This backports most of the Cdt (container data types) mechanism
from the ksh 93v- beta, based on ground work done by OpenSUSE:
https://build.opensuse.org/package/view_file/shells/ksh/ksh93-dttree-crash.dif
plus adaptations to match ksh 93u+m and an updated manual page
(src/lib/libast/man/cdt.3) added directly from the 93v- sources.
| Thu Dec 20 12:48:02 UTC 2012 - werner@suse.de
|
| - Add ksh93-dttree-crash.dif - Allow empty strings in (dt)trees
| (bnc#795324)
|
| Fri Oct 25 14:07:57 UTC 2013 - werner@suse.de
|
| - Rework patch ksh93-dttree-crash.dif
As usual, precious little information is available because the
OpenSUSE bug report is currently closed to the public:
https://bugzilla.opensuse.org/show_bug.cgi?id=795324
However, a cursory inspection suggests that this code contains
improvements to do with concurrent processing and related
robustness. The new cdt.3 manual page adds a lot about that.
This has been in production use on OpenSUSE for a long time,
so hopefully this will make ksh a little more stable again.
Only one way to find out: let's commit and test this...
BTW, to get a nice manual, use groff and ghostscript's ps2pdf:
$ groff -tman src/lib/libast/man/cdt.3 | ps2pdf - cdt.3.pdf
Commit 308696ec caused the build to fail on macOS Catalina.
src/cmd/INIT/iffe.sh:
- Fix a blatantly unportable practice of passing multiple
"|"-separated 'case' patterns through a variable. This was a way
of grepping for some headers including stdio.h, but it only works
this way on ksh93 and possibly the original Bourne shell, and not
on *any* other shell (not even pdksh or mksh) -- and the fact
that it works on ksh93 is arguably a bug. Fix by eliminating the
"noext" variable (which is init'ed once and never changes) and
using the pattern in the relevant 'case' statement directly.
src/cmd/builtin/features/pty:
- No matter what I try, including <stdio.h> causes the build to
fail on Gentoo Linux (i386) with mysterious "invalid identifier:
off64_t" errors -- this is probably some AST preprocessor hackery
gone awry, but I've no idea where to even begin with that. This
works around the problem by using AST sfio instead, which is
built and functional by the time this feature test is run.
- Remove explicit extern declaration for ptsname(2) that was never
used because it depended on an npt_ptsname feature test that
doesn't exist (or no longer exists).
- Add missing <fcntl.h>, <stdlib.h>, and <unistd.h> for open(2),
ptsname(2) and close(2), respectively.
src/lib/libast/features/float,
src/lib/libast/features/sfio,
src/lib/libast/features/stdio:
- Re-include <stdio.h>.
Fixes: https://github.com/ksh93/ksh/issues/164 (I hope)
This fixes annoying warnings from feature tests that show up when
building with IFFEFLAGS=-d1 (show compiler output from iffe), e.g.:
| In file included from <built-in>:367:
| <command line>:3:26: warning: missing terminating '"' character [-Winvalid-pp-token]
| #define _AST_git_commit \"a5c53a59\"
| ^
| 1 warning generated.
This means the double quotes were incorrectly escaped, which is
probably a bug in mamake -- but they're done correctly for the .c
files that actually need these flags. I may or may not trace the
mamake bug sometime.
src/*/*/Mamfile:
- Remove ${KSH_SHOPTFLAGS} en ${KSH_RELFLAGS} from the iffe
invocations; they are not relevant for feature tests, only when
actually compiling .c files (the $CC commands).
iffe --man documents that stdio.h is automatically pre-included for
all feature tests. Including it in the test code is not needed.
You'd think it shouldn't do any harm, but on a Gentoo i386 system,
this include turned out to be the cause of a mysterious 'unknown
type: off64_t' error while compiling the output{ ... }end block in
features/pty. I'm not going to bother with further tracing the
cause of that -- there is some hackery with off64_t defines in the
AST headers that probably has something to do with it.
src/cmd/builtin/features/pty,
src/lib/libast/features/float,
src/lib/libast/features/sfio,
src/lib/libast/features/stdio:
- Remove '#include <stdio.h>' from output{ ... }end blocks.
It was unreasonably hard to debug problems with iffe tests that
fail to compile where they should (particularly output{ ... }end
blocks that write esserntial headers).
In e72543a9 the problem was already somewhat mitigated by making
some of the failing output{ ... }end blocks emit #error directives
so that invalid/incomplete headers would cause an error at a
sensible point, and not a much harder to track error later.
This commit further mitigates the problem by making the Mamfiles
respect an IFFEFLAGS environmenet variable that is prefixed to
every iffe command's arguments. The typical use would be to export
IFFEFLAGS=-d1 to enable debug level 1: show compiler output for all
iffe tests. This now makes it reasonably feasible to detect
problems in the feature tests themselves.
src/**/Mamfile:
- Import IFFEFLAGS environment variable using setv.
