It seems that MOLD_MOSTLY_STATIC was required before to avoid packaging
every single dependency. But nowadays, mold uses whatever dependency it
can find and builds its own if it cannot.
Saves a small amount of compilation time.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/18881
Signed-off-by: Robert Marko <robimarko@gmail.com>
It's using cmake.mk so it needs CMake to build.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/18880
Signed-off-by: Robert Marko <robimarko@gmail.com>
Newer GCC and Clang default to C++17, which does not support register.
Just remove it.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/18859
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
New m4 release was made after a long time, for us most importantly
it ships newer gnulib so it compiles against GCC15 without forcing GNU17.
* Noteworthy changes in release 1.4.20 (2025-05-10) [stable]
** Fix a bug in the `eval' builtin where it does not suppress warnings
about division by zero that occurs within a more complex expression on
the right hand side of || or && (present since short-circuiting was
introduced in 1.4.8b).
** The `syscmd' and `esyscmd' builtins no longer mishandle a command line
starting with `-' or `+' (present since "the beginning").
** Fix regression introduced in 1.4.19 where trace output (such as with
`debugmode(t)') could read invalid memory when tracing a series of
pushed macros that are popped during argument collection.
** Fix regression introduced in 1.4.19 where the `format' builtin
inadvertently took on locale-dependent parsing and output of floating
point numbers as a side-effect of introducing message translations.
While it would be nice for m4 to be fully locale-aware, such a behavior
change belongs in a major version release such as 1.6, and not a minor
release.
** Fix regression introduced in 1.4.11 where the experimental `changeword'
builtin could cause a crash if given a regex that does not match all
one-byte prefixes of valid longer matches. As a reminder, `changeword'
is not recommended for production use, and will likely not be present
in the next major version release.
** On non-Unix platforms where binary files differ from text, loading a
frozen file (which should be cross-platform compatible) now correctly
uses binary mode.
** Several documentation improvements to the manual.
** Update to comply with newer C standards, and inherit portability
improvements from gnulib.
Link: https://github.com/openwrt/openwrt/pull/18821
Signed-off-by: Robert Marko <robimarko@gmail.com>
Update b43-tools to the latest upstream version to fix a
compilation error with C23:
util.h:25:15: error: 'bool' cannot be defined via 'typedef'
25 | typedef _Bool bool;
| ^~~~
Changelog:
c6fc53f replace custom bool typedef with <stdbool.h>
dadf30c fix format warning in compilation
2fe10ea b43-fwdump: Fix forwarding of arguments to disassembler
Link: https://github.com/openwrt/openwrt/pull/18708
Signed-off-by: Nick Hainke <vincent@systemli.org>
Update libtool to the current stable 2.5.4 release.
130-trailingslash.patch was upstream, 100 and 110 patches were
manually refreshed while rest was automatically refreshed via quilt.
Link: https://github.com/openwrt/openwrt/pull/18655
Signed-off-by: Robert Marko <robimarko@gmail.com>
13767a9 make_ext4fs: add missing space after LDFLAGS
Signed-off-by: Konstantin Demin <rockdrilla@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/18682
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
mold has seen a bunch of bugfixes and improvements since 2.33.0, so
lets update it.
Link: https://github.com/openwrt/openwrt/pull/18575
Signed-off-by: Robert Marko <robimarko@gmail.com>
Prevent accidentally including an old version of uuid.h, which causes
the build to fail because of missing definitions.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
5.8.1 (2025-04-03)
* Multithreaded .xz decoder (lzma_stream_decoder_mt()):
- Fix a bug that could at least result in a crash with
invalid input. (CVE-2025-31115)
- Fix a performance bug: Only one thread was used if the whole
input file was provided at once to lzma_code(), the output
buffer was big enough, timeout was disabled, and LZMA_FINISH
was used. There are no bug reports about this, thus it's
possible that no real-world application was affected.
* Avoid <stdalign.h> even with C11/C17 compilers. This fixes the
build with Oracle Developer Studio 12.6 on Solaris 10 when the
compiler is in C11 mode (the header doesn't exist).
* Autotools: Restore compatibility with GNU make versions older
than 4.0 by creating the package using GNU gettext 0.23.1
infrastructure instead of 0.24.
* Update Croatian translation.
Link: https://github.com/openwrt/openwrt/pull/18558
Signed-off-by: Robert Marko <robimarko@gmail.com>
drop upstream patches
$ git log --no-merges --oneline v2.7.6..v2.8
48ceda8 Version 2.8
b3d0c93 Regression in commit abe92e8010ab affecting MariaDB tests
29fced6 Count traditional diff pattern lines correctly
b5d2124 patch: fix --no-backup-if-mismatch regression
86ac7e2 Fix dodgy assert with side-effects in insert_cached_dirfd
7d87652 Declare an expected test failure on Haiku.
86baf97 build: update gnulib submodule to latest
1ba2c1b Fix two test failures on Haiku.
