Hi D6,
The least reliable part of VNAV is it's approach function. I've loaded the same approach to the same runway with the same speeds + altitudes on numerous occasions and watched in amazement as the aircraft has flown the approach differently each time.
Apart from that, VNAV actually works okay. It doesn't handle wind shifts - particularly on descent - very well, but I think that is common across most aircraft types. More accurate wind data would improve its performance in that regard.
Regarding your encounter with VNAV in the sim. . .
What you are saying is essentially correct, though I suspect you had entered the alt constraint as an "at or above" in the legs page. For example XXXXA, rather than just a hard altitude XXXX. In that case, the aircraft will continue climbing to the MCP altitude with VNAV SPD as the active pitch mode.
Had the alt constraint been entered as "XXXX" or "XXXXB" (at - or - below), then with the MCP altitude selected set to an altitude higher than the alt constraint, then the pitch mode will transition to VNAV PTH, and will remain so, until the waypoint or condition associated with the constraint is passed, after which the aircraft will continue climbing in VNAV SPD.
Regarding IRS failure . . .
The FMC position is a combination of IRS and navaid derived position (and GPS pos. where installed), so if the IRS position is drastically wrong, then the FMC position will also be wrong. There is no means of recalibrating the IRS position in flight.
The Capt's displays are fed by the left IRS. If this fails, then selecting "ALTN" selects the centre IRS as the source. No use of the "ATT" mode is required there. It is only selected when all 3 IRSs fail. Then, it provides only attitude and heading information, requiring entry of magnetic heading from the stby compass. All position information is lost.
WG