[Bug libstdc++/40331] New: uniform_real random number distribution produces wildly out of range values (TR1 & C++0x)
Using either the std::tr1 or C++0x std:: random number generators, the uniform_real distribution produces numbers that are completely wrong. For instance, using Boost.Random the test program produces this sequence of floating point numbers: Random real: 0.0975404 Random real: 0.547221 Random real: 0.278498 ... On the other hand, both the TR1 and C++0x random number generators produce something like Random real: 4.18933e+08 Random real: 2.35029e+09 Random real: 1.19614e+09 ... instead. This is using the mt19937 random number generator engine and the uniform_real distribution. I have tried both uniform_real and uniform_real and they both have this bug. -- Summary: uniform_real random number distribution produces wildly out of range values (TR1 & C++0x) Product: gcc Version: 4.3.3 Status: UNCONFIRMED Severity: normal Priority: P3 Component: libstdc++ AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: colin at gibibit dot com http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40331
[Bug libstdc++/40331] uniform_real random number distribution produces wildly out of range values (TR1 & C++0x)
--- Comment #1 from colin at gibibit dot com 2009-06-03 15:52 --- Created an attachment (id=17948) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=17948&action=view) Test case that generates random reals using Boost, TR1, and C++0x Uncomment exactly one of RNG_BOOST, RNG_TR1, or RNG_CXX0X to select the random number generator implementation to use. Only the Boost random number generator produces values from uniform_real in the correct range. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40331
[Bug other/30749] New: Optimizations cause ICE segfault w/ -march=pentium
When certain optimizations are enabled (but ones which are enabled with -O and
-O2), gcc crashes with an internal compiler error due to a segmentation fault.
I encountered this problem while building openssl v0.9.8d. I have reproduced
the problem on two different machines with svn builds of gcc trunk taken on
2007-02-08 and 2007-02-09 (revision 121606).
The openssl build crashed while compiling evp_pkey.c. I have included the
preprocessed and stripped of include file line numbers file e.i file that I am
using to reproduce the crash. Compiling it with the following command
demonstrates the crash:
$ gcc -O2 -march=pentium -pipe -c -o e.o e.i
e.i: In function 'EVP_PKEY2PKCS8_broken':
e.i:7903: warning: function called through a non-compatible type
e.i:7903: note: if this code is reached, the program will abort
e.i: In function 'dsa_pkey2pkcs8':
e.i:7999: warning: function called through a non-compatible type
e.i:7999: note: if this code is reached, the program will abort
e.i: In function 'EVP_PKEY2PKCS8_broken':
e.i:7937: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See http://gcc.gnu.org/bugs.html> for instructions.
The specific flags that seem to cause problems are different on different
compiles of gcc, however! On my one installation, I found I had to at least
combine -O with -fschedule-insns and -fschedule-insns2 to crash it, (and of
course -O2 would always crash), but when I rebuilt from the same source, with
the same build process, on a different machine, I found that either -fdefer-pop
or -ftree-lrs alone would crash the compile.
Compiling with -march=pentium or -march=pentium-mmx cause the crash, but no
other architecture that I tested causes this problem.
--
Summary: Optimizations cause ICE segfault w/ -march=pentium
Product: gcc
Version: 4.3.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: other
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: colin at gibibit dot com
GCC build triplet: i686-pc-linux-gnu
GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30749
[Bug other/30749] Optimizations cause ICE segfault w/ -march=pentium
--- Comment #1 from colin at gibibit dot com 2007-02-09 18:58 --- Created an attachment (id=13027) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=13027&action=view) File causing crash: evp_pkey.c preprocessed and stripped of include-file line number clutter This is the evp_pkey.c file from openssl, preprocessed. It causes the crash, though I have successfully build glibc and a couple dozen other packages without incident. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30749
[Bug rtl-optimization/29841] [4.2/4.3 regression] ICE with scheduling and __builtin_trap
--- Comment #15 from colin at gibibit dot com 2007-02-09 20:38 --- Is the patch mentioned by Maxim checked in now? Because this bug still occurs. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29841
