| Message ID | 20240303212411.1139387-1-fontaine.fabrice@gmail.com |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [1/1] package/fluent-bit: fix build with BR2_TIME_BITS_64 | expand |
On Sun, 3 Mar 2024 22:24:11 +0100 Fabrice Fontaine <fontaine.fabrice@gmail.com> wrote: > Indeed, this LFS workaround was present since the addition of the > package in commit 6a0f7c39bcb48fc13aa2ce3fc4996baf1be66483 and is > probably not needed anymore This package was added just a year ago, so I believe "probably not needed anymore" is most likely not sufficiently assertive. Can we get some more clarity on this, by reproducing the original issue that lead to this FLUENT_BIT_CFLAGS += -U_FILE_OFFSET_BITS magic, and then figure out how it got resolved, if it did? Thomas
Le dim. 3 mars 2024 à 23:17, Thomas Petazzoni <thomas.petazzoni@bootlin.com> a écrit : > > On Sun, 3 Mar 2024 22:24:11 +0100 > Fabrice Fontaine <fontaine.fabrice@gmail.com> wrote: > > > Indeed, this LFS workaround was present since the addition of the > > package in commit 6a0f7c39bcb48fc13aa2ce3fc4996baf1be66483 and is > > probably not needed anymore > > This package was added just a year ago, so I believe "probably not > needed anymore" is most likely not sufficiently assertive. Can we get > some more clarity on this, by reproducing the original issue that lead > to this FLUENT_BIT_CFLAGS += -U_FILE_OFFSET_BITS magic, and then figure > out how it got resolved, if it did? Comment above this "magic" is pointing to a policycoreutils defect opened 14 years ago in 2010. I assume that it could be linked to glibc < 2.23 as for the other packages. However, I'll let Thomas answers on why this "magic" was needed. > > Thomas > -- > Thomas Petazzoni, co-owner and CEO, Bootlin > Embedded Linux and Kernel engineering and training > https://bootlin.com Best Regards, Fabrice
Hi all, Op zo 3 mrt 2024 om 23:34 schreef Fabrice Fontaine <fontaine.fabrice@gmail.com>: > > Le dim. 3 mars 2024 à 23:17, Thomas Petazzoni > <thomas.petazzoni@bootlin.com> a écrit : > > > > On Sun, 3 Mar 2024 22:24:11 +0100 > > Fabrice Fontaine <fontaine.fabrice@gmail.com> wrote: > > > > > Indeed, this LFS workaround was present since the addition of the > > > package in commit 6a0f7c39bcb48fc13aa2ce3fc4996baf1be66483 and is > > > probably not needed anymore > > > > This package was added just a year ago, so I believe "probably not > > needed anymore" is most likely not sufficiently assertive. Can we get > > some more clarity on this, by reproducing the original issue that lead > > to this FLUENT_BIT_CFLAGS += -U_FILE_OFFSET_BITS magic, and then figure > > out how it got resolved, if it did? > > Comment above this "magic" is pointing to a policycoreutils defect > opened 14 years ago in 2010. > I assume that it could be linked to glibc < 2.23 as for the other packages. > However, I'll let Thomas answers on why this "magic" was needed. I added this "magic" in my way of getting fluent-bit compiled for all the buildroot targets. The typical './utils/test-pkg -p fluent-bit -a' script exposed that problem. But I forgot the target/architecture that needed this hack. So if all architectures do compile now, then I guess that we can just drop it. (To be checked.) Thomas > > > > Thomas > > -- > > Thomas Petazzoni, co-owner and CEO, Bootlin > > Embedded Linux and Kernel engineering and training > > https://bootlin.com > > Best Regards, > > Fabrice >
Le lun. 4 mars 2024 à 11:55, Thomas Devoogdt <thomas@devoogdt.com> a écrit : > > Hi all, > > Op zo 3 mrt 2024 om 23:34 schreef Fabrice Fontaine <fontaine.fabrice@gmail.com>: > > > > Le dim. 3 mars 2024 à 23:17, Thomas Petazzoni > > <thomas.petazzoni@bootlin.com> a écrit : > > > > > > On Sun, 3 Mar 2024 22:24:11 +0100 > > > Fabrice Fontaine <fontaine.fabrice@gmail.com> wrote: > > > > > > > Indeed, this LFS workaround was present since the addition of the > > > > package in commit 6a0f7c39bcb48fc13aa2ce3fc4996baf1be66483 and is > > > > probably not needed anymore > > > > > > This package was added just a