From 27600805e62f800bacf990354632eae4e487d34c Mon Sep 17 00:00:00 2001 From: Yazen Ghannam Date: Thu, 3 Sep 2026 10:43:25 -0500 Subject: x86/amd_node: Fix PCI device reference counting in amd_smn_init() The local "root" pointer is a temporary variable used during the device search. Therefore, refcount related to the search iterators should be cleaned up after the search is complete. Use the __free() cleanup macro to ensure the refcount is decremented when the temporary pointer goes out of scope. Additionally, increment the refcount when caching a root pointer. This ensures the in-use refcount is separate from the temporary search refcounting. Finally, drop the redundant "root = NULL" before the second search loop. The pci_get_class() iterator always decrements the refcount of its "from" argument, so the first loop can only fall through with "root" already NULL. Fixes: 0a4b61d9c2e4 ("x86/amd_node: Fix AMD root device caching") Closes: https://sashiko.dev/#/patchset/20260806160159.230453-1-jason.andryuk%40amd.com Reported-by: Sashiko Assisted-by: LLM Signed-off-by: Yazen Ghannam Signed-off-by: Borislav Petkov (AMD) Reviewed-by: Mario Limonciello (AMD) Cc: Link: https://patch.msgid.link/20260903154325.74343-1-yazen.ghannam@amd.com --- lib/cpumask.c | 168 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 168 insertions(+) create mode 100644 lib/cpumask.c (limited to 'lib/cpumask.c') diff --git a/lib/cpumask.c b/lib/cpumask.c new file mode 100644 index 000000000..5adb9874f --- /dev/null +++ b/lib/cpumask.c @@ -0,0 +1,168 @@ +// SPDX-License-Identifier: GPL-2.0 +#include +#include +#include +#include +#include +#include +#include + +/* These are not inline because of header tangles. */ +#ifdef CONFIG_CPUMASK_OFFSTACK +/** + * alloc_cpumask_var_node - allocate a struct cpumask on a given node + * @mask: pointer to cpumask_var_t where the cpumask is returned + * @flags: GFP_ flags + * @node: memory node from which to allocate or %NUMA_NO_NODE + * + * Only defined when CONFIG_CPUMASK_OFFSTACK=y, otherwise is + * a nop returning a constant 1 (in ). + * + * Return: TRUE if memory allocation succeeded, FALSE otherwise. + * + * In addition, mask will be NULL if this fails. Note that gcc is + * usually smart enough to know that mask can never be NULL if + * CONFIG_CPUMASK_OFFSTACK=n, so does code elimination in that case + * too. + */ +bool alloc_cpumask_var_node(cpumask_var_t *mask, gfp_t flags, int node) +{ + *mask = kmalloc_node(cpumask_size(), flags, node); + +#ifdef CONFIG_DEBUG_PER_CPU_MAPS + if (!*mask) { + printk(KERN_ERR "=> alloc_cpumask_var: failed!\n"); + dump_stack(); + } +#endif + + return *mask != NULL; +} +EXPORT_SYMBOL(alloc_cpumask_var_node); + +/** + * alloc_bootmem_cpumask_var - allocate a struct cpumask from the bootmem arena. + * @mask: pointer to cpumask_var_t where the cpumask is returned + * + * Only defined when CONFIG_CPUMASK_OFFSTACK=y, otherwise is + * a nop (in ). + * Either returns an allocated (zero-filled) cpumask, or causes the + * system to panic. + */ +void __init alloc_bootmem_cpumask_var(cpumask_var_t *mask) +{ + *mask = memblock_alloc_or_panic(cpumask_size(), SMP_CACHE_BYTES); +} + +/** + * free_cpumask_var - frees memory allocated for a struct cpumask. + * @mask: cpumask to free + * + * This is safe on a NULL mask. + */ +void free_cpumask_var(cpumask_var_t mask) +{ + kfree(mask); +} +EXPORT_SYMBOL(free_cpumask_var); + +/** + * free_bootmem_cpumask_var - frees result of alloc_bootmem_cpumask_var + * @mask: cpumask to free + */ +void __init free_bootmem_cpumask_var(cpumask_var_t mask) +{ + memblock_free(mask, cpumask_size()); +} +#endif + +/** + * cpumask_local_spread - select the i'th cpu based on NUMA distances + * @i: index number + * @node: local numa_node + * + * Return: online CPU according to a numa aware policy; local cpus are returned + * first, followed by non-local ones, then it wraps around. + * + * For those who wants to enumerate all CPUs based on their NUMA distances, + * i.e. call this function in a loop, like: + * + * for (i = 0; i < num_online_cpus(); i++) { + * cpu = cpumask_local_spread(i, node); + * do_something(cpu); + * } + * + * There's a better alternative based on for_each()-like iterators: + * + * for_each_numa_hop_mask(mask, node) { + * for_each_cpu_andnot(cpu, mask, prev) + * do_something(cpu); + * prev = mask; + * } + * + * It's simpler and more verbose than above. Complexity of iterator-based + * enumeration is O(sched_domains_numa_levels * nr_cpu_ids), while + * cpumask_local_spread() when called for each cpu is + * O(sched_domains_numa_levels * nr_cpu_ids * log(nr_cpu_ids)). + */ +unsigned int cpumask_local_spread(unsigned int i, int node) +{ + unsigned int cpu; + + /* Wrap: we always want a cpu. */ + i %= num_online_cpus(); + + cpu = sched_numa_find_nth_cpu(cpu_online_mask, i, node); + + WARN_ON(cpu >= nr_cpu_ids); + return cpu; +} +EXPORT_SYMBOL(cpumask_local_spread); + +static DEFINE_PER_CPU(int, distribute_cpu_mask_prev); + +/** + * cpumask_any_and_distribute - Return an arbitrary cpu within src1p & src2p. + * @src1p: first &cpumask for intersection + * @src2p: second &cpumask for intersection + * + * Iterated calls using the same srcp1 and srcp2 will be distributed within + * their intersection. + * + * Return: >= nr_cpu_ids if the intersection is empty. + */ +unsigned int cpumask_any_and_distribute(const struct cpumask *src1p, + const struct cpumask *src2p) +{ + unsigned int next, prev; + + /* NOTE: our first selection will skip 0. */ + prev = __this_cpu_read(distribute_cpu_mask_prev); + + next = cpumask_next_and_wrap(prev, src1p, src2p); + if (next < nr_cpu_ids) + __this_cpu_write(distribute_cpu_mask_prev, next); + + return next; +} +EXPORT_SYMBOL(cpumask_any_and_distribute); + +/** + * cpumask_any_distribute - Return an arbitrary cpu from srcp + * @srcp: &cpumask for selection + * + * Return: >= nr_cpu_ids if the intersection is empty. + */ +unsigned int cpumask_any_distribute(const struct cpumask *srcp) +{ + unsigned int next, prev; + + /* NOTE: our first selection will skip 0. */ + prev = __this_cpu_read(distribute_cpu_mask_prev); + next = cpumask_next_wrap(prev, srcp); + if (next < nr_cpu_ids) + __this_cpu_write(distribute_cpu_mask_prev, next); + + return next; +} +EXPORT_SYMBOL(cpumask_any_distribute); -- cgit v1.3.1