iwlwifi: fix use after free bug for paged rx
In the paged rx patch (4854fde2), I introduced a bug that could possibly
touch an already freed page. It is fixed by avoiding the access in this
patch. I've also added some comments so that other people touching the
code won't make the same mistake. In the future, if we cannot avoid
access the page after being handled to the upper layer, we can use
get_page/put_page to handle it. For now, it's just not necessary.
It also fixed a debug message print bug reported by Stanislaw Gruszka
<sgruszka@redhat.com>.
Signed-off-by: Zhu Yi <yi.zhu@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
diff --git a/drivers/net/wireless/iwlwifi/iwl3945-base.c b/drivers/net/wireless/iwlwifi/iwl3945-base.c
index 5977a57..8b08bdc 100644
--- a/drivers/net/wireless/iwlwifi/iwl3945-base.c
+++ b/drivers/net/wireless/iwlwifi/iwl3945-base.c
@@ -1124,6 +1124,7 @@
struct iwl_rx_mem_buffer *rxb;
struct page *page;
unsigned long flags;
+ gfp_t gfp_mask = priority;
while (1) {
spin_lock_irqsave(&rxq->lock, flags);
@@ -1135,13 +1136,13 @@
spin_unlock_irqrestore(&rxq->lock, flags);
if (rxq->free_count > RX_LOW_WATERMARK)
- priority |= __GFP_NOWARN;
+ gfp_mask |= __GFP_NOWARN;
if (priv->hw_params.rx_page_order > 0)
- priority |= __GFP_COMP;
+ gfp_mask |= __GFP_COMP;
/* Alloc a new receive buffer */
- page = alloc_pages(priority, priv->hw_params.rx_page_order);
+ page = alloc_pages(gfp_mask, priv->hw_params.rx_page_order);
if (!page) {
if (net_ratelimit())
IWL_DEBUG_INFO(priv, "Failed to allocate SKB buffer.\n");
@@ -1410,8 +1411,8 @@
if (priv->rx_handlers[pkt->hdr.cmd]) {
IWL_DEBUG_RX(priv, "r = %d, i = %d, %s, 0x%02x\n", r, i,
get_cmd_string(pkt->hdr.cmd), pkt->hdr.cmd);
- priv->rx_handlers[pkt->hdr.cmd] (priv, rxb);
priv->isr_stats.rx_handlers[pkt->hdr.cmd]++;
+ priv->rx_handlers[pkt->hdr.cmd] (priv, rxb);
} else {
/* No handling needed */
IWL_DEBUG_RX(priv,
@@ -1420,11 +1421,18 @@
pkt->hdr.cmd);
}
+ /*
+ * XXX: After here, we should always check rxb->page
+ * against NULL before touching it or its virtual
+ * memory (pkt). Because some rx_handler might have
+ * already taken or freed the pages.
+ */
+
if (reclaim) {
/* Invoke any callbacks, transfer the buffer to caller,
* and fire off the (possibly) blocking iwl_send_cmd()
* as we reclaim the driver command queue */
- if (rxb && rxb->page)
+ if (rxb->page)
iwl_tx_cmd_complete(priv, rxb);
else
IWL_WARN(priv, "Claim null rxb?\n");