bgcolor # Type Status Created By Subsys Changed Assigned Svr Pri Title _Description _Remarks #f2dcdc 53 code new 2004 Nov anonymous xds 2004 Nov rse 2 1 on certain architectures assignment va_list a = b; fails As in the summary: on some architectures (ppc for example) the code:{linebreak} va_list a;{linebreak} va_list b;{linebreak} b = a;{linebreak} will fail. After a quick googling I've found a solution: {linebreak} __va_copy(b,a); {linebreak} is the proper code - on some architectures va_list is defined differently and the simple substitution will fail. On such architectures the xds build fails with:{linebreak} xds.c: In function `xds_vencode':{linebreak} xds.c:386: error: incompatible types in assignment{linebreak} xds.c:391: warning: passing arg 6 of pointer to function from incompatible pointer type{linebreak} xds.c: In function `xds_vdecode':{linebreak} xds.c:486: warning: passing arg 6 of pointer to function from incompatible pointer type{linebreak} make: *** [xds.lo] Error 1{linebreak} {linebreak} the warnings are fixed with (va_list *) in front of the argument in question - I checked and it seems that the argument is proper anyways... #f2dcdc 208 code new 2023 Nov anonymous uuid 2023 Nov rse 1 1 IP D9:2B:67:4C:AF:65:0F:58:3C:BF:50:7E:59:71:B2:AF #f2dcdc 207 code new 2023 Oct anonymous uuid 2023 Oct rse 1 1 Build error with MinGW cross-compilation A MinGW build fails with libtool: install: ./shtool install -c -m 755 -s .libs/uuid /nix/store/r8clhx2d35c4iy3y299ifx1i8lpaphcx-libossp-uuid-x86_64-w64-mingw32-1.6.2/bin/uuid cp: cannot stat '.libs/uuid': No such file or directory The problem is that the built executable file has a =.exe= suffix which needs to be accounted for. The following patch fixes this in a cross-platform way: diff --git a/Makefile.in b/Makefile.in --- a/Makefile.in +++ b/Makefile.in @@ -56,6 +56,7 @@ CP = cp RMDIR = rmdir SHTOOL = $(S)/shtool LIBTOOL = $(C)/libtool +EXEEXT = @EXEEXT@ TRUE = true POD2MAN = pod2man PERL = @PERL@ @@ -253,7 +254,7 @@ install: -@if [ ".$(WITH_CXX)" = .yes ]; then \ $(LIBTOOL) --mode=install $(SHTOOL) install -c -m 644 $(CXX_NAME) $(DESTDIR)$(libdir)/; \ fi - @$(LIBTOOL) --mode=install $(SHTOOL) install -c -s -m 755 uuid $(DESTDIR)$(bindir)/ + @$(LIBTOOL) --mode=install $(SHTOOL) install -c -s -m 755 $(PRG_NAME)$(EXEEXT) $(DESTDIR)$(bindir)/ $(SHTOOL) install -c -m 644 $(S)/uuid.1 $(DESTDIR)$(mandir)/man1/ -@if [ ".$(WITH_PERL)" = .yes ]; then \ (cd $(S)/perl && $(MAKE) $(MFLAGS) install DESTDIR=$(DESTDIR)); \ #f2dcdc 206 code new 2022 May anonymous uuid 2022 May rse 3 4 Data::UUID compat create_from_name() is broken Hi, the create_from_namespace() method actually never uses the provided namespace (parameter $nsid). --- a/Data/UUID.pm +++ b/Data/UUID.pm @@ -61,7 +61,7 @@ my ($self, $nsid, $name) = @_; my $uuid = OSSP::uuid->new; my $nsiduuid = OSSP::uuid->new; - $nsiduuid->import('bin', $nsiduuid); + $nsiduuid->import('bin', $nsid); $uuid = OSSP::uuid->new; $uuid->make('v3', $nsiduuid, $name); return $uuid->export('bin'); Not sure if fixing this is actually a good idea. Doing so would mean the code would start generating different UUIDs for the same input. OTOH the current state makes this non-interoperable with other implementations. #f2dcdc 205 code new 2021 Feb anonymous uuid 2021 Feb rse 5 2 typo in uuid.so There is a typo "contant" in uuid.so. It should probably read "constant"?{linebreak} Like: $ perl -MOSSP::uuid -le 'print OSSP::uuid->new->typo' unknown contant OSSP::uuid::typo at -e line 1. TIA && Cheers #f2dcdc 204 event new 2019 Aug anonymous uuid 2019 Aug 1 1 Unable to connect to ftp When clicking on any link leading to uuid software (not just v 1.6.2), the connection to the ftp cannot be established. #f2dcdc 203 code new 2017 Nov anonymous uuid 2017 Nov rse 1 2 Trouble configuring with clang++ on android platform, arch-arm64 I am configuring with CC environment variable pointing to clang++, specifically from the android-ndk-r15c release. Also I am running the configure script with --host aarch64-none-linux-android although I also tried the triplet x86_64-linux-android with no success. The configure script is stopping with checking host system type... Invalid configuration `x86_64-linux-android`: system `android` not recognized Is there a workaround for this? I would be OK with using a regular C compiler however I chose clang++ because the ndk examples on android are all using it, and I need to gain more experience before attempting to build examples and existing code with regular C. Thank you. #f2dcdc 202 doc new 2016 Jun anonymous uuid 2016 Jun rse 3 1 export uuid Existe a possibildiade de usar a biblioteca uuid em aplicativos em pascal/delphi? #f2dcdc 201 doc new 2016 Jun anonymous uuid 2016 Jun rse 3 1 export uuid Existe a possibildiade de usar a biblioteca uuid em aplicativos em pascal/delphi? #f2dcdc 200 code new 2015 Sep anonymous uuid 2015 Sep rse 3 3 uuid.pc.in does not use the right autoconf variables for paths uuid.pc.in does not use the correct variables to define exec_prefix, includedir and libdir. As a result, if a build system overrides the default, the exported location hints are wrong. Eg: https://github.com/Homebrew/homebrew/issues/43843 OSX's HomeBrew passes "--includedir=#{include}/ossp", but the generated uuid.pc does NOT use it, so the header location is wrong. This simple patch fixes it: --- uuid.pc.in 2015-09-12 04:03:34.921192349 +0100 +++ uuid.pc.in 2015-09-12 04:13:53.226386329 +0100 @@ -30,5 +30,5 @@ prefix=@prefix@ -exec_prefix=${prefix} -includedir=${prefix}/include -libdir=${exec_prefix}/lib +exec_prefix=@exec_prefix@ +includedir=@includedir@ +libdir=@libdir@ #f2dcdc 199 code new 2015 Apr anonymous uuid 2015 Apr rse 1 1 aggregate 'uuid_t uu' has incomplete type and cannot be defined I try to install libdap 3.14.0 on CentOS 7. The system is fresh new meaning no 'libuuid-devel' is installed. I want to manually install uuid, and it is built successfuly. But when I built libdap, I encountered the following error:
DODSFilter.cc: In member function 'virtual void libdap::DODSFilter::dataset_constraint_ddx(libdap::DDS&, libdap::ConstraintEvaluator&, std::ostream&, const string&, const string&, bool) const': DODSFilter.cc:895:12: error: aggregate 'uuid_t uu' has incomplete type and cannot be defined uuid_t uu; ^ DODSFilter.cc:896:21: error: 'uuid_generate' was not declared in this scope uuid_generate(uu); ^ DODSFilter.cc:898:30: error: 'uuid_unparse' was not declared in this scope uuid_unparse(uu, &uuid[0]);#f2dcdc 198 code new 2015 Mar anonymous uuid 2015 Mar rse 2 2 Cannot cross compile for ARM architecture Executing: ./configure --host=arm-linux CC=arm-linux-gnueabihf-gcc CXX=arm-linux-gnueabihf-g++ Generate error from configure: ... checking for accept in -lsocket... no checking for va_copy() function... configure: error: cannot run test program while cross compiling See `config.log' for more details. The problem is AC_CHECK_VA_COPY() from uuid.ac file. Workaround: ac_cv_va_copy=C99 ./configure --host=arm-linux CC=arm-linux-gnueabihf-gcc CXX=arm-linux-gnueabihf-g++ #f2dcdc 196 code new 2013 Dec anonymous uuid 2013 Dec rse 1 1 Unable to cross compile due to AC_TRY_RUN When trying to cross compile uuid for arm configure stops with following error: checking for va_copy() function... configure: error: cannot run test program while cross compiling This is due to AC_TRY_RUN(__va_copy_test($2), [ac_cv_va_copy="$1"]) in aclocal.m4. I worked around this problem by defining ac_cv_va_copy=C99 as an environment variable before configure. It would be nice to have cross compilation working out of the box #f2dcdc 190 code new 2012 Jun anonymous uuid 2013 May rse 1 1 v1 uuids not working on Linux or with MAC addrs with high-bit set I can't get a v1 uuid on CentOS 6.2 or Fedora 16 with 1.6.2. [philipp@builder ~]$ uuid -v1{linebreak} 639519c8-b019-11e1-9b4b-ff91eac3a44f{linebreak} [philipp@builder ~]$ uuid -d `uuid -v1`{linebreak} encode: STR: 6aed0654-b019-11e1-abf4-8f6127f75e13{linebreak} SIV: 142128870312796779101177586876736298515{linebreak} decode: variant: DCE 1.1, ISO/IEC 11578:1996{linebreak} version: 1 (time and node based){linebreak} content: time: 2012-06-06 20:51:41.241506.0 UTC{linebreak} clock: 11252 (usually random){linebreak} node: 8f:61:27:f7:5e:13 (local multicast){linebreak} [philipp@builder ~]$ ip -0 link show{linebreak} 1: lo:
... ./libtool --quiet --mode=compile /opt/studio/SOS11/SUNWspro/bin/cc -c -xO3 -xarch=v9 -I/opt/csw/include mm_vers.c ./libtool --quiet --mode=link /opt/studio/SOS11/SUNWspro/bin/cc -o libmm.la mm_global.lo mm_alloc.lo mm_core.lo mm_lib.lo mm_vers.lo \ -rpath /opt/csw/lib/64 -version-info `./shtool version -l c -d libtool mm_vers.c` ./libtool --quiet --mode=compile /opt/studio/SOS11/SUNWspro/bin/cc -c -xO3 -xarch=v9 -I/opt/csw/include mm_test.c ./libtool --quiet --mode=link /opt/studio/SOS11/SUNWspro/bin/cc -o mm_test mm_test.lo libmm.la ld: fatal: file .libs/mm_test.o: wrong ELF class: ELFCLASS64 ld: fatal: File processing errors. No output written to .libs/mm_test gmake[2]: *** [mm_test] Error 1This is because the CFLAGS turning on 64 bit are not passed to libtool. _2009-Sep-09 16:45:33 by anonymous:_ {linebreak} This patch solves the problem:
diff -Naur mm-1.4.2.orig/Makefile.in mm-1.4.2.patched/Makefile.in --- mm-1.4.2.orig/Makefile.in 2006-08-10 21:05:55.000000000 +0200 +++ mm-1.4.2.patched/Makefile.in 2009-09-09 16:25:45.161331795 +0200 @@ -75,14 +75,14 @@ all: $(LIBS) $(MANS) $(TSTS) libmm.la: $(OBJS) - $(LIBTOOL) --quiet --mode=link $(CC) -o $@ $(OBJS) \ + $(LIBTOOL) --quiet --mode=link $(CC) $(CFLAGS) -o $@ $(OBJS) \ -rpath $(libdir) -version-info `$(SHTOOL) version -l c -d libtool mm_vers.c` mm_alloc.c mm_core.c mm_global.c: mm.h mm_vers.c mm.h: config.h mm_test: mm_test.lo libmm.la - $(LIBTOOL) --quiet --mode=link $(CC) -o $@ mm_test.lo libmm.la + $(LIBTOOL) --quiet --mode=link $(CC) $(CFLAGS) -o $@ mm_test.lo libmm.la mm.3: mm.pod V1=`$(SHTOOL) version -l c -d short mm_vers.c`; \#f2dcdc 71 new new 2005 Nov anonymous mm 2005 Nov rse 1 1 Enhancement request - Make test results parsable using a parser. I was trying to automate the mm test cases. For this I need to parse the test results. Right now only one message is displayed at the end of 'make test' command. Since there are 5 different test cases available for mm, Instead of single message, if we can display either success or fail message after each test completion it would be very easy to parse the test results automatically. I have made changes in mm_test.c file to implement the above change. I am sending the changed file as a seperate email . Please consider incorporating this change to mm_test.c . Thanks & Regards,{linebreak} Mugunth{linebreak} spikesource.com #f2dcdc 65 code new 2005 Sep anonymous mm 2005 Sep rse 4 3 mm_display_info shows wrong memory limits In mm_display_info the following casting is needed to show correct memory limits: {linebreak} {linebreak} fprintf(stderr, " memory area.......(unsigned long)((unsigned long)mm+mm->mp_size)... {linebreak} instead of: {linebreak} fprintf(stderr, " memory area.......(unsigned long)(mm+mm->mp_size)... {linebreak} {linebreak} and: {linebreak} {linebreak} fprintf(stderr, " chunk.....(unsigned long)((unsigned long)mc+mc->mc_size... {linebreak} instead of: {linebreak} fprintf(stderr, " chunk.....(unsigned long)(mc+mc->mc_size).. #f2dcdc 59 code new 2005 Feb anonymous mm 2005 Feb rse 2 3 Problem Linking 64-Bit Version Hello, I've found an little bug in the Makefile. The Linker-Flags don't effect the libtool, because they are not used. here is a little patch: {linebreak}
--- Makefile.in 2002-12-19 10:19:38.000000000 +0100 +++ ../../PESlibmm/Makefile.in.linkerpatch 2005-02-15 13:20:26.615497000 +0100 @@ -74,14 +74,14 @@ all: $(LIBS) $(MANS) $(TSTS) libmm.la: $(OBJS) - $(LIBTOOL) --quiet --mode=link $(CC) -o $@ $(OBJS) \ + $(LIBTOOL) --quiet --mode=link $(CC) $(LDFLAGS) -o $@ $(OBJS) \ -rpath $(libdir) -version-info `$(SHTOOL) version -l c -d libtool mm_vers.c` mm_alloc.c mm_core.c mm_global.c: mm.h mm_vers.c mm.h: config.h mm_test: mm_test.lo libmm.la - $(LIBTOOL) --quiet --mode=link $(CC) -o $@ mm_test.lo libmm.la + $(LIBTOOL) --quiet --mode=link $(CC) $(LDFLAGS) -o $@ mm_test.lo libmm.la mm.3: mm.pod V1=`$(SHTOOL) version -l c -d short mm_vers.c`; \Greetings Ralf #f2dcdc 57 code new 2005 Jan anonymous mm 2005 Jan rse 1 1 Bad interpreter Error message when installing [root@BASLIN2 mm-1.3.1]# ./configure : bad interpreter: No such file or directory What can I do? _2006-Mar-03 15:15:42 by anonymous:_ {linebreak} you most probably had a CR at the end of the #! line #f2dcdc 46 code new 2004 Apr anonymous mm 2004 Apr rse 1 1 Cannot find peer certificate chain I'm using the following components: *: apache 1.3.29 *: mod_ssl 2.8.16-1.3.29 *: openssl 0.9.7d *: mm 1.3.0 *: Solaris 8 I've configured: SSLSessionCache shmcb:/opt/slt/ses/apache/run/ssl_scache(512000) The problem only occures if we use client certs. If we do multiple requests on the same ssl session then I get an error the *first time* the request is handled by the *same apache child* that has stored the SSL session ID in the cache. All other childs can acces the cache