- Prefix ${IFFEFLAGS} to every iffe command.
src/**/features/*:
- Amend the new fail error messages to recommend exporting
IFFEFLAGS=-d1 to show the cause of the failure.
README.md, TODO:
- Updates.
There is a feature test for brk(2)/sbrk(2), but it was not checked
for in one place in vmbest.c, causing libdll to fail to build on
FreeBSD aarch64 because the features/dll output{...}end block
failed to link. This commit allows libdll to build on that system,
though another mysterious build failure apparently remains.
https://github.com/ksh93/ksh/issues/154
src/lib/libast/include/vmalloc.h,
src/lib/libast/vmalloc/vmbest.c:
- Add missing '#if _mem_sbrk' directives to disable uses of sbrk(2)
on systems that have removed this deprecated interface.
src/cmd/builtin/features/pty,
src/lib/libast/features/common,
src/lib/libast/features/float,
src/lib/libast/features/lib,
src/lib/libast/features/sfio,
src/lib/libast/features/sizeof:
- Add a fail clause to more 'tst - output{' blocks so they write an
informative #error directive if they fail to compile and write
required header identifiers. This should avoid much more obscure
compile errors later on. (re: e20c0c6b)
.gitignore:
- Add pattern for emacs #backup# files.
The only proper documentation of the MAM language is in Glenn
Fowler's paper, which is unfortunately copyrighted so we can't
include it. But we can at least provide a link to it.
src/**/Mamfile:
- Add header comment.
src/cmd/INIT/mamake.c:
- Re-enable clang warnings on unused values (there aren't any).
This post-Korn AT&T commit from Feburary 2020 broke the build at
least on Slackware 14.2 with gcc 5.5.0 and glibc 2.23 if vmalloc
was disabled by defining _std_malloc or _AST_ksh_release (see
35672208). So building with vmalloc disabled has always been broken
on 93u+m on at least this version of Linux.
As usual, AT&T did not document the reason for applying this
change. It was also part of a commit that I already have little
trust in (I reverted another part of it in 16e4824c). So let's just
revert this and see what happens.
Hmm. The Linux __malloc_initialize_hook(3) manual page says it's
deprecated and was to be removed from glibc as of 2.24, whereas
Slackware 14.2 uses glibc 2.23. This would explain why this change
didn't break Linux with newer glibc versions, as the feature test
won't detect it and it won't be used at all.
src/lib/libast/features/vmalloc, src/lib/libast/vmalloc/malloc.c:
- Revert change in definition of __malloc_initialize_hook. It now
conforms again with the spec in the Linux man page.
The build error caused by this change was:
| + cc -D_BLD_DLL -fPIC -D_BLD_ast '-D_AST_git_commit="e3f6d2d0"' -Os -g -D_std_malloc -I. -I/usr/local/src/ksh/src/lib/libast -Icomp -I/usr/local/src/ksh/src/lib/libast/comp -Ivmalloc -I/usr/local/src/ksh/src/lib/libast/vmalloc -Iinclude -I/usr/local/src/ksh/src/lib/libast/include -Istd -I/usr/local/src/ksh/src/lib/libast/std -D_PACKAGE_ast -c /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c
| /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c: In function '_ast_mallopt':
| /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c:1089:58: warning: implicit declaration of function 'mallopt' [-Wimplicit-function-declaration]
| extern int F2(_ast_mallopt, int,cmd, int,value) { return mallopt(cmd, value); }
| ^
| /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c: At top level:
| /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c:1093:22: error: return type is an incomplete type
| extern Mallinfo_t F0(_ast_mallinfo, void) { return mallinfo(); }
| ^
| /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c:72:19: note: in definition of macro 'F0'
| #define F0(f,t0) f(t0)
| ^
| /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c: In function '_ast_mallinfo':
| /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c:1093:52: warning: implicit declaration of function 'mallinfo' [-Wimplicit-function-declaration]
| extern Mallinfo_t F0(_ast_mallinfo, void) { return mallinfo(); }
| ^
| /usr/local/src/ksh/src/lib/libast/vmalloc/malloc.c:1093:52: warning: 'return' with a value, in function returning void
| mamake [lib/libast]: *** exit code 1 making malloc.o
A build failure on HP-UX B.11.11 was introduced when O_cloexec was
changed to O_CLOEXEC (which is POSIX standard) in the backported
93v- code. The lowercase variant is conditionally defined by libast
in src/lib/libast/features/fcntl.c precisely for compatibility with
systems that do not have O_CLOEXEC.
src/lib/libast/tm/tvtouch.c:
- Revert to using the AST O_cloexec flag when calling open(2).
The build system is adapted to make SHOPT_* compile-time options
editable without nmake. We can now easily change ksh's compile-time
options by editing src/cmd/ksh93/SHOPT.sh. The bin/package script
is adapted to turn these into compile flags. This resolves the most
important drawback of not using nmake.
