Sorry to reply to myself, but given the comments regarding the e1000 driver, I
should take the same changes into account here. New version of the patch, much
cleaner than previously.
Update e1000e driver to not allow dma beyond the end of the allocated sk
Signed-off-by: Neil Horman <[email protected]>
netdev.c | 14 +++++++++++++-
1 file changed, 13 insertions(+), 1 deletion(-)
diff --git a/drivers/net/e1000e/netdev.c b/drivers/net/e1000e/netdev.c
index c3105c5..84e1b5d 100644
--- a/drivers/net/e1000e/netdev.c
+++ b/drivers/net/e1000e/netdev.c
@@ -2332,6 +2332,16 @@ static void e1000_setup_rctl(struct e1000_adapter
*adapter)
/* Setup buffer sizes */
rctl &= ~E1000_RCTL_SZ_4096;
rctl |= E1000_RCTL_BSEX;
+ /*
+ * We need to normalize the rx_buffer_len here
+ * since the hardware only knows about 7 discrete
+ * frame lengths here. To accomodate that we need
+ * to set the rx length in the hardware to the next highest
+ * size over the rx_buffer_len, then increase rx_buffer_len
+ * to match it, so that we can get a full mtu sized frame
+ */
+ adapter->rx_buffer_len = roundup_pow_of_two(adapter->rx_buffer_len);
+
switch (adapter->rx_buffer_len) {
case 256:
rctl |= E1000_RCTL_SZ_256;
@@ -2346,7 +2356,6 @@ static void e1000_setup_rctl(struct e1000_adapter
*adapter)
rctl &= ~E1000_RCTL_BSEX;
break;
case 2048:
- default:
rctl |= E1000_RCTL_SZ_2048;
rctl &= ~E1000_RCTL_BSEX;
break;
@@ -2359,6 +2368,9 @@ static void e1000_setup_rctl(struct e1000_adapter
*adapter)
case 16384:
rctl |= E1000_RCTL_SZ_16384;
break;
+ default:
+ panic("Bad rx_buffer_len size\n");
+ break;
}
/*
------------------------------------------------------------------------------
Join us December 9, 2009 for the Red Hat Virtual Experience,
a free event focused on virtualization and cloud computing.
Attend in-depth sessions from your desk. Your couch. Anywhere.
http://p.sf.net/sfu/redhat-sfdev2dev
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel
