As an aside, the SSD firmwares -could- detect pages filled with ones or zeros and use that knowledge to transparently do a "TRIM light", at the cost of doubling

As an aside, the SSD firmwares -could- detect pages filled with ones or zeros and use that knowledge to transparently do a "TRIM light", at the cost of doubling the size of the "block in use" bitmap (to track 'free', 'in use', 'all zeroes', 'all ones') and added code complexity. For the SSD makers it probably isn't worth the effort and debugging, since TRIM support is well on its way to ubiquity. Should be an interesting experiment though, if nobody has done it yet.