Also, mamake now has support for indented Mam (Make Abstract
Machine) code. Only one type of block (make...done) is supported in
Mamfiles, so they are easy to indent automatically. A script to
(re)do this is included.
Since nmake is not going to be restored (it has too many problems
that no one is interested in fixing), this at least makes mamake
significantly easier to work with.
The Makefiles are deleted. They may still be handy for reference to
understand the Mamfiles, but they haven't actually matched the
Mamfiles for a while -- and you can still look in the git history.
Deleting them requires some adaptations to bin/package and mamake.c
because, even though they do not use those files, they still looked
for them to decide whether to build code in a directory.
Finally, this commit incorporates some #pragmas for clang to
suppress annoying warnings about the coding style used in this
historic code base. (gcc does not complain so much.)
src/cmd/ksh93/SHOPT.sh:
- Added.
bin/package, src/cmd/INIT/package.sh:
- cd into our own directory in case we were run from another dir.
- $makefiles: only look for Mamfiles.
- Add ksh compile-options via KSH_SHOPTFLAGS. Include SHOPT.sh.
- make_recurse(): Do not write a missing Makefile.
- finalize environment: Look for Mamfiles instead of Makefiles.
src/cmd/INIT/mamake.c:
- Tell clang to suppress annoying warnings about coding style.
- Update version string and self-documentation.
- input(): Add support for indented Mam code by skipping initial
whitespace on each input line.
- files[]: Instead of looking for various of Makefiles to decide
where to build, only look for Mamfiles.
src/Makefile, src/cmd/INIT/Makefile, src/cmd/Makefile,
src/cmd/builtin/Makefile, src/cmd/ksh93/Makefile, src/lib/Makefile,
src/lib/libast/Makefile, src/lib/libcmd/Makefile,
src/lib/libdll/Makefile, src/lib/libsum/Makefile:
- Removed.
src/Mamfile, src/cmd/INIT/Mamfile, src/cmd/Mamfile,
src/cmd/builtin/Mamfile, src/cmd/ksh93/Mamfile, src/lib/Mamfile,
src/lib/libast/Mamfile, src/lib/libcmd/Mamfile,
src/lib/libdll/Mamfile, src/lib/libsum/Mamfile:
- Indent the code with tabs.
- In ksh93/Mamfile, add ${KSH_SHOPT_FLAGS} to every $CC command.
- In ksh93/Mamfile, add "prev SHOPT.sh" for every *.o file
so they are rebuilt whenever SHOPT.sh changes.
bin/Mamfile_indent:
- Added, in case someone wants to re-indent a Mamfile.
src/cmd/INIT/proto.c, src/cmd/INIT/ratz.c, src/cmd/INIT/release.c,
src/lib/libast/features/common, src/lib/libast/include/ast.h:
- Tell clang to suppress annoying warnings about coding style that
it disapproves of (mainly concerning the use of parentheses).
src/cmd/INIT/cc.darwin, src/cmd/INIT/cc.freebsd,
src/cmd/INIT/cc.openbsd:
- Remove now-redundant clang warning suppression flags.
Resolves: https://github.com/ksh93/ksh/issues/60
This incorporates the last changes in the tm library before AT&T
laid off the AST developers. It contains mostly time zone and
locale related changes/fixes.
I was hoping these would fix#52 (locale-based 'printf %T' output
is broken), but no such luck. This is probably good to have anyway.
For example, this changes 'typeset -Q' (a bad option) from:
| ksh: typeset: -Q: unknown option
| Usage: typeset [-bflmnprstuxACHS] [-a[type]] [-i[base]] [-E[n]]
| [-F[n]] [-L[n]] [-M[mapping]] [-R[n]] [-X[n]]
| [-h string] [-T[tname]] [-Z[n]] [name[=value]...]
| Or: typeset [ options ] -f [name...]
to:
| ksh: typeset: -Q: unknown option
| Usage: typeset [-bflmnprstuxACHS] [-a[type]] [-i[base]] [-E[n]]
| [-F[n]] [-L[n]] [-M[mapping]] [-R[n]] [-X[n]]
| [-h string] [-T[tname]] [-Z[n]] [name[=value]...]
| Or: typeset -f [name...]
| Help: typeset [ --help | --man ] 2>&1
src/lib/libast/misc/optget.c: args():
- Revert the changes done in 6916a873 and ae92cd89. The --help and
--man labels weren't added consistently (they did not show up in
the example above) whereas they did show up unnecessarily in the
manual page itself.
- In the usage section and usage messges, only show an [ options ]
label on the first usage line; don't redundantly repeat on second
and further ("Or:") lines.
- In usage and --help (but not --man), add a new "Help:" line
telling the user about the --help and --man options. This
replaces the reverted changes. Show the 2>&1 redirection as a
reminder that you need to do this to pipe it into a pager, as
everything is written to standard error!
