SH7263 Datasheet, PDF (323/1862 Pages) Renesas Technology Corp – Renesas 32-Bit RISC Microcomputer SuperHTM RISC engine Family / SH7260 Series

Section 9 Bus State Controller (BSC)
9.5 Operation
9.5.1 Endian/Access Size and Data Alignment
This LSI supports both big endian, in which the most significant byte (MSB) of data is that in the
direction of the 0th address, and little endian, in which the least significant byte (LSB) is that in
the direction of the 0th address. In the initial state after a power-on reset, all areas will be in big
endian mode. Little endian cannot be selected for area 0. However, the endian of areas 1 to 7 can
be changed by the setting in the CSnBCR register setting as long as the target space is not being
accessed.
Three data bus widths (8 bits, 16 bits, and 32 bits) are selectable for areas 1 to 7, allowing the
connection of normal memory and of SRAM with byte selection. Two data bus widths (16 bits and
32 bits) are available for SDRAM. Two data bus widths (8 bits and 16 bits) are available for the
PCMCIA interface. For MPX-I/O, the data bus width can be fixed to either 8 or 16 bits, or made
selectable as 8 bits or 16 bits by one of the address lines. The data bus width for burst MPX-I/O is
fixed at 32 bits. Data alignment is in accord with the data bus width selected for the device. This
also means that four read operations are required to read longword data from a byte-width device.
In this LSI, data alignment and conversion of data length is performed automatically between the
respective interfaces. The data bus width of area 0 is fixed to 16 bits or 32 bits by the MD pin
setting at a power-on reset.
Tables 9.5 to 9.10 show the relationship between device data width and access unit. Note that the
correspondence between addresses and strobe signals for the 32- and 16-bit bus widths depends on
the endian setting. For example, with big endian and a 32-bit bus width, WE3 corresponds to the
0th address, which is represented by WE0 when little endian has been selected. Area 0 cannot be
set to little endian mode. In addition, fetching instructions from a little endian area can be difficult
because 32-bit and 16-bit accesses are mixed, so big endian mode should be used for instruction
execution.
Rev. 2.00 Mar. 14, 2008 Page 289 of 1824
REJ09B0290-0200