| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
| |
This is an upstream bugfix (with slight alterations):
Patch-ID: bash42-053
A combination of nested command substitutions and function importing from
the environment can cause bash to execute code appearing in the environment
variable value following the function definition.
Change-Id: I660e59c76d0d50de52d48a0d399fe1dfef3bded8
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is an upstream bugfix:
Patch-ID: bash42-052
When bash is parsing a function definition that contains a here-document
delimited by end-of-file (or end-of-string), it leaves the closing delimiter
uninitialized. This can result in an invalid memory access when the parsed
function is later copied.
Change-Id: I033a2062a3e8265ceda3ed4a61d2afb53581f71f
|
|
|
|
|
|
|
|
|
|
|
| |
Patch from upstream (with slight changes):
Patch-ID: bash42-051
There are two local buffer overflows in parse.y that can cause the shell
to dump core when given many here-documents attached to a single command
or many nested loops.
Change-Id: I018cb4e7c120cd9d0e6827c23c65ab4875583056
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is upstream's patch:
Patch-ID: bash42-050
This patch changes the encoding bash uses for exported functions to avoid
clashes with shell variables and to avoid depending only on an environment
variable's contents to determine whether or not to interpret it as a shell
function.
Change-Id: I594d4267e726453839d4ced9d36978a6fd84ee8d
|
|
|
|
|
|
|
|
|
|
| |
Patch-ID: bash41-013
Under certain circumstances, bash can incorrectly save a lookahead character and
return it on a subsequent call, even when reading a new line.
Change-Id: I29dcf444011f77e104b6a82e8cada731e3917646
ref: http://seclists.org/oss-sec/2014/q3/734
ref: http://seclists.org/oss-sec/2014/q3/685
|
|
|
|
|
|
|
|
| |
Patch-ID: bash30-017
Under certain circumstances, bash will execute user code while processing the
environment for exported function definitions.
Change-Id: Iccac7b4ae914354978079783a9fe50b3b38ddad5
|
|
|
|
|
|
|
| |
This upgrades bash to from 4.1-rc to 4.2-release. See CWRU/changelog for
changes.
Change-Id: I926269c300cf44fa25964b5b375a148fcf11c4b7
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In get_working_directory(), the getcwd call in the __ANDROID__
workaround case was obviously wrong, having sizeof(char *) as a length
argument instead of the length of the malloced string itself. Also, no
check for malloc failure was there.
With this fix, tested on maguro, using bash with the usual prompt
containing pwd references, as well as the pwd bash builtin itself,
no longer spew these annoying messages mentioned in the title.
Also made a small change to sh_invalidnum() to fix compile warnings.
Change-Id: I9546e07731a251f80e410935619ddcf0873694af
|
|
|
|
| |
Change-Id: Ie64b10dc325386fbb73c16d984462963161900fc
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
without "auto", scripting fails e.g.
for i in `ls /`;do
if [ -e /$i ];
then
echo pass
else
echo fail
fi
done
Change-Id: Id2563143ed2c359933c5b861cc2299f387e34d9c
|
|
|
|
| |
Change-Id: If7507e41c5f67e19b5f2879a3b3896678e1fadf2
|
|
|
|
| |
Change-Id: I973a89ca9ddd4d5b20a5be695319bb233744f7a2
|
|
|
|
| |
Change-Id: I7f64112c5449c27b1216c0ddf6a7416b1b011ad6
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|