- Add some comments clarifying what I think this code does...
src/cmd/ksh93/tests/builtins.sh:
- Update to match changes in getopts usage output.
src/cmd/ksh93/bltins/typeset.c:
- The new_argv[] array was one item too short (should be argc+2).
- Use AST stakalloc(3) to allocate it instead of a dynamic array;
this restores compatibility with ISO C90.
src/lib/libast/features/standards, src/cmd/INIT/cc.unixware.i386:
- Add support for UnixWare.
- Do not define any standards macros on this system, as on FreeBSD
and DragonFly BSD.
This makes ksh build on Alpine Linux which uses this C library.
src/lib/libast/include/ast_std.h:
- Define __DEFINED_FILE to hide FILE internals from the Korn
shell's SFIO.
src/lib/libast/features/wchar:
- Include wchar.h before redefining iswalpha() to avoid mangling
the C library's declaration.
src/lib/libast/features/lib:
- Test whether off64_t and off_t are actually distinct types before
using the former.
Fixes: #3
Same idea as in the referenced commit.
src/lib/libast/comp/conf.sh:
- If an output header file has not changed after rerunning conf.sh,
still update the output file's timestamp using touch(1) to signal
that the test has already been run.
AIX on ibm.risc comes with a broken version of ksh88 as /bin/sh
where the following causes breakage in the parser (spurious syntax
errors):
(set -o posix) 2>/dev/null && set -o posix
However, prefixing it with 'command' (while keeping the subshell)
circumvents the problem. So, why not.
(command set -o posix) 2>/dev/null && set -o posix
A common cause of build failures on some systems is that the output
block in the dll feature test silently fails to compile. This leads
to very-hard-to-trace compiler errors about missing identifiers
later on. iffe syntax does not allow aborting compilation if a
block does not compile, however, it does let us produce alternative
output from a shell script if compilation fails. This can be used
to generate an informative #error directive that is inserted in
place of the missing identifiers.
src/lib/libdll/features/dll:
- Add fail block to output block that produces an #error directive.
Instead, we now link to the libm system math library where needed
by adding -lm to the relevant compile commands in the Mamfiles.
This is not needed on every system but never does any harm.
(This adds more custom edits to the Mamfiles, which were originally
generated from the nmake Makefiles. This takes us further from
restoring nmake, but that already wasn't going to happen anyway,
due to its many problems... the path forward will be to translate
the Mamfiles to some other, current make system such as GNU make.)
bin/package, src/cmd/INIT/package.sh:
- Remove LDFLAGS=-lm hack for DragonFly BSD, NetBSD and Solaris.
src/cmd/builtin/Mamfile,
src/cmd/ksh93/Mamfile,
src/lib/libdll/Mamfile:
- Add -lm where linking failed on any of the three mentioned OSs.
src/lib/libdll/features/dll:
- In the output test program, add missing #include <math.h>, fixing
unknown identifier errors on NetBSD (ldexp, ldexpl).
src/cmd/builtin/features/pty:
- Add missing #include <stdio.h> to make printf work on all systems
(this is just a feature test, no need to bother with sfio here).
src/lib/libast/features/stdio:
- Undef __FILE_T to avoid interference from system headers on QNX.
(There are still other problems preventing a build on QNX 6.5.0.
The shipped version of gcc seems to be broken.)
This now makes ksh build on DragonFly BSD.
bin/package, src/cmd/INIT/package.sh:
- DragonFly also needs the -lm hack for LDFLAGS.
src/cmd/ksh93/sh/main.c, src/cmd/ksh93/tests/basic.sh:
- fixargs() doesn't work on DragonFly either
(re: 9b7c392a, 159fb9ee, cefe087d).
The following are backported from:
https://github.com/att/ast/issues/26#issuecomment-313927854https://github.com/att/ast/pull/19
src/lib/libast/comp/setlocale.c:
- Add missing #include <errno.h> since errno is used.
src/lib/libast/features/standards:
- Do not set any standards macros (_POSIX_SOURCE etc) on FreeBSD or
DragonflyBSD; they disable too much functionality on those.
src/lib/libast/features/wchar:
- Set _STDFILE_DECLARED on DragonFly, too.
src/lib/libast/include/sfio.h, src/lib/libast/include/sfio_t.h,
src/lib/libast/sfio/_sfopen.c, src/lib/libast/sfio/sfclrlock.c,
src/lib/libast/sfio/sfhdr.h, src/lib/libast/sfio/sfnew.c,
src/lib/libast/sfio/sfset.c:
- Rename SF_* macros to SFIO_* to avoid a conflict with system
headers.
src/lib/libast/string/strexpr.c:
- Rename error() to err() to avoid a conflict.
bin/package, src/cmd/INIT/package.sh:
- CCFLAGS overwrites the autodetected optimisation flags (e.g. -Os)
if set. Unfortunately, that also happened when we added something
to CCFLAGS for a release build or to add an extra flag needed by
Solaris. The fix is to use a new flags variable (KSH_RELFLAGS)
instead. This needs to be done in a different place as it needs
to be added to the mamake command as an assignment argument.