1da6bf8 Check for newlines only when creating a file name
30ee610 Gnulib renamed some modules
6dbc381 maint: update bootstrap from gnulib
05ac924 build: update gnulib submodule to latest
95e0092 maint: make update-copyright
5bac274 Set --no-backup-if-mismatch when in --posix mode
910fecf Add missing feature tests to the test suite
be8b3c6 Disable release-prep
c61485b Fix "make release" to handle alpha releases
499916f Add announce-gen module for "make release"
adb1ebc Pacify gcc -fsanitize=address
6bdae94 Fix memory leak when malformed unidiff patch
72a146c Port to clang address sanitizer
e2e6820 Refactor argc+argv processing
606c091 Omit needless get_some_switches code
ee3cc40 Revert "Remove obsolete require_gnu_diff function"
8cae4fc Remove obsolete require_gnu_diff function
164b529 savebuf can return a null pointer
91c1e4f Spelling fixes
a03e1ba Port other reject-format test to non-GNU diff
9ba5eb0 Don’t be fooled by NUL bytes in diff directives
79dd5e7 Don’t be fooled by "\000" in file name
8492a6a Port to quasi-GNU diff
f6f2c6f In previous patch, make w_q static
0525681 Pacify -Wunterminated-string-initialization
301411d Spelling fixes
4615206 Remove double semicolon
923e0ef Prefer angle bracket headers
5d17ca0 Update more old copyright notices
5f4edd3 Simplify memory allocation of files to delete
802511c Report patch read errors more immediately
a93b50d Port fflush usage to OpenBSD 7.5
55e224b Update old copyright notices
33a7fd8 Fix gl_gcc_warnings typo in ‘configure’
2313b37 build: update gnulib submodule to latest
37fec39 Remove some dependencies no longer needed
cc87173 Update NEWS as per recent changes
7887622 Update POSIX citations
18f4dd6 Use “Gruenbacher” in international contexts
638675c Adjust libs to match recent Gnulib
53400a1 Pacify clang -Wbitwise-conditional-parentheses
e8e1bcb Remove “support” for nested critical sections
b963510 Simplify critical section code in util fns
85949fb Shrink critical sections
448ff9b Defer signals by hand with sigatomic_t
b95a603 output_file_later avoid a malloc+free pair
2663228 Defend against closed stdin/stdout/stderr
fc77964 Stop using Gnulib ‘execute’ module
470699c Move defer_signals up
f696e4e Make sigs, NUM_SIGS local
4d3a4ab Rename block_signals
b3bb925 Trade a bit of space for time in parse_c_string
070d859 Allocate first patchbuf statically
f46a90f Use bigger buffer size by default
2e64cfd Preinitialize fatal_act.sa_hander
45de0d9 Prefer EXIT_SUCCESS etc. to literal integers
82c4940 Fix some signal handling races
eceea61 Fix --set-utc TZ setting
9c55d3f Improve logic for when rename removes source
61c72f0 Fix signal race when renaming file
705c9bc Simplify traverse_another_path via last_component
77f21a6 Stop using Gnulib dirname module
fe8ffd9 Fix implausible overflow when reading symlinks
5e84bda Access checks should use effective, not real
8d4ca49 Don’t assume AT_FDCWD != -1
fe5d4a0 Copy input to output attributes via fd if possible
ca4c431 Simplify timestamp epoch checking
0ad4347 Check for ftello failures
c49a16d Remove format_linenum
22efdee Fix "with multiple words" line number
5a70a1b Drop Plan B
c0d465f Prefer ximemdup0 to xmemdup0
b91aab2 Refactor ifetch API
55c8a5c Do not attempt huge I/Os
e0e121e Use outfd when setting file attributes
241e57e Don’t assume Linux-like S_IFREG
0f8c628 Use STDOUT_FILENO etc
f2c3676 Fix unlikely glitch with ed diffs
cec6407 Use fds to copy attrs in create_backup_copy
6b7b01b Be more careful about (time_t) -1
0433553 X == -1 → X < 0
0a66dee Let set_file_attributes use fds not names
36ff2c9 Port to narrow unsigned uid_t
3951496 Check for output errors more systematically
6429630 Report input error right away
03cb187 Simplify EOF testing
cc7cde7 Prefer other types to ‘int’
4c6650b Detect unlikely integer overflow in size calcs
abe92e8 Prefer idx_t, ptrdiff_t to lin
30449e2 Fix compatibility issue with blanks in patches
9228a8c pch_swap return type cleanup
c1c438d Fix unlikely int overflow in hunk counts
5d3f41f Use char for char in plan_a
fb056f2 Cache cwd_is_root dev, ino
84b5f34 Avoid ‘unsigned’ in safe.c
d2e113e Simplify get_sha1
f73718b Avoid casts in patch.c
79eef3e Prefer idx_t in util.c
11588d0 Prefer idx_t in pch.c
e16037d Prefer idx_t in patch.c
388926f Prefer idx_t in list.h
3582fdb Prefer idx_t in inp.c
4a47c00 build: update gnulib submodule to latest
4f8c4b8 Don’t limit strip counts etc. to INT_MAX
0a810b6 Fix unlikely integer overflows in patch.c
bac3b6d Fix unlikely integer overflows in pch.c
8fb784b Fix unlikely integer overflows in inp.c
e10f3ca Promote minmax.h to common.h
57e2165 Avoid some memory allocation by not using ‘const’
eb18b39 Remove unnecessary char * casts in inp.c
d60cb72 Pacify gcc -Wunused-parameter when !USE_XATTR
05ef886 Pacify gcc -Wunused-parameter when !USE_XATTR
1f8d192 maint: stop using alloca
ba92722 Don’t assume O_RDONLY == 0
0f98e03 Avoid syscall when nested signal block
1235ccc Add signal comment
43ee674 build: update gnulib submodule to latest
e381947 Update NEWS, README-prereq
2c2a83b Omit _Noreturn when easy
a13c2ea Replace __attribute__ with attribute.h
6eb2d13 Switch from ctype.h to c-ctype.h
d1a6847 Simplify warning configuration
d6631b3 Prefer ATTRIBUTE_* to _GL_ATTRIBUTE_*
bc6899d Pacify gcc -Wno-unused-parameter
da25985 Improve ‘git diff’ output if desired
248ef13 Prefer strerror to perror
6cb321a A bit more long-string fixing
47bc09d Prefer nullptr to NULL
7608746 More fixing of printing of very long strings
323da0d Don’t assume string sizes fit in int when printing
53d1014 Avoid fprintf INT_MAX overflow when merging
59681c8 Avoid sprintf INT_MAX overflow
4278b91 Reject output file names containing '\n'
34b45bc Update man page a bit.
d18c05d Update copyright notices
af828e5 Fix some races involving signals
b3a6c95 Don’t attempt to remove files we didn’t create
9abc949 Omit goto in try_safe_open
90e62d5 Pacify clang re obsolete O_CREAT test
2b87c1e Allow nested block/unblock of signals
7aa1c3b Adjust to new Gnulib bootstrap post imports
99c0c0b maint: remove generated file lib/Makefile.am
1c087d6 Rely on Gnulib inttypes module
7214f8d Update main locals more consistently
6785b2c Use struct outfile * in function args
72d7ed0 Refactor temp names into struct
abf6fb1 Simplify by using Gnulib sigaction
d3816ac Avoid unnecessary freeing in output_files
346d3ac Clean up cleanup
ff2317b Port better to GNU/Hurd
c2d9792 Don’t say empty backups are unreadable
8c27a03 Spelling fixes
d46d729 Change manywarnings usage to be more like coreutils
924698b Pacify clang, which dislikes n + "y"
8939519 Pacify -Wstrict-overflow in pch.c
531cc2b Pacify -Wsuggest-attribute=format in util.c
ff13fea Port to non-VLA C compilers
3d5c0d1 Rename vars to pacify gcc -Wshadow
56788ce Stop including stdbool.h
c10da77 Recommend 64-bit time_t on 32-bit platforms
1e21767 Remove pch_sha1
39005cf Move skip_spaces
755712d Remove pch_timestamp function
04f0eeb Prefer extern inline to static inline for list.h
f06c123 maint: pacify gcc 14 -Wcast-align
aab6e7b maint: pacify -Wanalyzer-null-argument
d1d32c9 maint: work around GCC bug 109839
7575694 maint: pacify gcc -Wmissing-variable-declarations
8f78b09 maint: pacify gcc -Winline
bb841fd maint: port _FORTIFY_SOURCE to Ubuntu
4887683 maint: assume STDC_HEADERS
5b8ecde maint: spruce up our .m4 files a bit
009a424 maint: omit obsolete macro calls
299167f maint: simplify .gitignore
3ec44a4 build: update gnulib submodule to latest
68cb529 build: update gnulib submodule to latest
f144b35 build: Enable the 'subdir-objects' Automake option.
faafc79 build: update gnulib submodule to latest
c835ecc Pass the correct stat to backup files
24f81be maint: modernize README-{hacking,prereq}
7623b2d Fix test for presence of BASH_LINENO[0]
0993940 gnulib: update to latest
78ed9de Add missing-section tests to context-format test case
76e7758 Fix failed assertion 'outstate->after_newline'
15b158d Avoid invalid memory access in context format diffs
dce4683 Don't follow symlinks unless --follow-symlinks is given
61d7788 Don't crash when RLIMIT_NOFILE is set to RLIM_INFINITY
b7b028a Abort when cleaning up fails
a5b442c Skip "ed" test when the ed utility is not installed
2b584ae Improve support for memory leak detection
9c98635 Fix swapping fake lines in pch_swap
ff81775 Make the (debug & 2) output more useful
369dccc Don't leak temporary file on failed multi-file ed-style patch
1959988 Don't leak temporary file on failed ed-style patch
f322a7e Request 'alloca' module from gnulib.
