xfsprogs-7.2.1 non-parallel-safe make and some gcc-16.2.0 warnings

xfsprogs-7.2.1 non-parallel-safe make and some gcc-16.2.0 warnings

[Dr. Thomas Orgis]

Hi linux-xfs,

I noticed that xfsprogs-7.2.1 `make install` fails for me with make
-j13 on a (oldish Xeon) box with 12 physical cores on two sockets (24
hyperthreads). Plus, there are some warnings from gcc-16.2.0 in the
default build that you might be interested in.

The full build/install log is at
	https://src.rrz.uni-hamburg.de/extra/paste/xfsprogs-7.2.1.log.bz2

The relevant excerpts:

1. make install error with -j13

(after successful `make -j13`)

Installing include-install
../install-sh -o root -g root -m 755 -d /usr/include/xfs
../../install-sh -o root -g root -m 644 xfs_db.8 /usr/share/man/man8/xfs_db.8
Installing libfrog-install-dev
../../install-sh -o root -g root -m 644 xfs_estimate.8 /usr/share/man/man8/xfs_estimate.8
Installing libxfs-install-dev
Installing libxlog-install-dev
gmake[1]: *** No rule to make target 'xfs_rt', needed by 'defer_item.lo'.  Stop.
gmake[1]: *** Waiting for unfinished jobs....
gmake[1]: *** No rule to make target 'xfs_rt', needed by 'defer_item.lo'.  Stop.
gmake[1]: *** Waiting for unfinished jobs....
make: *** [Makefile:139: libxfs-install-dev] Error 2
make: *** Waiting for unfinished jobs....
make: *** [Makefile:139: libxfs-install-dev] Error 2
make: *** Waiting for unfinished jobs....

Re-running `make install` with -j1 works.


2. gcc warnings

xfs_alloc.c: In function '__xfs_free_extent':
xfs_alloc.c:4017:41: warning: variable 'busy_flags' set but not used [-Wunused-but-set-variable=]
 4017 |         unsigned int                    busy_flags = 0;
      |                                         ^~~~~~~~~~

xfs_bmap.c: In function 'xfs_bmapi_convert_one_delalloc':
xfs_bmap.c:4360:33: warning: variable 'flags' set but not used [-Wunused-but-set-variable=]
 4360 |         uint16_t                flags = 0;
      |                                 ^~~~~

xfs_bmap.c: In function 'xfs_bmap_btree_to_extents':
xfs_bmap.c:598:22: warning: 'cbp' may be used uninitialized [-Wmaybe-uninitialized]
  598 |         if ((error = xfs_btree_check_block(cur, cblock, 0, cbp)))
      |                      ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
xfs_bmap.c:568:34: note: 'cbp' was declared here
  568 |         struct xfs_buf          *cbp;   /* child block's buffer */
      |                                  ^~~

xfs_log_rlimit.c: In function 'libxfs_log_get_max_trans_res':
xfs_log_rlimit.c:162:33: warning: variable 'i' set but not used [-Wunused-but-set-variable=]
  162 |         unsigned int            i;
      |   
                              ^
xfs_copy.c: In function 'main':
xfs_copy.c:562:25: warning: variable 'num_threads' set but not used [-Wunused-but-set-variable=]
  562 |         int             num_threads = 0;
      |                         ^~~~~~~~~~~

dino_chunks.c: In function 'process_uncertain_aginodes':
dino_chunks.c:1349:33: warning: variable 'cnt' set but not used [-Wunused-but-set-variable=]
 1349 |         int                     cnt;
      |                                 ^~~

prefetch.c: In function 'do_inode_prefetch':
prefetch.c:1027:33: warning: variable 'queues_started' set but not used [-Wunused-but-set-variable=]
 1027 |         int                     queues_started = 0;
      |                                 ^~~~~~~~~~~~~~

btheight.c: In function 'construct_records_per_block':
btheight.c:158:11: warning: assignment discards 'const' qualifier from pointer target type [-Wdiscarded-qualifiers]
  158 |         p = strchr(tag, ':');
      |           ^

Please just ignore me if this is irrelevant with current development
versions, I do not have time to invest in this, but thought it is more
polite to say something than to just ignore it outright.


Alrighty then,

Thomas

-- 
Dr. Thomas Orgis
HPC @ Universität Hamburg

Re: xfsprogs-7.2.1 non-parallel-safe make and some gcc-16.2.0 warnings

[Christoph Hellwig]

Hi Thomas,

all the warnings should be fixed in current git for-next.

The build systems in xfsprogs is a major pain and we'll need to find
time to replace it with something modern.