This is an automated email from the ASF dual-hosted git repository.
acassis pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/nuttx.git
The following commit(s) were added to refs/heads/master by this push:
new 767599cd822 mm/Kconfig: Use a spinlock for the page allocator with SMP.
767599cd822 is described below
commit 767599cd82218318a61b83f578fe8c1a6c47e458
Author: Marco Casaroli <[email protected]>
AuthorDate: Mon Oct 12 16:37:06 2026 +0200
mm/Kconfig: Use a spinlock for the page allocator with SMP.
The page allocator takes the granule lock once for each page: fork()
copies a process one page at a time, and the destruction of an
address environment frees it one page at a time. With SMP and a
semaphore, unlock gives the lock to a waiter that is only made ready.
When all CPUs run tasks of the same priority, that waiter can wait a
full round-robin interval for a CPU while it holds the lock, and every
page allocation in the system waits behind it.
On qemu-armv8a with 4 CPUs and 4 shells that run programs in a loop,
the LPWORK thread took up to 631 ms to destroy an address environment
(1 ms when idle). The freed pages fell behind until the page pool was
empty. With GRAN_INTR, the destroy took 2 ms and the pool stayed full.
Make GRAN_INTR the default for MM_PGALLOC with SMP.
Assisted-by: Claude Code:claude-opus-5-5
Signed-off-by: Marco Casaroli <[email protected]>
---
mm/Kconfig | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/mm/Kconfig b/mm/Kconfig
index 70a79bd1dad..fd2feac0be1 100644
--- a/mm/Kconfig
+++ b/mm/Kconfig
@@ -276,7 +276,7 @@ if GRAN
config GRAN_INTR
bool "Interrupt level support"
- default n
+ default MM_PGALLOC && SMP
---help---
Normally mutual exclusive access to granule allocator data is
assured
using a semaphore. If this option is set then, instead, mutual
@@ -284,6 +284,11 @@ config GRAN_INTR
invasive to system performance, it will also support use of the
granule
allocator from interrupt level logic.
+ The page allocator takes the lock once for each page. With
SMP, a
+ semaphore can be given to a thread that waits for a CPU, and all
+ page allocations then wait for that thread. This option is the
+ default for MM_PGALLOC with SMP.
+
config DEBUG_GRAN
bool "Granule Allocator Debug"
default n
