I used to do this solvation by doing these steps:
1. found the box volume information from gromacs by choosing the distance between the outer side of protein and the box randomly (I set it to 22 Angstrom as my '-d' for gmx editconf due to my very high salt concentration)
2. solvated the box to get to know the amount of water molecules needed to solvate the system
3. calculated the amount of each added Na+ and Cl- ions with this formula= (the desired concentration) x (avogadro constant) x (the volume of the box)
4. the initial amount of water molecules in a system with no salt is substracted by the total amount of total ions added to the system --> this value would be a final amount of water molecule to put in input files of packmol
but when I did the energy minimization, this did not work; the energy minimization was too high. I had problems with ensuring the salt concentration after I solvated the system in packmol and finding the right box size for my protein. What is your suggestion to solve this?