* Y2038: Sanity checker for non-64-bit time and file functions
@ 2022-12-08 7:11 Ola x Nilsson
2022-12-08 7:11 ` [PATCH RFC 1/3] glibc: Add ppoll fortify symbol for 64 bit time_t Ola x Nilsson
` (2 more replies)
0 siblings, 3 replies; 22+ messages in thread
From: Ola x Nilsson @ 2022-12-08 7:11 UTC (permalink / raw)
To: openembedded-core
Here is my sanity checker that check for glibc symbols that use 32-bit
time_t or file offsets.
I include one glibc patch that fixes a problem where the redirect for
ppoll did not work when fortify_source was enabled.
Also include some distro settings for the -F_TIME_BITS=64 flags that I
have used while testing this.
I tried to make a very clever INSANE_SKIP mechanism that allowed for
suppression of individual symbols in individual ELF files. That is
probably overkill but can be useful in this phase.
--
Ola Nilsson
^ permalink raw reply [flat|nested] 22+ messages in thread
* [PATCH RFC 1/3] glibc: Add ppoll fortify symbol for 64 bit time_t
2022-12-08 7:11 Y2038: Sanity checker for non-64-bit time and file functions Ola x Nilsson
@ 2022-12-08 7:11 ` Ola x Nilsson
2022-12-08 7:11 ` [PATCH RFC 2/3] Add QA check for 32 bit time and file offset functions Ola x Nilsson
2022-12-08 7:11 ` [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags Ola x Nilsson
2 siblings, 0 replies; 22+ messages in thread
From: Ola x Nilsson @ 2022-12-08 7:11 UTC (permalink / raw)
To: openembedded-core
Patch so ppoll is properly redirected to it's 64bit time variant even
when source fortification is active.
Signed-off-by: Ola x Nilsson <olani@axis.com>
---
...fortify-symbol-for-64-bit-time_t-BZ-.patch | 528 ++++++++++++++++++
meta/recipes-core/glibc/glibc_2.36.bb | 1 +
2 files changed, 529 insertions(+)
create mode 100644 meta/recipes-core/glibc/glibc/0001-Linux-Add-ppoll-fortify-symbol-for-64-bit-time_t-BZ-.patch
diff --git a/meta/recipes-core/glibc/glibc/0001-Linux-Add-ppoll-fortify-symbol-for-64-bit-time_t-BZ-.patch b/meta/recipes-core/glibc/glibc/0001-Linux-Add-ppoll-fortify-symbol-for-64-bit-time_t-BZ-.patch
new file mode 100644
index 0000000000..678dc2af2c
--- /dev/null
+++ b/meta/recipes-core/glibc/glibc/0001-Linux-Add-ppoll-fortify-symbol-for-64-bit-time_t-BZ-.patch
@@ -0,0 +1,528 @@
+From 69318192d16adc7d57adcd69fbbb1e3e107b2783 Mon Sep 17 00:00:00 2001
+From: Adhemerval Zanella <adhemerval.zanella@linaro.org>
+Date: Fri, 4 Nov 2022 16:02:52 -0300
+Subject: [PATCH] Linux: Add ppoll fortify symbol for 64 bit time_t (BZ# 29746)
+
+Similar to ppoll, the poll.h header needs to redirect the poll call
+to a proper fortified ppoll with 64 bit time_t support.
+
+The implementation is straightforward, just need to add a similar
+check as __poll_chk and call the 64 bit time_t ppoll version. The
+debug fortify tests are also extended to cover 64 bit time_t for
+affected ABIs.
+
+Unfortunately it requires an aditional symbol, which makes backport
+tricky. One possibility is to add a static inline version if compiler
+supports is and call abort instead of __chk_fail, so fortified version
+will call __poll64 in the end.
+
+Another possibility is to just remove the fortify support for
+_TIME_BITS=64.
+
+Checked on i686-linux-gnu.
+
+Upstream-Status: Backport
+
+Signed-off-by: Ola Nilsson <olani@axis.com>
+---
+ debug/Makefile | 41 ++++++++++++------
+ include/sys/poll.h | 5 ++-
+ io/bits/poll2.h | 25 +++++++++++
+ sysdeps/unix/sysv/linux/Versions | 5 +++
+ sysdeps/unix/sysv/linux/arm/be/libc.abilist | 1 +
+ sysdeps/unix/sysv/linux/arm/le/libc.abilist | 1 +
+ sysdeps/unix/sysv/linux/csky/libc.abilist | 1 +
+ sysdeps/unix/sysv/linux/hppa/libc.abilist | 1 +
+ sysdeps/unix/sysv/linux/i386/libc.abilist | 1 +
+ .../sysv/linux/m68k/coldfire/libc.abilist | 1 +
+ .../unix/sysv/linux/m68k/m680x0/libc.abilist | 1 +
+ .../sysv/linux/microblaze/be/libc.abilist | 1 +
+ .../sysv/linux/microblaze/le/libc.abilist | 1 +
+ .../sysv/linux/mips/mips32/fpu/libc.abilist | 1 +
+ .../sysv/linux/mips/mips32/nofpu/libc.abilist | 1 +
+ .../sysv/linux/mips/mips64/n32/libc.abilist | 1 +
+ sysdeps/unix/sysv/linux/nios2/libc.abilist | 1 +
+ .../linux/powerpc/powerpc32/fpu/libc.abilist | 1 +
+ .../powerpc/powerpc32/nofpu/libc.abilist | 1 +
+ sysdeps/unix/sysv/linux/ppoll.c | 3 +-
+ sysdeps/unix/sysv/linux/ppoll_chk.c | 42 +++++++++++++++++++
+ .../unix/sysv/linux/s390/s390-32/libc.abilist | 1 +
+ sysdeps/unix/sysv/linux/sh/be/libc.abilist | 1 +
+ sysdeps/unix/sysv/linux/sh/le/libc.abilist | 1 +
+ .../sysv/linux/sparc/sparc32/libc.abilist | 1 +
+ 25 files changed, 125 insertions(+), 15 deletions(-)
+ create mode 100644 sysdeps/unix/sysv/linux/ppoll_chk.c
+
+diff --git a/debug/Makefile b/debug/Makefile
+index 456b349c4d..075a135a2b 100644
+--- a/debug/Makefile
++++ b/debug/Makefile
+@@ -120,8 +120,10 @@ CPPFLAGS-tst-read-chk-cancel.c += -D_FORTIFY_SOURCE=2
+ # CFLAGS for the file.
+
+ tests-all-chk = tst-fortify
+-tests-c-chk =
+-tests-cc-chk =
++tests-c-def-chk =
++tests-cc-def-chk =
++tests-c-time64-chk =
++tests-cc-time64-chk =
+
+ CFLAGS-tst-fortify.c += -Wno-format -Wno-deprecated-declarations -Wno-error
+
+@@ -130,11 +132,11 @@ define cflags-default
+ endef
+
+ define cflags-lfs
+-CFLAGS-tst-fortify-$(1)-lfs-$(2).$(1) += -D_FILE_OFFSET_BITS=64
++CFLAGS-tst-fortify-$(1)-lfs-$(2)-$(3).$(1) += -D_FILE_OFFSET_BITS=64
+ endef
+
+ define cflags-nongnu
+-CFLAGS-tst-fortify-$(1)-nongnu-$(2).$(1) += -D_LARGEFILE64_SOURCE=1
++CFLAGS-tst-fortify-$(1)-nongnu-$(2)-$(3).$(1) += -D_LARGEFILE64_SOURCE=1
+ endef
+
+ src-chk-nongnu = \#undef _GNU_SOURCE
+@@ -145,12 +147,12 @@ src-chk-nongnu = \#undef _GNU_SOURCE
+ # And they also generate warnings from warning attributes, which
+ # cannot be disabled via pragmas, so require -Wno-error to be used.
+ define gen-chk-test
+-tests-$(1)-chk += tst-fortify-$(1)-$(2)-$(3)
+-CFLAGS-tst-fortify-$(1)-$(2)-$(3).$(1) += -D_FORTIFY_SOURCE=$(3) -Wno-format \
++tests-$(1)-$(4)-chk += tst-fortify-$(1)-$(2)-$(3)-$(4)
++CFLAGS-tst-fortify-$(1)-$(2)-$(3)-$(4).$(1) += -D_FORTIFY_SOURCE=$(3) -Wno-format \
+ -Wno-deprecated-declarations \
+ -Wno-error
+-$(eval $(call cflags-$(2),$(1),$(3)))
+-$(objpfx)tst-fortify-$(1)-$(2)-$(3).$(1): tst-fortify.c Makefile
++$(eval $(call cflags-$(2),$(1),$(3),$(4)))
++$(objpfx)tst-fortify-$(1)-$(2)-$(3)-$(4).$(1): tst-fortify.c Makefile
+ ( echo "/* Autogenerated from Makefile. */"; \
+ echo "$(src-chk-$(2))"; \
+ echo "#include \"tst-fortify.c\"" ) > $$@.tmp
+@@ -159,19 +161,29 @@ endef
+
+ chk-extensions = c cc
+ chk-types = default lfs nongnu
++# 64 bit time_t implies LFS.