- Remove the Solaris CCFLAGS hack; see features/common below.
src/*/*/Mamfile:
- Add ${KSH_RELFLAGS} to all the compiler commands.
src/lib/libast/features/common:
- Enable POSIX standard on Solaris (i.e.: if __sun is defined) by
defining _XPG6 directly in the feature test that generates
ast_std.h, which is indirectly included by everything. This
removes the need to pass -D_XPG6 via CCFLAGS. (Doing so
automatically with gcc was not otherwise possible.)
src/cmd/INIT/cc.sol11.*:
- No longer pass -D_XPG6, as per above.
That patch didn't work for non-gcc, non-clang compilers -- at least
Solaris Studio cc. It doesn't prefix error messages with "error:".
As a result, it caused the build to fail on Solaris with native cc.
src/lib/libast/comp/conf.sh:
- Use a sed formula that should catch error messages prefixed by
"line xx:" while still removing warnings and suggestions. This
works on at least clang, gcc, Solaris Studio cc.
src/lib/libast/tm/tvsleep.c:
- Since the 'sleep' builtin was backported/fixed from ksh93v- and
ksh2020, it makes sense to use the latest/last tvsleep(3), too.
Looks like this added an interrupt check (errno == EINTR).
Also, new fallback versions for systems without nanosleep(2).
Documentation: src/lib/libast/man/tv.3 (unchanged)
src/lib/libast/features/float:
- libast attempts to determine the binary representation of Inf and
NaN to use as a fall-back code path for systems that do not
support fpclassify(). The libast feature detection did not get
consistent signatures between builds. To fix this, zero the
memory before determining the signature.
src/lib/libast/sfio/sfcvt.c:
- The fall-back code path is broken because there are multiple
representations for NaN - the important thing is to check the
exponent and for a non-zero significand. The trailing bits can be
random or left over from interim operations. For that reason, to
ensure we never end up using the fall-back code path, explicitly
generate a compile error if we end up there.
Based on a patch from @citrus-it:
8bf59a9a8f
but uses POSIX memset(3) instead of deprecated bzero(3).
conf.sh checks for undefined symbols by parsing compiler output and
looking for strings of capital letters and underscores. Modern gcc
produces suggestions for replacement variables too, for example:
error: '_SC_CLOCKRES_MIN' undeclared here (not in a function); did you mean _POSIX_CLOCKRES_MIN?
_SC_CLOCKRES_MIN,
^~~~~~~~~~~~~~~~
_POSIX_CLOCKRES_MIN
This causes good variables to be excluded along with bad, causing differences
between the builtin and system getconf commands.
src/lib/libast/comp/conf.sh:
- Only use lines containing 'error:' and ignore everything starting
from 'did you mean:'. (Note this scripts sets the locale to C.)
Patch from @citrus-it:
061a4b1da1
This upstreams a Solaris patch:
https://github.com/oracle/solaris-userland/blob/master/components/ksh93/patches/050-CR7065478.patch
src/lib/libast/comp/setlocale.c:
- Add wide_wctomb() wrapper for wctomb(3). It changes an invalid
character (wctomb returns -1) to a single byte with length 1.
- set_ctype(): Use wide_wctomb() instead of wctomb(3) as the
conversion discipline function (ast.mb_conv). Effectively this
means there are no invalid characters. Perhaps this is necessary
for compatibility with ASCII. Sadly, no public info available.
This change is pulled from here:
https://github.com/oracle/solaris-userland/blob/master/components/ksh93/patches/275-20855453.patchhttps://github.com/att/ast/issues/30
George Lijo wrote on 17 Feb 2017:
> Here's a reproducible testcase on a Solaris11 host running
> ksh93u+(2012-08-01).
> $ cat a.sh
> #!/bin/sh
>
> AAA="aaa"
> echo 'insert character'
> BBB=`echo ${AAA} | sed "s/aaa/bbb/g"`
> logger "variable BBB = ${BBB}"
>
> $ cat t.sh
> #!/bin/ksh
>
> sleep 10
> /bin/ksh ./a.sh
> exit 0
>
> $
>
> $ ./t.sh
>
> The expected result is:
>
> Apr 9 12:43:34 lab user: [ID 702911 user.notice] variable BBB = bbb
>
> because variable "BBB" is supposed to be set to 'bbb' in a.sh.
>
> But if the parent shell is terminated, the variable is wrongly set.
>
> user@xxxxx$ telnet lab
> ...
> $ ./t.sh & <--- Run t.sh in background.
> [1] 2067
> $ logout <--- CTRL + D to exit while t.sh is running.
