On 02/10/2011 06:17 PM, jon.schell wrote: > 1. I've been reading some other support threads that you answer, and > wow, you are really unimaginative and constrained. Yes, the pattern > can be done from userspace, very slowly. Too slowly to be useful in > fact. When someone asks a question, the best thing to do is answer > the question and don't try to second-guess their motives or say "don't > do it that way" when you don't know why they're doing it that way, or > what the big picture of what they're doing is, which most people > aren't going to be telling you for a simple question.
The fact that Greg is paying attention to your questions at all is a sign that he's trying to help. Greg is a very experienced kernel expert. When he asks a question about what you are doing, it is so that he can provide the best help possible. When he says that something can be done another way better, it's based on what you've described so far in your e-mails. Very, very often when people inexperienced with the kernel ask how to do something, they are missing knowledge about a kernel system that already supports their need. But also very often, people aren't very forthcoming with what their exact need really is. > 2. No, I'm not creating a general kernel API, I'm creating something > that only I will be using, for my purpose, so it can be "incorrect" or > "not quite right" if I want. As I said, if you know of some other way > to get a block of binary data into the driver, feel free to suggest > it, but this way works great and will end my headache, so I don't care > if it's proper. If you don't plan to maintain the code, or push it to mainline, or release binaries including it to anyone - you are of course free to do anything you want. If you do plan any of the above, it could save you time and trouble to listen to Greg. There are LOTS of ways to move binary data between user space and the kernel. Which one is appropriate depends on a LOT of considerations. > 3. Proprietary code is not published. If you don't plan to ship the kernel to anyone (for example, if it's just part of some internal test framework), then that's right. Otherwise, that's wrong. -- Tim > On Feb 10, 11:57�am, Greg KH <gre...@gmail.com> wrote: >> On Thu, Feb 10, 2011 at 11:42 AM, jon.schell <jon.sch...@kyocera.com> wrote: >>> Yes, I'm sure it can't handle it. �Not a PWM, doing something like: >>> Turn on for 3 ms, turn off for 5ms, turn on for 1 ms, turn off for 1 >>> ms, turn on for 2 ms, turn off for 8 ms, etc. for any arbitrary >>> pattern you want to do. >> >> That type of pattern is easy to do from userspace, please do it that >> way and not from within the kernel, that's not the proper place to do >> it. >> >>> I'm not breaking anything with userspace as it's not a device anyone >>> will/can access except for my app. >> >> You are creating a user/kernel API which is important to get correct. >> >>> �And it won't be released into the code base. >> >> What do you mean by this? �All kernel code must be published when >> the product is shipped. >> >> thanks, >> >> greg k-h > -- ============================= Tim Bird Architecture Group Chair, CE Linux Forum Senior Staff Engineer, Sony Network Entertainment ============================= -- unsubscribe: android-kernel+unsubscr...@googlegroups.com website: http://groups.google.com/group/android-kernel
