2004-09-07 18:56:53

by Krzysztof Halasa

[permalink] [raw]
Subject: unneeded #include <version.h> in many places ?

Hi,

I noticed some kernel .c files #include <version.h> which typically
contains something like:

#define UTS_RELEASE "2.6.9-rc1"
#define LINUX_VERSION_CODE 132617
#define KERNEL_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c))

However, those files don't reference the macros.

The question is: are these includes completely unneeded, so I can
remove them, or do they serve some special purpose?

Another one: there are drivers using constructs like:

#if LINUX_VERSION_CODE > 0x20115
...
#endif

I understand they can be somehow useful for authors supporting many
kernel versions with a single set of files, however the gain isn't
clear to me. Should such conditional code be a) removed, b) left
in place, c) dealt with each case individually?
--
Krzysztof Halasa


2004-09-07 19:34:55

by Adrian Bunk

[permalink] [raw]
Subject: Re: unneeded #include <version.h> in many places ?

On Tue, Sep 07, 2004 at 06:17:53PM +0200, Krzysztof Halasa wrote:

> Hi,

Hi Krzysztof,

> I noticed some kernel .c files #include <version.h> which typically
> contains something like:
>
> #define UTS_RELEASE "2.6.9-rc1"
> #define LINUX_VERSION_CODE 132617
> #define KERNEL_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c))
>
> However, those files don't reference the macros.
>
> The question is: are these includes completely unneeded, so I can
> remove them, or do they serve some special purpose?

they don't serve any special purpose.

But please doublecheck that they are really not needed.

> Another one: there are drivers using constructs like:
>
> #if LINUX_VERSION_CODE > 0x20115
> ...
> #endif
>
> I understand they can be somehow useful for authors supporting many
> kernel versions with a single set of files, however the gain isn't
> clear to me. Should such conditional code be a) removed, b) left
> in place, c) dealt with each case individually?

c)

Many driver authors share their code this way between different major
kernel versions, which makes it easier for them to maintain their code.

#ifdef's for kernel 2.4 are often still actively maitained.

#ifdef's for kernel 2.2 are only very rarely actively maitained.

#ifdef's for kernel 2.0 (as in your example) are most likely completely
obsolete.

> Krzysztof Halasa

cu
Adrian

--

"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed