Chinese gūlu < *kʰaːroːɡ is probably not onomatopoeic, especially if it came from (PIE) *kʷékʷlos "wheel" (related to English "circle") as Bauer suggests.
Japanese guruguru "round and round" is standard onomatopoeic reduplication, like boroboro "old and tattered", wakuwaku "excited" etc. It's not very old at all; the oldest citation in the Kokugo Daijiten is 1672 (in guruguru-wage, a wrapped bun hairstyle).
guruguru must be related to gururi (1603) and kururi (1151), both with a meaning of "turning around, going around". An intermediate form may be kururi-kururi (ca. 14c) by dropping the -ri, cf. guru-tto < gururi-to. The morpheme in meguru (8c), "surround, go around", is probably the same, and with normal Japanese vowel mutation, also the one in korogaru "roll", koro-koro "rollingly" etc. (note kururi and korogaru are written with the same kanji, 転, Mandarin zhuǎn).
In summary, guru-guru is a late instance of a whole family of Japanese words circling around a core morpheme kuru/koro, with the meaning of "around, round and round, roll".
A connection of this productive "around" morpheme to Chinese *kʰaːroːɡ "wheel" (and thus to PIE) cannot be discarded, but I don’t think there’s any concrete evidence for it. The old Japanese orthographies for this morpheme didn’t use the characters 軲轆, but rather kana, or semantic characters like 回. This means that, if kuru comes from *kʰaːroːɡ, the transmission would have to have been ancient and prehistoric, so that by the point writing arrives, the connection was not apparent to Japanese speakers anymore.
Another possibility is that kuru is a native J morpheme but its productivity was later reinforced by gūlu in some form. But again, if that happened, I would expect more historical writers to have represented it with 軲轆 (or at least 轆 -ro/-ru, which is used in Japanese, marginally); but that didn't happen.