> -----Original Message----- > From: Felipe Balbi [mailto:ba...@ti.com] > Sent: Tuesday, January 27, 2015 8:38 PM > To: Krzysztof Opasiak > Cc: ba...@ti.com; linux-usb@vger.kernel.org; > gre...@linuxfoundation.org; bige...@breakpoint.cc; > s.wa...@samsung.com; k.lewando...@samsung.com; > m.szyprow...@samsung.com; andrze...@samsung.com > Subject: Re: [PATCH] usb: gadget: composite: Provide list of > registered functions > > Hi, > > On Tue, Jan 27, 2015 at 08:20:58PM +0100, Krzysztof Opasiak wrote: > > > > > > > -----Original Message----- > > > From: Felipe Balbi [mailto:ba...@ti.com] > > > Sent: Tuesday, January 27, 2015 7:45 PM > > > To: Krzysztof Opasiak > > > Cc: ba...@ti.com; linux-usb@vger.kernel.org; > > > gre...@linuxfoundation.org; bige...@breakpoint.cc; > > > s.wa...@samsung.com; k.lewando...@samsung.com; > > > m.szyprow...@samsung.com; andrze...@samsung.com > > > Subject: Re: [PATCH] usb: gadget: composite: Provide list of > > > registered functions > > > > > > Hi, > > > > > > On Tue, Jan 27, 2015 at 06:24:42PM +0100, Krzysztof Opasiak > wrote: > > > > > > So I must have misunderstood something. > > > > > > > > > > > > I'm not sure if this is a good idea. > > > > > > Some userspace depends on assumption that > sys/kernel/config > > > > > directory > > > > > > is empty and it's default place for mounting configfs. > > > > > > > > > > and that's fine, they can certainly assume that. Once > configfs > > > is > > > > > mounted, we will have a new /sys/kernel/config/usb- > functions > > > > > directory. > > > > > > > > > > Inside that directory we should have a file which contains > your > > > list > > > > > of available functions :-) I don't see what the problem is > with > > > that > > > > > :-s > > > > > > > > > > > > > Please hold done I don't get it. Let's clarify. > > > > > > _________________________________________ > > > / Hello! I'm a programmer cow who can \ > > > | explain ConfigFS-related questions much | better than Felipe > > > Balbi. > > > | What we need | > > > | in this case is just for the our USB | > > > | Gadget ConfigFS port to create the file | > > > | when that is mounted. Just have the | > > > | file come with /usb_gadget. Not ProcFS, | not SysFS, but with > > > ConfigFS > > > | and we're | > > > | good to go. Let me know if I can | > > > \ explain anything further / > > > ----------------------------------------- > > > \ ^__^ > > > \ (Oo)\_______ > > > (__)\ )\/\ > > > ||----w | > > > || || > > > > > > > Would you like to register a separate usb-functions subsystem > in > > > > configfs only to expose there a list of available functions? > As > > > far as > > > > I know it's the only way of creating anything in configfs > root. > > > > > > > > So you would get: > > > > > > > > $ ls /sys/kernel/config > > > > usb-gadget usb-functions > > > > > > > > > > > > Or it is just a typo and you would like to place a usb- > functions > > > > attribute in usb-gadget directory? > > > > In this option we will get: > > > > > > > > $ ls /sys/kernel/config > > > > usb-gadget > > > > $ ls /sys/kernel/config/usb-gadget usb-function $ cat usb- > function > > > > acm ecm ... > > > > > > ______________________________________ > > > / Hello again. This is exactly what I \ > > > | have in mind. Thank you. Let's just | call it a more > descriptive > > > name > > > | like | > > > \ 'available_functions' / > > > -------------------------------------- > > > \ ^__^ > > > \ (Oo)\_______ > > > (__)\ )\/\ > > > ||----w | > > > || || > > > > > > > I have written this few times in previous emails but our new > friend > > could miss those emails so once again esp. for beautiful cow from > a > > rubber duck: > > > > IMPORTANT______________________________________________ > > / \ > > | Adding a file to usb_gadget directory is an *ABI break*.| > > an ABI break would be removing an existing file, not adding a new > one.
ABI break is a change of behavior without backward compatibility > > > \_________________ _____________________________________ / > > // > > // > > ..---..// > > .' _ `. > > __..' (o) : > > `..__ ; > > `. / > > ; `..---...___ > > .' `~-. .-') > > . ' _.' > > : : > > \ ' > > + J > > `._ _.' > > `~--....___...---~' mh > > > > Currently each valid file name is also suitable for gadget. > > There is no additional restrictions. > > > > No matter what name we will give to this new file this name could > be > > used previously by someone in userspace. > > His program could be broken after kernel upgrade. > > the first thing a user of usb-gadget has to do, is create a > directory, not a file. This: $ modprobe libcomposite $ mount none -t configfs /sys/kernel/config/usb-gadget $ mkdir available_functions $ cd available_functions $ mkdir functions/ecm.eth0 $ mkdir configs/c.1 $ ln -s functions/ecm.eth0 configs/c.1 $ echo `ls /sys/class/udc | tail -n 1` > UDC works fine with current kernel and will *fail* after adding available_functions file due to name already taken even immediately after system boot. > > > Moreover some broken userspace programs may relay on fact that > > usb-gadget directory contains only gadget dirs. And some other > things > > there is a lot of possible breaks in userspace. > > if there are any users who rely on the fact that *only* directories > exist under usb-gadget, that's wrong. I know that it's wrong but that's what Documentation/ABI says about usb-gadget directory. > > Imagine is sysfs would be forbidden of adding new files and > directories. > That's inside, it's crazy-talk :-) > Objects in sysfs are managed by kernel. Kernel creates all files there (even if you are dealing with gpio you have to write name/id to export file and kernel will create attribute for you) and *kernel will handle their naming*. This means that user cannot add there an arbitrary file/directory. He may only read/write existing one. In this case adding a new file doesn't change the behavior but only adds a new attribute. This doesn't break the ABI. In configfs all naming issues are *handled by user*. User decides what is the name for his gadget. Adding a file to usb-gadget directory effectively adds a new restriction for gadget name. This may cause some programs which are currently working to fail with newer kernel. Best regards, -- Krzysztof Opasiak Samsung R&D Institute Poland Samsung Electronics k.opas...@samsung.com -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