> Connection to lab closed by foreign host.
>
> Again, access the system and check the output:
>
> user@xxxxx$ telnet lab
> ...
> $ tail -f /var/adm/messages
> :
> Apr 9 12:47:47 lab user: [ID 702911 user.notice] variable BBB =
> insert character <--- !!!
> Apr 9 12:47:47 lab bbb
> <--- !!!
>
> Thus the variable is wrongly set. (The previous echo string was
> not cleared.)
>
> The issue happens because the EIO error during the logout is not
> handled properly.
src/cmd/ksh93/sh/io.c,
src/lib/libast/include/error.h:
- Amend the ERROR_PIPE() macro to check for EIO as well as EPIPE
and ECONNRESET.
As of this commit, ksh 93u+m has a standard semantic version number
<https://semver.org/>, beginning with 1.0.0-alpha. This is added to
the version string in a way that should be compatible with scripts
parsing ${.sh.version} or $(ksh --version). This addition does not
replace the release date and does not affect $((.sh.version)).
For non-release builds, the version string will be:
FORK/VERSION+HASH YYYY-MM-DD
e.g.: 93u+m/1.0.0-alpha+41ef7f76 2021-01-03
For release builds, it will be:
FORK/VERSION YYYY-MM-DD
e.g.: 93u+m/1.0.0 2021-01-03
It is now automatically decided by bin/package whether to build a
release or development build. When building from a directory that
is not a git repository, or if the current git branch name starts
with a number (e.g. '1.0'), the release build is enabled; otherwise
a development build is the default. This is arranged by adding -D
flags to $CCFLAGS as described below. These flags are prepended to
$CCFLAGS, so they can be overridden by adding your own -D or -U
flags via the environment.
In addition, AST vmalloc is disabled for release builds as of this
commit, forcing the use of the OS's standard malloc(3). In 2021,
this is generally more reliable, faster, and more economical with
memory than AST vmalloc. Several memory leaks and crashing bugs are
avoided, e.g.: <https://github.com/ksh93/ksh/issues/95>.
For development builds, vmalloc stays enabled (along with its known
bugs) because this allows the use of the vmstat builtin, making it
much more efficient to test for memory leaks. For more info, see
the regression test script: src/cmd/ksh93/tests/leaks.sh
bin/package, src/cmd/INIT/package.sh:
- Add flags for build type. In $CCFLAGS, define _AST_ksh_release if
we're not on any git branch or on a git branch whose name starts
with a number. Otherwise, define _AST_git_commit as the first 8
characters of the current git commit hash.
src/lib/libast/features/vmalloc:
- If _AST_ksh_release is defined, disable vmalloc and force use of
the operating system's malloc. Discussion:
https://github.com/ksh93/ksh/issues/95
src/cmd/ksh93/include/version.h:
- Define new format for version string, adding a semantic version
number as well as (for non-release builds) the git commit hash.
src/cmd/ksh93/sh/init.c: e_version[]:
- Add a 'v' to the ${.sh.version} feature string if ksh was
compiled with vmalloc enabled. This allows scripts, such as
regression tests, to detect this.
src/cmd/ksh93/data/builtins.c: sh_optksh[]:
- Add a copyright line crediting the contributors to ksh 93u+m.
Resolves: https://github.com/ksh93/ksh/issues/95
Red Hat erratum: https://bugzilla.redhat.com/1221766
"Previously, the gcc utility optimized out a non-NULL test in the
ksh implementation of the strdup() function. This caused an
unexpected termination when ksh was executed in a clean chroot
environment. With this update, ksh compilation parameters have been
updated to prevent optimizing out a non-NULL test, and ksh no
longer crashes in clean chroot environments."
The optimizer bug occurs in that function's single-line body:
return (s && (t = oldof(0, char, n = strlen(s) + 1, 0))) ? (char*)memcpy(t, s, n) : (char*)0;
So it must be the test for non-NULL 's' that fails. And 's' is
declared in the function definition, as follows:
extern char*
strdup(register const char* s)
So that makes me wonder if we can work around the bug by simply
removing the 'const' (and the 'register' while we're at it).
However, I have no easy way to verify that at the moment. The Red
Hat patch instead tells gcc to disable optimization for this
function using a #pragma directive.
I have no idea if that gcc optimiser bug has been fixed in the
meantime, but experience from c258a04f has shown that we cannot
trust that it has been fixed (that other optimizer bug is at least
a decade old and still not fixed). So, in it goes, until someone
shows evidence that we no longer need it.
Original patch:
642af4d6/f/ksh-20120801-badgcc.patch
src/lib/libast/string/strdup.c:
- Tell GCC to disable all optimisations for strdup().
_sfcvt(), "convert a floating point value to ASCII", did not adjust
for negative decimal place movement as what happens with leading
zeroes. This caused ksh's 'printf %f' formatter to fail to round
floating point values correctly.
src/lib/libast/sfio/sfcvt.c:
- Removed constraint of <1e-8 for doubles by matching what was done
for long doubles having <.1.
