dissecting a LOOX 720

[rsainio]

As I mentioned in my previous post; I have several LOOX 720 PDA and
one of them is broken, so I could dissect it if you like to know more
about different pin layouts. I have hot-air soldering equipment so I
could desolder the chips and thus have access to the BGA soldering
points.
Anyone there interested in JTAG ??
Background:I have also bunch of LOOX 600 PDAs and with them I was able
to get uboot running and booting directly. This was mainly done with
JTAG.

[Martin]

I'd be interested in knowing whether the DMARQ and DMACK signals of
the PCMCIA socket (pins 43 and 44, see http://www.allpinouts.org/index.php?title=CompactFlash)
are connected to any processor GPIOs. If you find that DMARQ is
connected to one of the DREQ<2:0> processor inputs (see PXA27x
Processor Family Developer's Manual), we can use DMA mode on DMA
capable CF cards. The PXA ATA driver in the mainline kernel is DMA
capable, but we cannot use that feature without knowing which DREQ
line to use.

This could give us a twofold performance boost when using CF. DMA not
only improves throughput, but it also removes most of the data
forwarding burden from the CPU.

Martin

[rsainio]

I did not de-solder the CPU yet, but I did some research for DREQ[0-2]
as they share pins with some other functions. According to Developers
Manual we get following:

Pin AFI1 AFI2 AFI3 AFO1
AFO2 AFO3
GPIO20 DREQ0 MBREQ nSDSCS2
GPIO80 DREQ1 MBREQ CS4 PWM3
GPIO97 KP_DKIN4 DREQ1 KP_MKIN4
GPIO100 KP_MKIN0 DREQ2 FFCTS
GPIO115 DREQ0 CIF_DD MBREQ


GPIO20 unlikely as it might be used for SDRAM selection (we have 4
SDRAMS)
GPIO100 unlikely as we use ttyS0
GPIO115 unlikely as CIF_DD == quick capture interface


So, to me only GPIO80 or GPIO97 are usable if the KeyPad interface is
not in use.
Has somebody catched with HARET the GPIO-register settings when
Windows runs and there is a CF-card inserted ??

[Martin]

On 9 Feb., 19:56, rsainio <sai...@t-online.de> wrote:
> Has somebody catched with HARET the GPIO-register settings when
> Windows runs and there is a CF-card inserted ??

I just checked with HARET. GPIOs 80 and 97 are both used for the
keypad interface, unfortunately. Changing the alternate function of
any of these two to DREQ1 makes some of the buttons not work anymore.
There's also GPIO85 with DREQ2 as AFI2. But we are again not lucky, as
that pin is programmed to AFI3: CIV_LV, so it belongs to the quick
capture interface.

Another possibility would be to use DMA in PIO mode, triggering the
DMA transfers from the interrupt handler. But this would probably
make sense only when using "multiple sectors per interrupt" transfers
because of the DMA setup overhead. Multiple sectors per interrupt are
unfortunately not supported by my CF card, but maybe there are CF
cards that support this (you can check with "hdparm -i /dev/sda" or
whatever your CF device is, the value in question is "MaxMultSect").

Martin