Debian Patches

Status for firebird4.0/4.0.5.3140.ds6-12

Patch Description Author Forwarded Bugs Origin Last update
upstream/std-c++-17.patch pass -std=c++17 in CXXFLAGS Fixes build with libre2-dev 20240501 which requires at least c++-14
Cherry-picked from the upstream commit which is for Firebird 5
yes debian
upstream/freebsd_ppc64el.patch commit 96931200340c8e0abcd3529b8e443d5ab62adfa5

Add support for FreeBSD/powerpc64le
Piotr Kubaj <pkubaj@FreeBSD.org> no 2022-07-28
upstream/mips64el.patch commit 22846322070ddd9a6f4801a3605f690d9f70049c

Add linux/mips64el support

Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Jiaxun Yang <jiaxun.yang@flygoat.com> no 2022-08-03
upstream/freebsd_ppc64.patch commit 4cd5c17b29210d3da3eebc72a997d0d3b4bf163e

Add support for FreeBSD/powerpc64
Piotr Kubaj <pkubaj@FreeBSD.org> no 2022-12-13
upstream/freebsd_ppc.patch commit a6ab775f97d7ae5b2a25b6dac93b79e5d24de633

Add support for FreeBSD/powerpc
Piotr Kubaj <pkubaj@FreeBSD.org> no 2022-12-21
upstream/loongarch.patch commit fcc471e420525dbd4a1dda59ed51487d5ead5921

Add LoongArch support
loongson <loongson@loongson-pc> no 2023-03-31
upstream/thread-start-semaphore.patch commit 49e38630499cdab70bc15c255c89fc38048a57a9

Fix race in shutdown thread start (#8380)

Previously we assign value to shutdownSemaphore after shutdownThread is started, where it is already needed. So we can have situation where shutdownThread instantly leaving due to shutdownSemaphore == nullptr, and we are left with a server that can only be stopped with kill -9.

Co-authored-by: Artyom Ivanov <artyom.ivanov@red-soft.ru>
TreeHunter <60896014+TreeHunter9@users.noreply.github.com> no 2025-01-10
deb/system-libabseil.patch use system-wide libabseil This is only needed on big-endian platforms. It remains to be discovered
whether the resulting server/client is network-protocol compatible with
little-endian client/server. This is why the patch is so crude (no
--with-system-abseil configure switch) and not forwarded upstream.
Damyan Ivanov <dmn@debian.org> not-needed
out/honour-buildflags.patch improved support for build flags The first change makes linking makeHeader use the same CPP/CXX/LD flags as the
rest of the sources.
The second change stops btyacc/Makefile from ignoring CFLAGS from the
environment.
The third change stops overriding CXXFLAGS
The result is using hardening flags from the environment when compiling and
linking.
Damyan Ivanov <dmn@debian.org> no
out/no-copy-from-icu.patch Link isql with ICU instead of embedding part of it in the source Damyan Ivanov <dmn@debian.org> no
out/cloop-honour-build-flags.patch make cloop build honor compiler/linker flags from the environment Damyan Ivanov <dmn@debian.org> no
out/spelling.patch spelling error (perfrom -> perform) Damyan Ivanov <dmn@debian.org> yes
out/fbserver-startup.patch fix fbguard to exit to the OS only after fbserver is ready Exiting right after forking causes a race condition making it possible that
the service says it is startew, but the fbserver process isn't ready yet to
serve requests, because it hasn't called listen() yet.
.
The patch makes fbserver send SIGUSR1 to the fbguard parent process, which
waits for SIGUSR1 before it returns the control to the OS.
Damyan Ivanov <dmn@debian.org> yes debian
honor-env.patch allow all directory locations to be overridden by $FIREBIRD environment variable This is needed because the compile-time settings for DIR_CONF, DIR_MSG etc are
hardwired into libfbclient and on debian there can be only one libfbclient,
but we need to use different directories for different versions.
.
This patch just lets all directories to be sought under $FIREBIRD/, which is
set by the fb_wrapper script.
.
Doesn't seem possible to integrate upstream, because it requires that
$FIREBIRD is set to a suitable firebird installation root.
.
Still, this was not communicated with upstream so there may be a better way.
Damyan Ivanov <dmn@debian.org> not-needed
out/cross-configure.patch improve cross buildability
* Rather than checking $build (the architecture we are building on), we should
be checking $host. Unfortunately, $host tends to lack the vendor part, so we
need a tricky sed expression for inserting it. For native builds, $host and
$build are equal.
* The check for whether sem_init works only aborts the build in case of
failure. Since the check cannot be performed during cross building, the only
sane way is to just assume sem_init to work.
* Replace a pile of AC_RUN_IFELSE with AC_CHECK_SIZEOF and AC_CHECK_ALIGNOF.
The latter macros have a slower fallback path for cross compilation that use
compiler bisection to determine the values.
Helmut Grohne <helmut@subdivi.de> no
no-binary-gbaks.patch # Author: Damyan Ivanov <dmn@debian.org>
# Description: Remove usage of binary .gbak files
# These can't be in Debian-shipped sources
#
# help.fbk is removed from Debian sources so we remove mentions of it,
# together with qli, which needs it
# Forwarded: no
no
packaged-boost.patch use system-wide boost headers Damyan Ivanov <dmn@debian.org> yes
deb/no-suse.init.patch remove reference to firebird.init.d.suse.in That file is not present in the repackaged source, since it cannot
be distributed by Debian -- has copyright statement without license.
Damyan Ivanov <dmn@debian.org> not-needed
deb/gen-ids.patch add configure option to re-generate 'codes' unconditionally We want to create all files from their source Damyan Ivanov <dmn@debian.org> yes
out/hppa-mod_loader.patch fix mod_loader on HPPA The problem is dlsym returns a function pointer on hppa when passed the name
of a function symbol. It points at a function descriptor for the function
when the plabel bit is set in the pointer. The descriptor has an entry which
points at the function after it has been bound. Otherwise, it points at
trampoline code to fix up the descriptor.
.
In any case, dladdr doesn't do the expected thing when passed a function
pointer. It needs the actual address of the function.
.
The attached patch fixes the build on hppa-linux. Another option would be to
adjust the configure check.
John David Anglin <dave.anglin@bell.net> no debian
out/hurd-maxpathlen.patch move fallback definition of MAXPATHLEN outside the WinNT block helps the hurd build move a bit further
.
Additionally, include common.h in gpre.h to get the MAXPATHLEN define
.
The effor to build firebird on hurd is tracked at
<https://bugs.debian.org/626931>
Damyan Ivanov no
out/baseline-i386.patch drop -march=i586 -mtune=i686 -march=i686 is the default since a decade Matthias Klose <doko@ubuntu.com> no debian

All known versions for source package 'firebird4.0'

Links