user/sven/linux.git/include/linux/cleanup.h, branch v6.15.1

Revert "Merge tag 'irq-msi-2025-03-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip"

2025-03-28T18:22:54Z

This reverts commit 36f5f026df6c1cd8a20373adc4388d2b3401ce91, reversing changes made to 43a7eec035a5b64546c8adefdc9cf96a116da14b. Thomas says: "I just noticed that for some incomprehensible reason, probably sheer incompetemce when trying to utilize b4, I managed to merge an outdated _and_ buggy version of that series. Can you please revert that merge completely?" Done. Requested-by: Thomas Gleixner Signed-off-by: Linus Torvalds

Merge tag 'timers-core-2025-03-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

2025-03-25T17:33:23Z

Pull timer core updates from Thomas Gleixner: - Fix a memory ordering issue in posix-timers Posix-timer lookup is lockless and reevaluates the timer validity under the timer lock, but the update which validates the timer is not protected by the timer lock. That allows the store to be reordered against the initialization stores, so that the lookup side can observe a partially initialized timer. That's mostly a theoretical problem, but incorrect nevertheless. - Fix a long standing inconsistency of the coarse time getters The coarse time getters read the base time of the current update cycle without reading the actual hardware clock. NTP frequency adjustment can set the base time backwards. The fine grained interfaces compensate this by reading the clock and applying the new conversion factor, but the coarse grained time getters use the base time directly. That allows the user to observe time going backwards. Cure it by always forwarding base time, when NTP changes the frequency with an immediate step. - Rework of posix-timer hashing The posix-timer hash is not scalable and due to the CRIU timer restore mechanism prone to massive contention on the global hash bucket lock. Replace the global hash lock with a fine grained per bucket locking scheme to address that. - Rework the proc/$PID/timers interface. /proc/$PID/timers is provided for CRIU to be able to restore a timer. The printout happens with sighand lock held and interrupts disabled. That's not required as this can be done with RCU protection as well. - Provide a sane mechanism for CRIU to restore a timer ID CRIU restores timers by creating and deleting them until the kernel internal per process ID counter reached the requested ID. That's horribly slow for sparse timer IDs. Provide a prctl() which allows CRIU to restore a timer with a given ID. When enabled the ID pointer is used as input pointer to read the requested ID from user space. When disabled, the normal allocation scheme (next ID) is active as before. This is backwards compatible for both kernel and user space. - Make hrtimer_update_function() less expensive. The sanity checks are valuable, but expensive for high frequency usage in io/uring. Make the debug checks conditional and enable them only when lockdep is enabled. - Small updates, cleanups and improvements * tag 'timers-core-2025-03-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (27 commits) selftests/timers: Improve skew_consistency by testing with other clockids timekeeping: Fix possible inconsistencies in _COARSE clockids posix-timers: Drop redundant memset() invocation selftests/timers/posix-timers: Add a test for exact allocation mode posix-timers: Provide a mechanism to allocate a given timer ID posix-timers: Dont iterate /proc/$PID/timers with sighand:: Siglock held posix-timers: Make per process list RCU safe posix-timers: Avoid false cacheline sharing posix-timers: Switch to jhash32() posix-timers: Improve hash table performance posix-timers: Make signal_struct:: Next_posix_timer_id an atomic_t posix-timers: Make lock_timer() use guard() posix-timers: Rework timer removal posix-timers: Simplify lock/unlock_timer() posix-timers: Use guards in a few places posix-timers: Remove SLAB_PANIC from kmem cache posix-timers: Remove a few paranoid warnings posix-timers: Cleanup includes posix-timers: Add cond_resched() to posix_timer_add() search loop posix-timers: Initialise timer before adding it to the hash table ...

Merge tag 'irq-msi-2025-03-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

2025-03-25T16:15:17Z

Pull MSI irq updates from Thomas Gleixner: - Switch the MSI descriptor locking to guards - Replace the broken PCI/TPH implementation, which lacks any form of serialization against concurrent modifications with a properly serialized mechanism in the PCI/MSI core code - Replace the MSI descriptor abuse in the SCSI/UFS Qualcom driver with dedicated driver internal storage * tag 'irq-msi-2025-03-23' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: genirq/msi: Rename msi_[un]lock_descs() scsi: ufs: qcom: Remove the MSI descriptor abuse PCI/TPH: Replace the broken MSI-X control word update PCI/MSI: Provide a sane mechanism for TPH PCI: hv: Switch MSI descriptor locking to guard() PCI/MSI: Switch to MSI descriptor locking to guard() NTB/msi: Switch MSI descriptor locking to lock guard() soc: ti: ti_sci_inta_msi: Switch MSI descriptor locking to guard() genirq/msi: Use lock guards for MSI descriptor locking cleanup: Provide retain_ptr() genirq/msi: Make a few functions static

cleanup: Provide retain_ptr()