without problems. trace output in ssl_engine_log (debug level does not provide better info): [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Handshake: start{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Loop: before/accept initialization{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Loop: SSLv3 read client hello A{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Loop: SSLv3 write server hello A{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Loop: SSLv3 write change cipher spec A{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Loop: SSLv3 write finished A{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Loop: SSLv3 flush data{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Loop: SSLv3 read finished A{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Handshake: done{linebreak} [21/Apr/2004 09:48:18 01201] [info] Connection: Client IP: 192.168.167.99, Protocol: TLSv1, Cipher: RC4-MD5 (128/128 bits){linebreak} [21/Apr/2004 09:48:18 01201] [info] Initial (No.1) HTTPS request received for child 0 (server airlock_baumi.ergon.ch:4442){linebreak} [21/Apr/2004 09:48:18 01201] [trace] Changed client verification type will force quick renegotiation{linebreak} [21/Apr/2004 09:48:18 01201] [info] Requesting connection re-negotiation{linebreak} [21/Apr/2004 09:48:18 01201] [trace] Performing quick renegotiation: just re-verifying the peer{linebreak} [21/Apr/2004 09:48:18 01201] [error] Cannot find peer certificate chain{linebreak} [21/Apr/2004 09:48:18 01201] [trace] OpenSSL: Write: SSL negotiation finished successfully{linebreak} [21/Apr/2004 09:48:18 01201] [info] Connection to child 0 closed with standard shutdown (server airlock_baumi.ergon.ch:4442, client 192.168.167.99) I wonder about the "Cannot find peer certificate chain" and then the "SSL negotiation finished successfully". hmmm. If we use dbm instead of shmcb then this problem does not occure. Thanks for your help{linebreak} Erwin Huber #f2dcdc 45 code new 2004 Apr anonymous mm 2004 Apr rse 1 1 how do you install mm-1.3.0.tar.gz I am having problem under standing the install instructions. #f2dcdc 22 code new 2003 Oct anonymous mm 2003 Oct rse 1 3 Apache/PHP with MM session support does SegFault 11 under load / Excuse me for my English / I have self-compiled enviroment with PHP 4.3.3 with MM as static module for Apache 1.3.28. When I test this config locally system worked well. But there are some problem appeared in real web enviroment. PHP scripts with session support sometimes caused Segmentation Fault 11. There are about 12-24 hours pass from apache restart to segfault appeared at about 1 request per minute scripts that uses sessions. Setting Apache 'MaxRequestsPerChild' options does not help (is this good idea?). There are configure scripts: PHP: ./configure --with-config-file-path=/etc/httpd --with-mysql=/usr/local --with-apache=../apache_1.3.28 --with-zlib --disable-cli --disable-cgi --disable-ipv6 --with-openssl --with-mysql=/usr/local --with-mysql-sock=/var/lock/mysql.sock --with-mm=/usr/local --enable-sockets --disable-xml --with-mnogosearch=/usr/local Apache: SSL_BASE=/usr/src/openssl-0.9.7b EAPI_MM=/usr/local ./configure --with-layout=star --activate-module=src/modules/php4/libphp4.a --disable-module=actions --disable-module=cgi --disable-module=include --disable-module=userdir --enable-module=rewrite --enable-module=usertrack --enable-module=ssl --server-uid=www --server-gid=www --disable-rule=SSL_COMPAT There are debugger output (gdb /usr/local/sbin/httpd -c /usr/local/core): GNU gdb 5.3 Copyright 2002 Free Software Foundation, Inc. GDB is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Type "show copying" to see the conditions. There is absolutely no warranty for GDB. Type "show warranty" for details. This GDB was configured as "i386-slackware-linux"... Core was generated by `/usr/local/sbin/httpd -DSSL'. Program terminated with signal 11, Segmentation fault. Reading symbols from /lib/libcrypt.so.1...done. Loaded symbols for /lib/libcrypt.so.1 Reading symbols from /usr/local/lib/libmm.so.13...done. Loaded symbols for /usr/local/lib/libmm.so.13 Reading symbols from /usr/local/lib/mysql/libmysqlclient.so.12...done. Loaded symbols for /usr/local/lib/mysql/libmysqlclient.so.12 Reading symbols from /usr/lib/libz.so.1...done. Loaded symbols for /usr/lib/libz.so.1 Reading symbols from /lib/libresolv.so.2...done. Loaded symbols for /lib/libresolv.so.2 Reading symbols from /lib/libm.so.6...done. Loaded symbols for /lib/libm.so.6 Reading symbols from /lib/libdl.so.2...done. Loaded symbols for /lib/libdl.so.2 Reading symbols from /lib/libnsl.so.1...done. Loaded symbols for /lib/libnsl.so.1 Reading symbols from /lib/libdb.so.2...done. Loaded symbols for /lib/libdb.so.2 Reading symbols from /usr/lib/libexpat.so.0...done. Loaded symbols for /usr/lib/libexpat.so.0 Reading symbols from /lib/libc.so.6...done. Loaded symbols for /lib/libc.so.6 Reading symbols from /lib/ld-linux.so.2...done. Loaded symbols for /lib/ld-linux.so.2 Reading symbols from /lib/libnss_compat.so.2...done. Loaded symbols for /lib/libnss_compat.so.2 Reading symbols from /lib/libnss_files.so.2...done. Loaded symbols for /lib/libnss_files.so.2 #0 0x080ec415 in ps_sd_destroy (data=0x8536028, sd=0x423b4e7c) at /usr/src/php-4.3.3/ext/session/mod_mm.c:170 170 for (prev = data->hash[slot]; prev->next != sd; prev = prev->next); And there are server configuration: Slackware 9.0 (glibc 2.3.1) with recompiled kernel 2.4.22 on Dual PIII 1000/1GB RAM #f2dcdc 18 code new 2003 May anonymous mm 2003 May rse 2 1 Maximum segment size under linux... The maximum segment size under linux is a kernel parameter (/proc/sys/kernel/shmmax) which can be changed during runtime. The maximum segment size of mm is calculated at compile time and depends on the max seg size of the system, one compiled the library. It would be more appropriate for the library to calculate the max seg size during initialization of the mm_core by geting this variable from the /proc subsystem. if this fails then it could just use a "hardcoded" value calculated at configuration stage of the library (or even use bits of the code used in the configure script to calculate the maximum available size on the fly...). #f2dcdc 15 code new 2003 Apr anonymous mm 2003 Apr anonymous 1 2 Apache 1.3.27 with mod_ssl compile fails with mm ,but is OK without it Problem occurs using Apache 1.3.27, mod_ssl-2.8.14-1.3.27, openssl-0.9.7a and mm-1.3.0 on linux with gcc 2.96 (Mandrake 8.1, kernel 2.4.18). All the packages were downloaded as source tgz, and compiled as instructed in mod_ssl using --with-mm=../mm-1.3.0. Removing this reference to mm allows everything to compile normally. The error is in the comiplation of Apache and seems to be because of missing header files: ./gen_test_char: error while loading shared libraries: libmm.so.13: cannot open shared object file: No such file or directory make[3]: *** [test_char.h] Error 127 Running "make" a second time creates the missing header with 0 bytes content and things get a bit further, but then it crashes again a bit later. #f2dcdc 13 event new 2003 Mar anonymous mm 2003 Mar rse 1 1 Apache install Attempting to install apache for an ACID install. I installed Libmm.so.11 or so I thought I continue to get the Msg Dependency Lib not installed. I have RedHat 8.0 #f2dcdc 12 event new 2003 Mar anonymous mm 2003 Mar rse 1 1 Apache install Attempting to install apache for an ACID install. I installed Libmm.so.11 or so I thought I continue to get the Msg Dependency Lib not installed. #f2dcdc 7 event new 2002 Dec anonymous mm 2002 Dec rse 1 1 Segmentation fault at mod_mm.c:186 Hello, I have compiled apache the latest apache 1.3 with php 4.2.3 and lib mm. After several hours, I get a seg fault of apache processes which originate from mod_mm.c at line 186. It seems that php tries to retrieve a correct session id from libmm, but for some reason, it crashes in the for loop in mod_mm.c. Here is a gdb trace. Program received signal SIGSEGV, Segmentation fault. 0x40356f4c in ps_sd_lookup (data=0x8082690, key=0x816da24 "0922ecff590e004d7a75919ac38dde0a", rw=0) at mod_mm.c:186 186 for (prev = NULL, ret = data->hash[slot]; ret; prev = ret, ret = ret->next) (gdb) bt #0 0x40356f4c in ps_sd_lookup (data=0x8082690, key=0x816da24 "0922ecff590e004d7a75919ac38dde0a", rw=0) at mod_mm.c:186 #1 0x403572c8 in ps_read_mm (mod_data=0x403f8e50, key=0x816da24 "0922ecff590e004d7a75919ac38dde0a", val=0xbfffcc94, vallen=0xbfffcc98) at mod_mm.c:326 #2 0x40353f5b in php_session_initialize () at session.c:557 Best, Bastien Duclaux #f2dcdc 4 event new 2002 Oct anonymous mm 2002 Oct rse 1 2 mm 1.2.1 core dumps on Sparc/Solaris 8 Hi Ralf, I'm trying to compile mm1.2.1 on Sparc/Solaris 8. I installed gnu binutils 2.13, then bootstrapped