[#6592] PPI hardware/driver often capture corrupt images from image chip
Submitted By: Marius Kotsbak
Open Date
2011-05-11 05:30:27
Priority:
Medium Assignee:
Scott Jiang
Status:
Open Fixed In Release:
N/A
Found In Release:
2011R1 Release:
Trunk (ec45f5f)
Category:
Drivers Board:
Bluetechnix CM
Processor:
BF537 Silicon Revision:
Is this bug repeatable?:
Yes Resolution:
Assigned (Not Start)
Uboot version or rev.:
Toolchain version or rev.:
09r1-11
App binary format:
N/A
Summary: PPI hardware/driver often capture corrupt images from image chip
Details:
Often the images captured by the bfin_ppi linux module are corrupted, even if no error (under/overflow) is detected/reported. It seems like some pixels are moved to the previous line.
See the attached image for an example. Note the errors on the right side. The image is supposed to be light in the one quarter in the top left and darker in the rest of it. Clearly the stripe on the right is pixels from the left part.
I am not certain if this is caused by driver error, processor error or something else.
Follow-ups
--- Marius Kotsbak 2011-05-11 05:48:13
The image chip is a Micron MT9P031 and the settings used are:
{ CMD_PPI_SKIPPING, CFG_PPI_SKIP_DISABLE },
{ CMD_PPI_DATALEN, CFG_PPI_DATALEN_8 },
{ CMD_PPI_PACKING, CFG_PPI_PACK_ENABLE },
{ CMD_PPI_CLK_EDGE, CFG_PPI_CLK_EDGE_FALL },
{ CMD_PPI_TRIG_EDGE, CFG_PPI_TRIG_EDGE_RISE },
{ CMD_PPI_XFR_TYPE, CFG_PPI_XFR_TYPE_NON646 },
{ CMD_PPI_PORT_CFG, CFG_PPI_PORT_CFG_XSYNC23 },
{ CMD_PPI_SET_DIMS, CFG_PPI_DIMS_2D },
{ CMD_PPI_DELAY, 0 },
{ CMD_PPI_LINELEN, 2592 }
{ CMD_PPI_NUMLINES, 1944 }
--- Mike Frysinger 2011-05-11 13:21:35
didnt you already file this as [#6543] ?
--- Marius Kotsbak 2011-05-11 15:01:37
No, that is a similar bug, but not exactly the same symptoms, and that is for
EPPI in BF548 and this one PPI in BF537.
--- Scott Jiang 2011-10-21 03:28:39
It may caused by timing mismatch in hardware or software. The resolution seems
strange, you may use oscilloscope to monitor real timing.
--- Marius Kotsbak 2011-10-21 05:58:11
Yes, we have monitored this with oscilloscope too without finding any errors in
the signals. But now we have given up the whole platform and moved over to ARM
from TI as Analog Devices does not seem to be interested in solving neither the
problems with EPPI or PPI. There we don't have any reading problems with the
same image chip hardware and working Linux drivers are available.
Files
Changes
Commits
Dependencies
Duplicates
Associations
Tags
File Name File Type File Size Posted By
2011-05-03_13-48-52.bmp image/bmp 5039926 Marius Kotsbak