genirq/irqdesc: Honor caller provided affinity in alloc_desc()

commit edbbaae42a56f9a2b39c52ef2504dfb3fb0a7858 upstream.

Currently, whenever a caller is providing an affinity hint for an
interrupt, the allocation code uses it to calculate the node and copies the
cpumask into irq_desc::affinity.

If the affinity for the interrupt is not marked 'managed' then the startup
of the interrupt ignores irq_desc::affinity and uses the system default
affinity mask.

Prevent this by setting the IRQD_AFFINITY_SET flag for the interrupt in the
allocator, which causes irq_setup_affinity() to use irq_desc::affinity on
interrupt startup if the mask contains an online CPU.

[ tglx: Massaged changelog ]

Fixes: 45ddcecbfa ("genirq: Use affinity hint in irqdesc allocation")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/all/20240806072044.837827-1-shayd@nvidia.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This commit is contained in:
Shay Drory 2024-08-06 10:20:44 +03:00 коммит произвёл Greg Kroah-Hartman
Родитель a2ff348226
Коммит 0ad02834c2
1 изменённых файлов: 1 добавлений и 0 удалений

Просмотреть файл

@ -493,6 +493,7 @@ static int alloc_descs(unsigned int start, unsigned int cnt, int node,
flags = IRQD_AFFINITY_MANAGED | flags = IRQD_AFFINITY_MANAGED |
IRQD_MANAGED_SHUTDOWN; IRQD_MANAGED_SHUTDOWN;
} }
flags |= IRQD_AFFINITY_SET;
mask = &affinity->mask; mask = &affinity->mask;
node = cpu_to_node(cpumask_first(mask)); node = cpu_to_node(cpumask_first(mask));
affinity++; affinity++;