stev391@xxxxxxxxx wrote: > Steve, > > I have found the cause of my DMA timeouts, as per your suggestion I > checked the sram settings. > The cause of the issue was in SRAM_CH06 cdt, this was originally set to > 0x10480 and is currently set to 0x108d0. In changeset 7005:a6d2028a4aab > you introduced this as an alternative set of values and then in > changeset 7464:20a1412b4f1a it was all converted to these values. I was > wondering why this value was required to change? > > I have not yet had the time to analyse these values in detail, but the > following are possible options that I/we can persue: > > 1) Set the value back to 0x10480 (diff attached), the following > supported cards will use this value (from a quick glance): > CX23885_BOARD_HAUPPAUGE_HVR1800lp > CX23885_BOARD_HAUPPAUGE_HVR1800 > CX23885_BOARD_HAUPPAUGE_HVR1250 > CX23885_BOARD_HAUPPAUGE_HVR1500Q > CX23885_BOARD_HAUPPAUGE_HVR1500 > CX23885_BOARD_HAUPPAUGE_HVR1200 > CX23885_BOARD_HAUPPAUGE_HVR1700 > CX23885_BOARD_HAUPPAUGE_HVR1400 > CX23885_BOARD_DVICO_FUSIONHDTV_7_DUAL_EXP > > 2) Introduce another variable in struct cx23885_board to allow board > specific srams. The sram would not be duplicated in this struct, a > second version would be included in cx23885-core.c (similar to the 7005 > changeset except it is now manual configurable and not switching on PCI > id). The down side of this is the cx23885-core.c will be slightly > larger, with a larger memory footprint. > > 3) Reallocate entire sram, this would require a detailed look on my > behalf to see how much space each variable requires and reallocate it. > This will potentially break the cards mentioned in option 1 and will > take more time to implement and test. This is highly undesirable my > viewpoint. > > 4) Something else? Please suggest a solution... > > Regards, > > Stephen > > diff -Naur v4l-dvb/linux/drivers/media/video/cx23885/cx23885-core.c > v4l-dvb1/linux/drivers/media/video/cx23885/cx23885-core.c > --- v4l-dvb/linux/drivers/media/video/cx23885/cx23885-core.c > 2008-06-06 14:57:55.000000000 +1000 > +++ v4l-dvb1/linux/drivers/media/video/cx23885/cx23885-core.c > 2008-06-26 08:26:42.000000000 +1000 > @@ -142,7 +142,7 @@ > .name = "TS2 C", > .cmds_start = 0x10140, > .ctrl_start = 0x10680, > - .cdt = 0x108d0, > + .cdt = 0x10480, > .fifo_start = 0x6000, > .fifo_size = 0x1000, > .ptr1_reg = DMA5_PTR1, > > ----- Original Message ----- > From: "Steven Toth" > To: stev391@xxxxxxxxx > Subject: Re: cx23885 driver and DMA timeouts > Date: Mon, 23 Jun 2008 08:51:10 -0400 > > No need to try windows, if you have the driver already running > (pascoe's patches) then your chipset and hardware are fine. > > Sounds like you have a simple merge issue. > > Try to figure out which parts of the merge actually create the > problem then bring that issue back to this list for discussion. > > Regards, > > - Steve Let me go back and look at the existing layout. Something is obviously wrong. We don't want board specific values. - Steve _______________________________________________ linux-dvb mailing list linux-dvb@xxxxxxxxxxx http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
