By "ck" I'm assuming you're referring to the series solution's coefficients, i.e.

.
Differentiating gives


Substituting into the ODE yields


By extracting the first term (

) of the third series, you can consolidate the second and third series:

Next, since the first series starts with a constant term while the other two start with a quadratic term, remove the first two terms:

Now shift the index so that the series begins at the same starting index by replacing

with

.

So now the series can be joined into one:
![\displaystyle\sum_{k\ge1}\bigg[(k+3)(k+2)c_{k+3}-kc_k+c_k\bigg]x^{k+1}](https://tex.z-dn.net/?f=%5Cdisplaystyle%5Csum_%7Bk%5Cge1%7D%5Cbigg%5B%28k%2B3%29%28k%2B2%29c_%7Bk%2B3%7D-kc_k%2Bc_k%5Cbigg%5Dx%5E%7Bk%2B1%7D)
which means the ODE is equivalent to
![\displaystyle\sum_{k\ge1}\bigg[(k+3)(k+2)c_{k+3}-(k-1)c_k\bigg]x^{k+1}+c_0x+2c_2+6c_3x=0](https://tex.z-dn.net/?f=%5Cdisplaystyle%5Csum_%7Bk%5Cge1%7D%5Cbigg%5B%28k%2B3%29%28k%2B2%29c_%7Bk%2B3%7D-%28k-1%29c_k%5Cbigg%5Dx%5E%7Bk%2B1%7D%2Bc_0x%2B2c_2%2B6c_3x%3D0)
It follows that

and

. Using this, you solve the recurrence

with the initial values

and

.
Because

, it follows that

; in general,

for all

.
Also, because the RHS of the recurrence vanishes when

, it follows that

; or generally,

for

, leaving you with only one not-necessarily-zero term of

.
So the hard part is finding a solution for

when

is a multiple of 3. By finding the first few of these terms, you'll start to notice a pattern. Starting with

, you find




and so on, following a general pattern of


valid for

.
So the series solution is







So the two solutions are

and

.
To solve the IVP, notice that when

, you have


so the solution to the IVP is the same as that to the general solution with the unknowns replaced accordingly.