i had 9 of those smartphone / pocket-pc style devices back at the time: the absolute best one was the HTC Universal, as it was more like a hand-held clamshell laptop with built-in 3G. you _used_ to be able to get information about them on handhelds.org but we coordinated through #htc-linux (since taken over by android dummies) and used wiki.xda-developers.com (since taken over by android wannabe modders). [note to xda-developer forum users: i may be being slightly unfair though about the android dummies and wannabes: i apologise in advance to any of you that aren't so stupid as to be able to find and pay attention long enough to read slashdot :) ]
so you're going to have to dig... and you'll almost certainly need to begin with the 2.6 era linux kernel tree, which should give you a very very big hint about what you face, here. to give you an example: the fastest i've ever been able to reverse-engineer linux onto a device was 3 weeks and that was because it already had a [GPL-violating] linux kernel on it, where they had left some clues around and it was possible to poke around in /proc.
beyond that, the fastest i managed - and i could not get PM/wakeup to work because i could not locate the correct RAM/device re-initialisation parameters - was six to eight weeks on the HTC/Compaq Ipaq, i believe it was called the hw6915.
beyond _that_, the _longest_ i ever heard someone taking (and this was because it was worth it) to get full driver functionality was THREE YEARS, and that was for the HTC Universal (aka O2 "XDA III").
please please DO NOT underestimate how much work it takes to do reverse-engineering. these handhelds are actually far more complex pieces of kit, in engineering and in software terms, than any laptop or desktop PC you've ever encountered. the HTC Universal had SEVEN audio output paths for example, and over four audio input paths. there were over 110 GPIO pins on its Intel PXA ARM processor, but these were nowhere near enough, so they had to use an external GPIO IC (we called it ASIC3). but... they actually ran out of GPIO pins on that *as well*, so they ended up utilising the 16 pins of GPIO on the Ericsson 3G GSM modem (only contactable over USB!) in order to control some of the functions such as camera light.
so in many ways you are actually better off designing (and paying to have made) your own device. that is not a joke, in the slightest bit. it will take you less time and will cost you less in lost earnings from having to work full-time on the reverse-engineering. and before you splutter in disbelief, there are people who have done exactly that: Dr Schaeller did the GTA04 fairly recently (fits into a Neo FreeRunner case), and in that way he at least got to pick a) a modern-ish processor b) the best components that were available c) he got CONTROL OVER THE DEVICE DRIVERs, and he didn't have to _guess_ what the GPIO maps and memory maps are.
basically, what i'm trying to say is that if you cannot find a pre-existing project (you didn't mention what devices you actually have) that has done the reverse-engineering, unless you are actually thinking of learning reverse-engineering as a useful specialist marketable skill, either throw those devices into landfill, give them to someone who doesn't mind winceouch, or break them down for parts and sell the components on ebay. check beforehand to make sure that they're desirable parts of course.
of course... i say "throw them into landfill", which is directly and vehemently against our social responsibility, but unfortunately when actually buying these devices we make selfish decisions, not socially responsible ones, not least because they *aren't any alternatives*. now http://phonebloks.com/ is looking to change that in the smartphone space, and i'm looking to change that in the everything-else-device arena (starting here https://www.crowdsupply.com/eo...)