On Wed, Apr 23, 2025, 11:40 Chet Ramey wrote:
>
> On 4/23/25 11:30 AM, Александр Ушаков wrote:
>
> > I encountered an issue in Bash and would like to report it. buggyfile.txt
> > is attached to the email.
> >
> > Steps to reproduce
> >
> > $ CC=clang-19 CFLAGS="-fsanitize=address -g -O0" ./config
On Fri, Jan 17, 2025 at 5:34 PM Chet Ramey wrote:
>
> On 1/17/25 4:57 PM, Grisha Levit wrote:
> > On Fri, Jan 17, 2025 at 4:47 PM Grisha Levit wrote:
> >>
> >> On Fri, Jan 17, 2025 at 9:31 AM Chet Ramey wrote:
> >>>
> >>> On 1/15/25 4:0
On Fri, Jan 17, 2025 at 4:47 PM Grisha Levit wrote:
>
> On Fri, Jan 17, 2025 at 9:31 AM Chet Ramey wrote:
> >
> > On 1/15/25 4:08 PM, Grisha Levit wrote:
> > > Another undo list UAF. Related, I think, to edit-and-execute-command.
> > >
> > > HISTFIL
On Fri, Jan 17, 2025 at 9:31 AM Chet Ramey wrote:
>
> On 1/15/25 4:08 PM, Grisha Levit wrote:
> > Another undo list UAF. Related, I think, to edit-and-execute-command.
> >
> > HISTFILE= INPUTRC=/ bash --norc -in <<< $' \n\cP \cN\cP\cU\cX\cE\n\e<'
Another undo list UAF. Related, I think, to edit-and-execute-command.
HISTFILE= INPUTRC=/ bash --norc -in <<< $' \n\cP \cN\cP\cU\cX\cE\n\e<'
AddressSanitizer: heap-use-after-free on address 0xe09245621f48
READ of size 4 at 0xe09245621f48 thread T0
#0 0xb168908a6edc in rl_do_undo lib/readline/
If a key sequence ending with an isearch terminator is bound to a
function that longjmps, entering such a sequence during an isearch
causes readline to loop forever.
echo '".\e": end-kbd-macro' > irc
INPUTRC=irc bash --norc -in <<< $'X\n\cRX.\e'
---
lib/readline/util.c | 1 +
1 file chang
On Thu, Dec 12, 2024 at 5:03 PM Chet Ramey wrote:
>
> On 12/12/24 2:07 PM, Grisha Levit wrote:
>
> > FWIW there's still a use-after-free with something like:
> >
> > HISTFILE= INPUTRC=/ bash --norc -in <<< $'X\n\cPX\cR\n!\e3X\e^\n\cP'
>
>
On Mon, Dec 2, 2024 at 1:56 PM Chet Ramey wrote:
>
> On 11/18/24 10:22 PM, Grisha Levit wrote:
> > But here's a remaining one in combination with history-expand-line:
>
> Thanks for the report. I'm not sure what to do about this one yet.
>
> >
> > HISTFI
On Wed, Nov 13, 2024 at 11:13 AM Chet Ramey wrote:
>
> On 11/12/24 7:18 PM, Grisha Levit wrote:
>
> > The latest change,
> >
> > +Makefile.in
> > + - y.tab.h: move from CREATED_HEADERS to INSTALLED_HEADERS so we don't
> > + clean it
> >
> >
On Tue, Nov 5, 2024 at 11:20 AM Chet Ramey wrote:
>
> On 10/18/24 4:22 PM, Grisha Levit wrote:
> > There's some issue with undo list handling in history-search-* commands:
> >
> > Doing a successful search with a line that has an undo list causes the
> > un
On Wed, Nov 13, 2024 at 3:45 PM Chet Ramey wrote:
>
> On 11/12/24 8:17 PM, Grisha Levit wrote:
> > I'm not sure if this is the most idiomatic way to write the rule, but
> > if a libintl.h is generated, it needs to make it into the installed
> > headers directory to al
If you try using Cflags from the generated pc file, presumably with
something like:
cc $(pkg-config --cflags bash) -c o myload myload.c
the compilation is likely to fail because HAVE_CONFIG_H is not defined.
