| Commit message (Collapse) | Author |
|---|
|
On Sat, Aug 15, 2009 at 06:07:16PM +0000, Matthew Burgess wrote:
>
> My system has Coreutils-7.4 compiled with the i18n patch from
> http://cvs.fedoraproject.org/viewvc/devel/coreutils/coreutils-i18n.patch.
>
> Using this to compile dash, when in an en_GB.UTF-8 locale, I get the following error:
>
> if gcc -DHAVE_CONFIG_H -I. -I. -I.. -include ../config.h -DBSD=1 -DSHELL -DIFS_BROKEN -Wall -g -O2 -MT eval.o -MD -MP -MF ".deps/eval.Tpo" -c -o eval.o eval.c; \
> then mv -f ".deps/eval.Tpo" ".deps/eval.Po"; else rm -f ".deps/eval.Tpo"; exit 1; fi
> eval.c: In function ‘evalcommand’:
> eval.c:810: error: ‘EXECCMD’ undeclared (first use in this function)
> eval.c:810: error: (Each undeclared identifier is reported only once
> eval.c:810: error: for each function it appears in.)
> eval.c:812: error: ‘COMMANDCMD’ undeclared (first use in this function)
> make[3]: *** [eval.o] Error 1
>
> This is because the src/mkbuiltins script ends up generating an incomplete
> src/builtins.h file. This, in turn, is caused by 'sort -u -k 3,3' not
> working correctly.
>
> The attached patch fixes/works around things by setting LC_CTYPE=C, thus
> overriding the build environment (in a similar manner to the earlier call to
> 'sort' in that same script).
I've changed it to use LC_COLLATE.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Commit 55c46b7286f5d9f2d8291158203e2b61d2494420 ([BUILTIN] Honor
tab as IFS whitespace when splitting fields in readcmd) introduced
a bug where sometimes garbage would follow the last field preceding
the end-of-line. This was caused by an off-by-one error in the
string length calculation.
This patch fixes the bug.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Now that eval handles EV_TESTED correctly, we can remove the
SKIPEVAL hack and simply use EXEXIT for set -e.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
This patch fixes the case where the eval command is used with
set -e and as part of a construct that should not cause the
shell to abort, e.g., as part of the condition of an if statement.
This is achieved by propagating the EV_TESTED flag into the
evalstring function through evalcmd. As this alters the prototype
of evalcmd it is now invoked explicitly by evalbltin. The built-in
infrastructure has been changed to accomodate this special case.
In order to ensure that the EXIT trap is properly executed this
patch clears evalskip in exitshell. This wasn't needed before
because of the broken way evalstring worked where it always clears
evalskip when called by minusc.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Looks like in contrast to what the dash.1 manual page says, expansion
of PS{1,2,4} does work.
Here is a little patch to set LINENO. The ways in that it is less than
ideal mirror the ways that the line number error reporting is also
less than ideal.
For example if you run this:
(
x=$((1/0))
# Just to add another line
# And another
) # error reports this line
The error reported will be the closing parenthesis even though I
think most people would prefer the error to be the one where x was
set.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
When I try to split fields by tabs, dash doesn't honour multiple tabs
between fields as whitespace (at least that's how I interpret [1],
please correct me if I'm wrong).
#!/bin/sh
# "1\t2\t\t3"
TESTSTRING="1 2 3"
# only "\t"
IFS=" "
echo "$TESTSTRING" | while read p1 p2 p3; do
echo "p1=${p1}, p2=${p2}, p3=${p3}"
done
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
The change
[EXPAND] Do not quote back slashes in parameter expansions outside quotes
broke quote removal after parameter expansion. This is because
its effecte extended beyond that of quoted patterns.
This patch fixes this by limiting the change to just quoted
patterns.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
When I added savefd we may end up closing stderr if that is how
we get to the tty. This patch fixes by adding a second argument
to indicate what fd should be closed which lets jobs.c get around
the problem.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Rename sh to dash in the header and synopsis; remove reference to the
4.4 BSD release in the description, and replace the history information
with a reference to NetBSD's ash.
Suggested by jaalto through
http://bugs.debian.org/499838
Signed-off-by: Gerrit Pape <pape@smarden.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Instead of open-coding the newline loop, use the CHKNL flag to
get readtoken to eat the newlines before the in keyword for the
case statement.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
POSIX allows newlines before the "in" keyword in for statements
so we should too.
