Source-Changes-D archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: CVS commit: src/share/man/man9

>  >>  to handle PCI bus master devices that (via DMA) transfer word parameters
>  >>  in little endian even on big endian systems.
> Still sounds pretty awkward.

Well, the definition of endianness itself is always awkward.

> What about "little endian control data" instead of juts "little endian
> data"?

IMO, data itself doesn't have any property of endianness at all,
so "little endian (control) data" sounds weird for me.

"Endianness" is defined only if data is interpreted in a smaller unit
(byte access against word data, bit access against byte data etc).
If the first unit (smallest byte address or first bit in stream)
is the least significant byte/bit, such enconding method is called
little endian (or just literally "LSB first"), and vice versa.
If the data is read/written in its own size, endianness/byteorder
is not visible at all.

> It doesn't have to be perfect anyway; it's a historical note.

But there were several developers who asked why such APIs were needed,
and busmaster drivers are still major users of these functions
(and most x86 driver writers in the world won't consider about it).
That was the reason why I added the sentence.
Izumi Tsutsui

Home | Main Index | Thread Index | Old Index