Sections, Classes, and Groups

A section is a contiguous area of memory generated from code or data defined in a source module. Each section has a type, class, group, and a name. Sections are usually related to the hardware configuration. Therefore, portions of programs stored in ROM are in separate sections from the portions stored in RAM. Sections generated by the Keil C166 Compiler are given names which are composed of a prefix and the name of the source module.

A class is an arbitrarily ordered collection of sections that share a common attribute. The purpose of a class is to allow manipulation of a number of sections using only the class name. For instance, the near data objects from all modules in a program are combined into the NDATA class while the far constant objects are combined into the FCONST class.

A group is a shorthand which is used to reference a collection of sections that all reside within one data page (for data groups) or segment (for code groups). Groups define address-range limits for their references sections. Two group types are supported: CGROUP (code groups) and DGROUP (data groups). Groups are used by the linker to guarantee that classes within the group fit into the appropriate address range (16K for DATA or 64K for CODE) and that DPP register assignments are correctly performed in the startup code. The DPPUSE linker directive may be used to extend the address space possible for DATA groups.

Objects that the C166 Compiler generates (such as program code, program data, and constant data) are stored in sections that may be absolute or relocatable. The only exception is automatic variables that are stored on the USER STACK that is addressed by the R0 register. The compiler creates a separate section for each memory type used in the C source.

Section names created by the C166 Compiler are converted to and stored in uppercase and include a class-code and a module-name formatted as follows:

?class-code?module-name

Where

class-code	is an abbreviation for the class type.
module-name	is the name of the source file in which the object is declared.

The following table shows all possible section names with their associated class and group names. The class and group names are used by the linker to simplify address assignments and to ensure that objects from different modules are placed in valid address ranges.

Section Name	Group Name	Class Name	Section Contents
?PR?module-name	NCODE	NCODE	near function code
?PR?module-name	—	FCODE	far function code
?NC?module-name	NCONST	NCONST	const near objects
?FC?module-name	—	FCONST	const far objects
?HC?module-name	—	HCONST	const huge objects
?XC?module-name	—	XCONST	const xhuge objects
?ND?module-name	NDATA	NDATA	near objects with #pragma NOINIT
?ND0?module-name	NDATA	NDATA0	near objects with #pragma INIT
?SD?module-name	SDATA	SDATA	sdata objects with #pragma NOINIT
?SD0?module-name	SDATA	SDATA0	sdata objects with #pragma INIT
?ID?module-name	SDATA	IDATA	idata objects with #pragma NOINIT
?ID0?module-name	SDATA	IDATA0	idata objects with #pragma INIT
?BD?module-name	SDATA	BDATA	bdata objects with #pragma NOINIT
?BD0?module-name	SDATA	BDATA0	bdata objects with #pragma INIT
?BI?module-name	—	BIT	bit objects with #pragma NOINIT
?BI0?module-name	—	BIT0	bit objects with #pragma INIT
?FD?module-name	—	FDATA	far objects with #pragma NOINIT
?FD0?module-name	—	FDATA0	far objects with #pragma INIT
?HD?module-name	—	HDATA	huge objects with #pragma NOINIT
?HD0?module-name	—	HDATA0	huge objects with #pragma INIT
?XD?module-name	—	XDATA	xhuge objects with #pragma NOINIT
?XD0?module-name	—	XDATA0	xhuge objects with #pragma INIT
?C_INITSEC	—	—	Variable-initialization information
?C_MEMCLRSEC	—	—	RAM zero-initialization information
?C_STARTUP_CODE	—	ICODE	Startup code for the C main function

Note

• If the TINY memory model is used, no group names are assigned to the section names.