From: "Matthew Wilcox (Oracle)" <willy@infradead.org>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org>,
Zi Yan <ziy@nvidia.com>,
linux-mm@kvack.org, linux-fsdevel@vger.kernel.org,
Luis Chamberlain <mcgrof@kernel.org>
Subject: [PATCH] XArray: Set the marks correctly when splitting an entry
Date: Wed, 1 May 2024 16:31:18 +0100 [thread overview]
Message-ID: <20240501153120.4094530-1-willy@infradead.org> (raw)
If we created a new node to replace an entry which had search marks set,
we were setting the search mark on every entry in that node. That works
fine when we're splitting to order 0, but when splitting to a larger
order, we must not set the search marks on the sibling entries.
Reported-by: Luis Chamberlain <mcgrof@kernel.org>
Link: https://lore.kernel.org/r/ZjFGCOYk3FK_zVy3@bombadil.infradead.org
Fixes: c010d47f107f ("mm: thp: split huge page to any lower order pages")
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
---
lib/test_xarray.c | 14 +++++++++++++-
lib/xarray.c | 23 +++++++++++++++++++----
2 files changed, 32 insertions(+), 5 deletions(-)
diff --git a/lib/test_xarray.c b/lib/test_xarray.c
index 0aea6a85099d..ab9cc42a0d74 100644
--- a/lib/test_xarray.c
+++ b/lib/test_xarray.c
@@ -1788,9 +1788,11 @@ static void check_split_1(struct xarray *xa, unsigned long index,
unsigned int order, unsigned int new_order)
{
XA_STATE_ORDER(xas, xa, index, new_order);
- unsigned int i;
+ unsigned int i, found;
+ void *entry;
xa_store_order(xa, index, order, xa, GFP_KERNEL);
+ xa_set_mark(xa, index, XA_MARK_1);
xas_split_alloc(&xas, xa, order, GFP_KERNEL);
xas_lock(&xas);
@@ -1807,6 +1809,16 @@ static void check_split_1(struct xarray *xa, unsigned long index,
xa_set_mark(xa, index, XA_MARK_0);
XA_BUG_ON(xa, !xa_get_mark(xa, index, XA_MARK_0));
+ xas_set_order(&xas, index, 0);
+ found = 0;
+ rcu_read_lock();
+ xas_for_each_marked(&xas, entry, ULONG_MAX, XA_MARK_1) {
+ found++;
+ XA_BUG_ON(xa, xa_is_internal(entry));
+ }
+ rcu_read_unlock();
+ XA_BUG_ON(xa, found != 1 << (order - new_order));
+
xa_destroy(xa);
}
diff --git a/lib/xarray.c b/lib/xarray.c
index 1c87d871cacf..32d4bac8c94c 100644
--- a/lib/xarray.c
+++ b/lib/xarray.c
@@ -970,8 +970,22 @@ static unsigned int node_get_marks(struct xa_node *node, unsigned int offset)
return marks;
}
+static inline void node_mark_slots(struct xa_node *node, unsigned int sibs,
+ xa_mark_t mark)
+{
+ int i;
+
+ if (sibs == 0)
+ node_mark_all(node, mark);
+ else {
+ for (i = 0; i < XA_CHUNK_SIZE; i += sibs + 1)
+ node_set_mark(node, i, mark);
+ }
+}
+
static void node_set_marks(struct xa_node *node, unsigned int offset,
- struct xa_node *child, unsigned int marks)
+ struct xa_node *child, unsigned int sibs,
+ unsigned int marks)
{
xa_mark_t mark = XA_MARK_0;
@@ -979,7 +993,7 @@ static void node_set_marks(struct xa_node *node, unsigned int offset,
if (marks & (1 << (__force unsigned int)mark)) {
node_set_mark(node, offset, mark);
if (child)
- node_mark_all(child, mark);
+ node_mark_slots(child, sibs, mark);
}
if (mark == XA_MARK_MAX)
break;
@@ -1078,7 +1092,8 @@ void xas_split(struct xa_state *xas, void *entry, unsigned int order)
child->nr_values = xa_is_value(entry) ?
XA_CHUNK_SIZE : 0;
RCU_INIT_POINTER(child->parent, node);
- node_set_marks(node, offset, child, marks);
+ node_set_marks(node, offset, child, xas->xa_sibs,
+ marks);
rcu_assign_pointer(node->slots[offset],
xa_mk_node(child));
if (xa_is_value(curr))
@@ -1087,7 +1102,7 @@ void xas_split(struct xa_state *xas, void *entry, unsigned int order)
} else {
unsigned int canon = offset - xas->xa_sibs;
- node_set_marks(node, canon, NULL, marks);
+ node_set_marks(node, canon, NULL, 0, marks);
rcu_assign_pointer(node->slots[canon], entry);
while (offset > canon)
rcu_assign_pointer(node->slots[offset--],
--
2.43.0
next reply other threads:[~2024-05-01 15:31 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-05-01 15:31 Matthew Wilcox (Oracle) [this message]
2024-05-01 16:20 ` Luis Chamberlain
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20240501153120.4094530-1-willy@infradead.org \
--to=willy@infradead.org \
--cc=akpm@linux-foundation.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mcgrof@kernel.org \
--cc=ziy@nvidia.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox