Shakuhachi performer and researcher Gunnar Jinmei Linder has put his PhD thesis online: Deconstructing 'Tradition' in Japanese Music: A Study of Shakuhachi, Historical Authenticity and Transmission of Tradition. I'm still reading through it, but something on page 96 caught my eye:

In the Fudoki the xiao is referred to with the Chinese characters 殺古波乏 with the reading "sha - ku - ha - chi" (semantically the characters mean "kill - old - wave - poor"). This is probably the oldest extant reference to the word 'shakuhachi' in Japan.

And here's the footnote sourcing this claim:

Kurihara, Shakuhachi shikō (1918), 1975, 16. I have attempted to locate this in the original text without any success, both in hardcopy and through Internet sources. (風土記). If this reference is true, it implies that the word shakuhachi was not necessarily written with the semantically correct 尺八. This could be an indication of that the word 'shakuhachi' was known, but not necessarily with the commonly accepted Chinese characters. At this time, words were sometimes written using the phonetic value of the characters, disregarding their semantic meaning.

Now, it struck me as very unlikely that the Fudoki would actually include a reference to the shakuhachi, spelling it 殺古波乏. The first reason is pragmatic: if the shakuhachi was in the Fudoki, it would indeed be the oldest use of the word in recorded Japanese, and it would appear in every summary of the history of the shakuhachi. But there are other, linguistic reasons to doubt this. For one thing, there was no palatalized /sya/ syllable in Old Japanese. For another, in OJ the character 古 invariably represents the syllable /kwo/:

多迦比迦流 比能美古
takapikaru / pi no mikwo
"High-Shining / Child of the Sun" (trans. Cranston)

And while we're at it, 尺 is also found in the OJ corpus, with the "correct" meaning (a length of measure) and the reading /saka/:

吾嗟 八尺之嗟
wa-ga nageku / ya-saka no nageki
"the eight-foot sighs / that I sigh" (trans. me)

So, to summarize, it doesn't seem likely that 殺古波乏 represents a genuine Fudoki spelling at all, let alone that the word shakuhachi was known in that form at that time. But it did seem plausible that Kurihara had been taken in by some fake Fudoki material that was still in circulation at the time, so I dug up the reference in his book.

[...] 又和爾雅には尺八の疏註に「洞簫同堅笛也」とし、日本風土記には簫を殺古波乏（しやくはち）と訓じ、年山紀聞には「尺八の笛等々唐山にては洞簫といふよし」と載せあり [...]

[...] furthermore, in the Wa jiga [a late 17th-C. dictionary by Kaibara Yoshifuru 貝原好古 inspired by the Chinese Erya], the notes for the 'Shakuhachi' [entry] say "A flute of the same form as the dongxiao 洞簫"; in the Nihon fudoki, the character 簫 [xiao] is given the reading 殺古波乏 "sha-ku-ha-chi"; and the Nenzan kibun [an early 19th-C. book of essays by Andō Tameakira 安藤為章] includes the text "the shakuhachi flute ... is in China called the 'dongxiao'" [...]

Incidentally, this paragraph theoretically contains all that you need to solve the mystery of 殺古波乏 (it is neither hoax nor error), but I'll confess that it wasn't enough for me.

Now, I didn't find it very likely that Kurihara had read the whole Fudoki himself, looking for references to the instrument that interested him. He doesn't reference a source, but surely he had one; and given the time period, I reasoned, that source was highly likely to have been the Koji ruien 古事類苑, a Meiji leishu 類書 (sort of like an encyclopedia, but with more quotes from the classics). And, sure enough, the Koji ruien's entry for "shakuhachi" includes the 殺古波乏 quote, with a comment:

〔日本風土記　四　响器〕簫　殺古波乏（シヤクハチ）
○按ズル二、簫ヲシヤクハチト訓ズルハ、亦洞簫即チ尺八ナリト云ヘル、謬ヲ襲グ者ナリ

Nihon fudoki, vol. 4, "响器" [sound-making instruments]: 簫 = 殺古波乏 (shakuhachi)
○ Note: Applying the pronunciation shakuhachi to the character 簫 appears to be a repetition of the erroneous assertion that the dongxiao is [the same as] the shakuhachi.

(Incidentally, this quotation is sandwiched between the Wa jiga quotation and the Nenzan kibun quotation, so Kurihara was indeed probably working straight from the Koji ruien. No big deal; that's what it was for.)

But this reference in the Koji ruien packs so much oddness into so few characters. The Fudoki isn't generally divided into numbered volumes; the parts are referred to by the province they cover — the "Izumo fudoki", etc. And there ain't no province called 响器 — in fact, I'd never seen 响 in a Japanese context before.

That's when it hit me: the work I'm thinking of is a collection of fragments referred to collectively as the Fudoki. It isn't called the Nihon fudoki at all, and never has been, or at least not widely. The Koji ruien reference must be to a completely different book!

And indeed, it was. The Nihon fudoki, or Riben fengtuji, is a Chinese book about Japan written in the 16th century by Hou Jigao 侯継高. Waseda has two editions from the Edo period online, but neither of them include an entry for 簫; fortunately, the Kindai Digital Library has a different edition from the Taishō period which does include it:

Mystery solved: the pronunciation 殺古波乏 is indeed given for the character 簫 in a book called 日本風土記 — but that book is the 16th-century Chinese book by Hou Jigao (and, furthermore, the 殺古波乏 bit may be a much later addition), not the well-known Fudoki from the Nara period, and so it isn't a particularly exciting citation.