|
From: Kok, A. <auk...@in...> - 2008-03-31 18:41:53
|
Jeff Garzik wrote: > Auke Kok wrote: >> From: Bruce Allan <bru...@in...> >> >> The ndev_* printk's are too lenghty and we don't need to specify >> the adapter/netdev struct at all, making this a lot more readable. >> >> Signed-off-by: Bruce Allan <bru...@in...> >> Signed-off-by: Auke Kok <auk...@in...> >> --- >> >> drivers/net/e1000e/e1000.h | 27 ++++---- >> drivers/net/e1000e/ethtool.c | 38 +++++------ >> drivers/net/e1000e/netdev.c | 148 +++++++++++++++++++----------------------- >> drivers/net/e1000e/param.c | 31 +++------ >> 4 files changed, 108 insertions(+), 136 deletions(-) > > This stuff should just be removed. > > Either the code should be outputting "ethX: blahblah" via printk(), or > the code should be outputting the standard dev_printk() prefix [driver > name, PCI device info]. > > Also, it is recommended to use the NETIF_MSG_xxx bits to permit > fine-grained control over which messages are output, on a per-port basis > (controlled via ethtool). grep around for 'msg_enable' yes and I even tried pushing for a generic netdev printk that incorporates the msg_enable bits properly I also got absolutely unwelcome responses to that, if not none :) once we get such a thing, we can incorporate this into e1000 with this patch by changing 2 lines. so I think this patch is still a good thing, and it certainly removes a lot of wrapped lines. Auke |