> Mario Goebbels wrote: > > What's SUNWxwmod? What do these kernel modules do? > Xorg seems to run fine without them. > > The hardware cursor kernel module. Xorg doesn't > actually use it yet, but > porting the code to do so from Xsun to Xorg is on our > todo list. > > On systems with in-kernel fb drivers (i.e. SPARCs) > that support the right > ioctls, Xsun will push the hwc streams module onto > the stream coming out > of /dev/mouse - this module will then take all mouse > movements and translate > them into instructions to the fb driver to update the > cursor location on > screen - without waiting for the Xsun process to be > scheduled, the system > to context switch from kernel to Xsun user space, > Xsun to get back to it's > main event loop, select() the mouse device and > process the input from it. > On the SPARC workstations of the 90's, this made > cursor response much smoother > and gave users more natural feeling feedback. > > Whether we'll get x86 kernel graphics drivers like > nvidia or DRI to support > similar ioctls is another open question. One of > this year's X.Org projects > for Google Summer of Code involved investigating > solutions to smoother cursor > response for modern graphics hardware/software that > may result in a different > solution as well.
The increased responsiveness that gives is truly amazing. The only drawback (I don't know I'd really call it a fault, as such) is that the mouse cursor responsiveness remains even if the X server is wedged; it pretty much takes a hung kernel to make the mouse cursor responsiveness change. That makes it rather different to figure out what's gone wrong, depending on whether that feature is or isn't being used. Although if one pays close attention, one will see that when the X server is hung, the mouse cursor does not change shape in any of the various places (window borders, etc) it normally would. -- This message posted from opensolaris.org
