Framebuffer: client notification mecanism & PM
Review line 150
Quoted 10 lines · Benjamin Herrenschmidt
+++
linuxppc-2.5-benh/drivers/video/fbmem.c2003-07-23 20:05:02.000000000 -0400(quoted diff) +2
-0@@ -143,6 +143,8 @@ extern int sstfb_setup(char*); extern int i810fb_init(void); extern int i810fb_setup(char*); +extern int ibmlcdfb_init(void); +extern int ibmlcdfb_setup(char*); extern int ffb_init(void); extern int ffb_setup(char*); extern int cg6_init(void);
Oops... You probably did not want IBM LCD diffs in fbmem.c posted.
matroxfb tried to use a 'dead' for handling hot removal, but your code looks definitely saner.
Petr
Quoted 4 lines · Petr Vandrovec
Oops... You probably did not want IBM LCD diffs in
fbmem.cposted.matroxfb tried to use a 'dead' for handling hot removal, but your code looks definitely saner
Hrm... right, that ibmlcd bit can go ;) Hopefully someone else will get the real ibmlcd driver in sooner or later...
My code wasn't really intended to deal with hot-removal, more with "what happens if printk occurs during the Power Management suspend sequence", but I tried to keep the notification mecanism simple enough so it could be used for that as well. Also, indeed, the fbcon changes should deal with hot-removal in some way (though you surely also want to "deatch" from the fbdev's in this case).
Among other things that could be used for is live monitor insertion/ removal detection (some HW are able to do that), "pivot" monitor kind of things, etc... typically via the mode_changed hook.
Ben.
Quoted 11 lines
Quoted 3 lines
matroxfb tried to use a 'dead' for handling hot removal, but your code looks definitely saner
My code wasn't really intended to deal with hot-removal, more with "what happens if printk occurs during the Power Management suspend sequence", but I tried to keep the notification mecanism simple enough so it could be used for that as well. Also, indeed, the fbcon changes should deal with hot-removal in some way (though you surely also want to "deatch" from the fbdev's in this case).
Quoted 3 lines
Among other things that could be used for is live monitor insertion/ removal detection (some HW are able to do that), "pivot" monitor kind of things, etc... typically via the
mode_changedhook.
I knew it was a matter of time before "client" management would happen. Is this a 2.6.X thing tho or shoudl we wait for the next developement cycle. I don't mind working on experimental stuff.
Quoted 10 lines · James Simmons
Quoted 10 lines
Quoted 2 lines
matroxfb tried to use a 'dead' for handling hot removal, but your code looks definitely saner
My code wasn't really intended to deal with hot-removal, more with "what happens if printk occurs during the Power Management suspend sequence", but I tried to keep the notification mecanism simple enough so it could be used for that as well. Also, indeed, the fbcon changes should deal with hot-removal in some way (though you surely also want to "deatch" from the fbdev's in this case).
Quoted 7 lines
Quoted 3 lines
Among other things that could be used for is live monitor insertion/ removal detection (some HW are able to do that), "pivot" monitor kind of things, etc... typically via the
mode_changedhook.I knew it was a matter of time before "client" management would happen. Is this a 2.6.X thing tho or shoudl we wait for the next developement cycle. I don't mind working on experimental stuff.
We need that now for proper power management.
Ben.
Quoted 6 lines
Quoted 4 lines
I knew it was a matter of time before "client" management would happen. Is this a 2.6.X thing tho or shoudl we wait for the next developement cycle. I don't mind working on experimental stuff.
We need that now for proper power management.
Okay.
-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including Data Reports, E-commerce, Portals, and Forums are available now. Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01
Quoted 4 lines
Quoted 4 lines
I knew it was a matter of time before "client" management would happen. Is this a 2.6.X thing tho or shoudl we wait for the next developement cycle. I don't mind working on experimental stuff.
Note, BTW, there's also a locking issue. Typically, we must make sure the fbcon (which may be running another CPU) have finished doing whatever it was doing before returning, and we don't get any mode change or whatever ioctl initiated thing at that moment (or be in the middle of doing it).
Right now, my code assumes any operation on the fbdev is protected by the console sem. However, looking at the console & fbmem code, that is not always true. Typically, userland ioctls in fbmem aren't, and vt.c may call into consoles in some cases without holding it neither.
That should probably be fixed, though I haven't had time to work on it yet.
Ben.