Build fails on 32bit architectures

[igor-raits]

../src/wutil.cpp: In function 'int fd_check_is_remote(int)':
../src/wutil.cpp:311:14: error: narrowing conversion of '4283649346' from 'unsigned int' to 'int' [-Wnarrowing]
  311 |         case 0xFF534D42:  // CIFS_MAGIC_NUMBER
      |              ^~~~~~~~~~

[faho]

I'd assume this works if we just add a "u" to the end of the literal? I.e.

        case 0xFF534D42u:  // CIFS_MAGIC_NUMBER

[igor-raits]

Let me quickly test it.

[igor-raits]

@faho, no that does not help:

../src/wutil.cpp: In function 'int fd_check_is_remote(int)':
../src/wutil.cpp:311:14: error: narrowing conversion of '4283649346' from 'unsigned int' to 'int' [-Wnarrowing]
  311 |         case 0xFF534D42u: // CIFS_MAGIC_NUMBER
      |              ^~~~~~~~~~~

[igor-raits]

I guess we'd need to use statfs64()?

But I'm not very sure if that is not just a bug somewhere in glibc.

       The  original  Linux  statfs()  and  fstatfs() system calls were not designed with extremely large file sizes in mind.  Subsequently, Linux 2.6 added new statfs64() and
       fstatfs64() system calls that employ a new structure, statfs64.  The new structure contains the same fields as the original statfs structure, but the sizes  of  various
       fields are increased, to accommodate large file sizes.  The glibc statfs() and fstatfs() wrapper functions transparently deal with the kernel differences.

@fweimer, do you think this is some regression in new glibc or it is supposed to be this way?

[faho]

Maybe try adding the "u" to all cases in that switch?

The man page for fstatfs (linux man-pages 5.05) says:

The __fsword_t type used for various fields in the statfs structure definition is a glibc internal type, not intended for public use. This leaves the programmer in a bit of a conundrum when trying to copy or compare these fields to local variables in a program. Using unsigned int for such variables suffices on most systems.

And here it's still complaining that we're converting from unsigned int to int, so to me it seems like we're missing a cause for conversion.

[igor-raits]

I've tried both adding u suffix to all cases and switching to statfs64, none of which help ;(

[igor-raits]

diff --git a/src/wutil.cpp b/src/wutil.cpp
index 4971e0d63..aaa12a5ee 100644
--- a/src/wutil.cpp
+++ b/src/wutil.cpp
@@ -264,7 +264,7 @@ int fd_check_is_remote(int fd) {
     }
     // Linux has constants for these like NFS_SUPER_MAGIC, SMB_SUPER_MAGIC, CIFS_MAGIC_NUMBER but
     // these are in varying headers. Simply hard code them.
-    switch (buf.f_type) {
+    switch ((unsigned int)buf.f_type) {
         case 0x6969:      // NFS_SUPER_MAGIC
         case 0x517B:      // SMB_SUPER_MAGIC
         case 0xFF534D42:  // CIFS_MAGIC_NUMBER

But this patch fixes it.

[faho]

Applied with comments (and the "u") to add some clarity.

Thanks!

[zanchey]

What platform is this on? We're building 32-bit binaries and shipping them without problems on a number of Linux distributions.

[igor-raits]

@zanchey Fedora Rawhide. So this might be related to the new glibc/gcc

[fweimer]

The warning for case 0xFF534D42u looks like a compiler bug.

[igor-raits]

FTR I've opened bug against gcc in RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=1803333