(nuttx) 03/03: mm/mm_gran: Do not fail an allocation when a signal arrives.

[ac...@apache.org]

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

commit 15a3ee13cd7eec09084b7e64dce8a1e4fee8dbf2
Author: Marco Casaroli <ma...@gmail.com>
AuthorDate: Mon Oct 12 14:45:34 2026 +0200

    mm/mm_gran: Do not fail an allocation when a signal arrives.
    
    gran_enter_critical() returned the result of nxmutex_lock(), which is
    -EINTR when a signal interrupts the wait, and gran_alloc() then returned
    NULL as if the pool were full.  The page allocator uses a granule
    allocator, so fork(), sbrk() and the stack of a new task failed with
    ENOMEM while most of the page pool was free.
    
    Retry the lock when a signal interrupts it.
    
    Assisted-by: Claude Code:claude-opus-5-5
    Signed-off-by: Marco Casaroli <ma...@gmail.com>
---
 mm/mm_gran/mm_grancritical.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/mm/mm_gran/mm_grancritical.c b/mm/mm_gran/mm_grancritical.c
index 06481e1de06..5a01094dbfd 100644
--- a/mm/mm_gran/mm_grancritical.c
+++ b/mm/mm_gran/mm_grancritical.c
@@ -62,7 +62,17 @@ int gran_enter_critical(FAR struct gran_s *priv)
   priv->irqstate = spin_lock_irqsave(&priv->lock);
   return OK;
 #else
-  return nxmutex_lock(&priv->lock);
+  int ret;
+
+  /* A signal must not make the allocation fail */
+
+  do
+    {
+      ret = nxmutex_lock(&priv->lock);
+    }
+  while (ret == -EINTR);
+
+  return ret;
 #endif
 }