Hi Daniel,
This sounds like you may be hitting errata 1.3.24 (detailed in http://focus.ti.com/lit/er/sprz241e/sprz241e.pdf) - this relates to potential errors when multiple masters access ARM DTCM internal memory concurrently. - i.e. in your case, ARM and EDMA. ~roger ________________________________ From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of George Gu Sent: 10 January 2007 09:17 To: 'Daniel Cvetanov'; [email protected] Subject: RE: Strange EDMA behaviour Daniel, I think you are using EDMA controlled by DSP, so the problem maybe caused by Cache problem, you can try to flush DSP L2 Cache. George Gu (Xiangyu Gu) Wintech Digital System Co. Beijing, China Ph: 8610-8278-2828 ext.168 Fax: 8610-8278-0028 http://www.wintechdigtal.com <http://www.wintechdigtal.com/> ---------------------------- ________________________________ From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Daniel Cvetanov Sent: 2007年1月10日 16:55 To: [email protected] Subject: Fwd: Strange EDMA behaviour Friends, I'm using the EDMA to fill continuous 2D regions in the DDR2 memory with a given sample. The sample value is a 32-bit constant, setup in the internal RAM. I'm doing a very simple test, in which I'm filling continuously the framebuffer with black colour (0x10801080 sample value) I observe the following very strange behaviour of the EDMA: 1. When the EDMA source is in the internal RAM, 1 sample (4bytes) out of 207360 samples (for PAL framebuffer) is with a different colour, found on arbitrary position on the screen. 2. When the EDMA source is an IO register (I found one that can be used for the purpose), everything is OK. Have you observed such a thing? I suppose that this is something related to the bus speed, but I'm out of ideas how I can debug and track this. I already: * Checked that noone is modifying the source sample * Tried with different locations and banks within the internal RAM * Tried with different sample values but the result is always the same -> two nasty pixels on a cleared display Smiles, Daniel Cvetanov
_______________________________________________ Davinci-linux-open-source mailing list [email protected] http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