++chk-types-time64 = default nongnu
+ chk-levels = 1 2 3
+
+ $(foreach e,$(chk-extensions), \
+ $(foreach t,$(chk-types), \
+ $(foreach l,$(chk-levels), \
+- $(eval $(call gen-chk-test,$(e),$(t),$(l))))))
++ $(eval $(call gen-chk-test,$(e),$(t),$(l),def)))))
+
+-tests-all-chk += $(tests-c-chk) $(tests-cc-chk)
++tests-all-chk += $(tests-c-def-chk) $(tests-cc-def-chk)
++
++$(foreach e,$(chk-extensions), \
++ $(foreach t,$(chk-types-time64), \
++ $(foreach l,$(chk-levels), \
++ $(eval $(call gen-chk-test,$(e),$(t),$(l),time64)))))
++
++tests-all-time64-chk += $(tests-c-time64-chk) $(tests-cc-time64-chk)
+
+ define link-cc
+ LDLIBS-$(1) = -lstdc++
+ endef
+-$(foreach t,$(tests-cc-chk), $(eval $(call link-cc,$(t))))
++$(foreach t,$(tests-cc-def-chk), $(eval $(call link-cc,$(t))))
++$(foreach t,$(tests-cc-time64-chk), $(eval $(call link-cc,$(t))))
+
+ # backtrace_symbols only works if we link with -rdynamic. backtrace
+ # requires unwind tables on most architectures.
+@@ -201,6 +213,10 @@ tests = backtrace-tst \
+ tst-realpath-chk \
+ $(tests-all-chk)
+
++tests-time64 += \
++ $(tests-all-time64-chk) \
++ # tests-time64
++
+ ifeq ($(have-ssp),yes)
+ tests += tst-ssp-1
+ endif
+@@ -210,7 +226,7 @@ tests += tst-read-chk-cancel
+ endif
+
+ ifeq (,$(CXX))
+-tests-unsupported = $(tests-cc-chk)
++tests-unsupported = $(tests-cc-def-chk) $(test-cc-time64-chk)
+ endif
+
+ extra-libs = libpcprofile
+@@ -235,6 +251,7 @@ define chk-gen-locales
+ $(objpfx)$(1).out: $(gen-locales)
+ endef
+ $(foreach t, $(tests-all-chk), $(eval $(call chk-gen-locales,$(t))))
++$(foreach t, $(tests-all-time64-chk), $(eval $(call chk-gen-locales,$(t))))
+ endif
+
+ sLIBdir := $(shell echo $(slibdir) | sed 's,lib\(\|64\)$$,\\\\$$LIB,')
+diff --git a/include/sys/poll.h b/include/sys/poll.h
+index f904e21f89..228704fe79 100644
+--- a/include/sys/poll.h
++++ b/include/sys/poll.h
+@@ -2,13 +2,16 @@
+ # include <io/sys/poll.h>
+
+ #ifndef _ISOMAC
++#include <include/struct___timespec64.h>
++
+ extern int __poll (struct pollfd *__fds, unsigned long int __nfds,
+ int __timeout);
+ libc_hidden_proto (__poll)
+ libc_hidden_proto (ppoll)
+
+ # if __TIMESIZE == 64
+-# define __ppoll64 __ppoll
++# define __ppoll64 ppoll
++# define __ppoll64_chk __ppoll_chk
+ # else
+ # include <time.h>
+ # include <signal.h>
+diff --git a/io/bits/poll2.h b/io/bits/poll2.h
+index 3679d8451a..90f5574969 100644
+--- a/io/bits/poll2.h
++++ b/io/bits/poll2.h
+@@ -43,6 +43,30 @@ poll (struct pollfd *__fds, nfds_t __nfds, int __timeout)
+
+
+ #ifdef __USE_GNU
++# ifdef __USE_TIME_BITS64
++extern int __REDIRECT (__ppoll64_alias, (struct pollfd *__fds, nfds_t __nfds,
++ const struct timespec *__timeout,
++ const __sigset_t *__ss), __ppoll64);
++extern int __ppoll64_chk (struct pollfd *__fds, nfds_t __nfds,
++ const struct timespec *__timeout,
++ const __sigset_t *__ss, __SIZE_TYPE__ __fdslen)
++ __attr_access ((__write_only__, 1, 2));
++extern int __REDIRECT (__ppoll64_chk_warn, (struct pollfd *__fds, nfds_t __n,
++ const struct timespec *__timeout,
++ const __sigset_t *__ss,
++ __SIZE_TYPE__ __fdslen),
++ __ppoll64_chk)
++ __warnattr ("ppoll called with fds buffer too small file nfds entries");
++
++__fortify_function __fortified_attr_access (__write_only__, 1, 2) int
++ppoll (struct pollfd *__fds, nfds_t __nfds, const struct timespec *__timeout,
++ const __sigset_t *__ss)
++{
++ return __glibc_fortify (ppoll64, __nfds, sizeof (*__fds),
++ __glibc_objsize (__fds),
++ __fds, __nfds, __timeout, __ss);
++}
++# else
+ extern int __REDIRECT (__ppoll_alias, (struct pollfd *__fds, nfds_t __nfds,
+ const struct timespec *__timeout,
+ const __sigset_t *__ss), ppoll);
+@@ -65,6 +89,7 @@ ppoll (struct pollfd *__fds, nfds_t __nfds, const struct timespec *__timeout,
+ __glibc_objsize (__fds),
+ __fds, __nfds, __timeout, __ss);
+ }
++# endif
+ #endif
+
+ __END_DECLS
+diff --git a/sysdeps/unix/sysv/linux/Versions b/sysdeps/unix/sysv/linux/Versions
+index 65d2ceda2c..ca6eb1a49e 100644
+--- a/sysdeps/unix/sysv/linux/Versions
++++ b/sysdeps/unix/sysv/linux/Versions
+@@ -313,6 +313,11 @@ libc {
+ process_madvise;
+ process_mrelease;
+ }
++ GLIBC_2.37 {
++%ifdef TIME64_NON_DEFAULT
++ __ppoll64_chk;
++%endif
++ }
+ GLIBC_PRIVATE {
+ # functions used in other libraries
+ __syscall_rt_sigqueueinfo;
+diff --git a/sysdeps/unix/sysv/linux/arm/be/libc.abilist b/sysdeps/unix/sysv/linux/arm/be/libc.abilist
+index e0668a80cf..f28402fe03 100644
+--- a/sysdeps/unix/sysv/linux/arm/be/libc.abilist
++++ b/sysdeps/unix/sysv/linux/arm/be/libc.abilist
+@@ -513,6 +513,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 _Exit F
+ GLIBC_2.4 _IO_2_1_stderr_ D 0xa0
+ GLIBC_2.4 _IO_2_1_stdin_ D 0xa0
+diff --git a/sysdeps/unix/sysv/linux/arm/le/libc.abilist b/sysdeps/unix/sysv/linux/arm/le/libc.abilist
+index d28e7c60b7..e2f56880ed 100644
+--- a/sysdeps/unix/sysv/linux/arm/le/libc.abilist
++++ b/sysdeps/unix/sysv/linux/arm/le/libc.abilist
+@@ -510,6 +510,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 _Exit F
+ GLIBC_2.4 _IO_2_1_stderr_ D 0xa0
+ GLIBC_2.4 _IO_2_1_stdin_ D 0xa0
+diff --git a/sysdeps/unix/sysv/linux/csky/libc.abilist b/sysdeps/unix/sysv/linux/csky/libc.abilist
+index 922b05062f..319d92356e 100644
+--- a/sysdeps/unix/sysv/linux/csky/libc.abilist
++++ b/sysdeps/unix/sysv/linux/csky/libc.abilist
+@@ -2669,3 +2669,4 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+diff --git a/sysdeps/unix/sysv/linux/hppa/libc.abilist b/sysdeps/unix/sysv/linux/hppa/libc.abilist
+index 412144f94c..6450e17ebe 100644
+--- a/sysdeps/unix/sysv/linux/hppa/libc.abilist
++++ b/sysdeps/unix/sysv/linux/hppa/libc.abilist
+@@ -2618,6 +2618,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 __confstr_chk F
+ GLIBC_2.4 __fgets_chk F
+ GLIBC_2.4 __fgets_unlocked_chk F
+diff --git a/sysdeps/unix/sysv/linux/i386/libc.abilist b/sysdeps/unix/sysv/linux/i386/libc.abilist
+index 134393900a..0a24ec9afd 100644
+--- a/sysdeps/unix/sysv/linux/i386/libc.abilist
++++ b/sysdeps/unix/sysv/linux/i386/libc.abilist
+@@ -2802,6 +2802,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 __confstr_chk F
+ GLIBC_2.4 __fgets_chk F
+ GLIBC_2.4 __fgets_unlocked_chk F
+diff --git a/sysdeps/unix/sysv/linux/m68k/coldfire/libc.abilist b/sysdeps/unix/sysv/linux/m68k/coldfire/libc.abilist
+index 0604029c68..16243a7a92 100644
+--- a/sysdeps/unix/sysv/linux/m68k/coldfire/libc.abilist
++++ b/sysdeps/unix/sysv/linux/m68k/coldfire/libc.abilist
+@@ -514,6 +514,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 _Exit F
+ GLIBC_2.4 _IO_2_1_stderr_ D 0x98
+ GLIBC_2.4 _IO_2_1_stdin_ D 0x98
+diff --git a/sysdeps/unix/sysv/linux/m68k/m680x0/libc.abilist b/sysdeps/unix/sysv/linux/m68k/m680x0/libc.abilist
+index af2be5c80d..564a553b27 100644
+--- a/sysdeps/unix/sysv/linux/m68k/m680x0/libc.abilist
++++ b/sysdeps/unix/sysv/linux/m68k/m680x0/libc.abilist
+@@ -2745,6 +2745,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 __confstr_chk F
+ GLIBC_2.4 __fgets_chk F
+ GLIBC_2.4 __fgets_unlocked_chk F
+diff --git a/sysdeps/unix/sysv/linux/microblaze/be/libc.abilist b/sysdeps/unix/sysv/linux/microblaze/be/libc.abilist
+index e090b8d48f..e850f47b21 100644
+--- a/sysdeps/unix/sysv/linux/microblaze/be/libc.abilist
++++ b/sysdeps/unix/sysv/linux/microblaze/be/libc.abilist
+@@ -2718,3 +2718,4 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+diff --git a/sysdeps/unix/sysv/linux/microblaze/le/libc.abilist b/sysdeps/unix/sysv/linux/microblaze/le/libc.abilist
+index 8c5b2db243..37178c503f 100644
+--- a/sysdeps/unix/sysv/linux/microblaze/le/libc.abilist
++++ b/sysdeps/unix/sysv/linux/microblaze/le/libc.abilist
+@@ -2715,3 +2715,4 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+diff --git a/sysdeps/unix/sysv/linux/mips/mips32/fpu/libc.abilist b/sysdeps/unix/sysv/linux/mips/mips32/fpu/libc.abilist
+index 68847134a2..3b30b31466 100644
+--- a/sysdeps/unix/sysv/linux/mips/mips32/fpu/libc.abilist
++++ b/sysdeps/unix/sysv/linux/mips/mips32/fpu/libc.abilist
+@@ -2710,6 +2710,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 __confstr_chk F
+ GLIBC_2.4 __fgets_chk F
+ GLIBC_2.4 __fgets_unlocked_chk F
+diff --git a/sysdeps/unix/sysv/linux/mips/mips32/nofpu/libc.abilist b/sysdeps/unix/sysv/linux/mips/mips32/nofpu/libc.abilist
+index daa44e64fa..0e358570a2 100644
+--- a/sysdeps/unix/sysv/linux/mips/mips32/nofpu/libc.abilist
++++ b/sysdeps/unix/sysv/linux/mips/mips32/nofpu/libc.abilist
+@@ -2708,6 +2708,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 __confstr_chk F
+ GLIBC_2.4 __fgets_chk F
+ GLIBC_2.4 __fgets_unlocked_chk F
+diff --git a/sysdeps/unix/sysv/linux/mips/mips64/n32/libc.abilist b/sysdeps/unix/sysv/linux/mips/mips64/n32/libc.abilist
+index 6169188c96..59c598b98f 100644
+--- a/sysdeps/unix/sysv/linux/mips/mips64/n32/libc.abilist
++++ b/sysdeps/unix/sysv/linux/mips/mips64/n32/libc.abilist
+@@ -2716,6 +2716,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 __confstr_chk F
+ GLIBC_2.4 __fgets_chk F
+ GLIBC_2.4 __fgets_unlocked_chk F
+diff --git a/sysdeps/unix/sysv/linux/nios2/libc.abilist b/sysdeps/unix/sysv/linux/nios2/libc.abilist
+index 58e9b486b0..463e01ab84 100644
+--- a/sysdeps/unix/sysv/linux/nios2/libc.abilist
++++ b/sysdeps/unix/sysv/linux/nios2/libc.abilist
+@@ -2757,3 +2757,4 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libc.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libc.abilist
+index 8c9ca32cbe..405d40d11c 100644
+--- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libc.abilist
++++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/fpu/libc.abilist
+@@ -2772,6 +2772,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 _IO_fprintf F
+ GLIBC_2.4 _IO_printf F
+ GLIBC_2.4 _IO_sprintf F
+diff --git a/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libc.abilist b/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libc.abilist
+index 08a6604aab..ce89602b93 100644
+--- a/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libc.abilist
++++ b/sysdeps/unix/sysv/linux/powerpc/powerpc32/nofpu/libc.abilist
+@@ -2805,6 +2805,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 _IO_fprintf F
+ GLIBC_2.4 _IO_printf F
+ GLIBC_2.4 _IO_sprintf F
+diff --git a/sysdeps/unix/sysv/linux/ppoll.c b/sysdeps/unix/sysv/linux/ppoll.c
+index 1105e29b00..2e173b931d 100644
+--- a/sysdeps/unix/sysv/linux/ppoll.c
++++ b/sysdeps/unix/sysv/linux/ppoll.c
+@@ -67,7 +67,7 @@ __ppoll64 (struct pollfd *fds, nfds_t nfds, const struct __timespec64 *timeout,
+ libc_hidden_def (__ppoll64)
+
+ int
+-__ppoll (struct pollfd *fds, nfds_t nfds, const struct timespec *timeout,
++ppoll (struct pollfd *fds, nfds_t nfds, const struct timespec *timeout,
+ const sigset_t *sigmask)
+ {
+ struct __timespec64 ts64;
+@@ -77,5 +77,4 @@ __ppoll (struct pollfd *fds, nfds_t nfds, const struct timespec *timeout,
+ return __ppoll64 (fds, nfds, timeout ? &ts64 : NULL, sigmask);
+ }
+ #endif
+-strong_alias (__ppoll, ppoll)
+ libc_hidden_def (ppoll)
+diff --git a/sysdeps/unix/sysv/linux/ppoll_chk.c b/sysdeps/unix/sysv/linux/ppoll_chk.c
+new file mode 100644
+index 0000000000..65c4b56671
+--- /dev/null
++++ b/sysdeps/unix/sysv/linux/ppoll_chk.c
+@@ -0,0 +1,42 @@
++/* Fortify ppoll implementation. Linux version.
++ Copyright (C) 2022 Free Software Foundation, Inc.
++ This file is part of the GNU C Library.
++
++ The GNU C Library is free software; you can redistribute it and/or
++ modify it under the terms of the GNU Lesser General Public
++ License as published by the Free Software Foundation; either
++ version 2.1 of the License, or (at your option) any later version.
++
++ The GNU C Library is distributed in the hope that it will be useful,
++ but WITHOUT ANY WARRANTY; without even the implied warranty of
++ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
++ Lesser General Public License for more details.
++
++ You should have received a copy of the GNU Lesser General Public
++ License along with the GNU C Library; if not, see
++ <https://www.gnu.org/licenses/>. */
++
++#include <poll.h>
++
++int
++__ppoll64_chk (struct pollfd *fds, nfds_t nfds,
++ const struct __timespec64 *tmo, const __sigset_t *ss,
++ __SIZE_TYPE__ fdslen)
++{
++ if (fdslen / sizeof (*fds) < nfds)
++ __chk_fail ();
++
++ return __ppoll64 (fds, nfds, tmo, ss);
++}
++
++#if __TIMESIZE != 64
++int
++__ppoll_chk (struct pollfd *fds, nfds_t nfds, const struct timespec *tmo,
++ const __sigset_t *ss, __SIZE_TYPE__ fdslen)
++{
++ if (fdslen / sizeof (*fds) < nfds)
++ __chk_fail ();
++
++ return ppoll (fds, nfds, tmo, ss);
++}
++#endif
+diff --git a/sysdeps/unix/sysv/linux/s390/s390-32/libc.abilist b/sysdeps/unix/sysv/linux/s390/s390-32/libc.abilist
+index 009f22931e..5ca051a9eb 100644
+--- a/sysdeps/unix/sysv/linux/s390/s390-32/libc.abilist
++++ b/sysdeps/unix/sysv/linux/s390/s390-32/libc.abilist
+@@ -2770,6 +2770,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 _IO_fprintf F
+ GLIBC_2.4 _IO_printf F
+ GLIBC_2.4 _IO_sprintf F
+diff --git a/sysdeps/unix/sysv/linux/sh/be/libc.abilist b/sysdeps/unix/sysv/linux/sh/be/libc.abilist
+index afb5bc37b1..5b48168ec6 100644
+--- a/sysdeps/unix/sysv/linux/sh/be/libc.abilist
++++ b/sysdeps/unix/sysv/linux/sh/be/libc.abilist
+@@ -2625,6 +2625,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 __confstr_chk F
+ GLIBC_2.4 __fgets_chk F
+ GLIBC_2.4 __fgets_unlocked_chk F
+diff --git a/sysdeps/unix/sysv/linux/sh/le/libc.abilist b/sysdeps/unix/sysv/linux/sh/le/libc.abilist
+index 2b53a3cf92..c42b39cea8 100644
+--- a/sysdeps/unix/sysv/linux/sh/le/libc.abilist
++++ b/sysdeps/unix/sysv/linux/sh/le/libc.abilist
+@@ -2622,6 +2622,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 __confstr_chk F
+ GLIBC_2.4 __fgets_chk F
+ GLIBC_2.4 __fgets_unlocked_chk F
+diff --git a/sysdeps/unix/sysv/linux/sparc/sparc32/libc.abilist b/sysdeps/unix/sysv/linux/sparc/sparc32/libc.abilist
+index 43b9844a99..5a0a662dee 100644
+--- a/sysdeps/unix/sysv/linux/sparc/sparc32/libc.abilist
++++ b/sysdeps/unix/sysv/linux/sparc/sparc32/libc.abilist
+@@ -2765,6 +2765,7 @@ GLIBC_2.36 pidfd_open F
+ GLIBC_2.36 pidfd_send_signal F
+ GLIBC_2.36 process_madvise F
+ GLIBC_2.36 process_mrelease F
++GLIBC_2.37 __ppoll64_chk F
+ GLIBC_2.4 _IO_fprintf F
+ GLIBC_2.4 _IO_printf F
+ GLIBC_2.4 _IO_sprintf F
diff --git a/meta/recipes-core/glibc/glibc_2.36.bb b/meta/recipes-core/glibc/glibc_2.36.bb
index 45bc784b33..3010f783ac 100644
--- a/meta/recipes-core/glibc/glibc_2.36.bb
+++ b/meta/recipes-core/glibc/glibc_2.36.bb
@@ -54,6 +54,7 @@ SRC_URI = "${GLIBC_GIT_URI};branch=${SRCBRANCH};name=glibc \
file://0023-fix-create-thread-failed-in-unprivileged-process-BZ-.patch \
file://0024-Avoid-hardcoded-build-time-paths-in-the-output-binar.patch \
file://0025-startup-Force-O2.patch \
+ file://0001-Linux-Add-ppoll-fortify-symbol-for-64-bit-time_t-BZ-.patch \
"
S = "${WORKDIR}/git"
B = "${WORKDIR}/build-${TARGET_SYS}"
--
2.30.2
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH RFC 2/3] Add QA check for 32 bit time and file offset functions
2022-12-08 7:11 Y2038: Sanity checker for non-64-bit time and file functions Ola x Nilsson
2022-12-08 7:11 ` [PATCH RFC 1/3] glibc: Add ppoll fortify symbol for 64 bit time_t Ola x Nilsson
@ 2022-12-08 7:11 ` Ola x Nilsson
2022-12-17 11:17 ` [OE-core] " Richard Purdie
2022-12-08 7:11 ` [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags Ola x Nilsson
2 siblings, 1 reply; 22+ messages in thread
From: Ola x Nilsson @ 2022-12-08 7:11 UTC (permalink / raw)
To: openembedded-core
Check for known symbols that should have been redirected to 64bit
variants when -D_FILE_OFFSET_BITS=64 and -D_TIME_BITS=64 are set.