---
support/bash.pc.in | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff
I'm not sure if this is the most idiomatic way to write the rule, but
if a libintl.h is generated, it needs to make it into the installed
headers directory to allow loadables to build.
---
Makefile.in | 4
1 file changed, 4 insertions(+)
diff --git a/Makefile.in b/Makefile.in
index 00db8cbb.
On Thu, Nov 7, 2024 at 6:06 PM Chet Ramey wrote:
>
> On 11/4/24 9:44 AM, Mike Jonkmans wrote:
>
> > There are some issues though. Some less important.
> > 1) After 'make distclean', it doesn't build: some .o files need y.tab.h.
>
> y.tab.h isn't part of the devel branch, you need bison to build it
On Sat, Nov 9, 2024, 07:41 Clark Wang wrote:
> On Sat, Nov 9, 2024 at 1:48 AM Chet Ramey wrote:
>
> >
> > OK. You asked for the completions to be treated as filenames. When
> > readline displays the possible completions for filenames, it doesn't
> > display the full pathname -- just the portion
These are reported by make --warn-undefined-variables.
Most were being set previously (sometimes 20 years ago) and got left
behind in recepies after their definitions have been removed. Others
only get set in some configurations so it makes sense to prevent them
from inheriting stray values from t
The BASH_STRUCT_DIRENT* macros duplicate the caching, message printing,
and defining functionality already present in the AC_CHECK_MEMBERS macro
that gets called.
It's cosmetic, but causes the configure output to look like:
checking for struct dirent.d_ino... checking for struct dirent.d_ino.
Without an existing cache, if we run ./configure -C, we get the odd-
looking:
checking for usable strtoimax... checking for strtoimax... yes
checking whether strtoimax is declared... yes
yes
config.h is correct though:
#define HAVE_DECL_STRTOIMAX 1
#define HAVE_STRTOIMAX 1
H
On Sun, Oct 13, 2024, 14:58 Chet Ramey wrote:
> On 10/1/24 3:54 AM, konsolebox wrote:
> > # declare -A x
> > # echo ${x@a}
> > A
> > # set -u
> > # echo ${x@a}
> > bash: x: unbound variable
> >
> > Obvious workaround would be to disable `set -u` temporarily or assign
> > a temporary array value b
On Sun, Oct 20, 2024, 20:52 Zachary Santer wrote:
>
> Item 8 is just odd and is on the verge of being a dealbreaker. Not to
> go off on another tangent, but what on Earth?
>
AFAICT it's the non-POSIX-mode Bash behavior that is unusual.
While all shells will have treat single quotes as literal h
Sorry, previous patch should have been:
---
lib/readline/isearch.c | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/lib/readline/isearch.c b/lib/readline/isearch.c
index 9ddf9ce3..c6d8c8a8 100644
--- a/lib/readline/isearch.c
+++ b/lib/readline/isearch.c
@@ -742,10 +742,11
HISTFILE= INPUTRC=/ LC_ALL=C bash --norc -in <<< $'\cR.'
WARNING: MemorySanitizer: use-of-uninitialized-value
#0 _rl_isearch_dispatch isearch.c:745:31
#1 rl_search_historyisearch.c:926:11
#2 rl_reverse_search_historyisearch.c:135:11
#3 _rl_dispatch_subse
There's some issue with undo list handling in history-search-* commands:
Doing a successful search with a line that has an undo list causes the
undo entries from that list to leaked:
HISTFILE= INPUTRC=/ bash --norc -in <<< $'X\nX\e[5~'
=
On Fri, Oct 18, 2024 at 12:15 PM Chet Ramey wrote:
>
> On 10/18/24 11:51 AM, Chet Ramey wrote:
> > On 10/17/24 8:27 PM, Grisha Levit wrote:
> >> The mmap in history_do_write would usually fail when appending because
> >> the offset must be a multiple of the pag
$ bash --norc -in <<< $'\e&'
WARNING: MemorySanitizer: use-of-uninitialized-value
#0 rl_tilde_expand lib/readline/util.c:208:10
---
lib/readline/util.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/readline/util.c b/lib/readline/util.c
index 2bcc776a..d03
The mmap in history_do_write would usually fail when appending because
the offset must be a multiple of the page size.
---
lib/readline/histfile.c | 15 +++
1 file changed, 11 insertions(+), 4 deletions(-)
diff --git a/lib/readline/histfile.c b/lib/readline/histfile.c
index a5fb11d2..
This isn't a bug in bash. You might find some of the suggestions in
https://github.com/rstudio/rstudio/issues/8539 helpful.
On Sun, Oct 6, 2024, 18:02 Luis Vazquez de Lara via Bug reports for the GNU
Bourne Again SHell wrote:
> Reporting a bug after in RStudio terminal with the aliases
>
> Dr. L
help -d output assumes that long_doc[0] includes a newline, which is
not the case for loadable builtins:
$ enable ln rm
$ help -d ln rm
ln - Link files.rm - Remove files
---
builtins/help.def | 11 ---
1 file changed, 4 insertions(+), 7 deletions(-)
diff --git a/builtins/help
Fix typo in configure.ac causing:
src/bash/configure: 3892: test: =: unexpected operator
Update shopt reset code to use new define.
---
builtins/shopt.def | 2 +-
configure.ac | 6 +++---
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/builtins/shopt.def b/builtins/shopt
INTMAX_MAX=9223372036854775807
eval ": {$((INTMAX_MAX-1))..$((INTMAX_MAX))}"
braces.c:447:9: runtime error: signed integer overflow: 9223372036854775807 + 1
cannot be represented in type 'intmax_t' (aka 'long')
---
braces.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/b
On Wed, Aug 21, 2024, 11:27 Chet Ramey wrote:
> On 8/19/24 9:52 AM, Ángel wrote:
> > On 2024-08-18 at 11:21 +0700, Robert Elz wrote:
> >> Interactive shells with -n (noexec) set are pointless
> >
> > The man page states:
> >>-n Read commands but do not execute them. This may
This was actually caught by the test suite
---
builtins/shopt.def | 1 +
tests/shopt.right | 4
2 files changed, 1 insertion(+), 4 deletions(-)
diff --git a/builtins/shopt.def b/builtins/shopt.def
index 67bc0c22..37fda11e 100644
--- a/builtins/shopt.def
+++ b/builtins/shopt.def
@@ -357,6 +3
In a single-byte locale, if the prompt is wider than the screen width
and contains invisible characters on the last line, and the input line
is also wider than the screen width, redisplay gets messed up:
printf -v s '%*s' "$COLUMNS"
PS1=$s'%\[\e[0m\]' LC_ALL=C bash --norc -in <<< $s$'X\caY
On Mon, Aug 12, 2024, 11:04 Chet Ramey wrote:
> My question is why the (admittedly old) gnulib replacement strtod/strtold
> is messing things up.
>
Looks like printf(3) gets called with a `Lf' conversation specifier and a
double argument.
diff --git a/builtins/printf.def b/builtins/printf.def
i
The new read_mbchar code is missing an (unsigned char) cast, causing an
invalid continuation byte >0x7F to fail to be recognized as a delimiter
on platforms where char is signed.
$ printf '\317_' | { read -d _; echo "${REPLY@Q}"; }
$'\317'
$ printf '\317\360_' | { read -d $'\360'; echo
On e.g. Alpine Linux, the following are set by configure:
bash_cv_struct_winsize_ioctl='yes'
bash_cv_struct_winsize_termios='yes'
bash_cv_struct_winsize_header='ioctl_h'
ac_cv_func_tcgetwinsize='yes'
and so config.h ends up with:
#define STRUCT_WINSIZE_IN_SYS_IOCTL 1
/* #
On Tue, Aug 6, 2024, 14:19 alex xmb sw ratchev wrote:
> ~ $ alias tm='timemark+=( $EPOCHREALTIME )'
> ~ $ tm
>
> [Process completed (signal 11) - press Enter]
>
I believe this was reported in
https://lists.gnu.org/archive/html/bug-bash/2023-05/msg00146.html and fixed
in https://git.savannah.gnu.
