<div dir="ltr"><br><div class="gmail_extra"><br><br><div class="gmail_quote">On Mon, May 20, 2013 at 3:30 PM, Dag Lem <span dir="ltr"><<a href="mailto:dag@nimrod.no" target="_blank">dag@nimrod.no</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="im">Zhigang Gong <<a href="mailto:zhigang.gong@linux.intel.com">zhigang.gong@linux.intel.com</a>> writes:<br>
<br>
> In general, this patch looks good to me. Could you split it to two patches?<br>
> One is for clEnqueueReadImage/WriteImage. And the other is to refine the<br>
> clEnqueueMapImage. And you may consider to merge the clEnqueueMapImage refinement<br>
> patch and the previous's clEnqueueMapImage patch.<br>
><br>
<br>
</div>OK, I'm working on it.<br>
<div class="im"><br>
> Another issue about the clEnqueueMapImage is that if the image is a tiled<br>
> buffer, then we may need to use clMapBufferGTTIntel to map it to vma. Otherwise,<br>
> we may get a tiled mapping address which is not expected by user.<br>
<br>
</div>OK, so I should use clMapBufferIntel if cl_mem.tiled == CL_NO_TILE,<br>
clMapBufferGTTIntel if cl_mem.tiled == CL_TILE_Y, and throw an error<br>
otherwise (i.e. for the currently unused CL_TILE_X)?<br>
<br>
I guess clEnqueueUnmapMemObject should also be changed to call<br>
clUnmapBufferGTTIntel for tiled images?<br></blockquote><div> You are right. </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<span class="HOEnZb"><font color="#888888"><br>
--<br>
Best regards,<br>
<br>
Dag Lem<br>
</font></span><div class="HOEnZb"><div class="h5">_______________________________________________<br>
Beignet mailing list<br>
<a href="mailto:Beignet@lists.freedesktop.org">Beignet@lists.freedesktop.org</a><br>
<a href="http://lists.freedesktop.org/mailman/listinfo/beignet" target="_blank">http://lists.freedesktop.org/mailman/listinfo/beignet</a><br>
</div></div></blockquote></div><br></div></div>