Re: [PATCH v2] src: fix detached_mounts_propagation compile errors

["Yang Xu (Fujitsu)" <xuyang2018.jy@fujitsu.com>]

on 2023/04/25 9:59, Ziyang Zhang wrote:
> On 2023/4/25 09:28, Yang Xu (Fujitsu) wrote:
>>
> 
> [...]
> 
>>>
>>> Looks like not help [1]. The src/feature.c through global.h -> xfs/xfs.h
>>> -> xfs/linux.h -> linux/fs.h to include linux/mount.h, then it include
>>> "vfs/missing.h" manually. As you add "<sys/mount.h>" to vfs/missing.h,
>>> so there's a conflict.
>>
>> Yes, here is a complex problem than ltp because global.h includes
>> <linux/mount.h>. We don't include <sys/mount.h> or <linux/mount.h>
>> together on older glibc(glibc 2.36.6).  Newer glibc(glibc-2.37-1)[1]
>> seems has sloved this problem.
>>
>> [1]https://sourceware.org/git/?p=glibc.git;a=commit;h=774058d72942249f71d74e7f2b639f77184160a6
>>
>> I guess we can not use <sys/mount.h> header,
>> then compile info as below:
>> detached_mounts_propagation.c: In function 'main':
>> detached_mounts_propagation.c:129:15: warning: implicit declaration of
>> function 'mount' [-Wimplicit-function-declaration]
>>     129 |         ret = mount(NULL, base_dir, NULL, MS_REC | MS_SHARED,
>> NULL);
>>         |               ^~~~~
>> detached_mounts_propagation.c:176:23: warning: implicit declaration of
>> function 'umount2'; did you mean 'sys_umount2'?
>> [-Wimplicit-function-declaration]
>>     176 |                 ret = umount2(target, MNT_DETACH);
>>         |                       ^~~~~~~
>>         |                       sys_umount2
>>
>>
>> then we can use syscall wrapper directly instead of glibc wrapper
>> --- a/src/detached_mounts_propagation.c
>> +++ b/src/detached_mounts_propagation.c
>> @@ -20,7 +20,6 @@
>>    #include <stdio.h>
>>    #include <stdlib.h>
>>    #include <string.h>
>> -#include <sys/mount.h>
>>    #include <sys/stat.h>
>>    #include <sys/syscall.h>
>>    #include <sys/types.h>
>> @@ -127,7 +126,7 @@ int main(int argc, char *argv[])
>>           if (ret < 0)
>>                   exit_log("%m - Failed to create new mount namespace");
>>
>> -       ret = mount(NULL, base_dir, NULL, MS_REC | MS_SHARED, NULL);
>> +       ret = sys_mount(NULL, base_dir, NULL, MS_REC | MS_SHARED, NULL);
>>           if (ret < 0)
>>                   exit_log("%m - Failed to make base_dir shared mountpoint");
>>
>> @@ -174,7 +173,7 @@ int main(int argc, char *argv[])
>>                   }
>>                   close(fd_tree);
>>
>> -               ret = umount2(target, MNT_DETACH);
>> +               ret = sys_umount2(target, MNT_DETACH);
>>                   if (ret < 0) {
>>                           fprintf(stderr, "%m - Failed to unmount %s",
>> target);
>>                           exit_code = EXIT_FAILURE;
>>
>>
>> Best Regards
>> Yang Xu
>>
> 
> Hi Yang,
> 
> You patch works for me(kernel: 5.10.134, glibc:2.36). I think we could just
> modify detached_mounts_propagation.c to solve this problem.

OK, @Gao @Ziyang, so should I send a patch or you send it?

Best Regards
Yang Xu
> 
> Regards,
> Zhang