Signed-off-by: Ola x Nilsson <olani@axis.com>
---
meta/classes-global/insane.bbclass | 132 +++++++++++++++++++++++++++++
1 file changed, 132 insertions(+)
diff --git a/meta/classes-global/insane.bbclass b/meta/classes-global/insane.bbclass
index df2c40c3c5..b0e99c4c3f 100644
--- a/meta/classes-global/insane.bbclass
+++ b/meta/classes-global/insane.bbclass
@@ -506,6 +506,138 @@ def package_qa_check_symlink_to_sysroot(path, name, d, elf, messages):
trimmed = path.replace(os.path.join (d.getVar("PKGDEST"), name), "")
oe.qa.add_message(messages, "symlink-to-sysroot", "Symlink %s in %s points to TMPDIR" % (trimmed, name))
+QAPATHTEST[32bit-time] = "check_32bit_symbols"
+def check_32bit_symbols(path, packagename, d, elf, messages):
+ """
+ Check that ELF files do not use any 32 bit time APIs from glibc.
+ """
+ # TODO: We should have some DISTRO_FEATURE flag to check here
+ # if '-D_TIME_BITS=64' not in d.getVar('GLIBC_64BIT_TIME_FLAGS'):
+ # # No use checking if the feature is not enabled?
+ # return
+ import re
+ # This list is manually constructed by searching the image folder of the
+ # glibc recipe for __USE_TIME_BITS64. There is no good way to do this
+ # automatically.
+ api32 = {
+ # /usr/include/time.h
+ "clock_getres", "clock_gettime", "clock_nanosleep", "clock_settime",
+ "ctime", "ctime_r", "difftime", "gmtime", "gmtime_r", "localtime",
+ "localtime_r", "mktime", "nanosleep", "time", "timegm", "timelocal",
+ "timer_gettime", "timer_settime", "timespec_get", "timespec_getres",
+ # /usr/include/bits/time.h
+ "clock_adjtime",
+ # /usr/include/signal.h
+ "sigtimedwait",
+ # /usr/include/sys/time.h
+ "futimes", "futimesat", "getitimer", "gettimeofday", "lutimes",
+ "setitimer", "settimeofday", "utimes",
+ # /usr/include/sys/timex.h
+ "adjtimex", "ntp_adjtime", "ntp_gettime", "ntp_gettimex",
+ # /usr/include/sys/wait.h
+ "wait3", "wait4",
+ # /usr/include/sys/stat.h
+ "fstat", "fstat64", "fstatat", "fstatat64", "futimens", "lstat",
+ "lstat64", "stat", "stat64", "utimensat",
+ # /usr/include/sys/poll.h
+ "ppoll",
+ # /usr/include/sys/resource.h
+ "getrusage",
+ # /usr/include/sys/ioctl.h
+ "ioctl",
+ # /usr/include/sys/select.h
+ "select", "pselect",
+ # /usr/include/sys/prctl.h
+ "prctl",
+ # /usr/include/sys/epoll.h
+ "epoll_pwait2",
+ # /usr/include/sys/timerfd.h
+ "timerfd_gettime", "timerfd_settime",
+ # /usr/include/sys/socket.h
+ "getsockopt", "recvmmsg", "recvmsg", "sendmmsg", "sendmsg",
+ "setsockopt",
+ # /usr/include/sys/msg.h
+ "msgctl",
+ # /usr/include/sys/sem.h
+ "semctl", "semtimedop",
+ # /usr/include/sys/shm.h
+ "shmctl",
+ # /usr/include/pthread.h
+ "pthread_clockjoin_np", "pthread_cond_clockwait",
+ "pthread_cond_timedwait", "pthread_mutex_clocklock",
+ "pthread_mutex_timedlock", "pthread_rwlock_clockrdlock",
+ "pthread_rwlock_clockwrlock", "pthread_rwlock_timedrdlock",
+ "pthread_rwlock_timedwrlock", "pthread_timedjoin_np",
+ # /usr/include/semaphore.h
+ "sem_clockwait", "sem_timedwait",
+ # /usr/include/threads.h
+ "cnd_timedwait", "mtx_timedlock", "thrd_sleep",
+ # /usr/include/aio.h
+ "aio_cancel", "aio_error", "aio_read", "aio_return", "aio_suspend",
+ "aio_write", "lio_listio",
+ # /usr/include/mqueue.h
+ "mq_timedreceive", "mq_timedsend",
+ # /usr/include/glob.h
+ "glob", "glob64", "globfree", "globfree64",
+ # /usr/include/sched.h
+ "sched_rr_get_interval",
+ # /usr/include/fcntl.h
+ "fcntl", "fcntl64",
+ # /usr/include/utime.h
+ "utime",
+ # /usr/include/ftw.h
+ "ftw", "ftw64", "nftw", "nftw64",
+ # /usr/include/fts.h
+ "fts64_children", "fts64_close", "fts64_open", "fts64_read",
+ "fts64_set", "fts_children", "fts_close", "fts_open", "fts_read",
+ "fts_set",
+ # /usr/include/netdb.h
+ "gai_suspend",
+ }
+
+ ptrn = re.compile(
+ r'''
+ (?P<value>[\da-fA-F]+) \s+
+ (?P<flags>[lgu! ][w ][C ][W ][Ii ][dD ]F) \s+
+ (?P<section>\*UND\*) \s+
+ (?P<alignment>(?P<size>[\da-fA-F]+)) \s+
+ (?P<symbol>
+ ''' +
+ r'(?P<notag>' + r'|'.join(sorted(api32)) + r')' +
+ r'''
+ (@+(?P<tag>GLIBC_\d+\.\d+\S*)))
+ ''', re.VERBOSE
+ )
+
+ # elf is a oe.qa.ELFFile object
+ if elf is not None:
+ phdrs = elf.run_objdump("-tw", d)
+ syms = re.finditer(ptrn, phdrs)
+ usedapis = {sym.group('notag') for sym in syms}
+ if usedapis:
+ elfpath = package_qa_clean_path(path, d, packagename)
+ # Remove any .debug dir, heuristic that probably works
+ # At this point, any symbol information is stripped into the debug
+ # package, so that is the only place we will find them.
+ elfpath = elfpath.replace('.debug/', '')
+ allowed = (
+ d.getVarFlag(
+ 'INSANE_SKIP:' + d.getVar('PN'), elfpath.replace('/', '_')
+ ) or ''
+ ).split()
+ usedapis -= set(allowed)
+ if usedapis:
+ msgformat = elfpath + " uses 32-bit api '%s'"
+ for sym in usedapis:
+ oe.qa.add_message(messages, '32bit-time', msgformat % sym)
+ oe.qa.add_message(
+ messages, '32bit-time',
+ 'Suppress with INSANE_SKIP:%s[%s] = "%s"' % (
+ d.getVar('PN'), elfpath.replace('/', '_'),
+ ' '.join(usedapis)
+ )
+ )
+
# Check license variables
do_populate_lic[postfuncs] += "populate_lic_qa_checksum"
python populate_lic_qa_checksum() {
--
2.30.2
^ permalink raw reply related [flat|nested] 22+ messages in thread
* [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-08 7:11 Y2038: Sanity checker for non-64-bit time and file functions Ola x Nilsson
2022-12-08 7:11 ` [PATCH RFC 1/3] glibc: Add ppoll fortify symbol for 64 bit time_t Ola x Nilsson
2022-12-08 7:11 ` [PATCH RFC 2/3] Add QA check for 32 bit time and file offset functions Ola x Nilsson
@ 2022-12-08 7:11 ` Ola x Nilsson
2022-12-08 10:00 ` [OE-core] " Alexander Kanavin
` (3 more replies)
2 siblings, 4 replies; 22+ messages in thread
From: Ola x Nilsson @ 2022-12-08 7:11 UTC (permalink / raw)
To: openembedded-core
Signed-off-by: Ola x Nilsson <olani@axis.com>
---
meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
1 file changed, 23 insertions(+)
create mode 100644 meta/conf/distro/time64.conf
diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
new file mode 100644
index 0000000000..99eb06dc0f
--- /dev/null
+++ b/meta/conf/distro/time64.conf
@@ -0,0 +1,23 @@
+GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
+
+# TODO: Only for 32-bit architectures?
+TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
+TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
+TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
+
+GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
+GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
+# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
+# both 32 and 64 bit file APIs. But it does not handle the time side?
+# Needs further investigation
+GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
+GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
+
+INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so] = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
+INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
+INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64 setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime localtime_r"
+
+# libpulsedsp.so is a preload-library that hooks libc functions
+INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
+
+
--
2.30.2
^ permalink raw reply related [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-08 7:11 ` [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags Ola x Nilsson
@ 2022-12-08 10:00 ` Alexander Kanavin
2022-12-08 10:10 ` Richard Purdie
2022-12-09 10:36 ` Ola x Nilsson
2022-12-09 0:23 ` Khem Raj
` (2 subsequent siblings)
3 siblings, 2 replies; 22+ messages in thread
From: Alexander Kanavin @ 2022-12-08 10:00 UTC (permalink / raw)
To: Ola x Nilsson; +Cc: openembedded-core
On Thu, 8 Dec 2022 at 08:11, Ola x Nilsson <olani@axis.com> wrote:
> +# TODO: Only for 32-bit architectures?
> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
On the contrary, why not set this globally?
> +GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
> +GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
> +# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
> +# both 32 and 64 bit file APIs. But it does not handle the time side?
> +# Needs further investigation
> +GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
> +GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
> +
> +INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so] = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
> +INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
> +INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64 setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime localtime_r"
> +
> +# libpulsedsp.so is a preload-library that hooks libc functions
> +INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
Some of these perhaps need a ticket in bugzilla, and there should be a
meta-ticket for Y2038 that tracks individual issues (by depending on
them). Can you arrange that please?
Alex
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-08 10:00 ` [OE-core] " Alexander Kanavin
@ 2022-12-08 10:10 ` Richard Purdie
2022-12-09 10:31 ` Ola x Nilsson
2022-12-09 10:36 ` Ola x Nilsson
1 sibling, 1 reply; 22+ messages in thread
From: Richard Purdie @ 2022-12-08 10:10 UTC (permalink / raw)
To: Alexander Kanavin, Ola x Nilsson; +Cc: openembedded-core
On Thu, 2022-12-08 at 11:00 +0100, Alexander Kanavin wrote:
> On Thu, 8 Dec 2022 at 08:11, Ola x Nilsson <olani@axis.com> wrote:
> > +# TODO: Only for 32-bit architectures?
> > +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> > +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> > +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
>
> On the contrary, why not set this globally?
The 64 bit architectures don't need this and should "just work" so it
is probably better just to set this where needed. At the very least
that will keep the compiler command lines a bit cleaner.
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-08 7:11 ` [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags Ola x Nilsson
2022-12-08 10:00 ` [OE-core] " Alexander Kanavin
@ 2022-12-09 0:23 ` Khem Raj
2022-12-09 9:55 ` Ola x Nilsson
2022-12-15 16:32 ` Khem Raj
2022-12-16 18:44 ` Khem Raj
3 siblings, 1 reply; 22+ messages in thread
From: Khem Raj @ 2022-12-09 0:23 UTC (permalink / raw)
To: Ola x Nilsson, openembedded-core
Thanks for the patches
On 12/7/22 23:11, Ola x Nilsson wrote:
> Signed-off-by: Ola x Nilsson <olani@axis.com>
> ---
> meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> 1 file changed, 23 insertions(+)
> create mode 100644 meta/conf/distro/time64.conf
>
> diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> new file mode 100644
> index 0000000000..99eb06dc0f
> --- /dev/null
> +++ b/meta/conf/distro/time64.conf
> @@ -0,0 +1,23 @@
> +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
> +
> +# TODO: Only for 32-bit architectures?
> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
We should enable it across all 32bit systems
something like
TARGET_CC_ARCH += "${@'-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64' if
d.getVar('SITEINFO_BITS') == '32' else ''}"
would do it.
Also adding -D_LARGEFILE_SOURCE will ensure needed functions that LFS
needs is enabled as well, usually its automatically defined when
-D_GNU_SOURCE is set but that may not be case all the time.
> +
> +GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
> +GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
> +# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
> +# both 32 and 64 bit file APIs. But it does not handle the time side?
> +# Needs further investigation
> +GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
> +GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
> +
> +INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so] = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
> +INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
> +INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64 setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime localtime_r"
> +
> +# libpulsedsp.so is a preload-library that hooks libc functions
> +INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
> +
> +
>
>
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#174389): https://lists.openembedded.org/g/openembedded-core/message/174389
> Mute This Topic: https://lists.openembedded.org/mt/95533494/1997914
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [raj.khem@gmail.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-09 0:23 ` Khem Raj
@ 2022-12-09 9:55 ` Ola x Nilsson
2022-12-09 12:35 ` Richard Purdie
` (2 more replies)
0 siblings, 3 replies; 22+ messages in thread
From: Ola x Nilsson @ 2022-12-09 9:55 UTC (permalink / raw)
To: Khem Raj; +Cc: openembedded-core@lists.openembedded.org
On Fri, Dec 09 2022, Khem Raj wrote:
> Thanks for the patches
>
> On 12/7/22 23:11, Ola x Nilsson wrote:
>> Signed-off-by: Ola x Nilsson <olani@axis.com>
>> ---
>> meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
>> 1 file changed, 23 insertions(+)
>> create mode 100644 meta/conf/distro/time64.conf
>>
>> diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
>> new file mode 100644
>> index 0000000000..99eb06dc0f
>> --- /dev/null
>> +++ b/meta/conf/distro/time64.conf
>> @@ -0,0 +1,23 @@
>> +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
>> +
>> +# TODO: Only for 32-bit architectures?
>> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
>> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
>> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
>
> We should enable it across all 32bit systems
> something like
>
> TARGET_CC_ARCH += "${@'-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64' if
> d.getVar('SITEINFO_BITS') == '32' else ''}"
>
> would do it.
I was looking for a way to identify 32-bit platforms but did not find
SITEINFO_BITS. I'll look into using that instead.
> Also adding -D_LARGEFILE_SOURCE will ensure needed functions that LFS
> needs is enabled as well, usually its automatically defined when
> -D_GNU_SOURCE is set but that may not be case all the time.
As far as I can tell, _LARGEFILE_SOURCE and _LARGEFILE64_SOURCE are
independent of _TIME_BITS=64 and _FILE_OFFSET_BITS=64 (while
_TIME_BITS=64 requiress _FILE_OFFSET_BITS=64).
I thought the _LARGEFILE* options were controlled by the largefile
DISTRO_FEATURE but there seems to be some confusion about that.
It looks like you (Khem) made largefile default instead of using the
distro feature in all of the OE-core recipes, and the largefile distro
feature is no longer mentioned in docs. It is mentioned as removed in
the transition notes. But it is still in poky-tiny.conf and
POKY_DEFAULT_DISTRO_FEATURES. Probably an oversight?
I do not see the _LARGEFILE*_SOURCE options defined globally anywhere.
Am I missing something?
>> +
>> +GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
>> +GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
>> +# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
>> +# both 32 and 64 bit file APIs. But it does not handle the time side?
>> +# Needs further investigation
>> +GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
>> +GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
>> +
>> +INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so]
>> = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl
>> lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
>> +INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
>> +INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64
>> setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm
>> fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime
>> localtime_r"
>> +
>> +# libpulsedsp.so is a preload-library that hooks libc functions
>> +INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
>> +
>> +
>>
--
Ola x Nilsson
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-08 10:10 ` Richard Purdie
@ 2022-12-09 10:31 ` Ola x Nilsson
0 siblings, 0 replies; 22+ messages in thread
From: Ola x Nilsson @ 2022-12-09 10:31 UTC (permalink / raw)
To: Richard Purdie
Cc: Alexander Kanavin, openembedded-core@lists.openembedded.org
On Thu, Dec 08 2022, Richard Purdie wrote:
> On Thu, 2022-12-08 at 11:00 +0100, Alexander Kanavin wrote:
>> On Thu, 8 Dec 2022 at 08:11, Ola x Nilsson <olani@axis.com> wrote:
>> > +# TODO: Only for 32-bit architectures?
>> > +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
>> > +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
>> > +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
>>
>> On the contrary, why not set this globally?
>
> The 64 bit architectures don't need this and should "just work" so it
> is probably better just to set this where needed. At the very least
> that will keep the compiler command lines a bit cleaner.
It also changes the hashes for no good reason, right?
Khem suggested using SITEINFO_BITS, I'll see if that works.
--
Ola x Nilsson
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-08 10:00 ` [OE-core] " Alexander Kanavin
2022-12-08 10:10 ` Richard Purdie
@ 2022-12-09 10:36 ` Ola x Nilsson
2022-12-09 13:06 ` Alexandre Belloni
1 sibling, 1 reply; 22+ messages in thread
From: Ola x Nilsson @ 2022-12-09 10:36 UTC (permalink / raw)
To: Alexander Kanavin; +Cc: openembedded-core@lists.openembedded.org
On Thu, Dec 08 2022, Alexander Kanavin wrote:
> On Thu, 8 Dec 2022 at 08:11, Ola x Nilsson <olani@axis.com> wrote:
>> +GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
>> +GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
>> +# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
>> +# both 32 and 64 bit file APIs. But it does not handle the time side?
>> +# Needs further investigation
>> +GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
>> +GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
>> +
>> +INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so]
>> = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl
>> lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
>> +INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
>> +INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64
>> setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm
>> fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime
>> localtime_r"
>> +
>> +# libpulsedsp.so is a preload-library that hooks libc functions
>> +INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
>
> Some of these perhaps need a ticket in bugzilla, and there should be a
> meta-ticket for Y2038 that tracks individual issues (by depending on
> them). Can you arrange that please?
I can file tickets if that's the way we want to go. I'm not really at
that point yet, these are just for the packages that were included in
our products. I've just done one plain poky build to make sure this
stuff worked so far.