gcc 3.2 using the --with-gnu-ld & --with-gnu-as args. Now on to mm: ./config appears to work fine. I then do 'make' which also appears to work . I run make test and immediately get a core dump. I *have* successfully built mm before with a binary installed version of gcc, but using the sun (/usr/css/bin/) vesions of ld & as. Any thoughts on what may be wrong? My little ultra 1 compiled all night on gcc, but now I'm afraid it's all for naught if gcc is corrupt. Thanks for any help. Regards, D J Brooks #f2dcdc 197 code new 2014 Oct anonymous lmtp2nntp 2014 Oct thl 1 1 550 5.1.1 Recipient did not transform into group. I have a problem to bring this gateway to a new server. we have used this Software for many years on an old SuSe System. without problems. after porting this Gateway to the actual Opensuse this program should not working as expected. Both Systems using an MTA with Postfix. So the Konfiguration of the Gateway could leave as is. I dont understand what is going wrong. when I will send a Mail to the Gateway in the same manner as the old system I keep following Message:
--- lmtp2nntp_main.c 2004-05-07 16:52:03.000000000 +0200 +++ lmtp2nntp_main.c 2007-02-07 20:05:00.000000000 +0100 @@ -1315,7 +1315,7 @@ * RFC1893 3.5 Network and Routing Status X.5.2 Syntax error */ logbook(ctx->l2, L2_LEVEL_TRACE, "checking parameter syntax"); - if (str_parse(req->msg, "m/^RCPT To:\\s*(.+)$/i", &cp) <= 0) { + if (str_parse(req->msg, "m/^RCPT To:\\s*(\\S+)/i", &cp) <= 0) { res.statuscode = "501"; res.dsncode = "5.5.2"; res.statusmsg = "Syntax error in parameters.";=For me, lmtp2nntp covers much more than I need but is still easy to configure - thank you. #f2dcdc 70 code new 2005 Oct anonymous l2 2005 Oct anonymous 2 1 "incorrect" usage of va_list objects causes segfault on amd64 va_list objects has different internals on i386 and amd64. See the attachment. #f2dcdc 24 code new 2003 Oct anonymous l2 2003 Oct rse 2 3 l2 is not thread safe when used along with fsl l2 appears to be thread safe if used carefully by itself. However, when l2 is used with fsl, thread safety problems are evident when log lines start overwriting each other once the thread level becomes dense. The problem arises in the l2_channel_vlog call, which takes as an argument the l2 channel that fsl creates. From this channel, the function extracts the l2_env_t struct, which has a buffer that is used by a subsequent call to l2_util_format to store the formatted log message. The problem is that this l2_env_t struct is part of the l2 channel that fsl passed, which is global. The log line overwriting can occur anywhere after this buffer in l2_env_t is used becacuse any thread can write to it during the l2_util_format call while other threads read from it in downstream calls. The temporary solution that serves our purpose was to store the message in a local function variable before passing it to l2_util_format. The patch file used is attached. #f2dcdc 23 code new 2003 Oct anonymous l2 2003 Oct rse 3 4 Prefix channel buffer bug when '\n' character present When using the prefix channel along with fsl, any two subsequent calls to syslog in which the first one has length > n and the second one has lenght < n with the second one ending in a \n character produces three log lines: the first two are the expected log lines and the third one is the (lenght - n) characters of the first syslog call. It looks like some sort of buffer cleaning problem. It was not tested wheter or not this happens when prefix is used without fsl. To reproduce, a single program like this suffices: int main(void){ openlog("test", LOG_PID,LOG_LOCAL7); syslog(LOG_INFO, "000000000000000000000000000000000000000000000000000000000000"); syslog(LOG_INFO, "11111111111111111111111111111111111111111111111111111\n"); } which produces the output: (timestamp added by prefix channel) [2003-10-31, 11:40:22] 000000000000000000000000000000000000000000000000000000000000 {linebreak}[2003-10-31, 11:40:22] 11111111111111111111111111111111111111111111111111111 {linebreak}00000 Note: only the prefix channel and the file channel were specified int the config file for this test. The error does not appear when only the file channel is present. #f2dcdc 159 code new 2007 Sep anonymous js 2007 Sep rse 3 1 js portability issues for Windows I'm embedding OSSP js in a project and I found the following problems: 1. There's a bug in the function js_FreeRuntimeScriptState at jsscript.c:994. If you create a runtime and then destroy it before creating any contexts, the statement: JS_REMOVE_LINK(&sfp->links); causes a segfault because sfp is NULL. I didn't have the time or inclination to investigate this, but adding: if(! sfp) break; before that statement eliminated the crash. According to the SpiderMonkey team: "I don't know what version OSSP is using, but this sounds like bug 345350 (https://bugzilla.mozilla.org/show_bug.cgi?id=345350). Your patch just detects this in the loop, but that adds an if test to every iteration that runs when the table has not been finished. The patch in bug 345350 tests before the loop." 2. I had to build a Visual Studio solution for js since I need to deploy on Windows as well as UNIX (and mingw/cygwin were not an option). I ran into the following portability problems: a. prjmtime.c fails to compile because the two #include statements at line 61 are not sufficient; I got lots of errors about LPCONTEXT and such being undefined. I replaced these with a single "#include
we are experiencing random segfaults in our multithreaded application. i traced this back to the fact that JS is not compiled against NSPR. since we use glib/gthread for threading, JS native (asm) locks segfault too and only NSPR locks work correctly. the patch below implements a configure option to compile against a (std) installed "mozilla-devel/mozilla-nspr-devel" via mozilla-config or a custom install location. it also propagates the required changes to js.pc and js-config. --- js-1.6.20060820/configure.ac 2006-08-20 17:45:13.000000000 +0200 +++ js-1.6.20060820.nspr/configure.ac 2006-08-29 16:40:27.000000000 +0200 @@ -26,7 +26,7 @@ dnl ## configure.ac: GNU Autoconf source script dnl ## -AC_PREREQ(2.60) +AC_PREREQ(2.59) AC_INIT JS_VERSION=`./shtool version -l txt -d short VERSION` $ac_confdir/shtool echo -e \ @@ -55,6 +55,9 @@ *mingw* | *windows* | *winnt* ) CPPFLAGS="$CPPFLAGS -DXP_WIN" ;; * ) CPPFLAGS="$CPPFLAGS -DXP_UNIX" ;; esac + +CFG_CFLAGS="$CPPFLAGS" + CPPFLAGS="$CPPFLAGS -DEXPORT_JS_API" dnl # configure option --with-version @@ -160,11 +163,36 @@ CPPFLAGS="$CPPFLAGS -DJS_C_STRINGS_ARE_UTF8" fi +dnl # configure option --with-nspr[=path] +AC_ARG_WITH([nspr], + AC_HELP_STRING([--with-nspr], [build threadsafe library with nspr]), + [ac_cv_with_nspr=$withval], [ac_cv_with_nspr=no]) +AC_CACHE_CHECK([whether to build threadsafe with nspr], + [ac_cv_with_nspr], [ac_cv_with_nspr=no]) +if test ".$ac_cv_with_nspr" != ".no"; then + case "$ac_cv_with_nspr" in + /* ) + CPPFLAGS="$CPPFLAGS -DJS_THREADSAFE -DJS_USE_ONLY_NSPR_LOCKS -I$ac_cv_with_nspr/include" + LDFLAGS="$LDFLAGS -L$ac_cv_with_nspr/lib " + LIBS="$LIBS -lplc4 -lplds4 -lnspr4" + ;; + * ) + AC_PATH_PROG(MOZILLA_CONFIG, mozilla-config, no) + if test "$MOZILLA_CONFIG" != "no" ; then + CPPFLAGS="$CPPFLAGS -DJS_THREADSAFE -DJS_USE_ONLY_NSPR_LOCKS `$MOZILLA_CONFIG --cflags nspr`" + LIBS="$LIBS `$MOZILLA_CONFIG --libs nspr`" + fi + ;; + esac + CFG_CFLAGS="$CFG_CFLAGS -DJS_THREADSAFE" +fi + AC_SUBST(CLI_CFLAGS) AC_SUBST(CLI_CPPFLAGS) AC_SUBST(CLI_LDFLAGS) AC_SUBST(CLI_LIBS) AC_SUBST(CLI_OBJS) +AC_SUBST(CFG_CFLAGS) with_tags="" sinclude(libtool.m4) --- js-1.6.20060820/js.pc.in 2006-07-24 21:53:01.000000000 +0200 +++ js-1.6.20060820.nspr/js.pc.in 2006-08-29 16:41:02.000000000 +0200 @@ -7,4 +7,4 @@ Description: Mozilla JavaScript Engine Version: @JS_VERSION@ Libs: -L${libdir} -ljs @LIBS@ -Cflags: -DOSSP -DXP_UNIX -I${includedir}/js +Cflags: @CFG_CFLAGS@ -I${includedir}/js --- js-1.6.20060820/js-config.sh.in 2006-08-10 21:55:51.000000000 +0200 +++ js-1.6.20060820.nspr/js-config.sh.in 2006-08-29 16:41:44.000000000 +0200 @@ -109,7 +109,7 @@ output="$output $js_acdir" ;; --cppflags) - output="$output -DOSSP -DXP_UNIX -I$js_includedir/js" + output="$output @CFG_CFLAGS@ -I$js_includedir/js" ;; --cflags) : # none#f2dcdc 170 code new 2009 Feb anonymous ex 2009 Feb rse 1 1 [OSX] make install fails to install package with abstract message system: iMactel OSX10.5.6, xcode 3.0, gcc 4.0.1 When installing the package, a message is generated and no files are installed. The message is "make: `install' is up to date." A. #f2dcdc 157 code new 2007 Jun anonymous ex 2007 Aug rse 2 1 pthread library from ex(3) manpage causes segmentation fault The wrapper of type pthread_create_ex_t which saves the thread-function's address and its argument's address is allocated locally (on the stack) in pthread_create_ex() so in pthread_create_wrapper() this memory is not valid anymore. The applied patch fixes this. Greetings, Frank