Thanks to Maximilian Bernöcker for reporting this.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
As it stands if we fail to open /dev/tty we end up closing stderr
after saving it at a higher fd.
Thanks to David van Gorkom for reporting this.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Now that waitcmd no longer uses EXSIG we can remove it.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
This test program by Alexey Gladkov can cause dash to enter an
infinite loop in waitcmd.
#!/bin/dash
trap "echo TRAP" USR1
stub() {
echo ">>> STUB $1" >&2
sleep $1
echo "<<< STUB $1" >&2
kill -USR1 $$
}
stub 3 &
stub 2 &
until { echo "###"; wait; } do
echo "*** $?"
done
The problem is that if we get a signal after the wait3 system
call has returned but before we get to INTON in dowait, then
we can jump back up to the top and lose the exit status. So
if we then wait for the job that has just exited, then it'll
stay there forever.
I made the original change that caused this bug to fix pretty
much the same bug but in the opposite direction. That is, if
we get a signal after we enter wait3 but before we hit the kernel
then it too can cause the wait to go on forever (assuming the
child doesn't exit).
In fact this is pretty much exactly the scenario that you'll
find in glibc's documentation on pause(). The solution is given
there too, in the form of sigsuspend, which is the only way to
do the check and wait atomically.
So this patch fixes Alexey's race without reintroducing the old
bug by converting the blocking wait3 to a sigsuspend.
In order to do this we need to set a signal handler for SIGCHLD,
so the code has been modified to always do that.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
|
|
The file arith_yacc.h was missing from dash_SOURCES causing it
to be excluded from the generated tar file.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
|
|
Hi, Herbert and friends. I've created a small patch that allows dash
to be built on Mac OS X. I'm contributing it here with the hope that
it's suitable for inclusion in dash.
The changes in this patch are:
- __attribute__((__alias__())) is not supported, add an autoconf check
- open64 is not present although the stat64 family is, separate the
autoconf checks
- A syntax error had slipped into a non-glibc codepath
- mkbuiltins had a nonportable mktemp invocation for the case where
tempfile is not availalble
Nothing in this patch is actually Mac OS X-specific, so it might aid
portability to other platforms as well.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Hi, I propose to apply the following patch for dash. The problem is
alloca.h is absent on many platforms including NetBSD I'm running.
Also, NetBSD's version of mktemp doesn't work without temporary
filename pattern.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Sven Mascheck reported that we no longer accept the non-standard
for {} syntax but the manual page still refers to it. This patch
removes that reference.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
On Sat, Aug 02, 2008 at 01:24:16AM +0200, Peter Hartlich wrote:
>
> dash's debian/diff/0048--CD-Restored-warning-when-getcwd-fails.diff
> contains the following lines:
>
> - return getcwd(buf, sizeof(buf)) ? savestr(buf) : nullstr;
> +
> + if (getcwd(buf, sizeof(buf))
> + return savestr(buf);
>
> The if condition is missing a closing parenthesis, which probably went
> unnoticed because it occurs in a section only compiled if __GLIBC__ is
> not defined.
This patch adds the extra parenthesis. Thanks Peter.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
In section Redirection the following text misses a left brace:
where redir-op is one of the redirection operators mentioned previously.
Following is a list of the possible redirections. The [n] is an optional
number, as in \u20183\u2019 (not \u2018[3]\u2019, that refers to a file descriptor.
Reported by Jörg Sommer through
http://bugs.debian.org/481365
Signed-off-by: Gerrit Pape <pape@smarden.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
This patch adds support for the -l option (login shell) as required
by the LSB.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Per POSIX ENV needs to undergo parameter expansion.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
This function is no longer used anywhere else.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
The previous two changes were broken because t_lex uses global state.
This patch removes that by making t_wp local to t_lex.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Making these functions non-recursive is straightforward since they
carry no state.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
----- Forwarded message from Gerrit Pape <pape@smarden.org> -----
Subject: Bug#455828: dash: 4-argument test "test \( ! -e \)" yields an error
Date: Fri, 28 Dec 2007 08:53:29 +0000
From: Gerrit Pape <pape@smarden.org>
To: Vincent Lefevre <vincent@vinc17.org>, 455828@bugs.debian.org
On Thu, Dec 27, 2007 at 06:23:20PM +0100, Vincent Lefevre wrote:
> On 2007-12-27 16:00:06 +0000, Gerrit Pape wrote:
> > On Wed, Dec 12, 2007 at 02:18:47AM +0100, Vincent Lefevre wrote:
> > > According to POSIX[*], "test \( ! -e \)" is a 4-argument test and is
> > > here equivalent to "test ! -e". But dash (like ksh93 and bash) yields
> > > an error:
> > >
> > > $ test \( ! -e \) || echo $?
> > > test: 1: closing paren expected
> > > 2
> > > $ test ! -e || echo $?
> > > 1
> >
> > Hi Vincent,
> >
> > the -e switch to test takes an argument, a pathname.
>
> According to POSIX, in both above examples, "-e" is *not* a switch,
> just a string.
>
> test \( ! -e \)
>
> means: return true if the string "-e" is empty, otherwhise return false.
> The error in dash is that it incorrectly thinks that "-e" is a switch in
> this context.
I see, you're right. Thanks, Gerrit.
----- End forwarded message -----
This patch hard-codes the 3,4-argument cases in the way required by
POSIX.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Back in January an attempt was made to fix the interpretation of
quoted slashes in expmeta. However, this only fixed those cases
where the quoted slash is at the front of the word. The case of
non-leading slashes caused the previous directory part to gain a
back slash suffix which causes subsequent pattern matches to fail.
This patch fixes this by removing the back slash in that case.
Thanks to Romain Tartière fox reporting this bug.
Test case:
echo /*"/null"
Old result:
/*/null
New result:
/dev/null
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
The parser used to skip a byte when parsing the & and | operators, testcase:
$ dash -c 'echo $((7&1))'
$ dash -c 'echo $((7& 1))'
$ dash -c 'echo $((7&11))'
Signed-off-by: Gerrit Pape <pape@smarden.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
This gitignore covers most of the generated files in the src/ directory.
Signed-off-by: Dan McGee <dpmcgee@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
gcc generates bogus warnings about uninitialised variables in parser.c.
This patch borrows the uninitialized_var macro from Linux to silence
them.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
klibc does not have mempcpy, so system.h must be included where this is used
to provide the replacement. glob.h doesn't exist, so we need to guard this
include with the HAVE_GLOB definition. Finally, klcc didn't like the syntax
of the main definition in mksignames, and the resulting program segfaulted
when trying to dereference any part of the argv array. Updating the main
function definition solved the problem.
Signed-off-by: Dan McGee <dpmcgee@gmail.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
As it stands if we get an error before procargs gets called by main()
we'll try to print out a null pointer. This patch avoids this by
printing out "sh" instead.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Somewhere along the lines the warning when getcwd fails went missing.
This patch restores it.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
AC_GNU_SOURCE always defines _GNU_SOURCE so testing it is pointless.
This patch changes it to test __GLIBC__ instead.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
With klcc we get
klcc -DHAVE_CONFIG_H -I. -I.. -include ../config.h -DBSD=1 -DSHELL
-DIFS_BROKEN -Wall -D__CTYPE_NO_INLINE -MT exec.o -MD -MP -MF
.deps/exec.Tpo -c -o exec.o exec.c
exec.c: In function 'tryexec':
exec.c:160: warning: comparison with string literal results in unspecified behavior
Storing it in a local variable fixes the problem.
Thanks to Dan McGee for reporting this.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
These days dash is expected to build with libraries other than
glibc so we need to support the old way of calling getcwd again.
Thanks to Dan McGee for reporting this bug when dash is built with
klibc.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
(Herbert: for context, see
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=467065 )
This is a real bug in upstream dash, which has existed since at
least dash-0.5.1 (July 2004). It is a latent bug in cmdtxt()
(which I think applies only to pipes) as it handles "if" commands.
The attached patch fixes it for me.
It's possible this patch will fix one or more of #462414,
#462977, and #463649. I'll send messages there to see if
the submitters can test.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
The change
[EXPAND] Do not quote back slashes in parameter expansions outside quotes
triggered a latent bug in expmeta where the forward slashes when preceded
by a blackslash weren't recognised as directory separators. This was hidden
because a work-around was put in place for glob(3) which meant that we never
had any backslashes immediately before forward slashes.
