hi Eugene. Thanks for the replies so far.
I disabled CONFIG_FRAMEBUFFER_CONSOLE in kernel and the driver loads fine.
Here are the steps I performed:
#insmod llad.ko
#insmod em8xxx.ko
#osdbuf_control -c1280x768 -Ftc:32bpp
#insmod em8xxxfb.ko `osdbuf_control -p`
# osdbuf_control -i
===================== Buffer Infos ======================
Persistent OSD at 0x1763FC00 with 3932160 bytes
Format : tc:32bpp
Resolution : 1280x768
Color Space Conversion : YUV 601
Input Resolution : 4096x4096
Taps : 4
Anti Flicker Color : 2
Anti Flicker Alpha : 2
Direction estimation : enabled
Alpha0 : 128
Alpha1 : 0
Fading : disabled
OSD output channel status : enabled
=========================================================
#echo depth=16 >> /etc/directfbrc or /root/.directfbrc
I then applied the patch you suggested at fbdev.c.
With that, i can run some dfb tests -- dfbscreen, dfbinfo, etc but some
others -- such as dfb_knuckles etc just aborts:
Before Fix alignment 0x1763fc00 -> 0x2ae68000
Fix alignment 0x1763fc00 -> 0x2ae68c00
*) FBDev/Mode: Testing 1280x768 RGB16
(*) FBDev/Mode: Preparing switch to 1280x768 RGB16
(!) DirectFB/FBDev: Unsupported pixelformat: rgba 5/0, 6/5, 5/11, 0/0
(16bit)
DURAI : Forcing pixel format :
(*) FBDev/Mode: Switched to 1280x768 (1280x1536) at 16 bit ARGB (wanted
RGB16).
(*) FBDev/Surface: Allocated 1280x768 16bit RGB16 buffer at offset 0 and
pitch 2560.
(*) FBDev/Mode: (Post)Setting 1280x768 RGB16
(*) FBDev/Surface: Allocated 1280x768 16bit RGB16 buffer at offset 1966080
and pitch 2560.
(!) [ 772: 0.000] --> Caught signal 11 (at 0x2b228000, invalid
permissions) <--
(!!!) *** WARNING [still objects in 'Layer Region Pool'] *** [object.c:239
in fusion_object_pool_destroy()]
(!!!) *** WARNING [still objects in 'Layer Context Pool'] ***
[object.c:239 in fusion_object_pool_destroy()]
(!!!) *** WARNING [still objects in 'Surface Pool'] *** [object.c:239 in
fusion_object_pool_destroy()]
Aborted
#
is the dfb trying to access memory beyond what is allocated by the driver? :
Start address of driver: 0x2ae68000
number of bytes allocated by driver: 3932160
End address = 0x2b228000
This end address is where DFB aborts. is it that after the alignment, the
start address becomes 0x2ae68c00 but DFb still wants to access upto
0x2b228c00 (0x2ae68c00 + 3932160 bytes)?
Further, the address before the fix (0x2ae68000) seems to be aligned to the
4K boundary? i am using rev ES9 of the chip.
if i do not apply your patch, the OSD buffer gets deleted right after i run
any DFB sample (even dfbinfo)
Any thoughts?
Thanks a lot again,
--
View this message in context:
http://www.nabble.com/sigma-8634-8635-fix-tf4199088.html#a13266167
Sent from the DirectFB Dev mailing list archive at Nabble.com.
_______________________________________________
directfb-dev mailing list
[email protected]
http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
