[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20110930162649.GA32195@electric-eye.fr.zoreil.com>
Date: Fri, 30 Sep 2011 18:26:49 +0200
From: Francois Romieu <romieu@...zoreil.com>
To: Ben Hutchings <bhutchings@...arflare.com>
Cc: davem@...emloft.net, netdev@...r.kernel.org
Subject: Re: [PATCH next 2/7] 8139cp : removal of headers.
Ben Hutchings <bhutchings@...arflare.com> :
[...]
> > diff --git a/drivers/net/ethernet/realtek/8139cp.c b/drivers/net/ethernet/realtek/8139cp.c
> > index c77d5af..60c8de5 100644
> > --- a/drivers/net/ethernet/realtek/8139cp.c
> > +++ b/drivers/net/ethernet/realtek/8139cp.c
> > @@ -55,8 +55,6 @@
> >
> > #include <linux/module.h>
> > #include <linux/moduleparam.h>
> > -#include <linux/kernel.h>
>
> Needed for container_of.
>
> > -#include <linux/compiler.h>
>
> Needed for __packed. But it's probably safe to assume that
> <linux/kernel.h> will include it.
<linux/module.h> would include it too.
Being modular opens the doors for a lot of features. :o|
[...]
> > @@ -65,7 +63,6 @@
> > #include <linux/dma-mapping.h>
> > #include <linux/delay.h>
> > #include <linux/ethtool.h>
> > -#include <linux/gfp.h>
>
> Needed for GFP_KERNEL.
I would have assumed that <linux/dma-mapping.h> provides it as it
publicizes gfp_t dependant data.
I am sceptical about the value in including so much headers for
rather common things at the network device driver level. After all
the code is supposed to be cross-platform, modular or monolithic
built.
I can understand that even a single class of devices requires some
variation but including kernel.h, init.h, module.h and moduleparam.h
all around in drivers/net seems a bit silly. Same thing for including
both netdevice.h and etherdevice.h.
It's probably not worth the hassle. I'll drop it.
--
Ueimor
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists