After porting my display driver for the Discovery board based on STM32F100, I noticed that something is not quite right. The ILI9341 display seems to constantly miss the second command of the couple Column Address Set (0x2A) and Page Address Set (0x2B), irrelevant of the order they are issued.
As I bought a logic analyser for troubleshooting this sort of scenarios a while ago, I decided to fire it up and acquire signals on the SPI bus, along with control signals D/C and Reset.
Well, the above image shows clearly that D/C does not stay low for long enough when 0x2B is sent to the display (about 0.54 us instead of 1.5), hence the command is missed. At the moment this is always occurring, despite I poll the SPI status register to make sure a full byte was transmitted before asserting D/C. Oddly enough, such polling seems to work fine for the first command, 0x2A.