user/sven/linux.git/lib, branch v6.13.4

maple_tree: simplify split calculation

2025-02-17T10:36:55Z

commit 4f6a6bed0bfef4b966f076f33eb4f5547226056a upstream. Patch series "simplify split calculation", v3. This patch (of 3): The current calculation for splitting nodes tries to enforce a minimum span on the leaf nodes. This code is complex and never worked correctly to begin with, due to the min value being passed as 0 for all leaves. The calculation should just split the data as equally as possible between the new nodes. Note that b_end will be one more than the data, so the left side is still favoured in the calculation. The current code may also lead to a deficient node by not leaving enough data for the right side of the split. This issue is also addressed with the split calculation change. [Liam.Howlett@Oracle.com: rephrase the change log] Link: https://lkml.kernel.org/r/20241113031616.10530-1-richard.weiyang@gmail.com Link: https://lkml.kernel.org/r/20241113031616.10530-2-richard.weiyang@gmail.com Fixes: 54a611b60590 ("Maple Tree: add new data structure") Signed-off-by: Wei Yang Reviewed-by: Liam R. Howlett Cc: Sidhartha Kumar Cc: Lorenzo Stoakes Cc: Signed-off-by: Andrew Morton Signed-off-by: Greg Kroah-Hartman

atomic64: Use arch_spin_locks instead of raw_spin_locks

2025-02-17T10:36:26Z

commit 6c8ad3ab45ad0e94bfb7a9c71f2fa9c6cacea4b2 upstream. raw_spin_locks can be traced by lockdep or tracing itself. Atomic64 operations can be used in the tracing infrastructure. When an architecture does not have true atomic64 operations it can use the generic version that disables interrupts and uses spin_locks. The tracing ring buffer code uses atomic64 operations for the time keeping. But because some architectures use the default operations, the locking inside the atomic operations can cause an infinite recursion. As atomic64 implementation is architecture specific, it should not be using raw_spin_locks() but instead arch_spin_locks as that is the purpose of arch_spin_locks. To be used in architecture specific implementations of generic infrastructure like atomic64 operations. Note, by switching from raw_spin_locks to arch_spin_locks, the locks taken to emulate the atomic64 operations will not have lockdep, mmio, or any kind of checks done on them. They will not even disable preemption, although the code will disable interrupts preventing the tasks that hold the locks from being preempted. As the locks held are done so for very short periods of time, and the logic is only done to emulate atomic64, not having them be instrumented should not be an issue. Cc: stable@vger.kernel.org Cc: Mark Rutland Cc: Mathieu Desnoyers Cc: Andrew Morton Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: Linus Torvalds Cc: Andreas Larsson Link: https://lore.kernel.org/20250122144311.64392baf@gandalf.local.home Fixes: c84897c0ff592 ("ring-buffer: Remove 32bit timestamp logic") Closes: https://lore.kernel.org/all/86fb4f86-a0e4-45a2-a2df-3154acc4f086@gaisler.com/ Reported-by: Ludwig Rydberg Reviewed-by: Masami Hiramatsu (Google) Signed-off-by: Steven Rostedt (Google) Signed-off-by: Greg Kroah-Hartman

lockdep: Fix upper limit for LOCKDEP_*_BITS configs

2025-02-17T10:35:54Z

[ Upstream commit e638072e61726cae363d48812815197a2a0e097f ] Lockdep has a set of configs used to determine the size of the static arrays that it uses. However, the upper limit that was initially setup for these configs is too high (30 bit shift). This equates to several GiB of static memory for individual symbols. Using such high values leads to linker errors: $ make defconfig $ ./scripts/config -e PROVE_LOCKING --set-val LOCKDEP_BITS 30 $ make olddefconfig all [...] ld: kernel image bigger than KERNEL_IMAGE_SIZE ld: section .bss VMA wraps around address space Adjust the upper limits to the maximum values that avoid these issues. The need for anything more, likely points to a problem elsewhere. Note that LOCKDEP_CHAINS_BITS was intentionally left out as its upper limit had a different symptom and has already been fixed [1]. Reported-by: J. R. Okajima Closes: https://lore.kernel.org/all/30795.1620913191@jrobl/ [1] Cc: Peter Zijlstra Cc: Boqun Feng Cc: Ingo Molnar Cc: Waiman Long Cc: Will Deacon Acked-by: Waiman Long Signed-off-by: Carlos Llamas Signed-off-by: Boqun Feng Link: https://lore.kernel.org/r/20241024183631.643450-2-cmllamas@google.com Signed-off-by: Sasha Levin