458ac51 Fix 'ed-style' test failure.
1e9104c Fix check of return value of fwrite().
ae81be0 maint: avoid warnings from GCC8
2a32bf0 Minor cleanups in do_ed_script
ff1d3a6 Use gnulib execute module
3fcd042 Invoke ed directly instead of using the shell
123eaff Fix arbitrary command execution in ed-style patches (CVE-2018-1000156)
b5a91a0 Allow input files to be missing for ed-style patches
f290f48 Fix segfault with mangled rename patch
074e239 Test suite: fix Korn shell incompatibility
f6bc5b1 Test suite compatibility fixes
3bbebbb Avoid set_file_attributes sign conversion warnings
Signed-off-by: Russell Senior <russell@personaltelco.net>
Link: https://github.com/openwrt/openwrt/pull/18479
Signed-off-by: Robert Marko <robimarko@gmail.com>
Fedora 42 updated to GCC15 which now defaults to GNU23 as the default
instead of GNU17[1], and this breaks m4 compilation.
This looks like a gnulib issue, so until that is updated/fixed
lets simply set C language version back to GNU17.
[1] https://gcc.gnu.org/gcc-15/porting_to.html#c23
Link: https://github.com/openwrt/openwrt/pull/18506
Signed-off-by: Robert Marko <robimarko@gmail.com>
This reverts commit 21cece29e9.
Unfortunately, LibreSSL 4.0 causes the host APK tools to segfault under
fakeroot when .apk is being generated and it is completely breaking
building under Debian and Ubuntu hosts.
So, until this is fixed lets revert the update.
Signed-off-by: Robert Marko <robimarko@gmail.com>
Stable version. Odd versions are betas.
Added static patch to rename some ecdsa symbols that conflict with
u-boot's mkimage.
These symbols are not exported by default but because OpenWrt uses a
static libressl, they are present and conflict with mkimage's libecdsa.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/16901
Signed-off-by: Nick Hainke <vincent@systemli.org>
This reverts commit 160e0b7ad8.
It seems that this was not tested on MacOS and will fail to compile,
so revert it until a proper update is done.
Fixes: 160e0b7ad8 ("tools/coreutils: update to 9.6")
Signed-off-by: Robert Marko <robimarko@gmail.com>
Currently macOS builds are failing due to hash mismatch. Lets fix it by
using proper hash:
Here are the SHA1 and SHA256 checksums:
File: coreutils-9.6.tar.gz
SHA1 sum: 1da82e96486e0eedbd5257c8190f2cf9fcb71c2e
SHA256 sum: 2bec616375002c92c1ed5ead32a092b174fe44c14bc736d32e5961053b821d84
References: https://lists.gnu.org/archive/html/coreutils/2025-01/msg00049.html
Fixes: 160e0b7ad8 ("tools/coreutils: update to 9.6")
Reported-by: Mieczyslaw Nalewaj <namiltd@yahoo.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Fix the 000-relocatable.patch broken by e0f5ce9. The patch segment about
detecting STAGING_DIR_HOST was erroneously removed, as upstream had
deleted the previous bin/autoconf.as and had implemented it in perl
in bin/autoconf.in. Re-create the previous functionality in that.
Fixes: #18059
Fixes: e0f5ce974 ("tools/autoconf: bump to 2.72")
Signed-off-by: Hannu Nyman <hannu.nyman@iki.fi>
Link: https://github.com/openwrt/openwrt/pull/18073
Signed-off-by: Robert Marko <robimarko@gmail.com>
Switch to the git repository source so that we can use zstd
compression algorithm to create smaller package tarball. This
patch also corrected the license file name[1].
[1] 352cb28d12
Suggested-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: Shiji Yang <yangshiji66@qq.com>
Link: https://github.com/openwrt/openwrt/pull/17880
Signed-off-by: Robert Marko <robimarko@gmail.com>
This reverts commit b930ce3bf7.
Bump to 1.17 broke SDK portability as it partially reverted commit
("automake: portability fixes") which made Perl PATH "/usr/bin/env perl"
as that is portable, as otherwise automake will set the absolute path to
the Perl binary and this will then fail when using SDK on a different
system as that PATH is not present.
So, since fixing this would require backport of upstream commit
("configure: make perl path with whitespace a warning, not error.") which
requires autoreconf to be done in order for configure to get regenerated
we cannot do it because at that time we do not have automake built.
So, for now revert the bump until upstream makes a new release.
Signed-off-by: Robert Marko <robimarko@gmail.com>
Changelog included in package.
Signed-off-by: Jack Sun <sunjiazheng321521@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/17826
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
refresh patch:
001-no_doc.patch
002-fix-libmath.patch
Changelog included in package.
Signed-off-by: Jack Sun <sunjiazheng321521@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/17828
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The latest upstream version of Yafut builds on macOS and other
POSIX-compatible systems. Drop the custom OpenWRT patch applying
non-Linux compatibility fixes to the tool's source code.
Signed-off-by: Michał Kępień <openwrt@kempniu.pl>
Link: https://github.com/openwrt/openwrt/pull/18014
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The ccache build relies on doctest, which
is useless and can be disabled.
Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
Link: https://github.com/openwrt/openwrt/pull/17700
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
The ccache build requires xxhash, which is download from
the Internet. So add host-build package for xxhash.
Fixes: #16851
Signed-off-by: Chukun Pan <amadeus@jmu.edu.cn>
[ use git source proto ]
Link: https://github.com/openwrt/openwrt/pull/17700
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Update to include the latest upstream improvements and bugfixes, including
pahole now always encoding reproducibly.
Drop the local patch:
100-reproducible-builds.patch
Release Notes: https://lore.kernel.org/bpf/Z4-TDt42dTKZvCo6@x1/
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/17705
Signed-off-by: Robert Marko <robimarko@gmail.com>
Update to latest version. There are no patches that need to be
refreshed.
Signed-off-by: Shiji Yang <yangshiji66@qq.com>
Link: https://github.com/openwrt/openwrt/pull/17539
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Fixes since 4.10:
* Fixed detection of Fmt version 11 and newer.
* Fixed prefix command lookup from PATH.