POSIX says about the TZ variable:
If the dst field is specified and the rule field is not, it is
implementation-defined when the changes to and from DST occur.
musl seems to interpret `TZ=EST5EDT` as having DST always in effect,
causing the tests that rely on the glibc behavior (of defaul
On Mon, Jun 3, 2024 at 3:53 PM Chet Ramey wrote:
>
> On 5/31/24 5:07 PM, Grisha Levit wrote:
> > Two minor leak fixes for conditional command error conditions:
> >
> > If a WORD token is read when COND_AND, COND_OR, COND_END, or a binary
> > operator are expect
On Mon, Jun 3, 2024 at 3:53 PM Chet Ramey wrote:
>
> On 5/31/24 5:07 PM, Grisha Levit wrote:
> > Two minor leak fixes for conditional command error conditions:
> >
> > If a WORD token is read when COND_AND, COND_OR, COND_END, or a binary
> > operator are expect
The comment describing why this wasn't done has been there since the
start of the repo's history and AFAICT it is not accurate anymore, as
shell_execve only calls realloc when it's going to longjmp rather than
return.
Fixes leak in
bash -O execfail -c 'exec /; :'
---
builtins/exec.def | 4 --
Free temporary list allocated when exapnding `$@' in
bash -c 'IFS=:; : ${_+$@}' _ X
---
subst.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/subst.c b/subst.c
index 3faa4068..c56d2434 100644
--- a/subst.c
+++ b/subst.c
@@ -12144,6 +12144,7 @@ finished_with_string:
retu
When a named coproc is created, the name string and associated WORD_DESC
are leaked.
---
parse.y | 2 ++
1 file changed, 2 insertions(+)
diff --git a/parse.y b/parse.y
index d39d6737..fe5038be 100644
--- a/parse.y
+++ b/parse.y
@@ -1103,6 +1103,7 @@ coproc: COPROC shell_command
Two minor leak fixes for conditional command error conditions:
If a WORD token is read when COND_AND, COND_OR, COND_END, or a binary
operator are expected, the allocated WORD_DESC is leaked.
If a conditional command has a syntax error, the allocated COMMAND is
leaked.
---
parse.y | 14 ++
Avoid leaking expansion of `x' in `declare -Ar A; A[x]='
---
arrayfunc.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/arrayfunc.c b/arrayfunc.c
index e85ba6e5..aaa1963c 100644
--- a/arrayfunc.c
+++ b/arrayfunc.c
@@ -302,6 +302,7 @@ bind_assoc_variable (SHELL_VAR *entry, const char *name,
c
Avoid leaking ifs_chars on more return paths.
---
builtins/read.def | 16
1 file changed, 16 insertions(+)
diff --git a/builtins/read.def b/builtins/read.def
index 37328efc..69a1ef4c 100644
--- a/builtins/read.def
+++ b/builtins/read.def
@@ -635,6 +635,8 @@ read_builtin (WORD_LIS
The undo command fails if invoked via execute-named-command on a
line from the history list:
$ bash --norc -in <<< $'A\n\cPB\c_C'
bash-5.3$ A
bash-5.3$ AC
$ bash --norc -in <<< $'A\n\cPB\exundo\nC'
bash-5.3$ A
bash-5.3$ ABC
AFAICT the rl_maybe_replace_line in _rl_readstr_
This requires case-insensitive completion of two+ words where the the
prefix in a word contains the single-byte lowercase form of the multi-
byte character occuring in the same position in an earlier word.
