| Commit message (Collapse) | Author |
|---|
|
We need to flush at the very end in case we've generated any errors
before that. The flushall call cannot perform a longjmp so it's
safe there.
|
|
On Thu, Jun 07, 2007 at 12:20:27PM +0200, Denis Vlasenko wrote:
>
> When I start dash under Midnight Commander and then type 'exit', dash
> exits all right, but then MC is sent to background. It happens because
> dash does not restore current process group on exit.
>
> Attached patch fixes this. It also fixes another bug: setjobctl(0)
> must ignore tcsetpgrp errors, because there are cases when tty is
> destroyed under dash.
>
> Patch is run-tested.
I've fixed this slightly differently so that we don't need the
xtcsetpgrp change.
|
|
|
|
Add likely flag on REDIR_PUSH.
Remove sv check on REDIR_SAVEFD2 (it implies REDIR_PUSH).
Optimise NTOFD/NFROMFD same fd check.
|
|
As it stands if a redirection occurs on a file descriptor that was
previously closed, it won't be closed after that redirection goes
out of effect. This is because we don't keep track of closed file
descriptors properly as we do for open ones.
This patch fixes this by introducing two new states, CLOSED and
REALLY_CLOSED. The first represents an initially closed descriptor
which is now open while the second one represents an initally closed
descriptor which is now closed.
This patch is based on work by Rainer Weikusat.
|
|
Now that we're marking file descriptors as CLOEXEC in savefd, we no longer
need to close them on exec or in setinputfd.
|
|
There are two kinds of users to copyfd, those that want to copy an fd to
an exact value and those that want to move an fd to a value >= 10. The
former can simply use dup2 directly while the latter share a lot of common
code that now constitutes savefd.
|
|
Since we know that dup2 must succeed here we can call it directly.
|
|
No caller of copyfd need to ignore EMFILE so we can remove the special
case and just let it call sh_error on any error.
|
|
|
|
The value for CTLBACKQ | CTLQUOTE is now -123, not 133.
|
|
On Fri, Feb 16, 2007 at 04:24:55PM -0800, Dan Nicholson wrote:
> I was having a problem building dash where `:' wasn't being recognized
> as a builtin command. I won't bore you with the details of the
> debugging, but it came down to the fact that the generated struct in
> builtins.c wasn't being sorted correctly. The ":" name was coming
> before the "." name and then was never able to be found by bsearch().
>
> The issue turned out to be that I was building with a UTF-8 charset,
> causing the `sort' to come out differently from the mkbuiltins script.
> When I finally set LANG=C, it came out correct. Here's an example of
> the difference.
>
> $ LANG=C sort << "EOF"
> :
> .
> EOF
> .
> :
> $ LANG=en_US.ISO-8859-1 sort << "EOF"
> :
> .
> EOF
> :
> .
>
> My system has glibc-2.3.6, but I also tested it on an RHEL3 server at
> work. My solution was just to add LC_ALL=C to the mkbuiltins script.
> Maybe it's overkill to use LC_ALL. Patch against HEAD attached.
I've changed this patch so that we just set LC_COLLATE for the sort
command.
|
|
On Sat, Oct 21, 2006 at 02:19:18PM +0000, Gerrit Pape wrote:
> Hi Herbert, please see
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=387458
>
> On Thu, Sep 14, 2006 at 03:50:02PM +0200, Julien Danjou wrote:
> > I just found this bug which is easily reproductible:
> >
> > % dash -c 'command -v'
> > zsh: segmentation fault dash -c 'command -v'
Since bash's behaviour is minimalist here, I've decided to adopt its
behaviour here as well which is to return success silently.
|
|
Check getgroups() and fwrite() return code, required to build with
-D_FORTIFY_SOURCE=2.
|
|
The return value of ccmatch was being treated as 0 or 1 but
it's actually zero or non-zero. This broke inverted character
class matching.
