I'm writing a driver for OSRFX2 learning board to test bulkloop read/write. I think non-blocking write is good for such device due to the FX2 chipset has dual-buffer for bulk endpoints. I am just wondering if I also need to support blocking write IO due to I used to think from driver perspective, for example for the read/write operations, we need to support both blocking and nonblocking by default.
So you mean on this question, driver writer can determine if support one or both, then app programmer may have to consult with driver writer to know about this before writing their application? Looks like this is not a tech question ...
----------------------------------------
> Date: Thu, 11 Jul 2013 21:56:39 -0700
> From: gregkh@xxxxxxxxxxxxxxxxxxx
> To: unicorn_wang@xxxxxxxxxxx
> CC: oneukum@xxxxxxx; linux-usb@xxxxxxxxxxxxxxx
> Subject: Re: question on skel_read func of usb_skeleton.c
>
> On Fri, Jul 12, 2013 at 04:49:54AM +0000, WangChen wrote:
>> Hi, Oliver,
>> Regarding skel_write, I see your current desgin only refuse its
>> execution when> WRITES_IN_FLIGHT are on the fly, but this is not
>> blocking IO due to write() will not wait before callback returns,
>> right? Do you think it's unnecessary to support blocking IO on write
>> or any special reason else?
>
> Take a step back.
>
> What type of USB driver do you wish to write? For what type of device?
> That will answer usually all of these types of questions as the
> userspace api for the device will dictate this type of thing.
>
> thanks,
>
> greg k-h ےôè؛{.nا+‰·ں®‰†+%ٹثے±éف¶¥ٹwے؛{.nا+‰·¥ٹ{±ë)ي…وèw* jg¬±¨ ¶‰ڑژٹف¢jے¾«G«éے¢¸ ¢·¦j:+v‰¨ٹwèjطm¶ںےّ ¯ù ®w¥ٹàf£¢·hڑڈâْے†ظ¥