--
Ola x Nilsson
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-09 9:55 ` Ola x Nilsson
@ 2022-12-09 12:35 ` Richard Purdie
2022-12-09 13:09 ` Alexandre Belloni
2022-12-09 16:47 ` Khem Raj
2 siblings, 0 replies; 22+ messages in thread
From: Richard Purdie @ 2022-12-09 12:35 UTC (permalink / raw)
To: Ola x Nilsson, Khem Raj; +Cc: openembedded-core@lists.openembedded.org
On Fri, 2022-12-09 at 10:55 +0100, Ola x Nilsson wrote:
> On Fri, Dec 09 2022, Khem Raj wrote:
>
> > Thanks for the patches
> >
> > On 12/7/22 23:11, Ola x Nilsson wrote:
> > > Signed-off-by: Ola x Nilsson <olani@axis.com>
> > > ---
> > > meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> > > 1 file changed, 23 insertions(+)
> > > create mode 100644 meta/conf/distro/time64.conf
> > >
> > > diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> > > new file mode 100644
> > > index 0000000000..99eb06dc0f
> > > --- /dev/null
> > > +++ b/meta/conf/distro/time64.conf
> > > @@ -0,0 +1,23 @@
> > > +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
> > > +
> > > +# TODO: Only for 32-bit architectures?
> > > +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> > > +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> > > +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
> >
> > We should enable it across all 32bit systems
> > something like
> >
> > TARGET_CC_ARCH += "${@'-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64' if
> > d.getVar('SITEINFO_BITS') == '32' else ''}"
> >
> > would do it.
>
> I was looking for a way to identify 32-bit platforms but did not find
> SITEINFO_BITS. I'll look into using that instead.
>
> > Also adding -D_LARGEFILE_SOURCE will ensure needed functions that LFS
> > needs is enabled as well, usually its automatically defined when
> > -D_GNU_SOURCE is set but that may not be case all the time.
>
> As far as I can tell, _LARGEFILE_SOURCE and _LARGEFILE64_SOURCE are
> independent of _TIME_BITS=64 and _FILE_OFFSET_BITS=64 (while
> _TIME_BITS=64 requiress _FILE_OFFSET_BITS=64).
>
> I thought the _LARGEFILE* options were controlled by the largefile
> DISTRO_FEATURE but there seems to be some confusion about that.
>
> It looks like you (Khem) made largefile default instead of using the
> distro feature in all of the OE-core recipes, and the largefile distro
> feature is no longer mentioned in docs. It is mentioned as removed in
> the transition notes. But it is still in poky-tiny.conf and
> POKY_DEFAULT_DISTRO_FEATURES. Probably an oversight?
Yes. I just sent a patch to remove that.
> I do not see the _LARGEFILE*_SOURCE options defined globally anywhere.
>
> Am I missing something?
We used to control configure largefile options with the distro feature,
now we just turn it on anywhere we need to unconditionally. I suspect
we rely on software configuring by default to it in some cases. It
would be good to check if it isn't being enabled anywhere and fix those
cases somehow. Not sure if we need global options or not.
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-09 10:36 ` Ola x Nilsson
@ 2022-12-09 13:06 ` Alexandre Belloni
0 siblings, 0 replies; 22+ messages in thread
From: Alexandre Belloni @ 2022-12-09 13:06 UTC (permalink / raw)
To: Ola x Nilsson; +Cc: Alexander Kanavin, openembedded-core@lists.openembedded.org
On 09/12/2022 11:36:03+0100, Ola x Nilsson wrote:
>
> On Thu, Dec 08 2022, Alexander Kanavin wrote:
>
> > On Thu, 8 Dec 2022 at 08:11, Ola x Nilsson <olani@axis.com> wrote:
> >> +GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
> >> +GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
> >> +# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
> >> +# both 32 and 64 bit file APIs. But it does not handle the time side?
> >> +# Needs further investigation
> >> +GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
> >> +GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
> >> +
> >> +INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so]
> >> = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl
> >> lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
> >> +INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
> >> +INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64
> >> setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm
> >> fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime
> >> localtime_r"
> >> +
> >> +# libpulsedsp.so is a preload-library that hooks libc functions
> >> +INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
> >
> > Some of these perhaps need a ticket in bugzilla, and there should be a
> > meta-ticket for Y2038 that tracks individual issues (by depending on
> > them). Can you arrange that please?
>
> I can file tickets if that's the way we want to go. I'm not really at
> that point yet, these are just for the packages that were included in
> our products. I've just done one plain poky build to make sure this
> stuff worked so far.
I know there is pulseaudio that will fail too. I'll do a build including
time64.conf over the week end and I can file bugs where necessary.
>
--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-09 9:55 ` Ola x Nilsson
2022-12-09 12:35 ` Richard Purdie
@ 2022-12-09 13:09 ` Alexandre Belloni
2022-12-09 13:13 ` Alexander Kanavin
2022-12-09 16:36 ` Khem Raj
2022-12-09 16:47 ` Khem Raj
2 siblings, 2 replies; 22+ messages in thread
From: Alexandre Belloni @ 2022-12-09 13:09 UTC (permalink / raw)
To: Ola x Nilsson; +Cc: Khem Raj, openembedded-core@lists.openembedded.org
On 09/12/2022 10:55:16+0100, Ola x Nilsson wrote:
>
> On Fri, Dec 09 2022, Khem Raj wrote:
>
> > Thanks for the patches
> >
> > On 12/7/22 23:11, Ola x Nilsson wrote:
> >> Signed-off-by: Ola x Nilsson <olani@axis.com>
> >> ---
> >> meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> >> 1 file changed, 23 insertions(+)
> >> create mode 100644 meta/conf/distro/time64.conf
> >>
> >> diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> >> new file mode 100644
> >> index 0000000000..99eb06dc0f
> >> --- /dev/null
> >> +++ b/meta/conf/distro/time64.conf
> >> @@ -0,0 +1,23 @@
> >> +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
> >> +
> >> +# TODO: Only for 32-bit architectures?
> >> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> >> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> >> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
> >
> > We should enable it across all 32bit systems
> > something like
> >
> > TARGET_CC_ARCH += "${@'-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64' if
> > d.getVar('SITEINFO_BITS') == '32' else ''}"
> >
> > would do it.
>
> I was looking for a way to identify 32-bit platforms but did not find
> SITEINFO_BITS. I'll look into using that instead.
>
I don't think it is correct to assume all the 32bit platforms are
affected. riscv32 is not affected for example nor would be any newer
32bit architecture.
--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-09 13:09 ` Alexandre Belloni
@ 2022-12-09 13:13 ` Alexander Kanavin
2022-12-09 13:29 ` Richard Purdie
2022-12-09 16:36 ` Khem Raj
1 sibling, 1 reply; 22+ messages in thread
From: Alexander Kanavin @ 2022-12-09 13:13 UTC (permalink / raw)
To: alexandre.belloni
Cc: Ola x Nilsson, Khem Raj, openembedded-core@lists.openembedded.org
I'd rather not maintain a list of those that *are* affected, even if
we then include those that aren't. It's only build flags, and they
should be ignored when the platform has already done the right things
from the start.
Alex
On Fri, 9 Dec 2022 at 14:09, Alexandre Belloni via
lists.openembedded.org
<alexandre.belloni=bootlin.com@lists.openembedded.org> wrote:
>
> On 09/12/2022 10:55:16+0100, Ola x Nilsson wrote:
> >
> > On Fri, Dec 09 2022, Khem Raj wrote:
> >
> > > Thanks for the patches
> > >
> > > On 12/7/22 23:11, Ola x Nilsson wrote:
> > >> Signed-off-by: Ola x Nilsson <olani@axis.com>
> > >> ---
> > >> meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> > >> 1 file changed, 23 insertions(+)
> > >> create mode 100644 meta/conf/distro/time64.conf
> > >>
> > >> diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> > >> new file mode 100644
> > >> index 0000000000..99eb06dc0f
> > >> --- /dev/null
> > >> +++ b/meta/conf/distro/time64.conf
> > >> @@ -0,0 +1,23 @@
> > >> +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
> > >> +
> > >> +# TODO: Only for 32-bit architectures?
> > >> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> > >> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> > >> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
> > >
> > > We should enable it across all 32bit systems
> > > something like
> > >
> > > TARGET_CC_ARCH += "${@'-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64' if
> > > d.getVar('SITEINFO_BITS') == '32' else ''}"
> > >
> > > would do it.
> >
> > I was looking for a way to identify 32-bit platforms but did not find
> > SITEINFO_BITS. I'll look into using that instead.
> >
>
> I don't think it is correct to assume all the 32bit platforms are
> affected. riscv32 is not affected for example nor would be any newer
> 32bit architecture.
>
>
> --
> Alexandre Belloni, co-owner and COO, Bootlin
> Embedded Linux and Kernel engineering
> https://bootlin.com
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#174450): https://lists.openembedded.org/g/openembedded-core/message/174450
> Mute This Topic: https://lists.openembedded.org/mt/95533494/1686489
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [alex.kanavin@gmail.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-09 13:13 ` Alexander Kanavin
@ 2022-12-09 13:29 ` Richard Purdie
0 siblings, 0 replies; 22+ messages in thread
From: Richard Purdie @ 2022-12-09 13:29 UTC (permalink / raw)
To: Alexander Kanavin, alexandre.belloni
Cc: Ola x Nilsson, Khem Raj, openembedded-core@lists.openembedded.org
On Fri, 2022-12-09 at 14:13 +0100, Alexander Kanavin wrote:
> I'd rather not maintain a list of those that *are* affected,
This isn't actually too bad since these are architectures, not machines
or tunes and the list should remain static, nobody is creating new 32
bit platforms with the bug. So yes, it is a pain to write but once
written, it shouldn't change much.
> even if
> we then include those that aren't. It's only build flags, and they
> should be ignored when the platform has already done the right things
> from the start.
They do clutter up the compile logs and I'd prefer not to have them
where they aren't needed, extra noise causes different problems.
