Shall we do a 1.4.12 ?

[Gilles Espinasse]

----- Original Message ----- 
From: "Werner Koch" <wk at gnupg.org>
To: <gnupg-devel at gnupg.org>
Sent: Friday, January 06, 2012 2:22 PM
Subject: Shall we do a 1.4.12 ?


> Hi,
>
> the last 1.4 release is more than a year old (2010-10-08).  The repo has
> collected a few minor bug fixes and small enhancements.  Shall I do a
> new release next week?
>
>
> Shalom-Salam,
>
>    Werner
>


Tested using git tree with
./configure --prefix=/usr --build=i486-linux-gnu --disable-nls --enable-mini
mal -enable-noexecstack --enable-maintainer-mode
this compile and produce
===================
All 27 tests passed
===================

I spotted a few gliches

Like with 1.4.11, without setting in the environment
LDFLAGS    += -pie
it fail to compile using hardened compiler with this error
gcc  -Os -march=i486 -mtune=pentium -pipe -fomit-frame-pointer -Wall -Wcast-
align -Wshadow -Wstrict-prototypes -Wformat-nonliteral -Wno-pointer-sign  -W
l,--hash-style=gnu -o mpicalc mpicalc.o ../cipher/libcipher.a
../mpi/libmpi.a ../util/libutil.a
/usr/bin/ld: warning: creating a DT_TEXTREL in a shared object.

The compiler I use is gcc-4.4.5 modified with hardening by default, reason
why you don't see -fPIE.
I could live setting LDFLAGS

Compilation produce those new warnings, not seen in 1.4.11
miscutil.c:238: warning: format not a string literal, format string not
checked
estream-printf.c:1056: warning: format not a string literal, argument types
not checked
estream-printf.c:1059: warning: format not a string literal, argument types
not checked
so something is wrong, not yet understood what
Looking at miscutil.c, I checked the define in config.h and found
#define HAVE_STRFTIME 1
#define HAVE_NL_LANGINFO 1
LANG is not set inside the chroot
Same warnings happen if not using --disable-nls

The last thing I find is that clock_gettime syscall is not found in gnupg
and libgcrypt despite it is found by a small majority of other packages
(glibc-2.11.3 is in chroot). Looking at the compilation log :
grep -r clock_gettime log_i486 | grep yes
log_i486/02_base/gzip-1.4:checking for clock_gettime... yes
log_i486/02_base/diffutils-3.2:checking for clock_gettime... yes
log_i486/02_base/tar-1.26:checking for clock_gettime... yes
log_i486/02_base/coreutils-8.14:checking for clock_gettime... yes
log_i486/02_base/findutils-4.4.2:checking for clock_gettime... yes
log_i486/03_ipcop/cairo-1.10.2:checking for clock_gettime... yes
log_i486/03_ipcop/glib-2.26.1:checking for clock_gettime in -lrt... yes
log_i486/03_ipcop/libusb-1.0.8:checking for clock_gettime in -lrt... yes
log_i486/03_ipcop/ntp-4.2.6p5:checking for clock_gettime... yes
log_i486/03_ipcop/cpio-2.11:checking for clock_gettime... yes
log_i486/03_ipcop/wget-1.13.4:checking for clock_gettime... yes
log_i486/03_ipcop/wget-1.13.4:checking for clock_gettime... (cached) yes

grep -r clock_gettime log_i486 | grep no
log_i486/02_base/rsyslog-5.8.6:checking for clock_gettime... no
log_i486/02_base/gmp-5.0.2:checking for clock_gettime... no
log_i486/03_ipcop/db-4.8.30:checking for clock_gettime... no
log_i486/03_ipcop/db-4.8.30:checking for clock_gettime monotonic clock... no
log_i486/03_ipcop/libgcrypt-1.4.6:checking for clock_gettime... no
log_i486/03_ipcop/glib-2.26.1:checking for clock_gettime... no
log_i486/03_ipcop/lzo-2.04:checking for clock_gettime... no

Using AC_CHECK_FUNCS for clock_gettime is not enought as a test should be
made with -lrt

Gilles