xfsprogs: enable sparse checking

Enable "make C=1" or "make C=2" to do sparse checking. Blatantly ripped off from djwong's patch in e2fsprogs to do the same.

Note, this requires unreleased sparse after v0.5, which enables the CHAR_BIT definition; otherwise it chokes.

---

index f56aebd..bd0f9ee 100644
@@ -15,6 +15,20 @@ else
   Q = @
 endif
 
+CHECK=sparse
+CHECK_OPTS=-Wsparse-all -Wno-transparent-union -Wno-return-void -Wno-undef \
+	-Wno-non-pointer-null -D__linux__
+ifeq ("$(C)", "2")
+  CHECK_CMD=$(CHECK) $(CHECK_OPTS) -Wbitwise -D__CHECK_ENDIAN__ 
+else
+  ifeq ("$(C)", "1")
+    CHECK_CMD=$(CHECK) $(CHECK_OPTS)
+   else
+    CHECK_CMD=@true
+  endif
+endif
+export CHECK_CMD
+
 MAKEOPTS = --no-print-directory Q=$(Q)
 
 TOPDIR = .

index 399635e..feae376 100644
@@ -41,10 +41,12 @@ $(LTLIBRARY) : $(SUBDIRS) $(LTOBJECTS)
 %.lo: %.c
 	@echo "    [CC]     $@"
 	$(Q)$(LTCOMPILE) -c $<
+	$(Q)$(CHECK_CMD) $(CFLAGS) $<
 else
 %.o: %.c
 	@echo "    [CC]     $@"
 	$(Q)$(CC) $(CFLAGS) -c $<
+	$(Q)$(CHECK_CMD) $(CFLAGS) $<
 
 endif
 

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

I'm pretty sure I used to be able to do sparse checks on xfsprogs using

CC=cgcc make

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

Hm, so you can.

Ok, so maybe this is dumb, or is it convenient (to work like it does in the kernel tree?)

Thanks,
-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

The cgcc way seems easier for userspace, but I don't really have a strong opinion. It's not like your patch prevents me from using CC=cgcc

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

To get endian checking with your method, you'd need to edit the Makefile to add the -D define, right?

Thanks,
-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

That's a kernel special, I just did a test run using CC=cgcc and I get endianess warnings:

rdwr.c:190:21: warning: incorrect type in assignment (different base types)
rdwr.c:190:21: expected restricted __le32 [usertype] h_crc
rdwr.c:190:21: got restricted __be32 [usertype] <noident>

log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64 log_misc.c:621:59: warning: cast to restricted __be64

There's also tons of noise that should be easily upgradeable, but let me check latests xfsprogs first..

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

Oh, without an extra define? Hrm. Ok I guess I need to read up on what enables it.

I did send a couple patches to fix some of these. Sorry, didn't do it in a series:

[PATCH] xfsprogs: fix endian mishap in xfs_dialloc_ag() [PATCH] xfsprogs: fix harmless sparse endian nit

-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs

Ok, NAK to this one I guess, hch has sufficiently pointed out that I don't quite know what I'm doing here ;)

-Eric

_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs