> -----Original Message----- > From: Tomi Valkeinen [mailto:tomi.valkeinen@xxxxxxxxx] > Sent: Tuesday, July 27, 2010 1:14 PM > To: Taneja, Archit > Cc: linux-omap@xxxxxxxxxxxxxxx > Subject: RE: [PATCH v2] OMAP: DSS2: Fix error path in > omap_dsi_update() > > On Tue, 2010-07-27 at 09:34 +0200, ext Taneja, Archit wrote: > > > > > > Looks ok to me, I'll add it to my tree. Are you actually using > > > dsi_update_screen_l4? > > > > > > Tomi > > > > There is a usecase in OMAP4 where we can run 4 displays in > parallel, 3 > > can be run using the 3 dispc channels, and another DSI panel can be > > run using L4 updates. > > Ok. > > > It would probably be better if we can connect 2 or more > peripherals to > > one DSI PHY, but with l4 updates we don't need to introduce any > > hardware changes on the boards. > > > > So with l4 update we can free one dispc channel. I am not sure how > > much the > > 4 display usecase is going to be used. But having l4 > updates is still > > a handy feature, > > The current implementation of dsi_update_screen_l4() is very > bad performance-wise, for throughput and for being basically > a busy loop. > > Much better would be to use system DMA for the transfer. In > theory this is possible, and I've made some tests with it, > but I wasn't able to solve the problems how to inject DSI > commands (write_memory_start and write_memory_continue), and > how to get unpacked 24bit pixels from the memory changed into > packed 24bit pixels for the DSI transfer. I can try to look up more into this, I have no idea about it's improvement in terms of performance numbers, but preventing the cpu load makes it worth it. Archit-- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
