On Fri, 30 Dec 2005 19:26:10 -0500
Jeff Garzik <[EMAIL PROTECTED]> wrote:
> Stephen Hemminger wrote:
> > How come this hasn't made it upstream yet? The 2.6.15-rc7 code still
> > has the old version.
> >
>
> There were a bunch of not-for-rc changes, so it went into 2.6.16 pile:
>
> skge: avoid up/down on speed changes
> skge: avoid up/down on pause param changes
> skge: get rid of Yukon2 defines
> skge: version number (1.3)
Only the define's cleanup and version number are not bug fixes.
There were two classes of problems:
1. If skge driver was using up/down to do an internal restart
then if skge_up failed because of out of memory then the driver is in
a hosed state because netif thinks device is UP but it is not. This
also
leads to panic when device is then brought down.
I fixed some by checking the return value and forcing a down on failure.
[PATCH 3/6] skge: handle out of memory on MTU size changes
[PATCH 5/6] skge: handle out of memory on ring parameter change
[PATCH 7/6] skge: error handling on resume
Others by not doing up/down.
[PATCH 1/6] skge: avoid up/down on speed changes
[PATCH 2/6] skge: avoid up/down on pause param changes
2. In cases where device is restarted the multicast flags needed to be
reset or the device is deaf to broadcast unicast.
[PATCH] skge: don't forget multicast address
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
