Commit 48a30569 authored Sep 18, 2013 by Michael Chan Committed by David S. Miller Sep 18, 2013

cnic: Fix crash in cnic_bnx2x_service_kcq()



commit 104a43ed
    cnic: Use CHIP_NUM macros from bnx2x.h

changed the code to use the bnx2x macro NO_FCOE() to determine if FCoE
is supported or not.  There is another place in cnic that is still using
the old method to determine if FCoE is supported or not.  The 2 methods
may not yield the same result after the network interface is brought down
and up.  This will cause the crash as cnic_bnx2x_service_kcq() will access
the uninitialized cp->kcq2.

The fix is to consistently use the same macro CNIC_SUPPORTS_FCOE() which
uses the bnx2x NO_FCOE() macro.  As a follow-up, we can clean up the code
to remove the old method as it is no longer needed.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

parent f78afb35

drivers/net/ethernet/broadcom/cnic.c

+2 −1

Original line number	Diff line number	Diff line
		@@ -3135,6 +3135,7 @@ static void cnic_service_bnx2x_bh(unsigned long data)
		{
		struct cnic_dev dev = (struct cnic_dev ) data;
		struct cnic_local *cp = dev->cnic_priv;
		struct bnx2x *bp = netdev_priv(dev->netdev);
		u32 status_idx, new_status_idx;

		if (unlikely(!test_bit(CNIC_F_CNIC_UP, &dev->flags)))
		@@ -3146,7 +3147,7 @@ static void cnic_service_bnx2x_bh(unsigned long data)
		CNIC_WR16(dev, cp->kcq1.io_addr,
		cp->kcq1.sw_prod_idx + MAX_KCQ_IDX);

		if (cp->ethdev->drv_state & CNIC_DRV_STATE_NO_FCOE) {
		if (!CNIC_SUPPORTS_FCOE(bp)) {
		cp->arm_int(dev, status_idx);
		break;
		}