Package: intelrdfpmath
Version: 2.0u2-8
Severity: normal
Tags: patch
User: ubuntu-de...@lists.ubuntu.com
Usertags: origin-ubuntu mantic ubuntu-patch
Dear maintainers,
The intelrdfpmath package is failing to build on all architectures in Ubuntu
because some test cases are being built with -Werror, and this is currently
stricter on Ubuntu than on Debian:
[...]
make[2]: Entering directory '/<<PKGBUILDDIR>>/TESTS'
x86_64-linux-gnu-gcc -O0 -D__intptr_t_defined -DLINUX -Werror
-DDECIMAL_CALL_BY_REFERENCE=0 -DDECIMAL_GLOBAL_ROUNDING=0
-DDECIMAL_GLOBAL_EXCEPTION_FLAGS=0 -UBID_BIG_ENDIAN -UHPUX_OS -o readtest
readtest.c ../LIBRARY/libbidgcc000.a -lm
readtest.c: In function ‘print_mismatch’:
readtest.c:1397:24: error: ‘ ’ directive writing 1 byte into a region of size
between 0 and 1023 [-Werror=format-overflow=]
1397 | sprintf (line, "%s %s %s %s\n", func, op1, op2, op3);
| ^
readtest.c:1397:3: note: ‘sprintf’ output 5 or more bytes (assuming 3074) into
a destination of size 1024
1397 | sprintf (line, "%s %s %s %s\n", func, op1, op2, op3);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1: all warnings being treated as errors
make[2]: *** [makefile:126: readtest] Error 1
[...]
(https://launchpad.net/ubuntu/+source/intelrdfpmath/2.0u2-8/+build/26010059)
While not currently a build failure in Debian, we can expect it to become
one with a future compiler update.
Unfortunately, using snprintf() to enforce bounds checking of the target
array doesn't avert this error, because format-truncation is also applied to
the buffer in snprintf(). But the attached patch lets the package build, by
both using snprintf() for more explict safety wrt buffer handling, and
turning off this particular error from gcc.
I've uploaded this to Ubuntu to let the package build there.
Thanks for considering,
--
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.
Ubuntu Developer https://www.debian.org/
slanga...@ubuntu.com vor...@debian.org
diff -Nru intelrdfpmath-2.0u2/debian/patches/series
intelrdfpmath-2.0u2/debian/patches/series
--- intelrdfpmath-2.0u2/debian/patches/series 2023-02-08 06:20:15.000000000
-0800
+++ intelrdfpmath-2.0u2/debian/patches/series 2023-06-13 10:35:51.000000000
-0700
@@ -3,3 +3,4 @@
abs-declaration.patch
wchar_t.patch
mongo-inteldfp-s390x.patch
+snprintf-not-sprintf.patch
diff -Nru intelrdfpmath-2.0u2/debian/patches/snprintf-not-sprintf.patch
intelrdfpmath-2.0u2/debian/patches/snprintf-not-sprintf.patch
--- intelrdfpmath-2.0u2/debian/patches/snprintf-not-sprintf.patch
1969-12-31 16:00:00.000000000 -0800
+++ intelrdfpmath-2.0u2/debian/patches/snprintf-not-sprintf.patch
2023-06-13 10:39:53.000000000 -0700
@@ -0,0 +1,38 @@
+Description: use snprintf not sprintf to make the compiler happy
+ With the gcc currently in Ubuntu, the package build fails because the
+ compiler can't tell whether this sprintf() will overflow the target.
+ Just use snprintf() to ensure that it can't.
+ .
+ But -Werror=format-truncation triggers *anyway* even on snprintf, so
+ disable this error as well.
+Author: Steve Langasek <steve.langa...@ubuntu.com>
+Bug-Ubuntu: https://bugs.launchpad.net/bugs/2023683
+Last-Update: 2023-06-13
+Forwarded: no
+
+Index: intelrdfpmath-2.0u2/TESTS/readtest.c
+===================================================================
+--- intelrdfpmath-2.0u2.orig/TESTS/readtest.c
++++ intelrdfpmath-2.0u2/TESTS/readtest.c
+@@ -1394,7 +1394,7 @@
+
+ printf ("// Input operand strings: %s %s %s\n", istr1, istr2, istr3);
+ fail_res++;
+- sprintf (line, "%s %s %s %s\n", func, op1, op2, op3);
++ snprintf (line, 1024, "%s %s %s %s\n", func, op1, op2, op3);
+ printf ("// Ulp error: %e\n", ulp+ulp_add);
+ printf ("// Full input string: %s\n", full_line);
+ printf ("// Input string number: %d\n", line_counter);
+Index: intelrdfpmath-2.0u2/TESTS/makefile
+===================================================================
+--- intelrdfpmath-2.0u2.orig/TESTS/makefile
++++ intelrdfpmath-2.0u2/TESTS/makefile
+@@ -78,7 +78,7 @@
+ ifeq ($(CC),icc)
+ COPT = -Werror -Qoption,cpp,--extended_float_types
+ else
+-COPT = -Werror
++COPT = -Werror -Wno-error=format-truncation
+ endif
+ endif
+
