Re: [PATCH] usb gadget: Endpoint configuration

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Tue, Nov 15, 2011 at 10:23:09AM +0100, Praveena NADAHALLY wrote:
> Hello Felipe,
> 
> Can you check and provide your comments for 
> the mail below?
> 
> -----Original Message-----
> From: Praveena NADAHALLY
> Sent: Wednesday, October 26, 2011 7:03 PM
> To: 'balbi@xxxxxx'
> Cc: linux-usb@xxxxxxxxxxxxxxx; linus.walleij@xxxxxxxxxx; Sakethram
> BOMMISETTI
> Subject: RE: [PATCH] usb gadget: Endpoint configuration
> 
> > What's the rationale for that ?
> Already there are platform specific EP handling code for other
> platforms in epautoconf.c file in usb_ep_autoconfig().
> Adding a new  gadget_ops is based on a comment in 
> usb_ep_autoconfig(). Instead of changing the code in 
> epautoconf.c as done for few platforms, we thought it's a 
> good idea to add a gadget_ops so that each platform
> can handle their EP configuration stuffs in their 
> platform files.
> 
> > Why is ux500 so important that it needs
> > to handle endpoint configuration by itself ?
> > Why don't you just add a fifo_mode table for ux500 instead of adding
> > this hack ?
> 
> fifo_mode table is a onetime configuration during initialization. We
> can't optimize this table for different USB configurations(like MSC,
> serial, ECM as one config and ECM, serial as another config) 
> which user can change at run time by doing soft disconnect and 
> re-connect with a new configuration.
> 
> In ux500, we would like handle EP configurations separately 
> for the following reasons:
> * Allocate smaller than 512 bytes max packet EP to interrupt EP.
> * Allocate a particular EP for a particular class.
> This is mainly for allocating a double buffered and 
> DMA capable EP for a particular class like MSC...
> * Any other platform specific checks...

what we need with musb are patches which will benefit all users, not
only ux500 or omap or any other. There are a bunch of registers on musb
address space where you can ask the hardware how much FIFO space it has
and how many endpoints it has, based on that, we need a way to better
allocate FIFO space. Adding another function pointer which will only
benefit ux500 is a no-no.

All those fifo tables should go away and we should allocate fifo space
dynamically, that's why we have dynamic fifo sizing on musb to start
with.

-- 
balbi

Attachment: signature.asc
Description: Digital signature


[Index of Archives]     [Linux Media]     [Linux Input]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Old Linux USB Devel Archive]

  Powered by Linux