The words here start consist of one of the following followed by `-':
U+212A KELVIN SIG
On Thu, May 23, 2024 at 3:55 PM Chet Ramey wrote:
>
> On 5/23/24 7:43 AM, Michael Maurer wrote:
>
> > Bash Version: 5.2
> > Patch Level: 15
> > Release Status: release
> >
> > Description:
> > Calling a function that itself uses the keyword "time" to call
> > another function that performs
On Thu, May 23, 2024 at 4:11 PM Chet Ramey wrote:
>
> On 5/23/24 3:25 PM, Grisha Levit wrote:
> > On Thu, May 23, 2024 at 10:25 AM Chet Ramey wrote:
> >>
> >> On 5/21/24 2:42 PM, Grisha Levit wrote:
> >>> Avoid using (size_t)-1 as an offset.
> >
On Thu, May 23, 2024 at 10:25 AM Chet Ramey wrote:
>
> On 5/21/24 2:42 PM, Grisha Levit wrote:
> > Avoid using (size_t)-1 as an offset.
>
> I can't reproduce this on macOS. Where is the code that's using -1 as an
> offset?
The loop in rl_change_case does the follo
$ HISTFILE= bash --norc -in <<< $'#\n#\e\t'
bashline.c:3720:16: runtime error: null pointer passed as argument 1, which is
declared to never be null
/usr/include/stdlib.h:971:30: note: nonnull attribute specified here
SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior bashline.c:3720:16
---
bash --norc -in <<< $'A \e-\cXs'
---
bashline.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/bashline.c b/bashline.c
index 528f56e0..0a4e280e 100644
--- a/bashline.c
+++ b/bashline.c
@@ -1336,6 +1336,9 @@ bash_spell_correct_shellword (int count, int key)
int wbeg, wend;
char *text
Avoid leak when spell-correcting an empty line.
---
bashline.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/bashline.c b/bashline.c
index 525d065a..528f56e0 100644
--- a/bashline.c
+++ b/bashline.c
@@ -1348,7 +1348,10 @@ bash_spell_correct_shellword (int count, int key)
Avoid using (size_t)-1 as an offset. Also, not sure it makes sense to
change the case of an invalid byte cast to (wchar_t).
$ bash --norc -in <<< $'\300\e-1\eL'
lib/readline/text.c:1544:26: runtime error: addition of unsigned offset to
0x51100680 overflowed to 0x5110067f
ERROR: AddressSa
- free directory_part when completing command words like `~/bin/'
- free contents of matches when completing command words in old-style
command substitutions
---
bashline.c | 7 ++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/bashline.c b/bashline.c
index b638e001..525d065a
- free command if it is the empty string
- free command before calling bound function, in case bound function
does not return, like rl_abort
---
lib/readline/text.c | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/lib/readline/text.c b/lib/readline/text.c
index c5281efe
./bash --norc -in <<<$'XX\cXs'
bash-5.3$ XX==34016==WARNING: MemorySanitizer: use-of-uninitialized-value
#0 0xaafbbb80 in mindist /home/vm/src/bash/local/lib/sh/spell.c:138:22
#1 0xaafbb468 in spname /home/vm/src/bash/local/lib/sh/spell.c:90:11
#2 0xaafbbe78 in dirspell /hom
On Mon, May 20, 2024, 10:52 Chet Ramey wrote:
> On 5/20/24 10:42 AM, Grisha Levit wrote:
> > If bracketed paste input terminates prior to the paste end sequence,
>
> How would this happen? The terminal emulator (or whatever) performing
> the paste guarantees it.
>
Maybe ove
If bracketed paste input terminates prior to the paste end sequence,
the buffer passed to rl_insert_text never gets its null termination.
$ bash-asan --norc -in <<<$'\e[200~X'
==15989==ERROR: AddressSanitizer: heap-buffer-overflow on address
0x006167e51260 at pc 0x007e690b5374 bp 0x007ff50ab620 s
On Sun, May 19, 2024, 14:05 konsolebox wrote:
> Here's an output from devel branch bash and 5.2.26:
>
> $ bash-9 -c 'echo "$BASH_VERSION"; x="abc#xyz"; echo "${x/\#}"'
> 5.3.0(1)-alpha
> abc#xyz
> $ bash-5.2.26 -c 'echo "$BASH_VERSION"; x="abc#xyz"; echo "${x/\#}"'
> 5.2.26(1)-release
> abcxy
On Fri, May 17, 2024 at 3:06 PM Chet Ramey wrote:
>
> On 5/17/24 12:57 PM, Grisha Levit wrote:
> > The current cmp implementation for size and blocks subtracts the two
> > values and returns the difference as an int. This subtraction can
> > overflow, and the returned i
The current cmp implementation for size and blocks subtracts the two
values and returns the difference as an int. This subtraction can
overflow, and the returned int can end up having the wrong sign.
This also makes the qsort comparison function non-transitive. (Some
interesting discussion on that
Looks like `read -e -u N' creates (and leaks) a new stream for fd N.
If a second `read' tries to fdopen the same fd once more, Android's
FDSAN (on and enforcing defualt) causes the process to abort:
$ bash -c 'read -eu3; read -eu3' 3<&0
fdsan: failed to exchange ownership of file descriptor:
fd 3
If a completion function installs a new compspec and returns 124, and
the new compspec is for something other than the exact name currently
being completed (i.e. a new default compspec, or a compspec for the
basename of a command being completed that includes a slash), the newly
installed compspec
When building with glibc and an -std= arg in CFLAGS, asort and cat
fail to build since sigsetjmp is guarded by _POSIX_C_SOURCE
In file included from bashjmp.h:24,
from shell.h:25,
from examples/loadables/cat.c:30:
include/posixjmp.h:29:25: error: unknown type name
Since 9c430f6b changed some files to no longer include config.h,
building with glibc and an -std= in CFLAGS fails because getopt,
used in support/bashversion.c and mksyntax.c, is guarded by
_POSIX_C_SOURCE >= 2 || _XOPEN_SOURCE
support/bashversion.c:69:17: warning: implicit declaration of function
This test fails on systems with recent-ish BSD regex libs, where a
backslash followed by an alphabetic chatacter now throws REG_EESCAPE.
See https://reviews.freebsd.org/D10510
---
tests/cond-regexp3.sub | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/cond-regexp3.sub b/t
On Thu, Apr 25, 2024, 23:03 Oğuz wrote:
> On Fri, Apr 26, 2024 at 1:17 AM Grisha Levit
> wrote:
> > Actually, I see this on Ubuntu 22.04 but not on macOS.
> On Linux, O_EXCL and RX_EXPANDED (a flag used by bash to signal that
> the redirection word is already expanded) has
On Thu, Apr 25, 2024 at 6:05 PM Grisha Levit wrote:
>
> On Thu, Apr 25, 2024, 14:55 Chet Ramey wrote:
>>
>> On 4/25/24 12:18 PM, Clark Wang wrote:
>> > (I'm using the "devel" branch as the "bash-5.3-testing" branch failed
>> >
On Thu, Apr 25, 2024, 14:55 Chet Ramey wrote:
> On 4/25/24 12:18 PM, Clark Wang wrote:
> > (I'm using the "devel" branch as the "bash-5.3-testing" branch failed
> > to build for me. HEAD: 8c8daff1e3661c)
> >
> > To reproduce:
> >
> > $ bash53 --norc
> > bash53-5.3# touch /tmp/file
> > bash53-5.3#
On Wed, Apr 24, 2024 at 11:35 AM wrote:
>
> hello
>
> Apologies if I am missing some blatant point here
>
> I have noticed a difference in behavior of bash-5.2.26 and
> bash-5.3-alpha which isn't a problem of correctness, but may be wasn't
> intentional(?)
This is from https://lists.gnu.org/archi
On Fri, Mar 29, 2024, 09:37 Chet Ramey wrote:
> On 3/28/24 9:54 PM, Grisha Levit wrote:
> > The addition of stdckdint.h to CREATED_HEADERS in Makefile.in leads to
> > an error when installing loadable builtins on platforms that provide
> > the header:
>
> Thanks for th
The addition of stdckdint.h to CREATED_HEADERS in Makefile.in leads to
an error when installing loadable builtins on platforms that provide
the header:
install: cannot stat '/tmp/bash/stdckdint.h': No such file or directory
make[2]: *** [Makefile:903: install-headers] Error 1
make[2]: Leaving dire
Having funsubs in PS2 seems to cause input tokens that are split across
lines to be partially replaced with the last token of the last funsub
to be parsed while expanding the prompt string.
$ PS2='${ : ; }> '
$ printf '[%s]\n' ab\
> cd
[XXcd]
When \# or \! is decoded in a ${var@P} expansion embedded in a prompt
string, the value can be off by one from that of the same sequence
embedded directly in the prompt string.
$ H='\!'
$ PS1='(\!:${H@P}) $'
(3:2 501:500) $
* parse.y
- decode_prompt_string: accept new int flag to sign
On Tue, Feb 20, 2024 at 9:18 AM Chet Ramey wrote:
>
> Well, depending on when the terminal emulator sends the SIGWINCH, this is
> probably a bug in Dropbox or the File Provider code. Bash installs its
> SIGWINCH handler with SA_RESTART, and the default disposition is to
> discard, so even if a SIG
On Tue, Feb 20, 2024 at 4:35 AM Grisha Levit wrote:
>
> Though a similar test shows that the Readline part of the patch is wrong,
> and I'm not sure how to solve it, at least without changing existing
> applications' behavior (and requiring signal setup to happen bef
On Mon, Feb 19, 2024 at 5:10 PM Chet Ramey wrote:
>
> On 2/7/24 1:33 AM, Grisha Levit wrote:
> > I have some dotfiles symlinked to storage backed by a macOS File
> > Provider extension (e.g. Dropbox):
> >
> > $ realpath ~/.bash_profile
> > /User
On Sat, Feb 17, 2024, 12:29 Chet Ramey wrote:
> On 2/16/24 9:37 PM, Grisha Levit wrote:
> > After this change, `make -C examples/loadables others` fails building
> > necho.o with:
> >
> > use of undeclared identifier 'fflush_unlocked'
>
> I can&#
On Fri, Feb 16, 2024 at 9:02 PM Damien ISSANCHOU wrote:
> When expanding $* within a here-document, bash puts a space between
> each positional parameters while I would expect it to put the first
> character of the IFS variable (if it contains at least one character) as
> per POSIX.1-2017.
F
On Fri, Feb 16, 2024, 16:17 Marc Aurèle La France wrote:
>
> On Mon, 2021-Feb-01, Marc Aurèle La France wrote:
>
> > Currently, only the script's arguments are passed as positional
> > parameters. For compatibility reasons, $0 cannot be used to also pass the
> > script's filename, so I'm creating
After this change, `make -C examples/loadables others` fails building
necho.o with:
use of undeclared identifier 'fflush_unlocked'
I checked other object files to make sure the expected symbols are used
and they were, except for the loadable pushd.
Also added the dependency where needed in M
On Fri, Feb 16, 2024 at 11:32 AM Chet Ramey wrote:
>
> On 2/13/24 12:41 PM, Grisha Levit wrote:
> > On Wed, Jan 31, 2024, 14:10 Chet Ramey > <mailto:chet.ra...@case.edu>> wrote:
> >
> > Well, is this a "my arm hurts when I do this" problem
On Wed, Jan 31, 2024, 14:10 Chet Ramey wrote:
> Well, is this a "my arm hurts when I do this" problem, or should bash
> restrict the types of attributes that can be set on readonly variables?
>
For the attributes that only affect future assignments ([iluc]), I think
think it makes sense to eithe
Building readline.info fails in the devel branch:
$ make readline.info
rltech.texi:2399: @node seen before @end deftypevar
make: *** [readline.info] Error 1
---
lib/readline/doc/rltech.texi | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/readline/doc/rltech.texi b/lib/readlin
When assigning a variable as part of expanding a compound assignment
argument, any [aAgGiIluc] options are treated as either all being on
or all being off:
$ unset X; declare -a +i X=(2+2); declare -p X
declare -a X=([0]="4")
$ unset X; declare +i -u X=(foo); declare -p X
declare
On Mon, Feb 12, 2024 at 4:13 PM Chet Ramey wrote:
>
> On 2/7/24 2:23 AM, Grisha Levit wrote:
> > If a single declare command both sets and unsets the array or assoc
> > attribute for an existing scalar variable, the `value' member of the
> > SHELL_VAR is assigne
On Wed, Feb 7, 2024, 14:45 Kerin Millar wrote:
> On Wed, 7 Feb 2024 13:59:47 -0500
> Grisha Levit wrote:
>
> > +#LC_ALL=en_US.UTF-8
>
> Is this not merely adding a comment?
>
Sorry you're right, made the patch after commenting my change out to
confirm it was needed.
>
The tests in printf6.sub fail if `make check' is executed in the C locale.
diff --git a/tests/printf6.sub b/tests/printf6.sub
index fbacd4d5..382943c7 100644
--- a/tests/printf6.sub
+++ b/tests/printf6.sub
@@ -11,6 +11,8 @@
# You should have received a copy of the GNU General Public License
#
If a single declare command both sets and unsets the array or assoc
attribute for an existing scalar variable, the `value' member of the
SHELL_VAR is assigned an ARRAY* or HASH_TABLE* as appropriate, but
later ends up treated as a char*:
$ bash-asan -c 'X=Y; declare -A +A X; declare -p X'
If given both the -a and -A flags and an existing scalar variable,
declare will assign both attributes to the variable:
$ V=X; declare -Aa V; echo $?
0
$ echo ${V@a}
aA
$ (declare -p V)
Segmentation fault: 11
(Prompted by the report from Emanuele Torre in [1])
The value of currently_executing_command is garbage when there is an error
during assignment when popping the variable context:
$ bash -c 'declare -i SHELLOPTS; f() { local -; set -f; }; f' |& cat -v
bash: line 1: M-`^WM-R^N^\{: braceexp
I have some dotfiles symlinked to storage backed by a macOS File
Provider extension (e.g. Dropbox):
$ realpath ~/.bash_profile
/Users/levit/Library/CloudStorage/Dropbox/profile/.bash_profile
This normally works fine, except when my terminal emulator (tested
both Terminal.app and iTerm) re
Bash makes many calls to stdio functions that may have unlocked_stdio(3)
equivalents. Since the locking functionality provided by the regular
versions is only useful in multi-threaded applications, it probably makes
sense for Bash to use the *_unlocked versions where available.
E.g. in situations
On Sat, Feb 3, 2024 at 1:05 PM Chet Ramey wrote:
>
> On 2/2/24 6:33 PM, Grisha Levit wrote:
> > Is it necessary to check the error indicator if printf(3) just had a non-
> > negative return?
>
> I think printf is allowed to set the error flag that ferror checks even if
>
On Thu, Feb 1, 2024 at 7:41 PM Chet Ramey wrote:
>
> On 1/22/24 9:44 PM, Grisha Levit wrote:
> > The size of the buffer used for printf -v is tracked in an int but this
> > can overflow since the buffer can be built up by multiple vsnprintf(3)
> > calls, each of which
If currently_executing_command was a subshell, it can end up freed by
the time executing_line_number checks it.
$ bash-asan -c '${ (:); }/'
=
ERROR: AddressSanitizer: heap-use-after-free on address 0x000106208a40
at pc 0x000102c8bf38
The size of the buffer used for printf -v is tracked in an int but this
can overflow since the buffer can be built up by multiple vsnprintf(3)
calls, each of which can append up to INT_MAX bytes to the buffer:
$ INT_MAX=$(getconf INT_MAX)
$ printf -v VAR "%$((INT_MAX-1))s%$((INT_MAX-1))s"
On Mon, Jan 22, 2024, 01:54 Martin D Kealey wrote:
> You seem to have created an invalid executable. It seems that scripts
> without a #! can only be run with help from the debugger library;
That's not quite what happens. These scripts get executed by forking the
current bash process (without e
1 - 100 of 443 matches
Mail list logo