Suggested-by: Jen Wolf <jenwolf@protonmail.com>
Link: https://github.com/openwrt/openwrt/pull/17105
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Update to the most recent stable release,
including a new job scheduler and bug fixes.
Backport a patch to assist in reworking jobserver support.
The original patch for jobserver support requires reworking
due to the function CanRunMore() and its usage
being significantly changed, including the return
becoming an integer instead of a boolean.
AcquireToken() must now be used in CanRunMore()
in order to quantify how much the job searching loop
is able to run at any time, and in order to do so
CanRunMore() cannot be a constant function anymore.
Added:
- 010-backport-gtest.patch
Manually Adjust:
- 100-make_jobserver_support.patch
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/16693
Signed-off-by: Robert Marko <robimarko@gmail.com>
Add methods to skip the building and execution of Ninja by python
in order to allow Make to execute Ninja after the configure script.
This allows the user to build Ninja only once
if they already have an older version of Ninja built.
It also allows the user to test the jobserver functionality
by having Ninja built twice (clean then compile).
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/16693
Signed-off-by: Robert Marko <robimarko@gmail.com>
Use default CXXFLAGS and LDFLAGS variables.
Directly add --verbose flag without a variable.
Split up lines for script arguments.
Delete unnecessary lines.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/16693
Signed-off-by: Robert Marko <robimarko@gmail.com>
Bump to the latest 1.22 version which allows dropping our only patch.
Changelog:
Version 1.22
* The CMake-based build system now implements a workaround for gcc being paired
with a too-old binutils version. This can prevent build errors.
Version 1.21
* Fixed build error on x86 with gcc 8.1 and gcc 8.2.
* Fixed build error on x86 when gcc 11 is paired with a binutils version that
doesn't support AVX-VNNI, e.g. as it is on RHEL 9.
* Fixed build error on arm64 with gcc 6.
* Fixed build error on arm64 with gcc 13.1 and later with some -mcpu options.
* Enabled detection of dotprod support in Windows ARM64 builds.
Link: https://github.com/openwrt/openwrt/pull/16617
Signed-off-by: Robert Marko <robimarko@gmail.com>
In some versions of Xcode, some C++ header has an include
of the standard C++ header "stack" while "." is in the include paths
which can conflict with the binary "stack" built by elfutils.
This leads to a decode error as the binary is interpreted as text.
Add an arbitrary dependency between stack and the C++ program.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
The type 'id_t' check has been moved to configure.ac in upstream commit:
819063eb4d27 ("test for id_t with autoconf instead of blind typedef, fixes FTBFS on FreeBSD")
changelog:
e4780b1aad/debian/changelog
Signed-off-by: Shiji Yang <yangshiji66@qq.com>
Link: https://github.com/openwrt/openwrt/pull/16530
Signed-off-by: Nick Hainke <vincent@systemli.org>
After digging in the history of the e2fsck exit code patch, it seems like
there is no reason for us to carry this patch so lets drop it.
Link: https://github.com/openwrt/openwrt/pull/16500
Signed-off-by: Robert Marko <robimarko@gmail.com>
Currently, e2fsprogs is being patched to avoid detecting host crond, but
instead of doing that we can simply pass --with-crond-dir=no as an
argument and drop the patch.
Link: https://github.com/openwrt/openwrt/pull/16500
Signed-off-by: Robert Marko <robimarko@gmail.com>
Considering that this patch to not build e4defrag was added 12 years ago
because:
Compilation fails on older systems due to missing syscalls
I think its safe to say that enough time is passed so that whatever those
older systems were are either updated or not used anymore so lets drop
the patch.
Link: https://github.com/openwrt/openwrt/pull/16500
Signed-off-by: Robert Marko <robimarko@gmail.com>
flex currently leaks the path of m4 as found on the build host.
While it is possible to override this using the M4 environment
variable (which we always did for autotools based builds) when
using CMake or Ninja the M4 variable is not set.
One easy fix is to make flex take STAGING_DIR_HOST into account
and expect m4 there if that variable is set in the environment.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Another instance of files in build_dir symlinking to staging_dir. While
the symlinks do not currently cause any bugs in the libtool package,
such symlinks were found to make the build more fragile, as writing to
the symlink may accidentally modify the shared file in staging_dir. Pass
--copy to bootstrap to disable the symlinking.
Link: https://github.com/openwrt/openwrt/pull/15825
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Update to a more recent stable release.
Most notably, this version includes
some fixes for building on an OS like Alpine.
This allows for the removal of hacks
that fixed building on Alpine,
but broke building on ARM archs.
Manually adjust:
- 7-zip-flags.patch
- 7-zip-musl.patch
Link: https://7-zip.org/history.txt
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15991
Signed-off-by: Robert Marko <robimarko@gmail.com>
Define the version in one variable,
and use Make functions to use variations of it elsewhere.
Correct the CPE ID.
Override the default tar directory flag
by adding a new value to the tar options
instead of defining the entire command.
Use variables for adjusting build recipes
instead of adding custom build recipes.
Remove unnecessary lines and add spacing.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15991
Signed-off-by: Robert Marko <robimarko@gmail.com>
0cfd1043e6c4 zytrx: add ZyXEL LTE7490-M904
3d09357fc7b5 asusuimage: new tool for creating TRX-images compatible with AsusWRT
f35781fcd2c3 asusuimage: fix compile error with old GCC release
Link: https://github.com/openwrt/openwrt/pull/15840
Signed-off-by: Robert Marko <robimarko@gmail.com>
It seems that util-linux enables NLS support by default, this worked for
almost all platforms except for macOS on x86 where it seems that libintl is
preinstalled and thus it will link against gettext with libintl for NLS
support.
This would the later cause e2fsprogs and mtd-utils to fail:
Undefined symbols for architecture x86_64:
"_libintl_gettext", referenced from:
_random_tell_source in libuuid.a[13](libuuid_la-randutils.o)
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
Issue appeared after I converted the tool to use --disable-all-programs and
accidentally dropped the --disable-nls from the args.
Fixes: 54115ec22d ("tools: util-linux: use --disable-all-programs")
Link: https://github.com/openwrt/openwrt/pull/15909
Signed-off-by: Robert Marko <robimarko@gmail.com>
After changes to default host build recipes
and default variable definitions,
several custom definitions can be removed,
and the gnulib recipes replaced
with hooks to common recipes.
Also remove leftover PKG_INSTALL
which has no effect for host builds.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15853
Signed-off-by: Robert Marko <robimarko@gmail.com>
Replace the symlink in the build recipe for gnulib
with an extra include path flag in CPPFLAGS to the lib subdirectory
so that it is the last in the order of include paths,
and use a quote escape to make the flag a literal string
in order to use Make variables within it.
The original reason this is necessary is because the config.h header
provided by the project at the top-level build directory
calls another header eu-config.h
which is stored in the lib subdirectory instead of the top-level,
and building the gnulib library requires the config.h header.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15853
Signed-off-by: Robert Marko <robimarko@gmail.com>
We have to use curly braces on the exported STAGING_DIR_HOST env variable,
instead of evaluating it directly as we are not in Make but a separate
shell script.
