Hi there,

I have a CHD2WLANU (MGB100) that exhibits segmentation faults when doing large disk I/O, for instance when performing the timing tests from HDPARM or when using DD.

DMESG then reports spurious IRQ7, illegal operand and some other rather useless info.

I used the same firmware and config on a different system and there all is fine, so this is a hardware issue and not firmware related.

Before I scrap the PCB (I just ordered a new spare system from eBay - China); does anyone have a clue what could cause this?

Unfortunately after reflashing, the seg faults remain.

Just for fun see attached my JTAG setup I used today; a Dremel wrench holding the board and pressing pogo pins to the contacts for JTAG. I don't trust my soldering skills anymore on these tiny contacts. JTAG board (parallel) used from http://www.cdream5.com/htm/xilinx_JTAG.htm (around EUR 25,- on eBay).

To enable JTAG on this board where R24 is not present (see wiki) I used a small sewing needle soldered via a 4K7 resistor to a wire with a banana clip (the yellow clip in the picture) to the VCC pogo pin to pull R24 high - with a magnifying glass it worked at the first try. A kludge  but I was amazed it worked at the first try.



Ernst