<br><br><div class="gmail_quote">On 28 February 2011 11:33, Laurent Pinchart <span dir="ltr"><<a href="mailto:laurent.pinchart@ideasonboard.com">laurent.pinchart@ideasonboard.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<div><div></div><div class="h5">On Monday 28 February 2011 11:21:52 Hans Verkuil wrote:<br>
> On Monday, February 28, 2011 11:11:47 Laurent Pinchart wrote:<br>
> > On Saturday 26 February 2011 13:12:42 Hans Verkuil wrote:<br>
> > > On Friday, February 25, 2011 18:22:51 Linus Walleij wrote:<br>
> > > > 2011/2/24 Edward Hervey <<a href="mailto:bilboed@gmail.com">bilboed@gmail.com</a>>:<br>
> > > > > What *needs* to be solved is an API for data allocation/passing at<br>
> > > > > the kernel level which v4l2,omx,X,GL,vdpau,vaapi,... can use and<br>
> > > > > that userspace (like GStreamer) can pass around, monitor and know<br>
> > > > > about.<br>
> > > ><br>
> > > > I think the patches sent out from ST-Ericsson's Johan Mossberg to<br>
> > > > linux-mm for "HWMEM" (hardware memory) deals exactly with buffer<br>
> > > > passing, pinning of buffers and so on. The CMA (Contigous Memory<br>
> > > > Allocator) has been slightly modified to fit hand-in-glove with<br>
> > > > HWMEM, so CMA provides buffers, HWMEM pass them around.<br>
> > > ><br>
> > > > Johan, when you re-spin the HWMEM patchset, can you include<br>
> > > > linaro-dev and linux-media in the CC?<br></div></div></blockquote><div><br>Yepp..Johan will do that (his mail is fubar at the moment so I will answer instead :-) )<br> </div><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<div><div class="h5">
> > ><br>
> > > Yes, please. This sounds promising and we at linux-media would very<br>
> > > much like to take a look at this. I hope that the CMA + HWMEM<br>
> > > combination is exactly what we need.<br>
> ><br>
> > Once again let me restate what I've been telling for some time: CMA must<br>
> > be *optional*. Not all hardware need contiguous memory. I'll have a look<br>
> > at the next HWMEM version.<br></div></div></blockquote><div><br>hwmem API has Scattered memory support as well (not implemented yet though)<br> </div><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<div><div class="h5">
><br>
> Yes, it is optional when you look at specific hardware. On a kernel level<br>
> however it is functionality that is required in order to support all the<br>
> hardware. There is little point in solving one issue and not the other.<br>
<br>
</div></div>I agree. What I meant is that we need to make sure there's no HWMEM -> CMA<br>
dependency.<br>
<div><div></div><div class="h5"><br></div></div></blockquote><br>HWMEM has no CMA dependency, although hwmem is easily adapted ontop of CMA(once the speculative prefetch stuff in ARM arch is resolved)<br><br><br>BR<br>/Robert F<br>
<br>_______________________________________________<br><blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"><div><div class="h5">
st-ericsson mailing list<br>
<a href="mailto:st-ericsson@lists.linaro.org">st-ericsson@lists.linaro.org</a><br>
<a href="http://lists.linaro.org/mailman/listinfo/st-ericsson" target="_blank">http://lists.linaro.org/mailman/listinfo/st-ericsson</a><br>
</div></div></blockquote></div><br>