Re: [PATCH] src: fix detached_mounts_propagation compile errors

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



on 2023/04/19 16:45, Christian Brauner wrote:

> On Wed, Apr 19, 2023 at 11:29:55AM +0800, Gao Xiang wrote:
>> Newer glibc such as glibc 2.36 also defines 'struct mount_attr'
>> in addition to <linux/mount.h>.  It will report as below when
>> compiling with old linux kernel headers (without idmapped mounts,
>> such as kernel-headers 5.10.134) but with newer glibc (here checked
>> with glibc 2.36.6):
>>
>>      [CC]    detached_mounts_propagation
>> In file included from detached_mounts_propagation.c:29:
>> vfs/missing.h:115:8: error: redefinition of 'struct mount_attr'
>>    115 | struct mount_attr {
>>        |        ^~~~~~~~~~
>> In file included from detached_mounts_propagation.c:23:
>> /usr/include/sys/mount.h:210:8: note: originally defined here
>>    210 | struct mount_attr
>>        |        ^~~~~~~~~~
>> gmake[3]: *** [Makefile:102: detached_mounts_propagation] Error 1
>> gmake[2]: *** [include/buildrules:31: src] Error 2
>> make[1]: *** [Makefile:51: default] Error 2
>> make: *** [Makefile:49: default] Error 2
>>
>> Cc: Christian Brauner <brauner@xxxxxxxxxx>
>> Signed-off-by: Gao Xiang <hsiangkao@xxxxxxxxxxxxxxxxx>
>> ---
> 
> This header mount.h mess is a neverending nightmare...
> 
>>
>> Not quite look into that, but at a quick glance of tester reports
>> it seems the cases.
>>
>>   configure.ac                      | 5 ++++-
>>   src/detached_mounts_propagation.c | 1 -
>>   src/vfs/missing.h                 | 1 +
>>   3 files changed, 5 insertions(+), 2 deletions(-)
>>
>> diff --git a/configure.ac b/configure.ac
>> index 4687d8a3..517b4c18 100644
>> --- a/configure.ac
>> +++ b/configure.ac
>> @@ -74,7 +74,10 @@ AC_HAVE_FIEXCHANGE
>>   
>>   AC_CHECK_FUNCS([renameat2])
>>   AC_CHECK_FUNCS([reallocarray])
>> -AC_CHECK_TYPES([struct mount_attr], [], [], [[#include <linux/mount.h>]])
>> +AC_CHECK_TYPES([struct mount_attr], [], [], [[
>> +#include <linux/mount.h>
>> +#include <sys/mount.h>
> 
> I think we should only rely on sys/mount.h. linux/mount.h and
> sys/mount.h are incompatible because linux/mount.h might have a #define
> where sys/mount.h has an enum.

In fact, ltp also has this problem, but ltp has fixed this

We check whether mount_setattr() supported, if support, then use 
<sys/mount.h>,otherwise, use <linux/mount.h>

AC_CHECK_TYPES([struct mount_attr],,,[
#ifdef HAVE_MOUNT_SETATTR
# include <sys/mount.h>
#elif HAVE_LINUX_MOUNT_H
# include <linux/mount.h>
#endif

see 
https://github.com/linux-test-project/ltp/commit/30353a1eb3ee8940f38d6a0e8f5f39352fd8a179.

Best Regards
Yang Xu




[Index of Archives]     [Linux Filesystems Development]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux