On Wed, Sep 17, 2003 at 08:37:29PM +0100, James Simmons wrote:
quoted
- fbmem.c is responsible for managing the lifetime of the fb_info structure,
and the lifetime of the driver-private data is equal to the lifetime
of the fb_info structure.
This is not always the case. For dual headed graphics cards you can have
two struct fb_info and one driver-private data. If the fbdev driver
releases one framebuffer you still have one framebuffer left that is using
the driver-private data.
If this is not true, then how can anyone even be thinking about applying
the sysfs / driver model to the framebuffer drivers.
Unless this sort of data lifetime rules are clearly and unabiguously
defined, there is no way that a sysfs / driver model "port" will ever
come anywhere near being stable.
You can't get around these problems with module usage counting - module
usage refcounting is solely about code lifetimes. This is all about data
lifetimes - when is the data created, when can it be safely destroyed,
etc.
--
Russell King (rmk@arm.linux.org.uk) http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
2.6 ARM Linux - http://www.arm.linux.org.uk/
2.6 PCMCIA - http://pcmcia.arm.linux.org.uk/
2.6 Serial core
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
quoted
quoted
- fbmem.c is responsible for managing the lifetime of the fb_info structure,
and the lifetime of the driver-private data is equal to the lifetime
of the fb_info structure.
This is not always the case. For dual headed graphics cards you can have
two struct fb_info and one driver-private data. If the fbdev driver
releases one framebuffer you still have one framebuffer left that is using
the driver-private data.
If this is not true, then how can anyone even be thinking about applying
the sysfs / driver model to the framebuffer drivers.
Unless this sort of data lifetime rules are clearly and unabiguously
defined, there is no way that a sysfs / driver model "port" will ever
come anywhere near being stable.
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
quoted
quoted
- fbmem.c is responsible for managing the lifetime of the fb_info structure,
and the lifetime of the driver-private data is equal to the lifetime
of the fb_info structure.
This is not always the case. For dual headed graphics cards you can have
two struct fb_info and one driver-private data. If the fbdev driver
releases one framebuffer you still have one framebuffer left that is using
the driver-private data.
If this is not true, then how can anyone even be thinking about applying
the sysfs / driver model to the framebuffer drivers.
Unless this sort of data lifetime rules are clearly and unabiguously
defined, there is no way that a sysfs / driver model "port" will ever
come anywhere near being stable.
Sorry about the earlier email sent too early.
I'm attempting to look long term. We will have graphics cards with
built in video as well. Plus 3D stuff. So my attempt was to create a driver
core that was independent of any api. Then we would have mappers to
different api's (fbdev/v4l etc). Thus different api sturctures could share
the same core.
As for sysfs. I really don't see the need to export a par to userspace.
In fact it could be dangerous to mess with. Now fb_info is important to
export to userspace.
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf