aniani/elinks
1
0
Fork 0
mirror of https://github.com/rkd77/elinks.git synced 2024-06-20 00:15:31 +00:00
Code

Bug 1012: Compile with -fno-strict-overflow or -fwrapv.

Browse Source 
(cherry picked from commit b7312e6882)
This commit is contained in:
Kalle Olavi Niemitalo 2008-05-04 11:20:55 +03:00 committed by Kalle Olavi Niemitalo
parent 34bdddbd87
commit 2e1df1dc78
2 changed files with 56 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
NEWS
View File

@ -323,6 +323,7 @@ To be released as 0.11.4.
* bug 938: elinks -remote no longer needs a controlling tty
* bug 939: fix FSP directory listing (some compiler options left it empty)
* bug 978: Python's webbrowser.open_new_tab(URL) works since now
* bug 1012: compile with -fno-strict-overflow or -fwrapv if available
* minor bug 54, Debian bug 338402: don't force the terminal to 8 bits
with no parity, and don't disable XON/XOFF flow control either
* minor bug 951 in user SMJS: garbage-collect SMJS objects on 'File ->

55
configure.in
View File

@ -1451,6 +1451,61 @@ if test "x$ac_cv_c_compiler_gnu" = "xyes"; then
[AC_MSG_RESULT([no])])
EL_RESTORE_FLAGS
done
# Bug 1012: Some parts of ELinks do arithmetic with signed integers
# in such a way that an overflow is possible. GCC 4.2.1 warns
# "warning: assuming signed overflow does not occur when assuming
# that (X + c) > X is always true", which may be an incorrect
# optimization (although allowed by the standard), and breaks the
# build with -Werror.
#
# All of the following tests included -S -Wall -Wextra:
#
# GCC: (GNU) 4.0.4 20060507 (prerelease) (Debian 4.0.3-3)
# * gcc-4.0 -O0: OK, compares the values
# * gcc-4.0 -O2: assumes no overflow, does not warn
# * gcc-4.0 -O0 -fno-strict-overflow: unrecognized command line option
# * gcc-4.0 -O0 -fwrapv: OK, compares the values
# * gcc-4.0 -O2 -fwrapv: OK, compares the values
# * gcc-4.0 -O0 -ftrapv: OK, calls __addvsi3
# * gcc-4.0 -O2 -ftrapv: assumes no overflow, does not warn
# * gcc-4.0 -O0 -fwrapv -ftrapv: OK, calls __addvsi3
# * gcc-4.0 -O2 -fwrapv -ftrapv: compares the values
#
# GCC: (GNU) 4.1.3 20070812 (prerelease) (Debian 4.1.2-15)
# * gcc-4.1 -O0: assumes no overflow, does not warn
# * gcc-4.1 -O2: assumes no overflow, does not warn
# * gcc-4.1 -O0 -fno-strict-overflow: unrecognized command line option
# * gcc-4.1 -O0 -fwrapv: OK, compares the values
# * gcc-4.1 -O2 -fwrapv: OK, compares the values
# * gcc-4.1 -O0 -ftrapv: OK, calls __addvsi3
# * gcc-4.1 -O2 -ftrapv: compares the values
# * gcc-4.1 -O0 -fwrapv -ftrapv: OK, calls __addvsi3
# * gcc-4.1 -O2 -fwrapv -ftrapv: compares the values
#
# GCC: (GNU) 4.2.1 (Debian 4.2.1-5)
# * gcc-4.2 -O0: OK, compares the values
# * gcc-4.2 -O2: assumes no overflow, warns about it
# * gcc-4.2 -O0 -fno-strict-overflow: OK, compares the values
# * gcc-4.2 -O2 -fno-strict-overflow: OK, compares the values
# * gcc-4.2 -O0 -fwrapv: OK, compares the values
# * gcc-4.2 -O2 -fwrapv: OK, compares the values
# * gcc-4.2 -O0 -ftrapv: OK, calls __addvsi3
# * gcc-4.2 -O2 -ftrapv: compares the values
# * gcc-4.2 -O0 -fwrapv -ftrapv: OK, calls __addvsi3
# * gcc-4.2 -O2 -fwrapv -ftrapv: compares the values
#
# Apparently, -ftrapv does not work reliably at all.
for overflow_flag in -fno-strict-overflow -fwrapv; do
AC_MSG_CHECKING([whether $CC accepts $overflow_flag])
EL_SAVE_FLAGS
CFLAGS="$CFLAGS $overflow_flag"
AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[]], [[]])],
[AC_MSG_RESULT([yes])
break],
[AC_MSG_RESULT([no])])
EL_RESTORE_FLAGS
done
fi
EL_LOG_CONFIG(CFLAGS, [Compiler flags (CFLAGS)], [])