Linux Kernel Architecture

Mauerer runapp01.tex V1 - 09/04/2008 6:08pm Page 1137

Appendix A: Architecture Specifics

#define __le32_to_cpu(x) ((__force __u32)(__le32)(x))

#define __cpu_to_le16(x) ((__force __le16)(__u16)(x))

#define __le16_to_cpu(x) ((__force __u16)(__le16)(x))

#define __cpu_to_be64(x) ((__force __be64)__swab64((x)))

#define __be64_to_cpu(x) __swab64((__force __u64)(__be64)(x))

#define __cpu_to_be32(x) ((__force __be32)__swab32((x)))

#define __be32_to_cpu(x) __swab32((__force __u32)(__be32)(x))

#define __cpu_to_be16(x) ((__force __be16)__swab16((x)))

#define __be16_to_cpu(x) __swab16((__force __u16)(__be16)(x))

The names of the functions indicate their purpose. For example,__be32_to_cpusconverts a 32-bit big

endian value to the CPU-specific format, and__cpu_to_le64converts the CPU-specific format to little

endian for a 64-bit value.

The functions for big endian hosts are implemented using the same means (only the conversions made

are different).

A.9 Page Tables

To simplify memory management and to provide a memory model that abstracts the differences between

the various architectures, the ports must offer functions to manipulate page tables and their entries. These

are declared in<asm-arch/pgtable.h>. Chapter 3 discusses the most interesting definitions in this file,

so there’s no need to repeat them here.

A.10 Miscellaneous

This section covers three additional architecture-specific topics, which don’t fit into any of the previous

categories.

A.10.1 Checksum Calculation

Calculating checksums for packets is a key aspect of communication via IP networks and takes consid-

erable time. If possible, each architecture should therefore employ manually optimized assembler code

to calculate the checksums. The code needed to do this is declared in<asm-arch/checksum.h>.Two

functions are of prime importance:

❑ unsigned short ip_fast_csumcalculates the requisite checksum based on the IP header and the

header length.

❑ csum_partialcalculates the checksum for a packet from the fragments that are received one

by one.

A.10.2 Context Switch

The hardware-dependent part of context switchingtakes place once the scheduler has decided to instruct

the current process to relinquish the CPU so that anotherprocesscanrun.Forthispurpose,allarchitec-

tures must provide theswitch_tofunction or a corresponding macro with the following prototype in

<asm-arch/system.h>:

<asm-arch/system.h>

void switch_to(struct task_struct *prev, struct task_struct *next,

struct task_struct *last)