kernel/linux.git/arch/s390/include/asm/atomic_ops.h, branch linux-7.0.y

s390: Remove superfluous newlines from inline assemblies

2025-09-29T11:52:08+00:00

Remove superfluous newlines from inline assemblies. Compilers use the number of lines of inline assemblies as heuristic for the complexity and inline decisions. Therefore inline assemblies should only contain as many lines as required. A lot of inline assemblies contain a superfluous newline for the last line. Remove such newlines to improve compiler inlining decisions. Suggested-by: Juergen Christ Signed-off-by: Heiko Carstens Reviewed-by: Alexander Gordeev Reviewed-by: Juergen Christ Signed-off-by: Alexander Gordeev

s390/atomic_ops: Let __atomic_add_const() variants always return void

2025-03-04T16:34:04+00:00

Depending on MARCH_HAS_Z196_FEATURES __atomic_add_const() returns void or the previous value before the atomic variant. Make sure that for both cases void is returned so potential incorrect usage results in both cases in a compile error. Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens Signed-off-by: Vasily Gorbik

s390/atomic: Provide arch_atomic_*_and_test() implementations

2024-12-15T15:19:03+00:00

Provide arch_atomic_*_and_test() implementations which make use of flag output constraints, and allow the compiler to generate slightly better code. Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens Signed-off-by: Alexander Gordeev

s390: Remove superfluous new lines from inline assemblies

2024-12-15T15:19:03+00:00

GCC uses the number of lines of an inline assembly to calculate its length (number of instructions). This has an impact on GCCs inlining decisions. Therefore remove superfluous new lines from a couple of inline assemblies, so that their real size is reflected. Also use an "asm inline" statement for the fpu_lfpc_safe() inline assembly to enforce that GCC assumes the minimum size for this inline assembly, since it contains various statements which make it appear much larger than the resulting code is. Suggested-by: Juergen Christ Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens Signed-off-by: Alexander Gordeev

s390/atomic: Consistent layering between atomic.h and atomic_ops.h

2024-12-15T14:13:43+00:00

With commit c8a91c285d8c ("s390/atomic: move remaining inline assemblies to atomic_ops.h") all remaining atomic inline assemblies have been moved to atomic_ops.h. However the result is inconsistent: the functions in atomic_ops.h are supposed to be used with integral types like int and long pointers, while the functions in atomic.h work with atomic types. This layering got violated with the named commit. Therefore adjust this now, and also use consistent variable names in atomic_ops.h. Reviewed-by: Alexander Gordeev Signed-off-by: Heiko Carstens Signed-off-by: Alexander Gordeev

s390/atomic: Remove __atomic_cmpxchg() variants

2024-11-12T13:01:29+00:00

With users converted to the standard arch_cmpxchg() variants, remove the now unused __atomic_cmpxchg() and __atomic_cmpxchg_bool() variants. Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens

s390: Use MARCH_HAS_*_FEATURES defines

2024-09-07T15:12:42+00:00

Replace CONFIG_HAVE_MARCH_*_FEATURES with MARCH_HAS_*_FEATURES everywhere so code gets compiled correctly depending on if the target is the kernel or the decompressor. Reviewed-by: Sven Schnelle Signed-off-by: Heiko Carstens

s390/atomic_ops: Disable flag outputs constraint for GCC versions below 14.2.0

2024-07-23T14:02:31+00:00

GCC may die with an ICE if the flag outputs constraint is used in combination with other inline assemblies. This will be fixed with GCC 14.2.0. Therefore disable the use of the constraint for now. Link: https://gcc.gnu.org/git?p=gcc.git;a=commit;h=cd11413ff7c4353a3e336db415304f788d23a393 Signed-off-by: Heiko Carstens Signed-off-by: Sven Schnelle Signed-off-by: Vasily Gorbik

s390/atomic_ops: Make use of flag output constraint

2024-07-10T17:50:44+00:00

With gcc 14.1.0 support for flag output constraint was added for s390. Use this for __atomic_cmpxchg_bool(). This allows for slightly better code, since the compiler can generate code depending on the condition code which is the result of an inline assembly. The size of the kernel image is reduced by ~12kb. Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens Signed-off-by: Vasily Gorbik

s390/atomic_ops: Improve __atomic_set() for small values

2024-07-10T17:50:44+00:00

Use mvhi/mvghi for small constant values within the __atomic_set() inline assemblies. This avoids loading the specified value into a register. The size of the kernel image is reduced by ~1.2kb. Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens Signed-off-by: Vasily Gorbik