Re: libjava Divide_1 and pr6388 fail on 4.2.0 RC3 for several targets

9 msg	Mercurial repository set up for GCC that mirror...
4 msg	GIMPLE temporary variables
2 msg	Improper Frame Description Entry - DWARF2 debug...
4 msg	PR31264 on 4.2.0 branch
9 msg	MinGW, GCC Vista,
3 msg	tuples: call for help
4 msg	One more df-branch benchmarking on SPEC2000
13 msg	Combined tree builds for mingw32
17 msg	libffi & powerpc
16 msg	live insns deleted by delete_trivially_dead_ins...
2 msg	Getting the name of the file to compile
2 msg	[cxx0x-branch] Rvalue references vs RVO
1 msg	Missing features on Changes for GCC 4.3
6 msg	auto vectorization - should this work ?
10 msg	Successful build of GCC 4.2.0 RC3 on latest Cyg...
2 msg	including PR summary in gcc-cvs commit mails

libjava Divide_1 and pr6388 fail on 4.2.0 RC3 f...
\ Kaz Kojima (6 May 2007)
. \ David Daney (6 May 2007)
. . \ David Daney (6 May 2007)
. . . \ David Daney (6 May 2007)
. . . . \ Andrew Haley (6 May 2007)
. . . . . \ Eric Botcazou (6 May 2007)
. . . . . . \ Andrew Haley (6 May 2007)
. . \ Ian Lance Taylor (7 May 2007)
. . . \ Mark Mitchell (7 May 2007)
. . . . \ Ian Lance Taylor (8 May 2007)
. . . \ Tom Tromey (8 May 2007)
. . . . \ Ian Lance Taylor (8 May 2007)

3 msg	Regression in gfortran.fortran-torture/execute/...
1 msg	gcc-4.3-20070504 is now available
1 msg	Build status for GCC 3.4.5

Subject:	Re: libjava Divide_1 and pr6388 fail on 4.2.0 RC3 for several targets
Group:	Gcc
From:	Ian Lance Taylor
Date:	8 May 2007

Tom Tromey <tromey> writes:

> >>>>> "Ian" == Ian Lance Taylor <iant> writes:
>
> Ian> This is a bug in C++ code in libjava.
>
> Thanks. We enabled -fwrapv for the interpreter but, I think, thought
> that perhaps the other C++ code was safe.
> Would the new warning have caught this?

Yes. With -Wstrict-overflow:

../../../trunk/libjava/java/lang/natString.cc: In function ¡Æjint _Jv_FormatInt(jchar*, jint)¡Ç:
../../../trunk/libjava/java/lang/natString.cc:375: warning: assuming signed overflow does not occur when simplifying conditional to constant

It also warns about this sort of code which occurs a few times in that
file:

if (offset < 0 || count < 0 || offset + count < 0
|| offset + count > data_size)

Since if offset < 0 and count < 0, then VRP can assume that offset +
count < 0 is always true. If you want to code to correctly check for
signed overflow of offset + count, you will need to instead write
something along the lines of INT_MAX - offset <= count.

Ian