On 11/02/2010 06:29 AM, Frias Renato-B13784 wrote:
Hello Stefano,
Hi Ricardo,
However, in order to add support for the second display inteface on the imx51, I had to change the Framebuffer driver. Thus, support for vision2 will break, to restore it all is needed is to add couple arguments to mx51_fb_init call.
I see, you add the display interface to the parameter list of the probe() function.
However, it is really better to make the modification for the vision2 inside the same patchset. This guarantees that both boards work when your patches go to mainline.
Please review the patches and let me know if I should have done things in a different way.
ok
There is also a second issue where I would like to know your thoughts. Very early on system initialization, when LCD is enabled, there is a call to "lcd_setmem" from board.c. By that time, the video variables, "panel_info", are not set yet. Thus U-Boot doesn't reserve the appropriate amount of memory for the display. I was going to set "panel_info" variable on mx51evk.c, but I would like to know how you solved it for vision2 first.
Thanks for pointing out, I have already seen the point, but at the end I forget to fix it ;-). Probably because there is no side-effects on this board, but this does not mean that the issue should not fixed for vision2, too.
The problem arises from the fact that the setup of the display parameters was static and everything was solved at compile time. With the framebuffer for the MX51, I needed the possibility to change dinamically the parameters, because the board can have different LCD displays.
Consider this, I do not think the actual computation in lcd_setmem() is correct. We need to compute the maximum amount of memory to be reserved to the framebuffer, not the value requested by the current display interface. We could add a CONFIG_SYS_VIDEO_SIZE that contains the maximum amount of memory needed, because the value is strictly board-dependent, and change lcd_setmem to use it. Anatolij, what do you think about this ?
Stefano