Reported by Alexander Skwar.
|
|
Add to the configure.ac new option:
--with-libedit - Compile with libedit support.
|
|
The existing scheme of using the native char for syntax array indicies
makes cross-compiling difficult. Therefore it makes sense to choose
one specific sign for everyone.
Since signed chars are native to most platforms and i386, it makes more
sense to use that if we are to choose one type for everyone.
|
|
Now that the only thing protected by setjmp/longjmp is the saved string,
we can allocate it on the stack to get rid of the jump.
|
|
The parsebackquote flag is only used in a test where it always has the
value zero. So we can remove it altogether.
|
|
On Tue, Jan 10, 2006 at 10:56:23AM +0000, Gerrit Pape wrote:
> tags 347232 + patch
> quit
>
> On Mon, Jan 09, 2006 at 04:29:19PM +0100, Marco Nenciarini wrote:
> > The problem is here:
> >
> > # Set the kernel 2.6 option only for fresh install
> > test -z "$(GetMenuOpt "kopt" "")" && kopt_2_6="root=$root_device_2_6 ro"
> >
> > # Extract options for specific kernels
> > eval $(ExtractMenuOpts "\(kopt_[a-zA-Z0-9_]\+\)")
> >
> > If the first test fails and the eval argument is empty then dash
> > terminate with exitcode 1.
>
> > This is a simple testcase:
> > tm:~# bash -c "set -e ;/bin/false && : ; eval ''; echo 'END'"; echo $?
> > END
> > 0
> > tm:~# dash -c "set -e ;/bin/false && : ; eval ''; echo 'END'"; echo $?
> > 1
> >
> > if you insert any command with successfull exit status before the
> > empty eval, all work ok:
> > tm:~# bash -c "set -e ;/bin/false && : ; : ; eval ''; echo 'END'"; echo $?
> > END
> > 0
> > tm:~# dash -c "set -e ;/bin/false && : ; : ; eval ''; echo 'END'"; echo $?
> > END
> > 0
>
> Yes, I can confirm this is a bug in dash. The standard says
>
> EXIT STATUS
>
> If there are no arguments, or only null arguments, eval shall
> return a zero exit status; otherwise, it shall return the exit
> status of the command defined by the string of concatenated
> arguments separated by <space>s.
>
> Hi Herbert, please see http://bugs.debian.org/347232
Changed evalstring to return the exit status instead of evalskip. This
allows us to return zero if the string is empty.
|
|
|
|
On Thu, Nov 03, 2005 at 07:16:53PM +0100, Mike Hommey wrote:
> Here is a simple test case:
>
> #!/bin/dash
> echo test\\ test
> echo '\c'
> echo test\\ test
>
> it outputs:
> test\ test
> test\
>
> This is due to mis-usage of a global variable. See attached patch for a
> fix.
Instead of setting rval when \c is detected, this is now set in the
return value of conv_escape_str. This prevents the spillage reported
in http://bugs.debian.org/337294.
|
|
Instead of calling conv_escape_str when we detect a backslash we will call it
unconditionally. This helps get rid of some unnecessary code in echocmd.
|
|
Using char in the function argument causes unnecessary truncation to occur.
This patch changes it to an int.
|
|
tempfile(1) is Debian-specific so let's use mktemp(1) as a fallback.
|
|
Since nobody has compiled with JOBS turned off for quite a while, it has
bit-rotted. This patch makes it build again.
|
|
The implementation has an off-by-one bug. This results in random memory
corruption. In one particular case, it caused certain bits of a function
body to go missing.
|
|
Unfortunately klibc doesn't provide out-of-line versions of ctypes
functions such as isalpha. This is a nasty hack to create them.
|
|
Since we have a fallback implementation for stpcpy we can now use it
unconditionally in nodes.c.pat. This also fixes a link error with
klibc.
|
|
Add a dummy sysconf implementation that always fails for klibc.
|
|
Let's remove the support for standalone support from test for the same
reason as printf.
|
|
klibc doesn't have strtod or atof. So add an implementation that
always fails by setting the end pointer to the input string.
|
|
The support for standalone/csh printf only serves to complicate
maintainence.
|
|
klibc doesn't define the macro SSIZE_MAX.
|
|
Now that mkbuiltins correctly disabled histcmd we don't need the dummy
histcmd implementation anymore.
|
|
For systems without getrlimit (e.g., klibc) we will disable ulimit.
