linux/arch/s390/kernel/vdso, branch master

Merge tag 'timers-vdso-2026-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

2026-02-11T01:02:23Z

Pull VDSO updates from Thomas Gleixner: - Provide the missing 64-bit variant of clock_getres() This allows the extension of CONFIG_COMPAT_32BIT_TIME to the vDSO and finally the removal of 32-bit time types from the kernel and UAPI. - Remove the useless and broken getcpu_cache from the VDSO The intention was to provide a trivial way to retrieve the CPU number from the VDSO, but as the VDSO data is per process there is no way to make it work. - Switch get/put_unaligned() from packed struct to memcpy() The packed struct violates strict aliasing rules which requires to pass -fno-strict-aliasing to the compiler. As this are scalar values __builtin_memcpy() turns them into simple loads and stores - Use __typeof_unqual__() for __unqual_scalar_typeof() The get/put_unaligned() changes triggered a new sparse warning when __beNN types are used with get/put_unaligned() as sparse builds add a special 'bitwise' attribute to them which prevents sparse to evaluate the Generic in __unqual_scalar_typeof(). Newer sparse versions support __typeof_unqual__() which avoids the problem, but requires a recent sparse install. So this adds a sanity check to sparse builds, which validates that sparse is available and capable of handling it. - Force inline __cvdso_clock_getres_common() Compilers sometimes un-inline agressively, which results in function call overhead and problems with automatic stack variable initialization. Interestingly enough the force inlining results in smaller code than the un-inlined variant produced by GCC when optimizing for size. * tag 'timers-vdso-2026-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: vdso/gettimeofday: Force inlining of __cvdso_clock_getres_common() x86/percpu: Make CONFIG_USE_X86_SEG_SUPPORT work with sparse compiler: Use __typeof_unqual__() for __unqual_scalar_typeof() powerpc/vdso: Provide clock_getres_time64() tools headers: Remove unneeded ignoring of warnings in unaligned.h tools headers: Update the linux/unaligned.h copy with the kernel sources vdso: Switch get/put_unaligned() from packed struct to memcpy() parisc: Inline a type punning version of get_unaligned_le32() vdso: Remove struct getcpu_cache MIPS: vdso: Provide getres_time64() for 32-bit ABIs arm64: vdso32: Provide clock_getres_time64() ARM: VDSO: Provide clock_getres_time64() ARM: VDSO: Patch out __vdso_clock_getres() if unavailable x86/vdso: Provide clock_getres_time64() for x86-32 selftests: vDSO: vdso_test_abi: Add test for clock_getres_time64() selftests: vDSO: vdso_test_abi: Use UAPI system call numbers selftests: vDSO: vdso_config: Add configurations for clock_getres_time64() vdso: Add prototype for __vdso_clock_getres_time64()

s390/vdso: Disable kstack erase

2026-01-17T14:51:28Z

For some reason gcc 8, 9, 10, and 11 generate a dynamic relocation in vdso.so.dbg if CONFIG_KSTACK_ERASE is enabled: >> arch/s390/kernel/vdso/vdso.so.dbg: dynamic relocations are not supported make[3]: *** [arch/s390/kernel/vdso/Makefile:54: arch/s390/kernel/vdso/vdso.so.dbg] Error 1 $ readelf -rW arch/s390/kernel/vdso/vdso.so.dbg Relocation section '.rela.dyn' at offset 0x15c0 contains 1 entry: Offset Info Type Symbol's Value Symbol's Name + Addend 00000000000015f0 000000010000000b R_390_JMP_SLOT 0000000000000000 __sanitizer_cov_stack_depth + 0 Add $(DISABLE_KSTACK_ERASE) to vdso compile flags to fix this. Reported-by: kernel test robot Closes: https://lore.kernel.org/r/202601070505.xQcLr5KV-lkp@intel.com/ Signed-off-by: Heiko Carstens

vdso: Remove struct getcpu_cache

2026-01-14T07:56:40Z

The cache parameter of getcpu() is useless nowadays for various reasons. * It is never passed by userspace for either the vDSO or syscalls. * It is never used by the kernel. * It could not be made to work on the current vDSO architecture. * The structure definition is not part of the UAPI headers. * vdso_getcpu() is superseded by restartable sequences in any case. Remove the struct and its header. As a side-effect this gets rid of an unwanted inclusion of the linux/ header namespace from vDSO code. [ tglx: Adapt to s390 upstream changes */ Signed-off-by: Thomas Weißschuh Signed-off-by: Thomas Gleixner Acked-by: Arnd Bergmann Acked-by: Heiko Carstens # s390 Link: https://patch.msgid.link/20251230-getcpu_cache-v3-1-fb9c5f880ebe@linutronix.de

s390/vdso: Get rid of -m64 flag handling

2025-11-25T14:28:08Z

The compiler/assembler flag -m64 is added and removed at two locations. This pointless exercise is a leftover to keep the 31 and 64 bit vdso Makefiles as symmetrical as possible. Given that the 31 bit vdso code does not exist anymore, remove the -m64 flag handling. Suggested-by: Jens Remus Reviewed-by: Jens Remus Signed-off-by: Heiko Carstens

s390/vdso: Rename vdso64 to vdso

2025-11-25T14:28:07Z

Since compat is gone there is only a 64 bit vdso left. Remove the superfluous "64" suffix everywhere. Reviewed-by: Jens Remus Signed-off-by: Heiko Carstens