This patch fixes the metaflag loop to recognise forward slashes even when
they follow a backslash.
Thanks to Daniel Hahler for reporting this problem.
Test case:
echo "/"root*
Old result:
/root*
New result:
/root
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Once I fixed the previous problem it became apparent that we never dealt
with prompts with new-lines in them correctly. The problem is that we
showed a secondary prompt for each of them.
This patch disables prompt generation in expandstr.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Previously expandstr used the string "" to indicate that it needs to be
treated just like a here-doc except that there is no terminator. However,
the string "" is in fact a valid here-doc terminator so now that we deal
with it correctly expandstr no longer works in the presence of new-lines
in the prompt.
This patch introduces the FAKEEOFMARK macro which does not equal any
real EOF marker but is distinct from the NULL pointer which is used to
indicate non-here-doc contexts.
Thanks to Markus Triska for reporting this regression.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
Although in posix, imaxdiv() isn't implemented on Debian/alpha, causing
dash to fail to build. So use / and % operators if imaxdiv() isn't
available.
http://bugs.debian.org/456398
Signed-off-by: Gerrit Pape <pape@smarden.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
The dash(1) in Debian stable does not support "set +o" in the manner
specified by SUSv3:
|+o
| Write the current option settings to standard output in a format
| that is suitable for reinput to the shell as commands that
| achieve the same options settings.
(citation from
http://www.opengroup.org/onlinepubs/009695399/utilities/set.html)
Instead, dash's "set +o" prints the shell's options in a
human-readable format.
Here is a simple test program that exercises this feature; it works as
I believe is required under bash, but not under dash.
# Save the shell's options
set +o > /tmp/settings-commands
set -o | sort > /tmp/settings-before
# Change some options.
set -v
set -f
set -x
set +o emacs
set -o vi
# Try to restore our options.
. /tmp/settings-commands
set -o | sort > /tmp/settings-after
# Compare.
diff /tmp/settings-before /tmp/settings-after
I believe the following small patch adds this feature to dash, and
documents it in the manual page:
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
dash dies on sparc with a SIGBUS due to an arithmetic error introduced
with commit 03b4958, this patch fixes it.
---
> Hi Gerrit,
>
> dash 0.5.4-3 dies on sparc with a SIGBUS due to an arithmetic error
> introduced with the patch
> 0030-EXEC-Fixed-execing-of-scripts-with-no-hash-bang.diff. The
> attached
> patch fixes the problem.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
|
|
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
|
|
The herefd hack goes back more than a decade. it limits the amount of
memory we have to allocate when expanding here-documents by writing the
result out from time to time. However, it's no longer safe because the
stack is used to place intermediate results too and there we certainly
don't want to write them out should we be short on memory.
In any case, with today's computers we can afford to keep the entire
result in memory and write them out at the end.
|
|
On Tue, Oct 30, 2007 at 04:23:35AM +0000, Oleg Verych wrote:
>
> } 8<<""
> ======================
Actually this (the empty delim) only works with dash by accident.
I've tried bash and pdksh and they both terminate on the first
empty line which is what you would expect rather than EOF. The
real Korn shell does something completely different.
I've fixed this in dash to conform to bash/pdksh.
> In [0] it's stated, that delimiter isn't evaluated (expanded), only
> quoiting must be checked. That if() seems to be completely bogus.
OK I agree. The reason it was there is because the parser would
have already replaced the dollar sign by an internal representation.
I've fixed it properly with this patch.
Test case:
cat <<- $a
OK
$a
cat <<- ""
OK
echo OK
Old result:
dash: Syntax error: Illegal eof marker for << redirection
OK
echo OK
New result:
OK
OK
OK
|
|
The change
[PARSER] Recognise here-doc delimiters terminated by EOF
introduced a regerssion whereby lines starting with eofmark but are not equal
to eofmark would be corrupted. This patch fixes it.
Test case:
cat << _ACEOF
_ASBOX
_ACEOF
Old result:
SASBOX
New result:
_ASBOX
|
|
Debian's libc6 as of 2.6.1-6 has working glob(3)/fnmatch(3) support.
This patch adds the options --enable-glob and --enable-fnmatch to
the configure script. By default glob(3) and fnmatch(3) are still
unused. However, on distros where the glibc is known to work you
may enable these options.
|