Re: parisc 4.13-git, BUG: sleeping function called from invalid context at mm/slab.h:416

[James Bottomley <James.Bottomley@HansenPartnership.com>]

On Sun, 2017-07-30 at 14:21 -0400, John David Anglin wrote:
> On 2017-07-30, at 2:12 PM, Meelis Roos wrote:
> 
> > 
> > > 
> > > How about this?  There's no need to take the write lock when
> > > creating
> > > links anyway.
> > 
> > No effect.
> 
> Is the backtrace still the same?

There's another write lock over a symlink in
pdcs_register_pathentries() does moving it up as well finally make this
go away (cumulative patch below).  If that doesn't work, enabling
DEBUG_PREEMPT might show which lock is the problem.

James

---

diff --git a/drivers/parisc/pdc_stable.c b/drivers/parisc/pdc_stable.c
index 7147aa53e9a2..3e0c90420a1b 100644
--- a/drivers/parisc/pdc_stable.c
+++ b/drivers/parisc/pdc_stable.c
@@ -331,13 +331,13 @@ pdcspath_hwpath_write(struct pdcspath_entry *entry, const char *buf, size_t coun
 	/* Now, dive in. Write back to the hardware */
 	pdcspath_store(entry);
 	
+	write_unlock(&entry->rw_lock);
+
 	/* Update the symlink to the real device */
 	sysfs_remove_link(&entry->kobj, "device");
 	ret = sysfs_create_link(&entry->kobj, &entry->dev->kobj, "device");
 	WARN_ON(ret);
 
-	write_unlock(&entry->rw_lock);
-	
 	printk(KERN_INFO PDCS_PREFIX ": changed \"%s\" path to \"%s\"\n",
 		entry->name, buf);
 	
@@ -998,6 +998,7 @@ pdcs_register_pathentries(void)
 		/* kobject is now registered */
 		write_lock(&entry->rw_lock);
 		entry->ready = 2;
+		write_unlock(&entry->rw_lock);
 		
 		/* Add a nice symlink to the real device */
 		if (entry->dev) {
@@ -1005,7 +1006,6 @@ pdcs_register_pathentries(void)
 			WARN_ON(err);
 		}
 
-		write_unlock(&entry->rw_lock);
 		kobject_uevent(&entry->kobj, KOBJ_ADD);
 	}