Otherwise it would fail with:
staging_dir/host/bin/compile_et: line 6: STAGING_DIR_HOST: command not found
staging_dir/host/bin/mk_cmds: line 5: STAGING_DIR_HOST: command not found
And so when krb5 tries to build it will fail as compile_et and mk_cmds will
return an error.
Fixes: 55bda9863d ("tools/e2fsprogs: fix shell scripts under SDK")
Link: https://github.com/openwrt/openwrt/pull/15854
Signed-off-by: Robert Marko <robimarko@gmail.com>
Needed to fix users of libuuid.a as autoconf applies PIC to only shared
libraries by default.
Found when trying to build python3/host.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15852
Signed-off-by: Robert Marko <robimarko@gmail.com>
7c32295b00 exposed a problem where the SDK
builds these shell scripts with a nonsensical absolute path for the DIR
variable. Use sed to patch in $STAGING_DIR_HOST.
Also remove PKG_RELEASE as that is nonsensical for tools.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15841
Signed-off-by: Robert Marko <robimarko@gmail.com>
This was added to support building coreutils on host systems
that still only have 32-bit time support.
Because other tools now also require the flag for building
with 32-bit time when support for 64-bit time is not present,
this flag is now added to all host builds
on a variable basis using the same test before building,
so it can now be removed from specific tools.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15799
Signed-off-by: Robert Marko <robimarko@gmail.com>
Now that util-linux is building libuuid we can simply use that instead
of manually pointing to e2fsprogs libuuid so we can disable building
libuuid in e2fsprogs.
Link: https://github.com/openwrt/openwrt/pull/15806
Signed-off-by: Robert Marko <robimarko@gmail.com>
There is no need to manually only remove 2 binaries during cleanup and
leave rest of the e2fsprogs installed stuff untouched, so instead use
make uninstall to do the cleanup.
Link: https://github.com/openwrt/openwrt/pull/15806
Signed-off-by: Robert Marko <robimarko@gmail.com>
We are now using util-linux to provide libuuid so disable building it
in e2fsprogs and simply use the util-linux provided one.
Disabling libuuid removes the need for custom install recipe.
Link: https://github.com/openwrt/openwrt/pull/15806
Signed-off-by: Robert Marko <robimarko@gmail.com>
util-linux usually provides libuuid, and is preffered by mtd-utils so lets
enable building libuuid so we can disable it later in e2fsprogs.
Only the static version of library is intentionally built.
Link: https://github.com/openwrt/openwrt/pull/15806
Signed-off-by: Robert Marko <robimarko@gmail.com>
Instead of manually installing the binaries and removing them, we can now
simply rely on standard make install/uninstall as we are only building
tools we want to use.
This will be especially important when we start building libraries in
util-linux.
Link: https://github.com/openwrt/openwrt/pull/15806
Signed-off-by: Robert Marko <robimarko@gmail.com>
util-linux supports passing --disable-all-programs configure flag to
disable building anything that isnt then manually enabled.
So, lets switch to using that instead of manually having to disable all
tools we dont need.
However, current drawback is that there is no upstream support for enabling
building hexdump so I included a patch that is pending upstream[0].
[0] https://github.com/util-linux/util-linux/pull/3101
Link: https://github.com/openwrt/openwrt/pull/15806
Signed-off-by: Robert Marko <robimarko@gmail.com>
Passing --disable-tests does not do anything since upstream commit [0]
("Fix test binary installation") as that commit removed it since there is
already and existing --without-tests compile option to not compile
the tests at all as --disable-tests was just disabling their installation.
So, lets just pass --without-tests instead to disable test compilation.
[0] https://git.infraroot.at/mtd-utils.git/commit/?id=7170a28d46d5db1e7a9da24a5555a194a233ef0b
Fixes: 67efb6a661 ("tools: mtd-utils: Update to version 2.1.4")
Link: https://github.com/openwrt/openwrt/pull/15791
Signed-off-by: Robert Marko <robimarko@gmail.com>
Release notes:
https://lists.infradead.org/pipermail/linux-mtd/2024-March/104058.html
Manually refresh the portability and JFFS2 LZMA patches.
Since mtd-utils have converted most of the JFFS2 compressors to be compile
time configurable and manual refreshing of JFFS2 LZMA was needed I also
converted it to a compile time option and enabled the new --with-lzma
option.
Link: https://github.com/openwrt/openwrt/pull/15791
Signed-off-by: Robert Marko <robimarko@gmail.com>
Fakeroot uses an egrep configure check to look for the string "time64"
within a preprocessed include of the sys/stat.h header
in order to decide whether or not to create declarations
and internal functions used for wrapping the native functions
stat64_time64 fstat64_time64 lstat64_time64 and fstatat64_time64.
On specific older versions of glibc these functions are not included,
but there are some references to "time64" unrelated to functions,
like aliasing the time64_t typedef to the standard time_t typedef
when the size of a word is 64 bits or defining it if not.
In this case, a grep for "stat64"
of the preprocessed sys/stat.h header matches nothing,
however, the grep for "time64" in the configure check
of the preprocessed sys/stat.h header matches a line
that has nothing to do with the functions
that will be wrapped as a result of successful matching.
__extension__ typedef __int64_t __time64_t;
This causes the attempt to wrap the functions to occur,
which fails due to some of the corresponding macros being empty
since the native declarations they are based on do not exist,
causing a common error claiming that a part of the syntax is missing.
error: expected '=', ',', ';', 'asm' or '__attribute__' before ...
Fix this by making the testing regular expression more complex
in order to match actual function names ending in "_time64"
with or without a set of preceding underscores, but none after.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15773
Signed-off-by: Robert Marko <robimarko@gmail.com>
224d497dd94f srec2bin: drop unused "dum" variable
6777b2d51961 uimage_sgehdr: use "char" type for header struct strings
81db3025aac5 uimage_sgehdr: drop unused "ltmp" variable
bd7fcc74b43e pc1crypt: make decrypt/encrypt functions take void * as argument
6ac44974185a linksys: add magic header generation tool for e8350 v1
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
As a side-effect to adding a gnulib module for posix_fallocate(),
there are changes to the input file for fcntl.h which
are not handled here since autoreconf is not ran.
Skip updating the fcntl.h header from gnulib
and use the version shipped with the release.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
In order for linking the static libraries from elfutils to work,
other libraries need to be included to handle the references
to functions made in the library's objects that are not included
as they would already be if the library was a shared object instead.
A shared object library stores this list of libraries when it was made,
so that the dynamic linker can refer to that list at runtime,
but a static library has no such functionality so the list of libraries
for missing functions must be included at link time.
This information was already added to the pc file for libelf
using the definitions in src/Makefile.am,
so extend this to the rest of the pc files in the project.