year ago, so I believe "probably not > > > needed anymore" is most likely not sufficiently assertive. Can we get > > > some more clarity on this, by reproducing the original issue that lead > > > to this FLUENT_BIT_CFLAGS += -U_FILE_OFFSET_BITS magic, and then figure > > > out how it got resolved, if it did? > > > > Comment above this "magic" is pointing to a policycoreutils defect > > opened 14 years ago in 2010. > > I assume that it could be linked to glibc < 2.23 as for the other packages. > > However, I'll let Thomas answers on why this "magic" was needed. > > I added this "magic" in my way of getting fluent-bit compiled for all > the buildroot targets. > The typical './utils/test-pkg -p fluent-bit -a' script exposed that > problem. But I forgot the > target/architecture that needed this hack. So if all architectures do > compile now, then I > guess that we can just drop it. (To be checked.) After running ./utils/test-pkg -p fluent-bit -a, I got only 3 build failures raised by sourcery-arm toolchain: sourcery-arm-armv4t [40/45]: FAILED sourcery-arm [41/45]: FAILED sourcery-arm-thumb2 [42/45]: FAILED So, this is the same issue than libselinux and all the other packages. I'll send a v2 updating the commit log and another patch to drop this old toolchain. > > Thomas > > > > > > > Thomas > > > -- > > > Thomas Petazzoni, co-owner and CEO, Bootlin > > > Embedded Linux and Kernel engineering and training > > > https://bootlin.com > > > > Best Regards, > > > > Fabrice > > Best Regards, Fabrice
diff --git a/package/fluent-bit/fluent-bit.mk b/package/fluent-bit/fluent-bit.mk index e51322aad6..00a8904c55 100644 --- a/package/fluent-bit/fluent-bit.mk +++ b/package/fluent-bit/fluent-bit.mk @@ -12,8 +12,6 @@ FLUENT_BIT_CPE_ID_VENDOR = treasuredata FLUENT_BIT_CPE_ID_PRODUCT = fluent_bit FLUENT_BIT_DEPENDENCIES = host-bison host-flex libyaml openssl -FLUENT_BIT_CFLAGS = $(TARGET_CFLAGS) - FLUENT_BIT_CONF_OPTS += \ -DFLB_DEBUG=No \ -DFLB_RELEASE=Yes \ @@ -55,11 +53,6 @@ FLUENT_BIT_CONF_OPTS += \ FLUENT_BIT_CONF_OPTS += \ -DCMAKE_INSTALL_SYSCONFDIR="/etc/" -# Undefining _FILE_OFFSET_BITS here because of a "bug" with glibc fts.h -# large file support. -# https://bugzilla.redhat.com/show_bug.cgi?id=574992 -FLUENT_BIT_CFLAGS += -U_FILE_OFFSET_BITS - ifeq ($(BR2_PACKAGE_LIBEXECINFO),y) FLUENT_BIT_DEPENDENCIES += libexecinfo FLUENT_BIT_LDFLAGS += -lexecinfo @@ -76,8 +69,7 @@ FLUENT_BIT_LDFLAGS += -latomic endif FLUENT_BIT_CONF_OPTS += \ - -DCMAKE_EXE_LINKER_FLAGS="$(FLUENT_BIT_LDFLAGS)" \ - -DCMAKE_C_FLAGS="$(FLUENT_BIT_CFLAGS)" + -DCMAKE_EXE_LINKER_FLAGS="$(FLUENT_BIT_LDFLAGS)" define FLUENT_BIT_INSTALL_INIT_SYSV $(INSTALL) -D -m 0755 package/fluent-bit/S99fluent-bit \
Do not remove _FILE_OFFSET_BITS=64 from CFLAGS to avoid the following build failure with BR2_TIME_BITS_64 raised since commit 3c427c64726560ea1743282a3fdb78f5b28692eb: In file included from /home/thomas/autobuild/instance-1/output-1/host/microblaze-buildroot-linux-gnu/sysroot/usr/include/features.h:394, from /home/thomas/autobuild/instance-1/output-1/host/microblaze-buildroot-linux-gnu/sysroot/usr/include/bits/libc-header-start.h:33, from /home/thomas/autobuild/instance-1/output-1/host/microblaze-buildroot-linux-gnu/sysroot/usr/include/stdio.h:27, from /home/thomas/autobuild/instance-1/output-1/build/fluent-bit-2.1.7/tools/xxd-c/xxd-c.c:27: /home/thomas/autobuild/instance-1/output-1/host/microblaze-buildroot-linux-gnu/sysroot/usr/include/features-time64.h:26:5: error: #error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64" 26 | # error "_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64" | ^~~~~ Indeed, this LFS workaround was present since the addition of the package in commit 6a0f7c39bcb48fc13aa2ce3fc4996baf1be66483 and is probably not needed anymore Fixes: - http://autobuild.buildroot.org/results/ff5c60cd038550453ce138fe2a9383af2f5d6f2f Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com> --- package/fluent-bit/fluent-bit.mk | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-)