- Corrected a condition when the next power of 10 occurred and that
new 1 digit was being overwritten by a 0.
src/cmd/ksh93/tests/math.sh:
- Validate that typeset -E/F formatting matches that of their
equivalent printf formatting options as well as checking for
correct float scaling of the fractional parts.
To find the temporary files directory to use, the pathtemp()
function (generate a unique path to a temporary file) first checks
$TMPDIR and $TMPPATH, then falls back to /tmp, then to /usr/tmp as
a last resort. But all systems replaced /usr/tmp by /var/tmp
decades ago to allow mounting /usr as read-only, and a /usr/tmp
compatibility symlink is no longer commonly provided.
src/lib/libast/path/pathtemp.c:
- Change TMP2 definition from "/usr/tmp" to "/var/tmp".
src/lib/libast/features/mmap,
src/lib/libast/features/stdio:
- Change "/usr/tmp" to "/var/tmp" in feature tests.
All changed files:
- Put the shell in POSIX mode if it has an '-o posix' option.
- Remove nonsense disabling 'set -x' on bash. It's not broken.
bin/package, src/cmd/INIT/package.sh:
- Add check blocking native zsh mode (e.g., "$path" conflicts).
Using a 'sh -> zsh' symlink works, so recommend that.
- Remove old ksh93 version check for a supposed conflict with
libcmd. It was broken; it would revert to /bin/sh, but on illumos
distributions, /bin/sh is a ksh93 of a version that is supposedly
affected. It builds fine anyway.
- Rewrite checksh() to incorporate the shell compatibility checks
that were previously in two different places in 'package'.
bin/ignore, src/cmd/INIT/ignore.sh,
bin/silent, src/cmd/INIT/silent.sh:
- Change bad check for a full POSIX 'export' command (no, $RANDOM
has nothing to do with that) with a proper feature test.
This fixes a hanging bug that could occur on macOS when using the
'read' command to read from a FIFO and encountering end-of-file
without a final newline character. It also makes the 'read' command
perform 15-25% faster on macOS and Linux.
The previous version (ff385e5a) failed on SunOS/Solaris/Illumos
because those systems apparently don't (fully) support the POSIX
standard recv(2) syscall with MSG_PEEK[*], which is the feature
that iffe detects under the 'socket_peek' identifier. On Illumos,
using that methods causes a compilation failure (unknown identifier
MSG_PEEK); on Solaris 11.4, that method causes multiple regressions
in tests/io.sh, suggesting the method compiles but doesn't work at
all. Instead, SunOS/Solaris/Illumos requires the method using
ioctl(2)+I_PEEK and select(2). No other system that ksh currently
builds on requires this method, so it is now only used on
SunOS/Solaris/Illumos.
So far, this version of sfpkrd() has been tested to work correctly
on Linux, macOS, FreeBSD, NetBSD, OpenBSD, HP-UX, Solaris, and
OmniOS (an Illumos distribution).
It still fails to peek on Cygwin, but in the exact same way it
failed before, so that's no loss.
To test, run the 'io' test set: bin/shtests -p io
src/lib/libast/sfio/sfpkrd.c: sfpkrd():
- Remove long-obsolete Mac OS X and Solaris bug workarounds.
- Remove methods that are no longer needed.
On systems with a POSIX compliant recv(2), the only thing that
is required to avoid regressions is the code that was conditional
upon the socket_peek feature test, which tests for the correct
functioning of the recv(2) syscall. This has now been made
mandatory for non-SunOS/Solaris/Illumos systems (using an #error
directive if it is not detected), with the other methods removed.
The result performs 15-25% faster on macOS and Linux while
passing all the regression tests.
On macOS, avoiding the select(2) method fixes the hanging bug.
On SunOS/Solaris/Illumos (the '__sun' identifier), the method
using ioctl(2)+I_PEEK and select(2) (iffe feature IDs:
stream_peek and lib_select) is preserved.
Resolves: https://github.com/ksh93/ksh/issues/118 (again)
[*] https://pubs.opengroup.org/onlinepubs/9699919799/functions/recv.html
This fixes a hanging bug that could occur on macOS when using the
'read' command to read from a FIFO and encountering end-of-file
without a final newline character. It also makes the 'read' command
perform 15-25% faster on macOS and Linux and maybe other systems.
src/lib/libast/sfio/sfpkrd.c: sfpkrd():
- Get rid of the optional stuff that uses the poll(2) or select(2)
syscalls. The only thing that is required to avoid regressions is
the code that was conditional upon the socket_peek feature test,
which tests for the correct functioning of the recv(2) syscall.
This has now been made mandatory. The rest now uses what was
previously a fallback in plain C, resulting in a function that is
not only more readable, but actually faster than the syscalls.