For situations where the libraries may be used
without pkg-config setting the flags and library list,
this patch and the pc files can serve as a quick reference.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
Clang has support for weak aliases
despite no support for strong aliases,
but it only works with the #pragma directive.
Implementing weak aliases instead of none
is likely a more upstream-friendly solution
for supporting building on other platforms.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
The addition of LT_INIT as well as the adjustment of
the BUILD_STATIC and addition of the BUILD_SHARED conditionals
and their usage to block building of shared objects
and adjust the variables for building static libraries
is potentially upstream-friendly.
The use of a manifest file to keep a list
of the objects in each library instead of calling ar
is also potentially upstream-friendly.
Separate these changes from the macOS-specific hacks.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
Add a potentially upstream-friendly conditional
using the libtool configure variable "enable_shared"
in order to block building and installing of shared objects
and adjust the build of static libraries
instead of directly patching lines in or out.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
The use of ar to list the archive members in a library
in order to include them in another library is not portable.
On BSD and macOS, ar will also list
the special archive member "__.SYMDEF"
which is not a compiled object, rather it is
part of the metadata prepended to the library by ranlib.
Fix this by writing the list of unique objects used
to create the library into a separate "manifest" file
when the library is created, which will be read later
when the Makefiles of other subdirectories are ran.
Extend this to all other libraries whether or not they are linked
to another library for a shared object that is installed
so that it is possible for any of the libraries
to be statically built with more objects.
The use of the wildcard function to ignore the
special archive members which are only metadata
is no longer needed to prevent build errors.
Not using the wildcard function is preferred,
since errors should be caught during the build
instead of when linking something else or at runtime.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
Importing gnulib in order to have a local portable library
to link against for missing functions currently requires
using libtool to produce the libgnu.la library.
Ideally, linking would be simple if the rest of the libraries
built by elfutils were also built using libtool, as linking
them together would not require any manipulations of library paths.
However, upstream elfutils does not support building the libraries
statically with libtool, so using libtool comes at the cost
of creating a huge patch to introduce that functionality.
For building on macOS, it turns out that libgnu.la is only needed
for building the binaries, and that just one or two objects from libgnu
are needed to build the libraries, so in this case, it would be simple
to add the specific non-libtool-wrapped library and objects
to the link paths as needed, rather than use libtool to link
the libtool wrappers, which greatly reduces the need to patch.
Not using libtool also makes the original Makefile definitions for LIBADD
once again be the right ones to use. However, to be portable,
for libdw the wildcard function needs to be used in order to exclude
special archive members like "__.SYMDEF" which are not compiled objects
because some BSD-like versions of ar include that metadata in the list,
or because the library included may have objects from another subdirectory.
Also, the rest of the subdirectories have custom "LDLIBS" variables
meant for building shared objects only, so define the LIBADD variables
with objects from those existing definitions so that when building only
the static versions of the libraries, those objects can still be included.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
The version of posix_fallocate() patched into elfutils
for macOS using code from Mozilla is now patched into gnulib.
Import the fallocate-posix module and always link
the corresponding object to libraries whenever it is built.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
Add a module to gnulib to support posix_fallocate()
for macOS and other systems that are missing it.
Apple-specific code is sourced from Mozilla,
and the rest from glibc, both licensed under LGPL.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
Several changes to the elfutils source files
made during the process of figuring out how to
successfully build elfutils on macOS
turn out to not be necessary to do so,
and were most likely leftover bits during testing.
Remove the line changes that are not needed
and add some line changes to adapt to sources as is:
- Remove now unnecessary bump to autoconf version prereq
- AC_CONFIG_MACRO_DIRS is not necessary to define
as ACLOCAL_AMFLAGS is already defined in Makefiles
- let libtool "enable_static" variable also decide the value
of the local conditional BUILD_STATIC
- override configure variables instead of removing
checks for libraries or additions to LDFLAGS
- only exclude "hidden" attribute for macOS instead of deleting
- preserve original list of sources to build for libelf
- use openwrt Makefile to add gnulib headers
- use openwrt Makefile to add LIBADD variables
- remove deletion of variables and rules for shared objects
- prefer recursively expanded variables over muliple renames
each time that a word is added to its value
- remove changes to subdirectories that are not built
and remove changes to target files of those subdirectories
- prefer basic text rename over variables in cases where
there would be no line number difference
- give LT_INIT forced default values that match upstream
- move gl_EARLY and gl_INIT down relative to compiler checks
- reorganize some line changes to save some lines
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
The gnulib-tool script is written to have a fatal error
whenever the minimum required version of autoconf
for the project that gnulib is being imported into
as defined in configure.ac was less than
the minimum required version required by gnulib.
However, none of this matters if the version of autoconf
that we use is newer than both requirements.
Instead, use functions from the bootstrap script
to check for the version of autoconf being used
and print a warning whenever this case occurs.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
Overriding variables used in both the macros and the headers
like setting REPLACE_FCNTL to 0 while invoking Make causes the
function aliases like rpl_fcntl() to not be defined,
however the object may still be built with the fnctl() function.
Usually this is enough for building while avoiding
the need to link the resulting libgnu library
to every single other build target for a project
in order to include the gnulib copy of the function,
because in these cases we don't care which version
of the function is used.
However for functions like fcntl() this doesn't work
as it is designed to use either the alias or standard declaration
from gnulib headers in order to be
a wrapper for the native host copy of fcntl()
by containing recursive calls to fcntl() within itself
after undefining the gnulib function declaration.
Overriding the variables used by the header when invoking Make
causes the header's declarations to be blocked,
and this results in the gnulib version of fcntl()
to call itself recursively and indefinitely
leading to segmentation faults.
Fix this by using macros defined with those variables
in order to exclude the function during preprocessing.
While at it, do the same for reallocarray()
so that the configure option --avoid=reallocarray
and Make variable REPLACE_REALLOCARRAY set to 0
would have a similar effect.
In the future this patch can be expanded to include
more functions and some version of this may be upstreamable.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
Instead of backporting select 1.35 fixes to make tar work for us, lets
update to 1.35 now that we have identified the upstream fix for macOS.
Signed-off-by: Robert Marko <robimarko@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15743
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Backport an upstream patch series that adds backend elfutils support for
DWARF relocations in MIPS debug info. This support is needed by pahole to
generate BTF for modules in BTF-enabled kernel builds.
The problem first manifests as pahole warnings during build:
BTF [M] lib/libcrc32c.ko
die__process_unit: DW_TAG_compile_unit (0x11) @ <0x932d> not handled!
die__process_unit: tag not supported 0x11 (compile_unit)!
die__process: got compile_unit unexpected tag after DW_TAG_compile_unit!
die__process_unit: DW_TAG_compile_unit (0x11) @ <0x99a3> not handled!
die__process_unit: tag not supported 0x11 (compile_unit)!
die__process: got compile_unit unexpected tag after DW_TAG_compile_unit!
During system boot the problem then causes module loading failures, which
may result in many other runtime issues:
[ 13.169785] kmodloader: loading kernel modules from /etc/modules.d/*
[ ... ]
[ 17.422840] mac80211_hwsim: initializing netlink
[ 17.526518] PPP generic driver version 2.4.2
[ 17.550346] NET: Registered PF_PPPOX protocol family
[ 17.795353] kmodloader: 26 modules could not be probed
[ 17.796084] kmodloader: dependency not loaded nf_conntrack
[ 17.796737] kmodloader: - act_connmark - 1
[ 17.797402] kmodloader: dependency not loaded nf_conntrack
[ 17.798056] kmodloader: - act_ctinfo - 1
[ ... ]
Link: https://lore.kernel.org/bpf/ZlkoM6%2FPSxVcGM6X@kodidev-ubuntu/
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
Link: https://github.com/openwrt/openwrt/pull/15697
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Tar 1.34 ship an old version of paxlib with rtapelib.c that produce some
compilation warning. This library got updated in 1.35 but we still can't
use the new Tar version.
GCC 14 then made these compilarion warning errors.
Manually backport the fixes to rtapelib.c and patch the version shipped
in 1.34 to fix these compilation warning.
Fixes: #15692
Fixes: 2951e0a80e ("tools: tar: backport patches fixing broken --delete")
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
In experimenting with --delete for APK handling, it was discovered that
--delete is broken and corrupts the TAR in some case.
This is fixed in version 1.35 but 1.35 introduce some problem with MacOS
making it difficult to bump. Backport the 2 required patches to fix this
problem so --delete is usable again.
Link: https://github.com/openwrt/openwrt/pull/15543
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Since the Yafut tool is now used for both updating the kernel on
MikroTik devices with NAND flash and preparing firmware images for
MikroTik devices with NOR flash, remove the kernel2minor utility from
the tree as it is no longer used for anything.
Signed-off-by: Michał Kępień <openwrt@kempniu.pl>
Link: https://github.com/openwrt/openwrt/pull/13453
Signed-off-by: Robert Marko <robimarko@gmail.com>
The Yafut tool now has limited capabilities for working on filesystem
images stored in regular files. This enables preparing Yaffs2 images
for devices with NOR flash using upstream Yaffs2 filesystem code instead
of the custom kernel2minor tool.
Since minimizing the size of the resulting filesystem image size is
important and upstream Yaffs2 code requires two allocator reserve blocks
to be available when writing a file to the filesystem, a trick is
employed while preparing an OpenWRT image: the blank filesystem image
that Yafut operates on initially contains two extra erase blocks that
are chopped off after the kernel file is written. This is safe to do
because Yaffs2 has a true log structure and therefore only ever writes
sequentially (and the size of the kernel file is known beforehand).
While the two extra erase blocks are necessary for writes, Yaffs2 code
seems to be perfectly capable of reading back files from a "truncated"
filesystem that does not contain these extra erase blocks.
In terms of image size, this new approach is only marginally worse than
the current kernel2minor-based one: specifically, upstream Yaffs2 code
needs to write three object headers (each of which takes up an entire
data chunk) when the kernel file is written to the filesystem:
- an object header for the kernel file when it is created,
- an object header for the root directory when the kernel file is
created,
- an updated object header for the kernel file when the latter is
fully written (so that its new size can be recorded).
kernel2minor only writes two of these headers, which is the absolute
minimum required for reading the file back. This means that the
Yafut-based approach causes firmware images to be at most one erase
block (64 kB) larger than those created using kernel2minor, but only in
the very unfortunate scenario where the size of the kernel file is
really close to a multiple of the erase block size.
The rest of the calculations performed when the empty filesystem image
is first prepared stems from the Yaffs2 layout used by MikroTik NOR
devices: each 65,536-byte erase block contains 63 chunks, each of which
consists of 1024 bytes of data followed by 16-byte Yaffs tags without
ECC data; each such group of 63 chunks is then followed by 16 bytes of
padding, which translates to "-C 1040 -B 64k -E" in the Yafut
invocation. Yaffs2 checkpoints and summaries are disabled (using
Yafut's -P and -S switches, respectively) as they are merely performance
optimizations that require extra storage space. The -L and -M switches
are used to force little-endian or big-endian byte order (respectively)
in the resulting filesystem image, no matter what byte order the build
host uses. The tr invocation is used to ensure that the filesystem
image is initialized with 0xFF bytes (which are an indicator of unused
space for Yaffs2 code).
Signed-off-by: Michał Kępień <openwrt@kempniu.pl>
Link: https://github.com/openwrt/openwrt/pull/13453
Signed-off-by: Robert Marko <robimarko@gmail.com>
The Yafut tool has so far been used to update the kernel on devices with
NAND flash via MTD character devices. Recent upstream updates extended
the tool with limited support for working with filesystem images stored
in regular files. This enables Yafut to be used for preparing a Yaffs
filesystem image for a device with NOR flash on a build host and
subsequently flashing it to the target device without using Yafut
itself.
Add Yafut to tools/ so that it can be compiled and run on the host
building OpenWRT.
Signed-off-by: Michał Kępień <openwrt@kempniu.pl>
Link: https://github.com/openwrt/openwrt/pull/13453
Signed-off-by: Robert Marko <robimarko@gmail.com>
Trying to compile elfutils on Fedora 40 with GCC 14.1.1 will fail with:
/home/robimarko/Building/AX3600/qualcommax/staging_dir/host/bin/g++ -std=c++11 -D_GNU_SOURCE -DHAVE_CONFIG_H -DLOCALEDIR='"/home/robimarko/Building/AX3600/qualcommax/staging_dir/host/share/locale"' -DDEBUGPRED=0 -DSRCDIR=\"/home/robimarko/Building/AX3600/qualcommax/build_dir/host/elfutils-0.191/src\" -DOBJDIR=\"/home/robimarko/Building/AX3600/qualcommax/build_dir/host/elfutils-0.191/src\" -I. -I.. -I../libgnu -I../libgnu -I. -I. -I../lib -I.. -I./../libelf -I./../libebl -I./../libdw -I./../libdwelf -I./../libdwfl -I./../libasm -I../debuginfod -I/home/robimarko/Building/AX3600/qualcommax/staging_dir/host/include -std=c++11 -Wall -Wshadow -Wtrampolines -Wlogical-op -Wduplicated-cond -Wnull-dereference -Wimplicit-fallthrough=5 -Werror -Wunused -Wextra -Wstack-usage=262144 -D_FORTIFY_SOURCE=3 -c -o srcfiles.o srcfiles.cxx
In file included from /usr/include/c++/14/x86_64-redhat-linux/bits/os_defines.h:39,
from /usr/include/c++/14/x86_64-redhat-linux/bits/c++config.h:2521,
from /usr/include/c++/14/cstdlib:41,
from ../libgnu/gettext.h:56,
from ../libgnu/eu-config.h:62,
from ../config.h:2378,
from srcfiles.cxx:31:
/usr/include/features.h:414:4: error: #warning _FORTIFY_SOURCE requires compiling with optimization (-O) [-Werror=cpp]
414 | # warning _FORTIFY_SOURCE requires compiling with optimization (-O)
| ^~~~~~~
cc1plus: all warnings being treated as errors
So, lets do as the error says and pass -O2 in HOST_CXXFLAGS like we already
do by default in HOST_CFLAGS.
