kernel/linux.git/arch/s390/include/asm/bitops.h, branch v7.1

Merge tag 's390-6.18-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

2025-10-09T17:51:43+00:00

Pull more s390 updates from Alexander Gordeev: - Compile the decompressor with -Wno-pointer-sign flag to avoid a clang warning - Fix incomplete conversion to flag output macros in __xsch(), to avoid always zero return value instead of the expected condition code - Remove superfluous newlines from inline assemblies to improve compiler inlining decisions - Expose firmware provided UID Checking state in sysfs regardless of the device presence or state - CIO does not unregister subchannels when the attached device is invalid or unavailable. Update the purge function to remove I/O subchannels if the device number is found on cio_ignore list - Consolidate PAI crypto allocation and cleanup paths - The uv_get_secret_metadata() function has been removed some few months ago, remove also the function mention it in a comment * tag 's390-6.18-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: s390/uv: Fix comment of uv_find_secret() function s390/pai_crypto: Consolidate PAI crypto allocation and cleanup paths s390/cio: Update purge function to unregister the unused subchannels s390/pci: Expose firmware provided UID Checking state in sysfs s390: Remove superfluous newlines from inline assemblies s390/cio/ioasm: Fix __xsch() condition code handling s390: Add -Wno-pointer-sign to KBUILD_CFLAGS_DECOMPRESSOR

Merge tag 's390-6.18-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

2025-09-30T02:14:25+00:00

Pull s390 updates from Alexander Gordeev: - Refactor SCLP memory hotplug code - Introduce common boot_panic() decompressor helper macro and use it to get rid of nearly few identical implementations - Take into account additional key generation flags and forward it to the ep11 implementation. With that allow users to modify the key generation process, e.g. provide valid combinations of XCP_BLOB_* flags - Replace kmalloc() + copy_from_user() with memdup_user_nul() in s390 debug facility and HMC driver - Add DAX support for DCSS memory block devices - Make the compiler statement attribute "assume" available with a new __assume macro - Rework ffs() and fls() family bitops functions, including source code improvements and generated code optimizations. Use the newly introduced __assume macro for that - Enable additional network features in default configurations - Use __GFP_ACCOUNT flag for user page table allocations to add missing kmemcg accounting - Add WQ_PERCPU flag to explicitly request the use of the per-CPU workqueue for 3590 tape driver - Switch power reading to the per-CPU and the Hiperdispatch to the default workqueue - Add memory allocation profiling hooks to allow better profiling data and the /proc/allocinfo output similar to other architectures * tag 's390-6.18-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (21 commits) s390/mm: Add memory allocation profiling hooks s390: Replace use of system_wq with system_dfl_wq s390/diag324: Replace use of system_wq with system_percpu_wq s390/tape: Add WQ_PERCPU to alloc_workqueue users s390/bitops: Switch to generic ffs() if supported by compiler s390/bitops: Switch to generic fls(), fls64(), etc. s390/mm: Use __GFP_ACCOUNT for user page table allocations s390/configs: Enable additional network features s390/bitops: Cleanup __flogr() s390/bitops: Use __assume() for __flogr() inline assembly return value compiler_types: Add __assume macro s390/bitops: Limit return value range of __flogr() s390/dcssblk: Add DAX support s390/hmcdrv: Replace kmalloc() + copy_from_user() with memdup_user_nul() s390/debug: Replace kmalloc() + copy_from_user() with memdup_user_nul() s390/pkey: Forward keygenflags to ep11_unwrapkey s390/boot: Add common boot_panic() code s390/bitops: Optimize inlining s390/bitops: Slightly optimize ffs() and fls64() s390/sclp: Move memory hotplug code for better modularity ...

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/bitops: Switch to generic ffs() if supported by compiler

2025-09-24T14:24:18+00:00

Use generic ffs() / __builtin_ffs() if supported by the compiler. GCC 16 will have support for __builtin_ffs(). See gcc commit f50cff9766c5 ("s390: Implement clz and ctz for SI mode"). In the distant future when GCC 16 becomes the minimum supported version, this allows to get rid of the flogr inline assembly. Kernel image size is reduced by ~500 bytes (gcc 16 beta + defconfig). Acked-by: Alexander Gordeev Signed-off-by: Heiko Carstens Signed-off-by: Alexander Gordeev

s390/bitops: Switch to generic fls(), fls64(), etc.