2025-03-13T17:57:59Z

In cases where an allocation is consumed by another function, the allocation needs to be retained on success or freed on failure. The code pattern is usually: struct foo *f = kzalloc(sizeof(*f), GFP_KERNEL); struct bar *b; ,,, // Initialize f ... if (ret) goto free; ... bar = bar_create(f); if (!bar) { ret = -ENOMEM; goto free; } ... return 0; free: kfree(f); return ret; This prevents using __free(kfree) on @f because there is no canonical way to tell the cleanup code that the allocation should not be freed. Abusing no_free_ptr() by force ignoring the return value is not really a sensible option either. Provide an explicit macro retain_ptr(), which NULLs the cleanup pointer. That makes it easy to analyze and reason about. Signed-off-by: Thomas Gleixner Reviewed-by: Jonathan Cameron Link: https://lore.kernel.org/all/20250313130321.442025758@linutronix.de

posix-timers: Make lock_timer() use guard()

2025-03-13T11:07:17Z

The lookup and locking of posix timers requires the same repeating pattern at all usage sites: tmr = lock_timer(tiner_id); if (!tmr) return -EINVAL; .... unlock_timer(tmr); Solve this with a guard implementation, which works in most places out of the box except for those, which need to unlock the timer inside the guard scope. Though the only places where this matters are timer_delete() and timer_settime(). In both cases the timer pointer needs to be preserved across the end of the scope, which is solved by storing the pointer in a variable outside of the scope. timer_settime() also has to protect the timer with RCU before unlocking, which obviously can't use guard(rcu) before leaving the guard scope as that guard is cleaned up before the unlock. Solve this by providing the RCU protection open coded. [ tglx: Made it work and added change log ] Signed-off-by: Peter Zijlstra Signed-off-by: Thomas Gleixner Acked-by: Frederic Weisbecker Link: https://lore.kernel.org/all/20250224162103.GD11590@noisy.programming.kicks-ass.net Link: https://lore.kernel.org/all/20250308155624.087465658@linutronix.de

: Allow the passing of both iomem and non-iomem pointers to no_free_ptr()

2025-03-10T19:02:14Z

Calling no_free_ptr() for an __iomem pointer results in Sparse complaining about the types: warning: incorrect type in argument 1 (different address spaces) expected void const volatile *val got void [noderef] __iomem *__val [ The example is from drivers/platform/x86/intel/pmc/core_ssram.c:283 ] The problem is caused by the signature of __must_check_fn() added in: 85be6d842447 ("cleanup: Make no_free_ptr() __must_check") ... to enforce that the return value is always used. Use __force to allow both iomem and non-iomem pointers to be given for no_free_ptr(). Reported-by: kernel test robot Signed-off-by: Ilpo Järvinen Signed-off-by: Ingo Molnar Reviewed-by: Andy Shevchenko Reviewed-by: Dan Williams Cc: "H. Peter Anvin" Cc: Linus Torvalds Link: https://lore.kernel.org/r/20250310122158.20966-1-ilpo.jarvinen@linux.intel.com Closes: https://lore.kernel.org/oe-kbuild-all/202403050547.qnZtuNlN-lkp@intel.com/

headers/cleanup.h: Remove the if_not_guard() facility

2024-12-07T10:15:14Z

Linus noticed that the new if_not_guard() definition is fragile: "This macro generates actively wrong code if it happens to be inside an if-statement or a loop without a block. IOW, code like this: for (iterate-over-something) if_not_guard(a) return -BUSY; looks like will build fine, but will generate completely incorrect code." The reason is that the __if_not_guard() macro is multi-statement, so while most kernel developers expect macros to be simple or at least compound statements - but for __if_not_guard() it is not so: #define __if_not_guard(_name, _id, args...) \ BUILD_BUG_ON(!__is_cond_ptr(_name)); \ CLASS(_name, _id)(args); \ if (!__guard_ptr(_name)(&_id)) To add insult to injury, the placement of the BUILD_BUG_ON() line makes the macro appear to compile fine, but it will generate incorrect code as Linus reported, for example if used within iteration or conditional statements that will use the first statement of a macro as a loop body or conditional statement body. [ I'd also like to note that the original submission by David Lechner did not contain the BUILD_BUG_ON() line, so it was safer than what we ended up committing. Mea culpa. ] It doesn't appear to be possible to turn this macro into a robust single or compound statement that could be used in single statements, due to the necessity to define an auto scope variable with an open scope and the necessity of it having to expand to a partial 'if' statement with no body. Instead of trying to work around this fragility, just remove the construct before it gets used. Reported-by: Linus Torvalds Signed-off-by: Ingo Molnar Cc: David Lechner Cc: Peter Zijlstra Link: https://lore.kernel.org/r/Z1LBnX9TpZLR5Dkf@gmail.com

Merge tag 'locking-core-2024-11-18' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

2024-11-19T20:43:11Z

Pull locking updates from Ingo Molnar: "Lockdep: - Enable PROVE_RAW_LOCK_NESTING with PROVE_LOCKING (Sebastian Andrzej Siewior) - Add lockdep_cleanup_dead_cpu() (David Woodhouse) futexes: - Use atomic64_inc_return() in get_inode_sequence_number() (Uros Bizjak) - Use atomic64_try_cmpxchg_relaxed() in get_inode_sequence_number() (Uros Bizjak) RT locking: - Add sparse annotation PREEMPT_RT's locking (Sebastian Andrzej Siewior) spinlocks: - Use atomic_try_cmpxchg_release() in osq_unlock() (Uros Bizjak) atomics: - x86: Use ALT_OUTPUT_SP() for __alternative_atomic64() (Uros Bizjak) - x86: Use ALT_OUTPUT_SP() for __arch_{,try_}cmpxchg64_emu() (Uros Bizjak) KCSAN, seqlocks: - Support seqcount_latch_t (Marco Elver) : - Add if_not_guard() conditional guard helper (David Lechner) - Adjust scoped_guard() macros to avoid potential warning (Przemek Kitszel) - Remove address space of returned pointer (Uros Bizjak) WW mutexes: - locking/ww_mutex: Adjust to lockdep nest_lock requirements (Thomas Hellström) Rust integration: - Fix raw_spin_lock initialization on PREEMPT_RT (Eder Zulian) Misc cleanups & fixes: - lockdep: Fix wait-type check related warnings (Ahmed Ehab) - lockdep: Use info level for initial info messages (Jiri Slaby) - spinlocks: Make __raw_* lock ops static (Geert Uytterhoeven) - pvqspinlock: Convert fields of 'enum vcpu_state' to uppercase (Qiuxu Zhuo) - iio: magnetometer: Fix if () scoped_guard() formatting (Stephen Rothwell) - rtmutex: Fix misleading comment (Peter Zijlstra) - percpu-rw-semaphores: Fix grammar in percpu-rw-semaphore.rst (Xiu Jianfeng)" * tag 'locking-core-2024-11-18' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (29 commits) locking/Documentation: Fix grammar in percpu-rw-semaphore.rst iio: magnetometer: fix if () scoped_guard() formatting rust: helpers: Avoid raw_spin_lock initialization for PREEMPT_RT kcsan, seqlock: Fix incorrect assumption in read_seqbegin() seqlock, treewide: Switch to non-raw seqcount_latch interface kcsan, seqlock: Support seqcount_latch_t time/sched_clock: Broaden sched_clock()'s instrumentation coverage time/sched_clock: Swap update_clock_read_data() latch writes locking/atomic/x86: Use ALT_OUTPUT_SP() for __arch_{,try_}cmpxchg64_emu() locking/atomic/x86: Use ALT_OUTPUT_SP() for __alternative_atomic64() cleanup: Add conditional guard helper cleanup: Adjust scoped_guard() macros to avoid potential warning locking/osq_lock: Use atomic_try_cmpxchg_release() in osq_unlock() cleanup: Remove address space of returned pointer locking/rtmutex: Fix misleading comment locking/rt: Annotate unlock followed by lock for sparse. locking/rt: Add sparse annotation for RCU. locking/rt: Remove one __cond_lock() in RT's spin_trylock_irqsave() locking/rt: Add sparse annotation PREEMPT_RT's sleeping locks. locking/pvqspinlock: Convert fields of 'enum vcpu_state' to uppercase ...

deal with the last remaing boolean uses of fd_file()

2024-11-03T06:28:07Z

Reviewed-by: Christian Brauner Signed-off-by: Al Viro

cleanup: Add conditional guard helper

2024-10-25T08:01:51Z

Add a new if_not_guard() macro to cleanup.h for handling conditional guards such as mutext_trylock(). This is more ergonomic than scoped_guard() for most use cases. Instead of hiding the error handling statement in the macro args, it works like a normal if statement and allow the error path to be indented while the normal code flow path is not indented. And it avoid unwanted side-effect from hidden for loop in scoped_guard(). Signed-off-by: David Lechner Co-developed-by: Fabio M. De Francesco Signed-off-by: Fabio M. De Francesco Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Dan Williams Link: https://lkml.kernel.org/r/20241001-cleanup-if_not_cond_guard-v1-1-7753810b0f7a@baylibre.com