In order to achieve this, builtins.def is now produced by cpp which
allows us to use macros such as HAVE_GETRLIMIT in it.
Thie also means that we can get rid of the cflags parsing code in
mkbuiltins.
|
|
klibc doesn't have killpg. Since we only call it for valid values of
pid, we can call kill instead.
|
|
klibc doesn't have strsignal but it does have sys_siglist.
|
|
All users of mempcpy must include system.h.
|
|
Removed obsolete extern declaration on funcnest. This conflits with the
correct static definition.
Changed memtodest prototype to use char * instead of unsigned char *.
Perform the unsigned char cast inside memtodest instead.
|
|
klibc doesn't have and doesn't need getpwnam. This change creates
getpwhome which always returns NULL if getpwnam doesn't exist.
|
|
Added impelmentation of bsearch since klibc doesn't have it yet.
|
|
64-bit file calls such as stat64 don't exist on all architectures for
dietlibc, and they don't exist at all for klibc. In those cases the
normal calls such as stat are already 64-bit.
So simply define stat64 as stat if it doesn't exist. Do the same for
all other 64-bit calls as well.
|
|
This gets rid of the only reference of sysexits.h in dash which is
from commandcmd. This is needed for klibc support since it doesn't
have sysexits.h.
The only uses of sysexits.h in commandcmd is superfluous anyway. In
fact, it is overly sensitive about usages such as 'command -vV ls'.
By making its behaviour close to that of bash/ksh, we end up saving
a bit of space too.
|
|
klibc has bsd_signal instead of signal. So we will define signal as
bsd_signal if
1) signal does not exist.
2) bsd_signal exists.
|
|
This is the only instance of __P in the entire source so it makes sense
to get rid of it rather than making it work with klibc.
|
|
This change updates the BSD licence to the three-clause version since
NetBSD has already done so. This makes dash GPL-compatible.
It also adds Christos Zoulas (NetBSD ash maintainer) to the COPYING file.
I've added "copyright by Herbert Xu" to most files.
Finally all CVS IDs and inclusion of sys/cdefs.h have been removed.
The latter is needed for support of klibc.
|
|
Added unlikely markers.
Reduced switch coverage.
Removed p since we now erase NULs as soon as we see them.
Use more to store parselleft.
|
|
On Sun, Sep 25, 2005 at 07:50:54PM +0000, Gerrit Pape wrote:
> On Sat, Sep 03, 2005 at 02:56:00PM +0200, Martin Dickopp wrote:
> > When analysing the bug I came to the following conclusion: The loop
> > beginning at input.c:302 overwrites a single input line with itself,
> > skipping NUL characters. Therefore, after the loop the line buffer
> > has less characters than originally read if and only if the input line
> > contains NUL characters.
>
> yes. Thanks a lot for the details and patch.
>
> > The pointer that is used to read from the buffer, 'parsenextc' (cf.
> > input.h:66), is also used as the beginning of the next line in
> > input.c:296. This fails if the buffer contains less characters
> > than originally read into it due to NUL characters.
> >
> > The proposed patch (attached) keeps track of the number of skipped
> > characters and advances 'parsenextc' accordingly before processing
> > the next input line.
>
> Hi Herbert, please see
> http://bugs.debian.org/317516
Instead of moving the characters in the loop, we will do memmove every
time we see a NUL character. This hurts if there are a lot of NUL
characters, but should be a win in normal situations.
|
|
Let evaltree handle traps from cmdloop.
Reset evalskip after minusc is executed.
Stop executing traps once SKIPEVAL is seen.
|