I found the segfault bug in my 4bpp patch :-p
DirectFB (with fusion kernel patch) runs very well on preemption kernel. most of the benchmark results are increase by 5~10%. The speed of Blit with alpha channel is doubled!
Thank you very much for your help.
Regards
James Su
btw, you may find the preemption kernel patch at http://www.tech9.net/rml/linux/
James Su wrote:
The speed improves about 5~10%. Regards James SuDenis Oliver Kropp wrote:Quoting James Su ([EMAIL PROTECTED]):Hi, Today, I tried DirectFB 0.9.16 and GTK+-2.0.9 on a StrongARM SA1100 with kernel 2.4.18+preemption patch. The DirectFB demo program runs very well and faster than before. But when I run gtk program and try to drag a window, the program always crashed. The log is like:How much faster and why?(!) ---> CAUGHT SIGNAL 11 <--- (!) [50: 13749] DirectFB/core/fusion: (fusion_ref_zero_trylock) FUSION_REF_ZERO_TRYLOCK --> Resource temporarily unavailable (!) [51: 13751] DirectFB/core/fusion: (skirmish_prevail) FUSION_SKIRMISH_PREVAIL --> Bad address (!) [51: 13752] DirectFB/core/fusion: (fusion_ref_down) FUSION_REF_DOWN_GLOBAL --> Resource temporarily unavailable (!) [51: 13752] DirectFB/core/fusion: (skirmish_prevail) FUSION_SKIRMISH_PREVAIL --> Bad address (!) [51: 13753] DirectFB/core/fusion: (fusion_ref_down) FUSION_REF_DOWN_GLOBAL --> Resource temporarily unavailable (!) [51: 13753] DirectFB/core/fusion: (skirmish_prevail) FUSION_SKIRMISH_PREVAIL --> Bad address (!) [51: 13754] DirectFB/core/fusion: (fusion_ref_down) FUSION_REF_DOWN_GLOBAL --> Resource temporarily unavailable (!) [50: 13854] DirectFB/core/fusion: (fusion_ref_zero_trylock) FUSION_REF_ZERO_TRYLOCK --> Resource temporarily unavailable Does anyone can tell me what's the matter with it?Since you got a SIGSEGV you should ignore the error messages. The emergency cleanup code seems to need some polishing.
--
Info: To unsubscribe send a mail to [EMAIL PROTECTED] with "unsubscribe directfb-dev" as subject.
