Interesting segment. I think that the second half of the measure can be heard as a D-diminished-7th in third inversion, and thus functions pretty normally as vii-dim-4/2 leading to the I6/4 of the following measure that isn't in your example. The Db in the top voice admittedly complicates that analysis, but I don't think it's dispositive. The first half of the measure is more problematic. On the one hand, it's clearly just a sequence, with precisely the same pattern as the second half of the measure but written a half-step higher (spelled like F#-diminished 7, but just as likely heard as an Eb-diminished 7). On the other hand, it does seem to have a possible hearing (underlined by Rachmaninoff's or the engraver's choice of F#) as a secondary vii of a G chord. This is strengthened by the D natural in the upper voice on beat 2 which could make a secondary V of G. The ultimate resolution to Eb is thus both a perfectly normal vii-I progression, with a hint of a deceptive progression due to the first chord's implication of a temporary G as tonic.
That's probably a bit of a stretch, but I do think it adds a interesting layer to the progression. For the most part, it seems to be more of an ornamental chord moving in sequence down by half step to a vii-4/2.
[Thanks to MarkM for the correction about the chord qualities.]