On Sun, Apr 26, 2009 at 10:07:36PM +0200, Roland Mainz wrote: > Fedor Sergeev wrote: > > On Fri, Apr 24, 2009 at 08:43:55PM +0200, Roland Mainz wrote: > > > Chris Quenelle wrote: > > > > Roland Mainz wrote: > > > > > Does Sun Studio 12 have a warning for statements like... > > > > > -- snip -- > > > > > int pd; > > > > > char *ptr1, > > > > > *ptr2; > > > > > /* ... */ > > > > > pd = ptr2-ptr1; > > > > > -- snip -- > > > > > ... on 64bit targets ? > > > > > > > > > > I've spend today some hours to dig-out this kind of bug where a > > > > > pointer > > > > > difference (32bit on 32bit SPARC and 64bit on 64bit SPARC, e.g. a > > > > > |ptrdiff_t|) is stored in an |int| (32bit integer) the loss of bits > > > > > trigged a malfunction... > > > > > ... somehow I was wondering why Sun Studio didn't warn about this. > > > > > > > > There's a chapter in the C user's guide about porting programs > > > > from 32-bits to 64-bits. It mentions a lint flag > > > > > > > > -errchk=longptr64 flag checks assignments of pointer > > > > expressions and long integer expressions to plain > > > > integers, even when explicit casts are used. > > > > > > > > http://docs.sun.com/app/docs/doc/819-5265/bjami?a=view > > > > > > After some hacking with "lint vs. libshell" it seems that "lint" doesn't > > > complain about the issue even when -errchk=%all is enabled... ;-( > > > > That does not agree with my experience: > > > > ] cat ptrdiff.c > > int pd; > > char *ptr1, > > *ptr2; > > pd = ptr2-ptr1; > > > > return pd; > > } > > ] lint -V ptrdiff.c -errchk=%all -c > > lint: Sun C 5.9 SunOS_i386 Patch 124868-08 2008/11/25 > > lint: Sun C 5.9 SunOS_i386 Patch 124868-08 2008/11/25 > > (5) warning: assignment of 64-bit integer to 32-bit integer > > (5) warning: variable may be used before set: ptr2 > > (5) warning: possible ptrdiff_t overflow > > What is the name of the "error tag" for this warning ? > > > (5) warning: variable may be used before set: ptr1 > > ] > > Can you go to usr/src/lib/libshell/amd64/ and run $ make lintcheck # > there, please ? I don't get this warning there but there are several
make lintcheck results in a lint run that has -erroff=E_ASSIGN_INT_TO_SMALL_INT (and other -erroffs like E_PASS_INT_TO_SMALL_INT, E_CAST_INT_TO_SMALL_INT). Thus it is just switching off this particular diagnostics. After adding -errchk=%all -erroff=%none ] make lintcheck 2>&1 | grep E_ASSIGN_INT_TO_SMALL_INT "../common/bltins/read.c", line 147: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 153: warning: assignment of 32-bit integer to 16-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 158: warning: assignment of 32-bit integer to 16-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 179: warning: assignment of 32-bit integer to 16-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 286: warning: assignment of 32-bit integer to 16-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 300: warning: assignment of 32-bit integer to 8-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 302: warning: assignment of 32-bit integer to 8-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 340: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 344: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 348: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 356: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 356: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 359: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 378: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 399: warning: assignment of 32-bit integer to 16-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 414: warning: assignment of 32-bit integer to 16-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 426: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 428: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 445: warning: assignment of 32-bit integer to 8-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 479: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 542: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 544: warning: assignment of 64-bit integer to 32-bit integer (E_ASSIGN_INT_TO_SMALL_INT) "../common/bltins/read.c", line 667: warning: assignment of 32-bit integer to 16-bit integer (E_ASSIGN_INT_TO_SMALL_INT) ] Seems to be enough for starters ;) regards, Fedor. > statements (e.g. > http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/lib/libshell/common/bltins/read.c#359 > - "m = (end - var) + (c - (end - cur));") which are currently causing > problems with string data >= 2GB because the calculations use |int| > instead of |ptrdiff_t| (this is going to be fixed with the next > ksh93-integration update but I'd like to be sure that we caught all > incarnations of this kind of bug). > > "lint" and "cc" versions are: > -- snip -- > $ cc > -V > > cc: Sun C 5.9 SunOS_i386 Patch 124868-07 2008/10/07 > usage: cc [ options] files. Use 'cc -flags' for details > $ lint > -V > > lint: Sun C 5.9 SunOS_i386 Patch 124868-07 2008/10/07 > usage: lint [ options] files. Use 'lint -flags' for details > -- snip -- > > ---- > > Bye, > Roland > > -- > __ . . __ > (o.\ \/ /.o) roland.mainz at nrubsig.org > \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer > /O /==\ O\ TEL +49 641 3992797 > (;O/ \/ \O;)