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-09 13:09 ` Alexandre Belloni
2022-12-09 13:13 ` Alexander Kanavin
@ 2022-12-09 16:36 ` Khem Raj
1 sibling, 0 replies; 22+ messages in thread
From: Khem Raj @ 2022-12-09 16:36 UTC (permalink / raw)
To: Alexandre Belloni; +Cc: Ola x Nilsson, openembedded-core@lists.openembedded.org
On Fri, Dec 9, 2022 at 5:09 AM Alexandre Belloni
<alexandre.belloni@bootlin.com> wrote:
>
> On 09/12/2022 10:55:16+0100, Ola x Nilsson wrote:
> >
> > On Fri, Dec 09 2022, Khem Raj wrote:
> >
> > > Thanks for the patches
> > >
> > > On 12/7/22 23:11, Ola x Nilsson wrote:
> > >> Signed-off-by: Ola x Nilsson <olani@axis.com>
> > >> ---
> > >> meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> > >> 1 file changed, 23 insertions(+)
> > >> create mode 100644 meta/conf/distro/time64.conf
> > >>
> > >> diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> > >> new file mode 100644
> > >> index 0000000000..99eb06dc0f
> > >> --- /dev/null
> > >> +++ b/meta/conf/distro/time64.conf
> > >> @@ -0,0 +1,23 @@
> > >> +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
> > >> +
> > >> +# TODO: Only for 32-bit architectures?
> > >> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> > >> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> > >> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
> > >
> > > We should enable it across all 32bit systems
> > > something like
> > >
> > > TARGET_CC_ARCH += "${@'-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64' if
> > > d.getVar('SITEINFO_BITS') == '32' else ''}"
> > >
> > > would do it.
> >
> > I was looking for a way to identify 32-bit platforms but did not find
> > SITEINFO_BITS. I'll look into using that instead.
> >
>
> I don't think it is correct to assume all the 32bit platforms are
> affected. riscv32 is not affected for example nor would be any newer
> 32bit architecture.
yeah that's a good point, I overlooked that because these defines will
become redundant in those cases
>
>
> --
> Alexandre Belloni, co-owner and COO, Bootlin
> Embedded Linux and Kernel engineering
> https://bootlin.com
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-09 9:55 ` Ola x Nilsson
2022-12-09 12:35 ` Richard Purdie
2022-12-09 13:09 ` Alexandre Belloni
@ 2022-12-09 16:47 ` Khem Raj
2 siblings, 0 replies; 22+ messages in thread
From: Khem Raj @ 2022-12-09 16:47 UTC (permalink / raw)
To: Ola x Nilsson; +Cc: openembedded-core@lists.openembedded.org
On Fri, Dec 9, 2022 at 2:31 AM Ola x Nilsson <ola.x.nilsson@axis.com> wrote:
>
>
> On Fri, Dec 09 2022, Khem Raj wrote:
>
> > Thanks for the patches
> >
> > On 12/7/22 23:11, Ola x Nilsson wrote:
> >> Signed-off-by: Ola x Nilsson <olani@axis.com>
> >> ---
> >> meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> >> 1 file changed, 23 insertions(+)
> >> create mode 100644 meta/conf/distro/time64.conf
> >>
> >> diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> >> new file mode 100644
> >> index 0000000000..99eb06dc0f
> >> --- /dev/null
> >> +++ b/meta/conf/distro/time64.conf
> >> @@ -0,0 +1,23 @@
> >> +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
> >> +
> >> +# TODO: Only for 32-bit architectures?
> >> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> >> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> >> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
> >
> > We should enable it across all 32bit systems
> > something like
> >
> > TARGET_CC_ARCH += "${@'-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64' if
> > d.getVar('SITEINFO_BITS') == '32' else ''}"
> >
> > would do it.
>
> I was looking for a way to identify 32-bit platforms but did not find
> SITEINFO_BITS. I'll look into using that instead.
>
> > Also adding -D_LARGEFILE_SOURCE will ensure needed functions that LFS
> > needs is enabled as well, usually its automatically defined when
> > -D_GNU_SOURCE is set but that may not be case all the time.
>
> As far as I can tell, _LARGEFILE_SOURCE and _LARGEFILE64_SOURCE are
> independent of _TIME_BITS=64 and _FILE_OFFSET_BITS=64 (while
> _TIME_BITS=64 requiress _FILE_OFFSET_BITS=64).
>
> I thought the _LARGEFILE* options were controlled by the largefile
> DISTRO_FEATURE but there seems to be some confusion about that.
>
this was transitional it was removed because we wanted to enable it
globally by default
however, we might want to add --enable-largefile unconditionally to
autotools.bbclass
for some autotooled packages which still need it for others we have to define
-D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64, in this case 64bit off_t
is common need between both LFS and 64bit time_t
Also see this discussion on same topic
https://sourceware.org/pipermail/libc-alpha/2022-November/143386.html
> It looks like you (Khem) made largefile default instead of using the
> distro feature in all of the OE-core recipes, and the largefile distro
> feature is no longer mentioned in docs. It is mentioned as removed in
> the transition notes. But it is still in poky-tiny.conf and
> POKY_DEFAULT_DISTRO_FEATURES. Probably an oversight?
yes seems so, it should be removed from there.
>
> I do not see the _LARGEFILE*_SOURCE options defined globally anywhere.
>
> Am I missing something?
most of packages are enabling it indirectly via enabling _GNU_SOURCE macro
on glibc systems. So we are fine but it will be better to be explicit about it.
>
> >> +
> >> +GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
> >> +GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
> >> +# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
> >> +# both 32 and 64 bit file APIs. But it does not handle the time side?
> >> +# Needs further investigation
> >> +GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
> >> +GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
> >> +
> >> +INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so]
> >> = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl
> >> lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
> >> +INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
> >> +INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64
> >> setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm
> >> fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime
> >> localtime_r"
> >> +
> >> +# libpulsedsp.so is a preload-library that hooks libc functions
> >> +INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
> >> +
> >> +
> >>
>
> --
> Ola x Nilsson
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-08 7:11 ` [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags Ola x Nilsson
2022-12-08 10:00 ` [OE-core] " Alexander Kanavin
2022-12-09 0:23 ` Khem Raj
@ 2022-12-15 16:32 ` Khem Raj
2022-12-17 11:15 ` Richard Purdie
2022-12-16 18:44 ` Khem Raj
3 siblings, 1 reply; 22+ messages in thread
From: Khem Raj @ 2022-12-15 16:32 UTC (permalink / raw)
To: Ola x Nilsson; +Cc: openembedded-core
On Wed, Dec 7, 2022 at 11:11 PM Ola x Nilsson <olani@axis.com> wrote:
>
> Signed-off-by: Ola x Nilsson <olani@axis.com>
> ---
> meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> 1 file changed, 23 insertions(+)
> create mode 100644 meta/conf/distro/time64.conf
>
> diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> new file mode 100644
> index 0000000000..99eb06dc0f
> --- /dev/null
> +++ b/meta/conf/distro/time64.conf
I think this file should be called time64.inc to make it similar to
other distro fragments that distros can use.
> @@ -0,0 +1,23 @@
> +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
Also add -D_LARGEFILE_SOURCE here too so we have all dependent changes
in one place
> +
> +# TODO: Only for 32-bit architectures?
> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
I think it should be made generic across all 32bit arches as I
mentioned last time.
> +
> +GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
> +GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
> +# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
> +# both 32 and 64 bit file APIs. But it does not handle the time side?
> +# Needs further investigation
> +GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
> +GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
> +
> +INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so] = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
> +INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
> +INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64 setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime localtime_r"
> +
> +# libpulsedsp.so is a preload-library that hooks libc functions
> +INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
> +
Earlier I thought we won't need it on musl. Ideally that's true, but
many applications do rely on _FILE_OFFSET_BITS=64 so I think
we will need to include it on musl too even though these defines wont
means much from libc point of view.
> +
> --
> 2.30.2
>
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#174389): https://lists.openembedded.org/g/openembedded-core/message/174389
> Mute This Topic: https://lists.openembedded.org/mt/95533494/1997914
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [raj.khem@gmail.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-08 7:11 ` [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags Ola x Nilsson
` (2 preceding siblings ...)
2022-12-15 16:32 ` Khem Raj
@ 2022-12-16 18:44 ` Khem Raj
3 siblings, 0 replies; 22+ messages in thread
From: Khem Raj @ 2022-12-16 18:44 UTC (permalink / raw)
To: Ola x Nilsson; +Cc: openembedded-core
I ran a build with time64 enabled on meta-openembedded + few more
layers enabled on qemuarm/clang using yoe distro settings.
there are some failures as below.
https://errors.yoctoproject.org/Errors/Build/156858/
On Wed, Dec 7, 2022 at 11:11 PM Ola x Nilsson <olani@axis.com> wrote:
>
> Signed-off-by: Ola x Nilsson <olani@axis.com>
> ---
> meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> 1 file changed, 23 insertions(+)
> create mode 100644 meta/conf/distro/time64.conf
>
> diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> new file mode 100644
> index 0000000000..99eb06dc0f
> --- /dev/null
> +++ b/meta/conf/distro/time64.conf
> @@ -0,0 +1,23 @@
> +GLIBC_64BIT_TIME_FLAGS = "-D_TIME_BITS=64 -D_FILE_OFFSET_BITS=64"
> +
> +# TODO: Only for 32-bit architectures?
> +TARGET_CC_ARCH:append:arm = " ${GLIBC_64BIT_TIME_FLAGS}"
> +TARGET_CC_ARCH:append:armeb = " ${GLIBC_64BIT_TIME_FLAGS}"
> +TARGET_CC_ARCH:append:mips32el = " ${GLIBC_64BIT_TIME_FLAGS}"
> +
> +GLIBC_64BIT_TIME_FLAGS:pn-glibc = ""
> +GLIBC_64BIT_TIME_FLAGS:pn-glibc-tests = ""
> +# pipewire-v4l2 explicitly sets _FILE_OFFSET_BITS=32 to get access to
> +# both 32 and 64 bit file APIs. But it does not handle the time side?
> +# Needs further investigation
> +GLIBC_64BIT_TIME_FLAGS:pn-pipewire = ""
> +GLIBC_64BIT_TIME_FLAGS:pn-gcc-sanitizers = ""
> +
> +INSANE_SKIP:libstd-rs[_usr_lib_rustlib_armv7-poky-linux-gnueabihf_lib_libstd.so] = "clock_gettime gettime fcntl fstat64 fstatat64 getsockopt ioctl lstat64 nanosleep prctl recvmsg sendmsg setsockopt stat64"
> +INSANE_SKIP:librsvg[_usr_bin_rsvg-convert] = "fcntl fstat64 prctl stat64 clock_gettime"
> +INSANE_SKIP:librsvg[_usr_lib_librsvg-2.so.2.48.0] = "fcntl lstat64 setsockopt sendmsg fstat64 getsockopt ioctl nanosleep timegm fstatat64 prctl mktime gmtime_r recvmsg stat64 clock_gettime localtime_r"
> +
> +# libpulsedsp.so is a preload-library that hooks libc functions
> +INSANE_SKIP:pulseaudio[_usr_lib_pulseaudio_libpulsedsp.so] = "setsockopt fcntl"
> +
> +
> --
> 2.30.2
>
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#174389): https://lists.openembedded.org/g/openembedded-core/message/174389
> Mute This Topic: https://lists.openembedded.org/mt/95533494/1997914
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [raj.khem@gmail.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags
2022-12-15 16:32 ` Khem Raj
@ 2022-12-17 11:15 ` Richard Purdie
0 siblings, 0 replies; 22+ messages in thread
From: Richard Purdie @ 2022-12-17 11:15 UTC (permalink / raw)
To: Khem Raj, Ola x Nilsson; +Cc: openembedded-core
On Thu, 2022-12-15 at 08:32 -0800, Khem Raj wrote:
> On Wed, Dec 7, 2022 at 11:11 PM Ola x Nilsson <olani@axis.com> wrote:
> >
> > Signed-off-by: Ola x Nilsson <olani@axis.com>
> > ---
> > meta/conf/distro/time64.conf | 23 +++++++++++++++++++++++
> > 1 file changed, 23 insertions(+)
> > create mode 100644 meta/conf/distro/time64.conf
> >
> > diff --git a/meta/conf/distro/time64.conf b/meta/conf/distro/time64.conf
> > new file mode 100644
> > index 0000000000..99eb06dc0f
> > --- /dev/null
> > +++ b/meta/conf/distro/time64.conf
>
> I think this file should be called time64.inc to make it similar to
> other distro fragments that distros can use.
Agreed, it also needs to be in the include directory as it isn't a
distro called "time64". I've merged this patch but I moved the file to
distro/include/time64.inc. We can improve it from there with further
patches.
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 2/3] Add QA check for 32 bit time and file offset functions
2022-12-08 7:11 ` [PATCH RFC 2/3] Add QA check for 32 bit time and file offset functions Ola x Nilsson
@ 2022-12-17 11:17 ` Richard Purdie
2022-12-17 17:19 ` Khem Raj
0 siblings, 1 reply; 22+ messages in thread
From: Richard Purdie @ 2022-12-17 11:17 UTC (permalink / raw)
To: Ola x Nilsson, openembedded-core
On Thu, 2022-12-08 at 08:11 +0100, Ola x Nilsson wrote:
> Check for known symbols that should have been redirected to 64bit
> variants when -D_FILE_OFFSET_BITS=64 and -D_TIME_BITS=64 are set.
>
> Signed-off-by: Ola x Nilsson <olani@axis.com>
> ---
> meta/classes-global/insane.bbclass | 132 +++++++++++++++++++++++++++++
> 1 file changed, 132 insertions(+)
>
> diff --git a/meta/classes-global/insane.bbclass b/meta/classes-global/insane.bbclass
> index df2c40c3c5..b0e99c4c3f 100644
> --- a/meta/classes-global/insane.bbclass
> +++ b/meta/classes-global/insane.bbclass
> @@ -506,6 +506,138 @@ def package_qa_check_symlink_to_sysroot(path, name, d, elf, messages):
> trimmed = path.replace(os.path.join (d.getVar("PKGDEST"), name), "")
> oe.qa.add_message(messages, "symlink-to-sysroot", "Symlink %s in %s points to TMPDIR" % (trimmed, name))
>
> +QAPATHTEST[32bit-time] = "check_32bit_symbols"
> +def check_32bit_symbols(path, packagename, d, elf, messages):
> + """
> + Check that ELF files do not use any 32 bit time APIs from glibc.
> + """
> + # TODO: We should have some DISTRO_FEATURE flag to check here
> + # if '-D_TIME_BITS=64' not in d.getVar('GLIBC_64BIT_TIME_FLAGS'):
> + # # No use checking if the feature is not enabled?
> + # return
I did merge this patch with two fixes:
a) It needed the class name in the subject line so we know which code
the patch changes at a glance
b) I removed the above TODO since I don't think we should be adding a
distro feature for this, it can be handled by including a the include
file from the distro and setting WARN_QA and ERROR_QA as the distro
wishes.
Cheers,
Richard
^ permalink raw reply [flat|nested] 22+ messages in thread
* Re: [OE-core] [PATCH RFC 2/3] Add QA check for 32 bit time and file offset functions
2022-12-17 11:17 ` [OE-core] " Richard Purdie
@ 2022-12-17 17:19 ` Khem Raj
0 siblings, 0 replies; 22+ messages in thread
From: Khem Raj @ 2022-12-17 17:19 UTC (permalink / raw)
To: Richard Purdie; +Cc: Ola x Nilsson, openembedded-core
On Sat, Dec 17, 2022 at 3:17 AM Richard Purdie
<richard.purdie@linuxfoundation.org> wrote:
>
> On Thu, 2022-12-08 at 08:11 +0100, Ola x Nilsson wrote:
> > Check for known symbols that should have been redirected to 64bit
> > variants when -D_FILE_OFFSET_BITS=64 and -D_TIME_BITS=64 are set.
> >
> > Signed-off-by: Ola x Nilsson <olani@axis.com>
> > ---
> > meta/classes-global/insane.bbclass | 132 +++++++++++++++++++++++++++++
> > 1 file changed, 132 insertions(+)
> >
> > diff --git a/meta/classes-global/insane.bbclass b/meta/classes-global/insane.bbclass
> > index df2c40c3c5..b0e99c4c3f 100644
> > --- a/meta/classes-global/insane.bbclass
> > +++ b/meta/classes-global/insane.bbclass
> > @@ -506,6 +506,138 @@ def package_qa_check_symlink_to_sysroot(path, name, d, elf, messages):
> > trimmed = path.replace(os.path.join (d.getVar("PKGDEST"), name), "")
> > oe.qa.add_message(messages, "symlink-to-sysroot", "Symlink %s in %s points to TMPDIR" % (trimmed, name))
> >
> > +QAPATHTEST[32bit-time] = "check_32bit_symbols"
> > +def check_32bit_symbols(path, packagename, d, elf, messages):
> > + """
> > + Check that ELF files do not use any 32 bit time APIs from glibc.
> > + """
> > + # TODO: We should have some DISTRO_FEATURE flag to check here
> > + # if '-D_TIME_BITS=64' not in d.getVar('GLIBC_64BIT_TIME_FLAGS'):
> > + # # No use checking if the feature is not enabled?
> > + # return
>
> I did merge this patch with two fixes:
>
> a) It needed the class name in the subject line so we know which code
> the patch changes at a glance
>
> b) I removed the above TODO since I don't think we should be adding a
> distro feature for this, it can be handled by including a the include
> file from the distro and setting WARN_QA and ERROR_QA as the distro
> wishes.
Right, I think once it's sorted out and is working well, we should
probably include it
in poky distro conf and/or maybe even in defaultsetup.conf eventually.
>
> Cheers,
>
> Richard
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#174772): https://lists.openembedded.org/g/openembedded-core/message/174772
> Mute This Topic: https://lists.openembedded.org/mt/95533492/1997914
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [raj.khem@gmail.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>
^ permalink raw reply [flat|nested] 22+ messages in thread
end of thread, other threads:[~2022-12-17 17:19 UTC | newest]
Thread overview: 22+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2022-12-08 7:11 Y2038: Sanity checker for non-64-bit time and file functions Ola x Nilsson
2022-12-08 7:11 ` [PATCH RFC 1/3] glibc: Add ppoll fortify symbol for 64 bit time_t Ola x Nilsson
2022-12-08 7:11 ` [PATCH RFC 2/3] Add QA check for 32 bit time and file offset functions Ola x Nilsson
2022-12-17 11:17 ` [OE-core] " Richard Purdie
2022-12-17 17:19 ` Khem Raj
2022-12-08 7:11 ` [PATCH RFC 3/3] time64.conf: Include to enable 64 bit time flags Ola x Nilsson
2022-12-08 10:00 ` [OE-core] " Alexander Kanavin
2022-12-08 10:10 ` Richard Purdie
2022-12-09 10:31 ` Ola x Nilsson
2022-12-09 10:36 ` Ola x Nilsson
2022-12-09 13:06 ` Alexandre Belloni
2022-12-09 0:23 ` Khem Raj
2022-12-09 9:55 ` Ola x Nilsson
2022-12-09 12:35 ` Richard Purdie
2022-12-09 13:09 ` Alexandre Belloni
2022-12-09 13:13 ` Alexander Kanavin
2022-12-09 13:29 ` Richard Purdie
2022-12-09 16:36 ` Khem Raj
2022-12-09 16:47 ` Khem Raj
2022-12-15 16:32 ` Khem Raj
2022-12-17 11:15 ` Richard Purdie
2022-12-16 18:44 ` Khem Raj
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.