2025-09-24T14:24:17+00:00

Switch to generic fls(), fls64(), etc. which are implemented with __builtin_ctzl(), __builtin_clzl(). Those builtins are available for all supported compilers. Kernel image size is reduced by ~10kb (gcc 15.1.0 + defconfig). Acked-by: Alexander Gordeev Signed-off-by: Heiko Carstens Signed-off-by: Alexander Gordeev

s390/bitops: Cleanup __flogr()

2025-09-18T12:06:41+00:00

The flogr() inline assembly has no side effects and generates the same output if the input does not change. Therefore remove the volatile qualifier to allow the compiler to optimize the inline assembly away, if possible. Also remove the superfluous '\n' which makes the inline assembly appear larger than it is according to compiler heuristics (number of lines). Furthermore change the return type of flogr() to unsigned long and add the const attribute to the function. This reduces the kernel image size by 994 bytes (defconfig, gcc 15.2.0). Suggested-by: Juergen Christ Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens Signed-off-by: Alexander Gordeev

s390/bitops: Use assume() for flogr() inline assembly return value

2025-09-18T12:06:40+00:00

Use __assume() to tell compilers that the output operand of the __flogr() inline assembly contains a value in the range of 0..64. This allows to optimize the logical AND operation away. This reduces the kernel image size by 2804 bytes (defconfig, gcc 15.2.0). Suggested-by: Juergen Christ Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens Signed-off-by: Alexander Gordeev

s390/bitops: Limit return value range of __flogr()

2025-09-18T12:06:40+00:00

With the recent ffs() and ffs64() optimization a logical AND operation was removed, which allowed the compiler to tell the return value range of both functions. This may lead to compile warnings as reported by the kernel test robot: drivers/infiniband/hw/mlx5/mr.c: In function 'mlx5r_cache_create_ent_locked': >> drivers/infiniband/hw/mlx5/mr.c:840:31: warning: 'sprintf' may write a terminating nul past the end of the destination [-Wformat-overflow=] 840 | sprintf(ent->name, "%d", order); | ^ In function 'mlx5_mkey_cache_debugfs_add_ent', inlined from 'mlx5r_cache_create_ent_locked' at drivers/infiniband/hw/mlx5/mr.c:930:3: drivers/infiniband/hw/mlx5/mr.c:840:9: note: 'sprintf' output between 2 and 5 bytes into a destination of size 4 840 | sprintf(ent->name, "%d", order); | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Add the AND operation again to address the warning. From a correctness point of view the AND operation is not necessary, however there is no other way to tell the compiler that the returned value of the flogr inline assembly is in the range of 0..64. This increases the kernel image size by 566 bytes (defconfig, gcc 15.2.0). Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-kbuild-all/202508211859.UoYsJbLN-lkp@intel.com/ Fixes: de88e74889a3 ("s390/bitops: Slightly optimize ffs() and fls64()") Suggested-by: Juergen Christ Reviewed-by: Juergen Christ Signed-off-by: Heiko Carstens Signed-off-by: Alexander Gordeev

s390: Add __attribute_const__ to ffs()-family implementations

2025-09-08T21:58:52+00:00

While tracking down a problem where constant expressions used by BUILD_BUG_ON() suddenly stopped working[1], we found that an added static initializer was convincing the compiler that it couldn't track the state of the prior statically initialized value. Tracing this down found that ffs() was used in the initializer macro, but since it wasn't marked with __attribute__const__, the compiler had to assume the function might change variable states as a side-effect (which is not true for ffs(), which provides deterministic math results). Add missing __attribute_const__ annotations to S390's implementations of ffs(), __ffs(), fls(), and __fls() functions. These are pure mathematical functions that always return the same result for the same input with no side effects, making them eligible for compiler optimization. Build tested ARCH=s390 defconfig with GCC s390x-linux-gnu 14.2.0. Link: https://github.com/KSPP/linux/issues/364 [1] Acked-by: Heiko Carstens Acked-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/r/20250804164417.1612371-14-kees@kernel.org Signed-off-by: Kees Cook

s390/bitops: Optimize inlining

2025-08-20T14:38:24+00:00

GCC inlining heuristics prevent code growth due to inlining into cold paths. This causes GCC to emit a partially specialized version of __flogr for non-constant input for all occurrences on cold paths. This happens since the overhead seen during inlining includes setting up a union register_pair, calling flogr, and extracting and casting the result. This overhead is not removed until the function is lowered into RTL. But this happens after inlining. For -ftrivial-var-auto-init=zero builds, an additional initialization of the union register_pair adds another statement to be inlinined. This is unneeded since the even register is initialized anyway and the odd register is not an input register. It is only marked as such since the whole pair has to be marked as a read/write output register. Mark the union register_pair as uninitialized to get rid of this statement. This, however, does not change the code since the initialization happens when part of the register pair is written. Nevertheless, GCC function size approximation during inlining is reduced by one statement. Force inlining of flogr and also flatten some other functions that should be leaf functions but are called in cold context, like, e.g., __init functions. Acked-by: Heiko Carstens Signed-off-by: Juergen Christ Signed-off-by: Alexander Gordeev