set KERNEL "zImage"
set MTYPE 1157
set CMDLINE "mem=64M"
25Oct2006: first revision for HW6915 committed to cvs - MACH_TYPE_HW6900 for the HTC Sable
25Oct2006: first boot of linux on the IPAQ hw6915 - discovered that the boot parameter mem=64M was required (just like for HTC_Universal)
31oct2006: udc enabled, tested and working - sufficient to follow the HimalayaLinuxBootingViaRootNFS instructions, successfully.
01nov2006: hacked together a TI tsc2046 touchscreen driver, from htcuniversal_ts.c, by back-porting the original use of GPIO from h4700_ts.c. also successfully (once...) got /dev/mmcblk0p2 to be recognised - and mounted - but only when using nfs as the root device, not stand-alone.
02nov2006: got a first initrd boot (using the Universal_Installing_Linux page as a guide) and loaded asic3_mmc manually - and discovered that mmc cards are mounted read-only: presumably by a GPIO or ASIC3GPIO signal, which needs to be found... commented out the read-only check in mmc_block.c, for now, and the mmc works!
03nov2006: managed to get opie 0.8.4 working up to the touchscreen/cal bit - then it asked for language settings and wouldn't accept any further input. more investigation required. (cr2 says i should have used htcuniversal_ts2.c as the basis for the htcsable_ts.c code. oops)
04nov2006: got bluetooth driver working - identical copy of htcuniversal_bt.c except for some GPIO and commenting out the EGPIO enabling rubbish.
16nov2006: after making a stupid coding mistake, got the GSM initialisation sequence. the protocol sent over the serial link has not been encountered before however it's pretty obvious how it works. (update: actually... er... discovered that there was a crc on the last two bytes which wasn't obvious but a couple of hours pissing about found the algorithm to be crc16 of some description).
18nov2006: re-scanned the expected memory area for the ACX100, and found it looked right. tried writing a driver: partial success! the device is recognised, and firmware can be uploaded, but only if the wireless was running at the time that gnuharet boots into linux, from wince. guesses as to why this is are that the power-up sequence might not be right, but there could be half a dozen other reasons.
22nov2006: use htchimalaya_ts2.c as the basis for a touchscreen, and using SSP2 instead of SSP1 - it works. hooray.
22nov2006: whilst looking for audio gpios on asic3 and pxa i accidentally pulled low some of the ACX power pins. testing setting of these pins in htcsable_acx startup results in it working! we have wifi!
25nov2006: working with cr2 and psoko, managed to locate both the ASIC3 GPIO pins for the ak4641 codec reset and also to track down a strange bug in the i2s code. sound works!
26nov2006: lcd poweron / poweroff now possible. backlight controllable. working on suspend/resume. keyboard driver, using pxa27x_kbd.
30nov2006: been stuck on suspend/resume problem for four days. argh.
22dec2006: after abandoning suspend/resume finally there is hope: paul psokolovsky noticed that the device drivers hadn't been converted fully to use platform_device. can't test it myself as i don't have the charger for my hw6915 - so someone else needs to try it (and let me know...)