Slavic *rad- 'care'

The handbooks I got to hand (Vasmer, Pokorny, EIEC) derive Slavic *rad- ~
*rod- 'care, (make) effort' from PIE *roh1(-dH)- 'put in order' (to OInd.
ra:dhnóti 'achieves'). The derivation is not unproblematic, though.

First, *radi"ti ~ *rodi"ti is either (b) (SCr. S^tok. ráditi, râdi:m, C^ak.
ra:di``ti, rãdi:s^ 'work', Slovene (arch.) ró,dim 'care') or (c) (ORuss.
according to Zaliznjak), and *râdU ~ ro``dU seems to be (c) (SCr. S^tok.
râd, râda 'work'). (b) of the denominative verb would be completely
incompatible with the laryngeal of *roh1dH-, so it must be secondary,
replacing earlier (c) which reflects (c) of the noun. In turn, the (c) of
*râdU ~ ro``dU (with failed Hirt's Law) requires *u- or *C- stem, thus
something like *roh1dHú- as a proto-form. But does it look like a normal PIE
formation?

Secondly, how the *o of the variant forms *rodi"ti (OCS roditi 'care') and
ro``dU (ORuss. nerodU 'carelessness') can be accounted for? Pokorny suggests
a zero grade (r&dH-), but why would *rh1dH- be vocalized as *r&1dH- rather
than *r.h1dH-? If the first variant of vocalization were possible it would
probably solve the accentological problem as well, since the non-acute
Slavic *-a- < *a: could be easily explained away as a product of later
(Balto-)Slavic lengthening of the original short *a < *&1.

Sergei