alloc_tag: avoid current->alloc_tag manipulations when profiling is disabled

2025-02-08T09:02:05Z

[ Upstream commit 07438779313caafe52ac1a1a6958d735a5938988 ] When memory allocation profiling is disabled there is no need to update current->alloc_tag and these manipulations add unnecessary overhead. Fix the overhead by skipping these extra updates. I ran comprehensive testing on Pixel 6 on Big, Medium and Little cores: Overhead before fixes Overhead after fixes slab alloc page alloc slab alloc page alloc Big 6.21% 5.32% 3.31% 4.93% Medium 4.51% 5.05% 3.79% 4.39% Little 7.62% 1.82% 6.68% 1.02% This is an allocation microbenchmark doing allocations in a tight loop. Not a really realistic scenario and useful only to make performance comparisons. Link: https://lkml.kernel.org/r/20241226211639.1357704-1-surenb@google.com Fixes: b951aaff5035 ("mm: enable page allocation tagging") Signed-off-by: Suren Baghdasaryan Cc: David Wang <00107082@163.com> Cc: Kent Overstreet Cc: Yu Zhao Cc: Zhenhua Huang Signed-off-by: Andrew Morton Signed-off-by: Sasha Levin

rhashtable: Fix rhashtable_try_insert test

2025-02-08T09:01:41Z

[ Upstream commit 9d4f8e54cef2c42e23ef258833dbd06a1eaff89b ] The test on whether rhashtable_insert_one did an insertion relies on the value returned by rhashtable_lookup_one. Unfortunately that value is overwritten after rhashtable_insert_one returns. Fix this by moving the test before data gets overwritten. Simplify the test as only data == NULL matters. Finally move atomic_inc back within the lock as otherwise it may be reordered with the atomic_dec on the removal side, potentially leading to an underflow. Reported-by: Michael Kelley Fixes: e1d3422c95f0 ("rhashtable: Fix potential deadlock by moving schedule_work outside lock") Signed-off-by: Herbert Xu Tested-by: Michael Kelley Reviewed-by: Breno Leitao Tested-by: Mikhail Zaslonko Signed-off-by: Herbert Xu Signed-off-by: Sasha Levin

rhashtable: Fix potential deadlock by moving schedule_work outside lock

2025-02-08T09:01:34Z

[ Upstream commit e1d3422c95f003eba241c176adfe593c33e8a8f6 ] Move the hash table growth check and work scheduling outside the rht lock to prevent a possible circular locking dependency. The original implementation could trigger a lockdep warning due to a potential deadlock scenario involving nested locks between rhashtable bucket, rq lock, and dsq lock. By relocating the growth check and work scheduling after releasing the rth lock, we break this potential deadlock chain. This change expands the flexibility of rhashtable by removing restrictive locking that previously limited its use in scheduler and workqueue contexts. Import to say that this calls rht_grow_above_75(), which reads from struct rhashtable without holding the lock, if this is a problem, we can move the check to the lock, and schedule the workqueue after the lock. Fixes: f0e1a0643a59 ("sched_ext: Implement BPF extensible scheduler class") Suggested-by: Tejun Heo Signed-off-by: Breno Leitao Modified so that atomic_inc is also moved outside of the bucket lock along with the growth above 75% check. Signed-off-by: Herbert Xu Signed-off-by: Sasha Levin

alloc_tag: skip pgalloc_tag_swap if profiling is disabled

2025-01-16T05:15:43Z

When memory allocation profiling is disabled, there is no need to swap allocation tags during migration. Skip it to avoid unnecessary overhead. Once I added these checks, the overhead of the mode when memory profiling is enabled but turned off went down by about 50%. Link: https://lkml.kernel.org/r/20241226211639.1357704-2-surenb@google.com Fixes: e0a955bf7f61 ("mm/codetag: add pgalloc_tag_copy()") Signed-off-by: Suren Baghdasaryan Cc: David Wang <00107082@163.com> Cc: Kent Overstreet Cc: Yu Zhao Cc: Zhenhua Huang Cc: Signed-off-by: Andrew Morton

maple_tree: reload mas before the second call for mas_empty_area

2024-12-31T01:59:07Z

Change the LONG_MAX in simple_offset_add to 1024, and do latter: [root@fedora ~]# mkdir /tmp/dir [root@fedora ~]# for i in {1..1024}; do touch /tmp/dir/$i; done touch: cannot touch '/tmp/dir/1024': Device or resource busy [root@fedora ~]# rm /tmp/dir/123 [root@fedora ~]# touch /tmp/dir/1024 [root@fedora ~]# rm /tmp/dir/100 [root@fedora ~]# touch /tmp/dir/1025 touch: cannot touch '/tmp/dir/1025': Device or resource busy After we delete file 100, actually this is a empty entry, but the latter create failed unexpected. mas_alloc_cyclic has two chance to find empty entry. First find the entry with range range_lo and range_hi, if no empty entry exist, and range_lo > min, retry find with range min and range_hi. However, the first call mas_empty_area may mark mas as EBUSY, and the second call for mas_empty_area will return false directly. Fix this by reload mas before second call for mas_empty_area. [Liam.Howlett@Oracle.com: fix mas_alloc_cyclic() second search] Link: https://lore.kernel.org/all/20241216060600.287B4C4CED0@smtp.kernel.org/ Link: https://lkml.kernel.org/r/20241216190113.1226145-2-Liam.Howlett@oracle.com Link: https://lkml.kernel.org/r/20241214093005.72284-1-yangerkun@huaweicloud.com Fixes: 9b6713cc7522 ("maple_tree: Add mtree_alloc_cyclic()") Signed-off-by: Yang Erkun Signed-off-by: Liam R. Howlett Cc: Christian Brauner Cc: Chuck Lever says: Cc: Liam R. Howlett Cc: Signed-off-by: Andrew Morton

alloc_tag: fix module allocation tags populated area calculation

2024-12-19T03:04:46Z

vm_module_tags_populate() calculation of the populated area assumes that area starts at a page boundary and therefore when new pages are allocation, the end of the area is page-aligned as well. If the start of the area is not page-aligned then allocating a page and incrementing the end of the area by PAGE_SIZE leads to an area at the end but within the area boundary which is not populated. Accessing this are will lead to a kernel panic. Fix the calculation by down-aligning the start of the area and using that as the location allocated pages are mapped to. [gehao@kylinos.cn: fix vm_module_tags_populate's KASAN poisoning logic] Link: https://lkml.kernel.org/r/20241205170528.81000-1-hao.ge@linux.dev [gehao@kylinos.cn: fix panic when CONFIG_KASAN enabled and CONFIG_KASAN_VMALLOC not enabled] Link: https://lkml.kernel.org/r/20241212072126.134572-1-hao.ge@linux.dev Link: https://lkml.kernel.org/r/20241130001423.1114965-1-surenb@google.com Fixes: 0f9b685626da ("alloc_tag: populate memory for module tags as needed") Signed-off-by: Suren Baghdasaryan Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-lkp/202411132111.6a221562-lkp@intel.com Acked-by: Yu Zhao Tested-by: Adrian Huang Cc: David Wang <00107082@163.com> Cc: Kent Overstreet Cc: Mike Rapoport (Microsoft) Cc: Pasha Tatashin Cc: Sourav Panda Cc: Signed-off-by: Andrew Morton

mm/codetag: clear tags before swap

2024-12-19T03:04:46Z

When CONFIG_MEM_ALLOC_PROFILING_DEBUG is set, kernel WARN would be triggered when calling __alloc_tag_ref_set() during swap: alloc_tag was not cleared (got tag for mm/filemap.c:1951) WARNING: CPU: 0 PID: 816 at ./include/linux/alloc_tag.h... Clear code tags before swap can fix the warning. And this patch also fix a potential invalid address dereference in alloc_tag_add_check() when CONFIG_MEM_ALLOC_PROFILING_DEBUG is set and ref->ct is CODETAG_EMPTY, which is defined as ((void *)1). Link: https://lkml.kernel.org/r/20241213013332.89910-1-00107082@163.com Fixes: 51f43d5d82ed ("mm/codetag: swap tags when migrate pages") Signed-off-by: David Wang <00107082@163.com> Reported-by: kernel test robot Closes: https://lore.kernel.org/oe-lkp/202412112227.df61ebb-lkp@intel.com Acked-by: Suren Baghdasaryan Cc: Kent Overstreet Cc: Yu Zhao Cc: Signed-off-by: Andrew Morton