in my original drafts of this code [p10 dual colour led panel] i did use cs/ss signals to sync transmission.
it turned out that the time taken to clear/set cs pin and the impost of two extra pins was as waste of resources an an unnecessary
complication. it works fine with out it.
i don't have a k40 chip to test with and don't see why mcc insists on a ss pin . i developed code on a k22 chip
and did not use the ss2 pin , loss of a pin is not tragic.

sheldons comments are cluttering the story and some are just incorrect
this is what i would use

Code:
TRISB  = $F2 
TRISD  = $FF    ;ss2
ANSELD = $FE  ;ss2
ANSELB = $C0 ; just the spi pins


WPUD   = $01  ;ss2 
WPUB   = $30 ; just the spi pins




RB0PPS = $0F             ;RB0->MSSP1:SCK1     
RB2PPS = $10             ;RB2->MSSP1:SDO1     
RB3PPS = $12             ;RB3->MSSP2:SDO2 
SSP1CLKPPS = $08         ;RB0->MSSP1:SCK1     
SSP2DATPPS = $0D         ;RB5->MSSP2:SDI2  
SSP1DATPPS = $0C         ;RB4->MSSP1:SDI1     
SSP2CLKPPS = $09         ;RB1->MSSP2:SCK2     
SSP2SSPPS = $18          ;RD0->MSSP2:SS2    it seems to insist on an ss pin


SSP1STAT = $40  ;bog standard spi
SSP1CON1 = $2A  ;fosc/4 with ssp1add divider
SSP1ADD  = $01   ; clock divide 2
SSP2STAT = $40   ;bog standard spi
SSP2CON1 = $25 ;bog standard spi no ss

;spi pin allocations
'    sck1        PORTB.0       
'    sck2        PORTB.1      ' wired to portb.0
'    sdo1        PORTB.2      
'    sdo2        PORTB.3       
'    sdi2        PORTB.4      
'    sdi1        PORTB.5      
'    ss2         PORTD.2