Link: https://github.com/openwrt/openwrt/pull/15368
Signed-off-by: Robert Marko <robimarko@gmail.com>
The compiled library resulting from importing gnulib has been
linked to libelf in order to easily cover other link dependencies.
However, this is not appropriate for linking libelf to other programs
as it bloats the resulting libelf library, and may result in
multiple defintions of symbols based on whether or not
certain modules from gnulib are included while elfutils
already has it's own definition of a function.
This is not a problem while building elfutils, because gnulib has
it's own way of creating function aliases and special declarations
that allow the linker to ignore the original function definitions,
however, when libelf is used to link to something else,
this results in an error at link time.
The gnulib manual recommended linking the libraries directly,
but those who have written it may not have considered how this
can affect the ability to link that library in other builds,
they likely assume the build targets would not be a dependency.
Fix this by removing the linking between gnulib and libelf
and instead overriding Make variables in order to add linking
between gnulib and each of the binaries provided by elfutils,
using Make functions to avoid applying it to other subdirectories.
The function tdestroy() would still be missing on macOS,
but the existence of the gnulib tsearch object having been built
is an indicator of whether or not it is needed
because it is only built conditionally by gnulib,
so include linking that object only when it exists.
Block the unnecessary replacement of some functions by gnulib
so that future linking with libelf doesn't require
the associated gnulib "rpl" prefixed functions.
These replacements are very strict in order to correct
minor bugs that don't have a real impact in almost all cases
or new standards requirements that are not yet in effect or used.
Tested-by: Georgi Valkov <gvalkov@gmail.com> # macOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15368
Signed-off-by: Robert Marko <robimarko@gmail.com>
Install binaries that are not common with binutils
instead of none at all. This adds a negligible time to the build.
Building shared libraries is disabled, so the AM_LDFLAGS can be reset
without the rpath-link option which is unrecognized by clang.
Some of the binaries depend on functions that are defined
using a "strong alias" instead of a normal definition,
but this is disabled by our patches in order to work on macOS,
so use the identical function directly instead.
Add fnmatch from gnulib with GNU extensions
which is needed for usage of the FNM_EXTMATCH flag.
Handle a "Wunused-const-variable" error with the same
preprocessor conditional used to include the function
that the variable is used in.
Ref: f64bd4b6c ("tools/elfutils: only build required components")
Tested-by: Georgi Valkov <gvalkov@gmail.com> # macOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15368
Signed-off-by: Robert Marko <robimarko@gmail.com>
The gnulib fts header is meant to not be overwritten
in any way by the host system's copy of fts.h
and was therefore given a unique name instead.
This is fine if the built libgnu library is directly linked
with the target library, but if we want to keep them isolated
we end up having the definitions being mangled anyway
when the next object to link against included the fts.h header.
On some macOS platforms, the use of __DARWIN_INODE64
is messing with the link name for fts functions, resulting in:
Undefined symbols for architecture x86_64:
"_rpl_fts_close$INODE64", referenced from:
...
Create a local fts header for gnulib
that completely blocks the macOS host fts header.
An alternative and more upstream friendly fix would be
to rename fts_.h to fts.h and add the macOS-only
include guard to that file within it's own include guard,
but that would be a massive patch, so do this for now.
Tested-by: Georgi Valkov <gvalkov@gmail.com> # macOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15368
Signed-off-by: Robert Marko <robimarko@gmail.com>
Refresh all tools patches now that tools/refresh correctly works.
CI now checks for them and actively complain if tools have unrefreshed
patches.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
[ reword commit message ]
Link: https://github.com/openwrt/openwrt/pull/15524
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
meson is used to build it since 291b137. No need to patch Makefiles.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Link: https://github.com/openwrt/openwrt/pull/15524
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Now that Host/Prepare/Default is always defined, we can use that instead
of using raw commands to move files from the src directory to
HOST_BUILD_DIR.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Install files from HOST_BUILD_DIR instead of src. These files are now
correctly copied to HOST_BUILD_DIR and can be referenced from there.
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
New:
- mold is now up to 10% faster when linking very large, debug
info-enabled executables such as Blender (~1.8 GiB) or Clang (~3.8
GiB), thanks to several improvements we've made to the string merging
algorithm. (53ebcd8, d714301, 40f6b17, c9faf3d)
- -z start-stop-visibility=hidden is now supported so that
linker-synthesized __start_<section-name> and __stop_<section-name>
symbols can be completely hidden from other ELF modules. Previously,
only -z start-stop-visibility=protected was supported. (99a5b15)
- -Bsymbolic-non-weak and -Bsymbolic-non-weak-functions options are now
supported for compatibility with LLVM lld. Just like lld, these options
control which symbols are exported as dynamic symbols.
-Bsymbolic-non-weak makes the linker to export only weak symbols,
whereas -Bsymbolic-non-weak-functions makes it to export only weak
function symbols. (7d17aa8)
Bug fixes and compatibility improvements:
- Previously, if a linker script contains a newline character in the
beginning four bytes of a file, it was not recognized as a linker
script by mold. Now, mold allows newlines at the beginning of a file.
(ea054cc)
- Under rare circumstances, the INPUT linker script command may have
found a different file than GNU ld would. Now, mold's behavior aligns
with GNU ld's. (163975d)
- Previously, the --repro option produced corrupted tar files. Now the
bug has been fixed. (32c4a09)
- mold generally guarantees that its output is reproducible, meaning that
if you run the linker with the exact same command line options and
input files, the output is guaranteed to be bit-for-bit identical to
the previous outputs. However, under rare circumstances, it might
produce different output due to a bug. It's reported that this
nondeterminism caused random crashes for some programs (#1247). This
bug has been fixed. (6463a7c)
- mold no longer sets the address of the .text section as the entry point
address if --entry option is not given, just like LLVM lld. (020b1a7)
- [RISC-V] __global_pointer$ symbol is now exported from executables as
required by the processor-specific ABI. (3df7c8e)
- [ARM32] --long-plt option is now recognized as known option by mold.
mold ignores the option, though, because the PLTs generated by our
linker is always long. (d432e98)
Release Notes:
https://github.com/rui314/mold/releases/tag/v2.31.0
Signed-off-by: Sean Khan <datapronix@protonmail.com>
Link: https://github.com/openwrt/openwrt/pull/15403
Signed-off-by: Robert Marko <robimarko@gmail.com>
Some package might require to fix their pkg-config file to point to host
or hostpkg file. This is the case for glib2 library that provides with
pkg-config variables, tools to generates files from xml. Those tools
should use the host binary instead of the targets one to correctly build
packages that makes use of such tools.
Link: https://github.com/openwrt/openwrt/pull/15134
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
