[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CACRpkdYhZcNLfZ-EiJ6gmqriMdSkgAbLpQs7nGoSzLq_nMEEeg@mail.gmail.com>
Date: Wed, 2 May 2012 12:22:37 +0200
From: Linus Walleij <linus.walleij@...aro.org>
To: David Laight <David.Laight@...lab.com>
Cc: netdev@...r.kernel.org, linux-usb@...r.kernel.org,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
"David S. Miller" <davem@...emloft.net>,
Felipe Balbi <balbi@...com>,
Jussi Kivilinna <jussi.kivilinna@...et.fi>,
Haiyang Zhang <haiyangz@...rosoft.com>,
Wei Yongjun <yongjun_wei@...ndmicro.com.cn>,
Ben Hutchings <ben@...adent.org.uk>
Subject: Re: [PATCH 01/14 v3] usb/net: rndis: inline the cpu_to_le32() macro
On Wed, May 2, 2012 at 10:44 AM, David Laight <David.Laight@...lab.com> wrote:
>> [mailto:netdev-owner@...r.kernel.org] On Behalf Of Linus Walleij
>> The header file <linux/usb/rndis_host.h> used a number of #defines
>> that included the cpu_to_le32() macro to assure the result will be
>> in LE endianness. Inlining this into the code instead of using it
>> in the code definitions yields consolidation opportunities later
>> on as you will see in the following patches. The individual
>> drivers also used local defines - all are switched over to the
>> pattern of doing the conversion at the call sites instead.
>>
>> Signed-off-by: Linus Walleij <linus.walleij@...aro.org>
>> diff --git a/drivers/net/usb/rndis_host.c
>> b/drivers/net/usb/rndis_host.c
>> index c8f1b5b..05cad0b 100644
>> --- a/drivers/net/usb/rndis_host.c
>> +++ b/drivers/net/usb/rndis_host.c
>> @@ -78,10 +78,10 @@ static void rndis_msg_indicate(struct
>> usbnet *dev, struct rndis_indicate *msg,
>> dev->driver_info->indication(dev, msg, buflen);
>> } else {
>> switch (msg->status) {
>> - case RNDIS_STATUS_MEDIA_CONNECT:
>> + case cpu_to_le32(RNDIS_STATUS_MEDIA_CONNECT):
>> dev_info(udev, "rndis media connect\n");
>> break;
>
> I don't actually see how this helps!
It helps to unify the #defines, so we can consolidate them
in patch 2/13.
I.e today we have:
include/linux/usb/rndis_host.h
#define RNDIS_STATUS_MEDIA_CONNECT cpu_to_le32(0x4001000b)
drivers/usb/gadget/rndis.h
#define RNDIS_STATUS_MEDIA_CONNECT 0x4001000BU
drivers/net/hyperv/hyperv_net.h
#define RNDIS_STATUS_MEDIA_CONNECT (0x4001000BL)
Notice the odd bird on the top.
After this patch I can collapse these three defines into one.
As the overall goal is to start to unify the four different
RNDIS implementations in the kernel today.
> It seems to make the code more unreadable and more likely to
> be wrong (due to typos/coding slipups etc).
> I didn't see anything in the earlier versions of the patches
> where leaving the constants 'host endian' helped.
OK I guess this means I should simply wrap all the #defines
into cpu_to_le32() in the new <linux/rndis.h>?
I think the reason it looks as it does is that the rndis host is
the only thing that actually works on bigendian, such as when
Mac:s, SPARCs or PowerPC:s want to use RNDIS dongles.
I suspect using a BE system to produce a gadget driver
won't work, and that the hypervisor code is implicitly
LE and has never ever been designed with BE
in mind.
> I also actually suspect that the generated code for some of the
> switch statements would be better if the 'status' were byteswapped.
> In particular it might allow the compiler to use a jump table
> (when all the case labels are in one set of values).
Hm not sure if I'm following what you want me to do here...
I could cook a patch set with all defines wrapped in the cpu_to_le32()
macro to begin with.
Yours,
Linus Walleij
--
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