 and all other POSIX shells.
From Martijn Dekker (who also wrote the above paragraphs) back in 2016.
Thanks to Avi Halachmi for reminding us of the issue.
ok czarkoff deraadt kn
|
|
OK tb@
|
|
With the pipefail option set, the exit status of a pipeline is 0 if all
commands succeed, or the return status of the rightmost command that
fails. This can help stronger error checking, but is not a silver
bullet. For example, commands will exhibit a non-zero exit status if
they're killed by a SIGPIPE when writing to a pipe. Yet pipefail was
considered useful enough to be included in the next POSIX standard.
This implementation remembers the value of the pipefail option when
a pipeline is started, as described as option 1) in
https://www.austingroupbugs.net/view.php?id=789#c4102
Requested by ajacoutot@, ok millert@
|
|
terminate the shell when running under -e.
See also https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=269067 and
Bug reported including fix by Leah Neukirchen, Thanks!
ok millert@
|
|
Mixing up function and void pointers isn't defined by POSIX or the
C standard. POSIX only specifies that casting the result of dlsym(3) to
an appropriate function pointer works.
Avoid all this by using a typedef.
from Michael Forney, ok tb@
|
|
the environment at shell initialization time. During startup, ksh calls
'eval typeset -i TMOUT="${TMOUT:-0}"'. which allows command injection via
arithmetic expansion, e.g., by setting TMOUT to 'x[`/bin/echo Hi >&2`]'.
Problem noted by Andras Farkas and tj, inspired by a similar issue in
AT&T's ksh. Tested in snaps for two weeks.
"go for it" deraadt
|
|
from chohag
|
|
|
|
Using alloc.c for the history array brings no value and prevents
easy handling of memory shortage. Switch to plain reallocarray and
keep running if HISTSIZE is too big.
The allocation is still done upfront with no sanity checking, it would
probably be nicer to allocate space as needed.
Issue reported by thomas@habets.se who suggested a different approach.
|
|
Spotted by maya@netbsd
ok deraadt@
|
|
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.
|
|
ok millert nicm tb, etc
|
|
|
|
Reference the First Edition (1989) of Bolsky/Korn which is about ksh88,
the shell the OpenBSD ksh(1) descends from (via pdksh).
The Second Edition (1995) of the book is about ksh93 which we don't provide.
Pointed out by Andras Farkas on bugs@.
|
|
|
|
Previously they were only recognized in [[ ... ]] expressions. This
changes sh/ksh to be consistent with test(1) as well as shells like
bash and dash. OK jca@ jmc@
|
|
omission reported by Rudolf Sykora <rsykora at disroot dot org> on misc@;
tweak and OK jmc@
|
|
Slightly more useful for some, same defaults as bash.
No objection deraadt@ phessler@, ok tb@ kn@ benno@
|
|
never do substitution (neither parameter, nor command, nor arithmetic,
nor tilde substitution) on the values of any variables encountered
inside the expression, but do recursively perform arithmetical
evaluation of subexpressions as required. This makes behaviour
more consistent, without hindering any behaviour promised in the
manual page.
A quirk originally reported by Andy Chu <andychup at gmail dot com>
was that in the past, when encountering an array index, the shell
would not only do evaluation, but also substitution on the array
index, even though substitution would not be done on the expression
in general.
tobias@ contributed to initial efforts of understanding the quirk.
patch tested in a bulk build by naddy@
"please commit" deraadt@
|
|
|
|
1. Another off-by-one: if a mail file name ends in an (escaped)
percent sign, do not forget to check whether the next byte is the
percent sign introducing the message (MAILPATH='filename\%%msg').
2. If the message is empty, use the default message rather than
printing a blank line (MAILPATH='filename%').
3. If the file name is empty, don't bother with mballoc(): the
subsequent stat(2) can never succeed. (MAILPATH='%msg').
Found while reviewing the previous commit by tedu@.
OK tedu@.
|
|
ok deraadt schwarze tb
|
|
|
|
ok deraadt@
|
|
in "for name in [word ...]; do list; done" can be empty.
In sh(1), clarify what happens in that case.
In ksh(1), clarify how it can happen that the list is never executed.
OK jmc@ tb@
|
|
While the code intended to support both -s NAME and -s SIGNAME, the
tests performed were wrong. Replace convoluted code with less cryptic
conditionals. ok anton@
|
|
ok guenther@ deraadt@
|
|
|
|
interrupted.
Lots of back and forth with anton@
OK jca@, tb@, anton@
|
|
grep and compare the use in all programs..
|
