Angle-resolved upper critical field Hc2 provides an efficient tool to probe the gap symmetry of unconventional superconductors. We revisit the behavior of in-plane Hc2 in d-wave superconductors by considering both the orbital effect and Pauli paramagnetic effect. After carrying out systematic analysis, we show that the maxima of Hc2 could be along either nodal or antinodal directions of a dwave superconducting gap, depending on the specific values of a number of tuning parameters. This behavior is in contrast to the common belief that the maxima of in-plane Hc2 are along the direction where the superconducting gap takes its maximal value. Therefore, identifying the precise d-wave gap symmetry through fitting experiments results of angle-resolved Hc2 with model calculations at a fixed temperature, as widely used in previous studies, is difficult and practically unreliable. However, our extensive analysis of angle-resolved Hc2 show that there is a critical temperature T * : in-plane Hc2 exhibits its maxima along nodal directions at T < T * and along antinodal directions at T * < T < Tc. The concrete value of T * may change as other parameters vary, but the existence of π/4 shift of Hc2 at T * appears to be a general feature. Thus a better method to identify the precise d-wave gap symmetry is to measure Hc2 at a number of different temperatures, and examine whether there is a π/4 shift in its angular dependence at certain T * . We further show that Landau level mixing does not change this general feature. However, in the presence of Fulde-Ferrell-LarkinOvchinnikov state, the angular dependence of Hc2 becomes quite complicated, which makes it more difficult to determine the gap symmetry by measuring Hc2. Our results indicate that some previous studies on the gap symmetry of CeCu2Si2 are unreliable and need to be reexamined, and also provide a candidate solution to an experimental discrepancy in the angle-resolved Hc2 in CeCoIn5.