This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: linux/xattr.h and sys/xattr.h conflicts
- From: Allan McRae <allan at archlinux dot org>
- To: libc-alpha <libc-alpha at sourceware dot org>
- Date: Mon, 06 Jan 2014 08:45:53 +1000
- Subject: Re: linux/xattr.h and sys/xattr.h conflicts
- Authentication-results: sourceware.org; auth=none
- References: <52C96CC7 dot 4020006 at archlinux dot org> <20140105193242 dot GZ24286 at brightrain dot aerifal dot cx>
On 06/01/14 05:32, Rich Felker wrote:
> On Mon, Jan 06, 2014 at 12:31:35AM +1000, Allan McRae wrote:
>> Hi,
>>
>> The latest version of libcap (2.23) provides the header
>> <sys/capability.h> which now includes <linux/xattr.h>. From a quick
>> look, it appears the libcap now library uses various things in the
>> <linux/xattr.h> header that are not in the <sys/xattr.h> header.
>>
>>
>> Anyway... this has lead to bug reports in the form:
>>
>> 1) compiles fine:
>> #include <sys/xattr.h>
>> #include <linux/xattr.h>
>>
>> 2) generates compile errors:
>> #include <linux/xattr.h>
>> #include <sys/xattr.h>
>>
>>
>> <linux/xattr.h> has:
>>
>> #define XATTR_CREATE 0x1 /* set value, fail if attr already exists */
>> #define XATTR_REPLACE 0x2 /* set value, fail if attr does not exist */
>>
>>
>> <sys/xattr.h> has:
>>
>> enum
>> {
>> XATTR_CREATE = 1, /* set value, fail if attr already exists. */
>> #define XATTR_CREATE XATTR_CREATE
>> XATTR_REPLACE = 2 /* set value, fail if attr does not exist. */
>> #define XATTR_REPLACE XATTR_REPLACE
>> };
>>
>>
>> so if the <linux/xattr.h> header is included first, then XATTR_CREATE is
>> defined and that enum definition fails.
>>
>> I some questions on my way to fix this. Is there a point to that enum
>> that I am missing or can it go? If it needs to stay, can its member
>> names be altered?
>
> Inclusion of linux/*.h in userspace apps/libs is generally a bug in
> the absence of a really good reason to be doing so. Here I think
> libcap should just be fixed to use the libc header. If that's missing
> some important definitions, they should be added.
>
I agree. But I also think the incompatibility between headers is a bug
that needs fixed. So I'd like to fix that as the first step.
Allan