https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102405
Bug ID: 102405
Summary: Loop index limited by a function return value
considered to result in too big array subscript
Product: gcc
Version: 12.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: cazfi74 at gmail dot com
Target Milestone: ---
For a code where for-loop terminates before index variable gets value 25*8,
compiler gives a warning that the variable/8 used as array subscript has value
of 25 and is thus above array bounds.
Warning (turned to an error by -Werror):
In function 'pick_random_tech_to_lose',
inlined from 'update_bulbs' at ../../../src/server/techtools.c:621:42:
../../../src/server/techtools.c:674:271: error: array subscript 25 is above
array bounds of 'unsigned char[25]' [-Werror=array-bounds]
674 | if (BV_ISSET(eligible_techs, i)) {
|
^
In file included from ../../../src/common/player.h:31,
from ../../../src/common/game.h:36,
from ../../../src/server/techtools.c:27:
../../../src/common/tech.h: In function 'update_bulbs':
../../../src/common/tech.h:144:32: note: while referencing 'vec'
144 | BV_DEFINE(bv_techs, A_LAST);
| ^
cc1: all warnings being treated as errors
The code compiles fine with gcc-11 (and number of earlier versions). It has
failed for some weeks with the Debian's gcc-12 development snapshot, with
latest update just today:
$ /usr/lib/gcc-snapshot/bin/gcc --version
gcc (Debian 20210918-1) 12.0.0 20210918 (experimental) [master
r12-3644-g7afcb534239]
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
The loop termination is determined by a function call, like this:
'for (; _index < advance_count(); _index++) {'
If 'advance_count()' is replaced by '(25*8)', it compiles fine. The compiler
doesn't know what values advance_count() can return (it's implemented in a
different source file) - the compile fails even if it's implemented as simple
'return 25*8;'
There's a lot of macro magic around in the original code (if you think that the
warning is what one should get from the attached .i, then the bug is probably
preprocessor side)
The command line:
/usr/lib/gcc-snapshot/bin/gcc -v -save-temps -DHAVE_CONFIG_H -I.
-I../../../src/server -I../gen_headers -I../../../src/ai
-I../../../src/ai/classic -I../../../src/common -I../../../src/common/aicore
-I../../../src/common/scriptcore -I../../../src/utility
-I../../../src/server/advisors -I../../../src/server/generator
-I../../../src/server/scripting -DLOCALEDIR="/usr/local/share/locale"
-DBINDIR="/usr/local/bin" -DFC_CONF_PATH="/usr/local/etc/freeciv"
-DDEFAULT_DATA_PATH=".:data:~/.freeciv/2.6:/usr/local/share/freeciv"
-DDEFAULT_SAVE_PATH=".:~/.freeciv/saves"
-DDEFAULT_SCENARIO_PATH=".:data/scenarios:~/.freeciv/2.6/scenarios:~/.freeciv/scenarios:/usr/local/share/freeciv/scenarios"
-Wall -Wpointer-arith -Wcast-align -Werror -Wmissing-prototypes
-Wmissing-declarations -Wformat -Wformat-security -Wnested-externs -Wshadow
-Wold-style-declaration -Wno-tautological-compare -Wno-nonnull-compare -g -O2
-MT techtools.lo -MD -MP -MF .deps/techtools.Tpo -c
../../../src/server/techtools.c -o techtools.o
