Thank you for your answer.
I'm sorry my previous question is simplified, cause you misunderstand my needs. My exactly operating flow is:
(1) Firstly, get the camera video data from the IPU, the data format is packed U Y V Y U Y V Y......;
(2) Secondly, I need deal with the data, use some specific algorithm, such as yuv' = fun(yuv), to produce new yuv data, then convert the new yuv to rgb to display.
For this application, because the IPU cann't do the yuv' = fun(yuv) , I should put the packed UYVY data to GPU, with glTexDirectVIVMap()function ( or have other better method? please point out, thanks ), then sample the yuv data in fragment shader. I know i can use GL_VIV_UYVY format with glTexDirectVIVMap()function, but when use rgba = texture2D() in fragment shader, How rgba corresponding to yuv ? and because of the data is packed UYVY, is the MSAA can work correctly? We need anti-aliasing image.