Just FWIW-- The South Sulawesi langs. of Indonesia allowed proto finals *t k 
n N r l s, and probably *p m and maybe *G (> r mostly).

These merged along the way:
One subgroup: (p) t > t > ? (t reappears when a suffix is added)
(G) r l s ult. > ? (r s or t may reappear when suffixed)
k > k
(m) n > n and N > N (one lang. in the group _may_ have retained m, it's not 
and in one lang, there is sporadic -h < *G which IIRC > r when suffixed

Another subgroup: r l s n N retained, (p) t k > ? (k reappears)

Makassarese: r l s retained, add an echo-vowel that drops when suffixed
all stops > ? (phonemic /k/, which appears when suffixed)
all nasals > N

Buginese: all nasals > N, all stops+rls > ? (reappear variously as k r s 
when suffixed)
Lots of analogical re-shuffling !! What a mess.

It's similar to the Oceanic lang. situation, where suffixes to CVCV forms 
show retention of original final C, though not regularly (analogy again).

James Matisoff had a nice name for the gradual loss of finals in SE Asian 
langs-- "continuum of consonantal attrition"