Date: Tue, 10 Aug 2010 14:36:57 +0200 From: Guillaume Ballet <gballet@gmail.com> To: freebsd-arm@freebsd.org, ben.r.gray@gmail.com Subject: Re: OMAP3530 - Beagleboard and I2C problems Message-ID: <AANLkTimXvb4C_=ekd4hfXJ89-jPqyPhYy6E8kXVL9o%2BO@mail.gmail.com>
next in thread | raw e-mail | index | archive | help
Hey Ben, I have been working for a port myself, at http://code.google.com/p/beagleboard-freebsd/ Haven't had the time to work on it lately, though. > Message: 2 > Date: Mon, 09 Aug 2010 22:42:17 +0100 > From: Ben Gray <ben.r.gray@gmail.com> > Subject: OMAP3530 - Beagleboard and I2C problems > To: freebsd-arm@freebsd.org > Message-ID: <4C607639.9050506@gmail.com> > Content-Type: text/plain; charset=3DISO-8859-1; format=3Dflowed > > Hi, > > =A0 =A0I've been working on a port of FreeBSD to Texas Instruments OMAP35= 30 > for a while now. I have the basic drivers, Clocks, MMC, DMA, GPIO's, > etc. The kernel is coming up, however it crashes with a seg fault when > starting the init process, this is probably caused by the hacks I had to > put in the pmap code to get it to work with ARMv7 MMU's, but that's an > email for another day. > > =A0 =A0The problem I'm currently having is with the I2C code and it is > perhaps highlighting a more fundamental problem with my port so far. > Currently I have an 'omap3' bus device, which is the parent for all the > other peripheral drivers (I2C, MMC, DMA, etc). =A0These child bus drivers > are added during the attach call of the parent, but at that time IRQ's > are not enabled and nor is the system clocking code. So the problem I > have is that during the attach phase of the I2C device driver I want to > be able to send messages over the I2C bus, but my bus driver doesn't > work because IRQ's are still disabled. =A0So what is the correct solution > for this? =A0How do I delay device initialisation to after this? I had some basic intc controller that I tested with the timer. I also have an even more basic interrupt allocation support in the form of a bus. I also have some changes for the pmap you might be interested in. I can check at what I have tonight that you don't have and send it to you. I will help you with the merge of what you don't already have. > > =A0 =A0Is there documentation anyone can point me at that describes the > startup phases? or alternatively examples in the existing code base? > I've looked at the current ARM ports, and on the face of it they seem to > suffer the same problem. > I only took a quick look at your code (once again, will take a closer look at it tonight), and I didn't find the part where you are inserting the devices. As far as I recall (possibly incorrectly), you can specify in which order you insert devices. What I did was to insert the intc device first, so that it would be ready when inserting other drivers. > > =A0 =A0In case anyone is interested some of my work to date is available = on > googlecode here <http://code.google.com/p/beagleboard-freebsd/>. At the > moment it can't just be pulled into the tree, partially because it's > missing the patches needed for the pmap code, but mostly because large > parts of it don't work properly :-) I'll try it with my own pmap changes :) > > =A0 =A0And thanks to to the author of the freebsd-bgb googlecode project, > without that I probably wouldn't have got started. You are welcome. I'm still interested in the port even if I have less time to work on it for the moment - still I'm available for doing some work, though :) Guillaume
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?AANLkTimXvb4C_=ekd4hfXJ89-jPqyPhYy6E8kXVL9o%2BO>