Warning fixes for r1648

[Michael D. Adams]

Attached is a patch to fix a few warnings I found. Below is an explanation of the warnings and how they were fixed. After that are listed some warnings that I found that were not fixed in the patch.

FIXED WARNINGS:

ikarus-ffi.c:193: warning: ‘dump_stack’ defined but not used
  - guard with #ifdef DEBUG_FFI

ikarus-ffi.c:338: warning: ISO C forbids passing argument 2 of ‘ffi_call’ between function pointer and ‘void *’
  - use "void (*fn)() = FFI_FN(...)"

ikarus-numerics.c:804: warning: pointer of type ‘void *’ used in arithmetic
  - "(void*)res+disp_bignum_data" --> "(void*)(res+disp_bignum_data)"

ikarus-ffi.c:173: warning: comparison between signed and unsigned
ikarus-ffi.c:328: warning: comparison between signed and unsigned
ikarus-ffi.c:340: warning: comparison between signed and unsigned
ikarus-ffi.c:405: warning: comparison between signed and unsigned
  - make "i" into "unsigned int"

UNFIXED WARNINGS: The following are warnings I didn't know if or how they should be fixed

ikarus-runtime.c:126: warning: comparison between signed and unsigned
  - Maybe use "unsigned long int size" in function signature?
  - If change, ensure ik_munmap is consistent with ik_munmap_from_segmant.

ikarus-collect.c:1363: warning: comparison between signed and unsigned
  - the problem is "relative_distance != (int)relative_distance"
  - added in r1544
  - seems to be intentional but I'm not sure what the motive is or if there is a better way
  - if kept should have a comment explaining what it is about

[Abdulaziz Ghuloum (aghuloum)]

Thanks!, fixed in rev 1653.
Re:
ikarus-collect.c:1363: warning: comparison between signed and unsigned
  - the problem is "relative_distance != (int)relative_distance"
  - added in r1544
  - seems to be intentional but I'm not sure what the motive is or if there is a better way
  - if kept should have a comment explaining what it is about

I changed that test to be:
        if(((long int)relative_distance) != ((long)((int)relative_distance))){
which means: does the value change if cast to an int. I think it's as clear as C could be. :-)

BTW, and maybe I asked you this before, but what version of gcc do you use, and what flags do you give it?

[Michael D. Adams (mdmkolbe)]

I use GCC 4.2.4 and I pass it '-W -Wall -pedantic' by using "make CFLAGS='-W -Wall -pedantic'". The stderr output I capture and filter to remove some of the more common but useless errors (e.g. "C90 forbids mixing declarations and code").

[Michael D. Adams (mdmkolbe)]

Ok, a few problems cropped up when you applied the changes:

ikarus-ffi.c:191: #ifdef DEBUG_FFI has extra spaces after it
ikarus-ffi.c:325: you now have it as "void (*fn)() = (void*) ..." but assigning a void* to a void(*fn)() is not legal without a cast.
  Either change it to "void (*fn)() = (void (*fn)()) ..." to explicitly do the cast or use the FFI_FN macro provided by libffi to do the cast for you.

ikarus-runtime.c:215: type of mapsize needs to be "unsigned long int" to make the comparison with size in the assert on the next line happy.
ikarus-runtime.c:258: The printf needs to use %016lx form instead of %016x form at least for the second argument which has the unsigned long mapsize in it. (Compare this with line 235 which is ok.)

[Abdulaziz Ghuloum (aghuloum)]

Fixed again in revision 1656.

[leppie (leppie)]

Here is one that still comes up on Cygwin (not sure if present elsewhere):

gcc -DHAVE_CONFIG_H -I. -I.. -g -O2 -DNDEBUG -O3 -Wall -MT ikarus-runtime.o -MD -MP -MF .deps/ikarus-runtime.Tpo -c -o ikarus-runtime.o ikarus-runtime.c
ikarus-runtime.c: In function `ik_munmap':
ikarus-runtime.c:255: warning: passing arg 1 of `win_munmap' makes pointer from integer without a cast

[Abdulaziz Ghuloum (aghuloum)]

Fixed in 1660. Thanks.

(btw, you should reopen a closed bug if you think it still needs fixing, like in this case)

[leppie (leppie)]

> like in this case

Ok :)

[Michael D. Adams (mdmkolbe)]

When you fixed it, you still missed these two (maybe you are compiling with -DNDEBUG):

ikarus-runtime.c:215: type of mapsize needs to be "unsigned long int" to make the comparison with size in the assert on the next line happy.
ikarus-runtime.c:258: The printf needs to use %016lx form instead of %016x form at least for the second argument which has the unsigned long mapsize in it. (Compare this with line 235 which is ok.)

[Abdulaziz Ghuloum (aghuloum)]

Fixed in 1686.