x86/mm/pat: Do a small optimization and fix in reserve_memtype

From: Pan Xinhui <[email protected]>

It's more reasonable to unlock memtype_lock right after rbt_memtype_check_insert. All pat_rbtree operations need to be performed while holding the memtype_lock. But codes like kfree, pr_info, etc have nothing to do with this lock. So move spin_unlock a little ahead.

If *new* succeed to be stored into the rb-tree, we might hit panic. Because we access *new* in dprintk "cattr_name(new->type)". Data stored in the rb-tree might be freed at any possible time. It's obviously wrong to access such data without lock held.

We use actual_type instead of new->type in dprintk. Two benefits, 1) We don't touch *new* here now so panic can be avoided. 2) As new->type is same with *new_type, We needn't to print it again, what's more, this actual_type's output can help debug. When new_type is NULL, actual_type is equal to new->type, So no key information is lost.

Do a minor codes style improvement.

---
change from v3:

	output actual_type instead of new->type.
	codes style improvement.
	update comments.

change from v2:
update comments.
change from V1:
fix an access of *new* without memtype_lock held.

---
 arch/x86/mm/pat.c | 14 ++++----------
 1 file changed, 4 insertions(+), 10 deletions(-)

index 188e3e0..8fa1f07 100644
@@ -521,10 +521,7 @@ int reserve_memtype(u64 start, u64 end, enum page_cache_mode req_type,
 
 	is_range_ram = pat_pagerange_is_ram(start, end);
 	if (is_range_ram == 1) {
-
-		err = reserve_ram_pages_type(start, end, req_type, new_type);
-
-		return err;
+		return reserve_ram_pages_type(start, end, req_type, new_type);
 	} else if (is_range_ram < 0) {
 		return -EINVAL;
 	}

@@ -538,22 +535,19 @@ int reserve_memtype(u64 start, u64 end, enum page_cache_mode req_type,
 	new->type	= actual_type;
 
 	spin_lock(&memtype_lock);
-
 	err = rbt_memtype_check_insert(new, new_type);
+	spin_unlock(&memtype_lock);
+
 	if (err) {
 		pr_info("x86/PAT: reserve_memtype failed [mem %#010Lx-%#010Lx], track %s, req %s\n",
 			start, end - 1,
 			cattr_name(new->type), cattr_name(req_type));
 		kfree(new);
-		spin_unlock(&memtype_lock);
-
 		return err;
 	}
 
-	spin_unlock(&memtype_lock);
-
 	dprintk("reserve_memtype added [mem %#010Lx-%#010Lx], track %s, req %s, ret %s\n",
-		start, end - 1, cattr_name(new->type), cattr_name(req_type),
+		start, end - 1, cattr_name(actual_type), cattr_name(req_type),
 		new_type ? cattr_name(*new_type) : "-");
 
 	return err;

1.9.1

---
Robert Elliott, HP Server Storage

...

With each branch now just one line, the {} can be removed.

���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m����������?�I�

... and I had suggested it without the {}.

Regards/Gruss,
    Boris.

ECO tip #101: Trim your mails when you reply.
--

hi, Borislav
very kind of you. :)

hmm, sorry for my mistakes.
Now I have to send patch V5 to you, the version number 5 is a little big for a simple patch like this..

thanks
xinhui

hi, Elliot
thanks for your reply.

thanks for the nice tips. sorry for my mistakes.

thanks
xinhui