Resolves: https://github.com/ksh93/ksh/issues/118
A memory leak occurred upon leaving a virtual subshell if a
function was defined within it. If this was done more than 32766
(= 2^15-2 = the 'short' max value - 1) times, the shell crashed.
Discussion and reproducer: https://github.com/ksh93/ksh/issues/114
src/cmd/ksh93/sh/subshell.c: table_unset():
- A subshell-defined function was never freed because a broken
check for autoloaded functions (which must not be freed[*]). It
looked for an initial '/' in the canonical path of the script
file that defined the function, but that path is also stored for
regular functions. Now use a check that executes nv_search() in
fpathdict, the same method used in _nv_unset() in name.c for a
regular function unset.
src/cmd/ksh93/bltins/misc.c: b_dot_cmd():
- Fix an additional memory leak introduced in bd88cc7f, that caused
POSIX functions (which are run with b_dot_cmd() like dot scripts)
to leak extra. This fix avoids both the crash fixed there and the
memory leak by introducing a 'tofree' variable remembering the
filename to free. Thanks to Johnothan King for the patch.
src/lib/libast/include/stk.h,
src/lib/libast/misc/stk.c,
src/lib/libast/man/stk.3,
src/lib/libast/man/stak.3:
- Make the stack more resilient by extending the stack reference
counter 'stkref' from (signed) short to unsigned int. On modern
systems with 32-bit ints, this extends the maximum number of
elements on a stack from 2^15-1==32767 to 2^32-1==4294967295.
The ref counter can never be negative, so there is no reason for
signedness. sizeof(int) is defined as the size of a single CPU
word, so this should not affect performance at all.
On a 16-bit system (not that ksh still compiles there), this
doubles the max number of entries to 2^16-1=65535.
src/cmd/ksh93/tests/leaks.sh:
- Add leak regression tests for ksh functions, POSIX functions, dot
scripts run with '.', and dot scripts run with 'source'.
src/cmd/ksh93/tests/path.sh:
- Add an output builtin with a redirect to an autoloaded function
so that a crash[*] is triggered if the check for an autoloaded
function is ever removed from table_unset(), as was done in ksh
93v- (which crashed).
[*] Freeing autoloaded functions after leaving a virtual subshell
causes a crashing bug: https://github.com/att/ast/issues/803
Co-authored-by: Johnothan King <johnothanking@protonmail.com>
Fixes: https://github.com/ksh93/ksh/issues/114
As of aa4669ad, astconf("PATH") is implemented as a hardcoded AST
configuration variable that always has a value, instead of one that
falls back on the OS. Its value is now obtained from the OS (with a
fallback) at configure time and not at runtime. This means that any
fallback for astconf("PATH") is now never used.
src/cmd/ksh93/data/msg.c,
src/cmd/ksh93/include/shell.h:
- Remove e_defpath[]. (The path "/bin:/usr/bin:" made no sense as a
default path anyway, as the final empty element is wrong: default
utilities should never be sought in the current working dir.)
src/cmd/ksh93/sh/path.c,
src/lib/libast/path/pathbin.c:
- abort() if astconf("PATH") returns null.
src/lib/libast/comp/conf.tab: PATH:
- If no 'getconf' utility can be found, use a fallback path that
finds more utilities by also searching in 'sbin' directories.
On some systems, this is needed to find chown(1).
src/cmd/ksh93/sh.1:
- Update doc re default path.
There are convincing arguments why including '.' and '..' in the
result of pathname expansion is actively harmful. See:
https://www.austingroupbugs.net/view.php?id=1228https://github.com/ksh93/ksh/issues/58#issuecomment-653716846
pdksh, mksh and zsh already skip these special traversal names
in all cases. This commit makes ksh act like these shells.
Since passing '.' and especially '..' as arguments to commands like
'chmod -R' and 'cp -r' may cause harm, this change seems likely to
fix more legacy scripts than it breaks. I'm unaware of anyone ever
having come up with a concrete use case for the old behaviour.
This change also fixes the bug that '.' and '..' failed to be
ignored as documented if FIGNORE is set.
src/lib/libast/misc/glob.c: glob_dir():
- Explicitly skip any matching '.' and '..' in all cases.
src/cmd/ksh93/tests/glob.sh:
- Add test_glob() tests for '*' and '.*'.
src/cmd/ksh93/sh.1: File Name Generation:
- Update to match new behaviour.
Resolves: https://github.com/ksh93/ksh/issues/58
src/lib/libast/tm/tmxfmt.c:
- Making %l and %k aliases to %_I and %_H caused zero padding with
%0l and %0k to fail. Fix that by fully implementing %l and %k
without 'goto push'. This duplicates code from %I and %H, but it
is necessary for these formats to work correctly when zero padded.
src/cmd/ksh93/tests/builtins.sh:
- Add a regression test for manually specifying blank and zero
padding with sixteen different formats.
