Message ID | 1468182896-8355-1-git-send-email-yann.morin.1998@free.fr |
---|---|
State | Rejected |
Headers | show |
>>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes: > ... since it uses O_TMPFILE as an open(2) flag, and it was only > introduced in 3.11. > Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr> > Cc: Gabe Evans <gabe@hashrabbit.co> > Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com> Committed, thanks.
Peter, All, On 2016-07-11 21:57 +0200, Peter Korsgaard spake thusly: > >>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes: > > ... since it uses O_TMPFILE as an open(2) flag, and it was only > > introduced in 3.11. > > > Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr> > > Cc: Gabe Evans <gabe@hashrabbit.co> > > Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com> > Committed, thanks. Had you seen the other patch a bit later, with comments from Thomas, before applying this one? https://patchwork.ozlabs.org/patch/646796/ Regards, Yann E. MORIN.
>>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes: > Peter, All, > On 2016-07-11 21:57 +0200, Peter Korsgaard spake thusly: >> >>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes: >> > ... since it uses O_TMPFILE as an open(2) flag, and it was only >> > introduced in 3.11. >> >> > Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr> >> > Cc: Gabe Evans <gabe@hashrabbit.co> >> > Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com> >> Committed, thanks. > Had you seen the other patch a bit later, with comments from Thomas, > before applying this one? > https://patchwork.ozlabs.org/patch/646796/ I hadn't, no. I just did a test build with 3.10 headers and glibc 2.22, and it built without problems - So I've dropped this patch.
Hello, On Tue, 12 Jul 2016 10:47:31 +0200, Peter Korsgaard wrote: > I hadn't, no. I just did a test build with 3.10 headers and glibc 2.22, > and it built without problems - So I've dropped this patch. So the comment from Yann in https://patchwork.ozlabs.org/patch/646796/ that both were needed is incorrect? Thomas
Thomas, Peter, All, On 2016-07-16 14:26 +0200, Thomas Petazzoni spake thusly: > On Tue, 12 Jul 2016 10:47:31 +0200, Peter Korsgaard wrote: > > > I hadn't, no. I just did a test build with 3.10 headers and glibc 2.22, > > and it built without problems - So I've dropped this patch. Yes, it works because glibc 2.22 (which is >= 2.19) defines it even if not available in the kernel headers). > So the comment from Yann in https://patchwork.ozlabs.org/patch/646796/ > that both were needed is incorrect? So, I built a toolchain with Linux headers 3.10 and glibc 2.18. And systemd-importd does not build: src/import/export-raw.c: In function 'reflink_snapshot': src/import/export-raw.c:271:26: error: 'O_TMPFILE' undeclared (first use in this function) new_fd = open(d, O_TMPFILE|O_CLOEXEC|O_NOCTTY|O_RDWR, 0600); ^ src/import/export-raw.c:271:26: note: each undeclared identifier is reported only once for each function it appears in Makefile:16515: recipe for target 'src/import/systemd_export-export-raw.o' failed O_TMPFILE was added in Linux 3.11; glibc added the corresponding definition in glib-2.19. glibc \ linux | <= 3.10 | >= 3.11 | ----------------+-----------+-----------+ <= 2.18 | No | No* | >= 2.19 | Yes | Yes | *: headers 3.12 So, not sure where to go from there... Note: I have not tested uClibc or musl, because,well, systemd depends on glibc... Regards, Yann E. MORIN.
>>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes: Hi, > So, I built a toolchain with Linux headers 3.10 and glibc 2.18. And > systemd-importd does not build: > src/import/export-raw.c: In function 'reflink_snapshot': > src/import/export-raw.c:271:26: error: 'O_TMPFILE' undeclared (first use in this function) > new_fd = open(d, O_TMPFILE|O_CLOEXEC|O_NOCTTY|O_RDWR, 0600); > ^ > src/import/export-raw.c:271:26: note: each undeclared identifier is reported only once for each function it appears in > Makefile:16515: recipe for target 'src/import/systemd_export-export-raw.o' failed > O_TMPFILE was added in Linux 3.11; glibc added the corresponding > definition in glib-2.19. > glibc \ linux | <= 3.10 | >= 3.11 | > ----------------+-----------+-----------+ > <= 2.18 | No | No* | >> = 2.19 | Yes | Yes | > *: headers 3.12 > So, not sure where to go from there... > Note: I have not tested uClibc or musl, because,well, systemd depends on > glibc... I would just go for >= glibc 2.19. People are unlikely to use old glibc with new kernel headers (and it cannot be built with buildroot).
Peter, All, On 2016-07-17 09:02 +0200, Peter Korsgaard spake thusly: > >>>>> "Yann" == Yann E MORIN <yann.morin.1998@free.fr> writes: > > So, I built a toolchain with Linux headers 3.10 and glibc 2.18. And > > systemd-importd does not build: > > > src/import/export-raw.c: In function 'reflink_snapshot': > > src/import/export-raw.c:271:26: error: 'O_TMPFILE' undeclared (first use in this function) > > new_fd = open(d, O_TMPFILE|O_CLOEXEC|O_NOCTTY|O_RDWR, 0600); > > ^ > > src/import/export-raw.c:271:26: note: each undeclared identifier is reported only once for each function it appears in > > Makefile:16515: recipe for target 'src/import/systemd_export-export-raw.o' failed > > > O_TMPFILE was added in Linux 3.11; glibc added the corresponding > > definition in glib-2.19. > > > glibc \ linux | <= 3.10 | >= 3.11 | > > ----------------+-----------+-----------+ > > <= 2.18 | No | No* | > >> = 2.19 | Yes | Yes | > > > *: headers 3.12 > > > So, not sure where to go from there... > > > Note: I have not tested uClibc or musl, because,well, systemd depends on > > glibc... > > I would just go for >= glibc 2.19. People are unlikely to use old glibc > with new kernel headers (and it cannot be built with buildroot). So, should we add GLIBC_AT_LEAST_X_Y symbols? Regards, Yann E. MORIN.
diff --git a/package/systemd/Config.in b/package/systemd/Config.in index ddaf3e2..4a2aa33 100644 --- a/package/systemd/Config.in +++ b/package/systemd/Config.in @@ -144,6 +144,7 @@ config BR2_PACKAGE_SYSTEMD_HWDB config BR2_PACKAGE_SYSTEMD_IMPORTD bool "enable import daemon" depends on BR2_PACKAGE_LIBGPG_ERROR_ARCH_SUPPORTS # libgcrypt + depends on BR2_TOOLCHAIN_HEADERS_AT_LEAST_3_11 # O_TMPFILE select BR2_PACKAGE_LIBCURL select BR2_PACKAGE_LIBGCRYPT select BR2_PACKAGE_BZIP2 @@ -157,6 +158,10 @@ config BR2_PACKAGE_SYSTEMD_IMPORTD http://www.freedesktop.org/software/systemd/man/machinectl.html#Image%20Transfer%20Commands +comment "import daemon needs a toolchain w/ headers >= 3.11" + depends on BR2_PACKAGE_LIBGPG_ERROR_ARCH_SUPPORTS + depends on !BR2_TOOLCHAIN_HEADERS_AT_LEAST_3_11 + config BR2_PACKAGE_SYSTEMD_KDBUS bool "enable kdbus support" help
... since it uses O_TMPFILE as an open(2) flag, and it was only introduced in 3.11. Signed-off-by: "Yann E. MORIN" <yann.morin.1998@free.fr> Cc: Gabe Evans <gabe@hashrabbit.co> Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com> --- package/systemd/Config.in | 5 +++++ 1 file changed, 5 insertions(+)