*** pthread_ex.c.manPage 2007-06-15 14:56:55.000000000 +0200 --- pthread_ex.c 2007-06-15 14:57:45.000000000 +0200 *************** *** 1,5 **** --- 1,6 ---- #include#f2dcdc 25 code new 2003 Nov anonymous ex 2003 Nov rse 1 1 `make check' segmentation fault Linux spspare 2.4.22 #2 SMP Mon Sep 29 18:13:12 CST 2003 i686 GNU/Linux Debian GNU/Linux testing after ./configure and make make-check prints # make check Test Suite: OSSP ex (Exception Handling) __________________________________________________________________ Test: basic nested control flow ............................... OK Test: exception value passing ................................. OK Test: variable value preservation ............................. OK Test: exception deferring ..................................... OK Test: exception shielding ..................................... OK Test: cleanup handling ........................................ OK __________________________________________________________________ Test Summary: 6 tests (6 ok, 0 failed), 6 checks (6 ok, 0 failed) Test Suite: OK make: *** [check] Segmentation fault (core dumped) #f2dcdc 192 code new 2012 Nov anonymous cfg 2012 Nov rse 1 3 cfg_export dumps core on FreeBSD, Darwin When attempting to evaluate libcfg for a project I'm starting, I ran the test on my FreeBSD 9 and OS X which dumped core. It looks like there's some incorrect usage of =va_list= variables given that the =va_arg()= call that initializes =s= on line 908 of =cfg_fmt.c= returns invalid memory. I put a =va_copy()= at the top of =cfg_fmt_format()=, making a copy of =ap=, and changed =cfg_fmt_format()= to operate on that copy. The test then passes. I'm not sure if this is the right solution, but at least the code runs and passes its test. _2012-Nov-03 02:12:26 by anonymous:_ {linebreak} I should mention that the test runs and passes on Linux and and Solaris. #f2dcdc 28 code new 2004 Jan anonymous cfg 2004 Jan rse 1 1 SIGSEGV received while parsing config file When using fsl, logging configuration is read from a file with a well-known format using libcfg. If this config file has a block (see used config file..block starts at "ident (.+)/local7 q{" and ends in "};") that is longer than 1035, a segfault is caused. Attached you can find the file I used as a config file and a very simple test program, as well as the modified section of code from libcfg that helped me pinpooint the source. If you run this program with this config (compiilng with libfsl of course) you should get the same results. I traced the problem with gdb and managed to pin-point it to line 1233 in cfg_syn_scan.c (same as line 255 of cfg_syn_scan.l). This line is: yylval->cpString = strdup(caStr); I inserted some print statments to test and discovered that the cause is that yylval in this case is a NULL pointer or an invalid address.Whenever I delete a single character from my config file (in any config line of the section) the segfault goes away. That's how I deduced the limit to be 1035 chars in the section before the segfault (ie. 1036 causes a segfault). Strangely enough, if I add a couple of more chars.. my debug line doesn't even show! Any prompt help regarding a workaround or a fix would be most appreciated. #f2dcdc 16 code new 2003 Apr anonymous cfg 2003 Apr rse 1 1 parse error in cfg_data.c - GNU/Linux Compilation on a Debian GNU/Linux : gcc version 2.95.4 20011002 (Debian prerelease) gcc -DHAVE_CONFIG_H -O2 -pipe -c cfg_data.c -fPIC -DPIC -o .libs/cfg_data.lo cfg_data.c: In function `cfg_data_set': cfg_data.c:189: parse error before `*' cfg_data.c:189: parse error before `)' make: *** [cfg_data.lo] Error 1 [I do not have OSSP ex installed on this host, should I ? the autoconf has given no error/warning] thanks #f2dcdc 8 code new 2002 Dec anonymous cfg 2002 Dec rse 1 1 Build failed Under my FreeBSD 4.7-STABLE (both at home and at work) build process failed:{linebreak} gcc -DHAVE_CONFIG_H -O2 -pipe -c cfg_data.c -fPIC -DPIC -o .libs/cfg_data.lo{linebreak} cfg_data.c: In function `cfg_data_set':{linebreak} cfg_data.c:189: syntax error before `*'{linebreak} gmake: *** [cfg_data.lo] Error 1{linebreak} {linebreak} *:FreeBSD 4.7-STABLE i386 *: gcc version 2.95.4 20020320 [FreeBSD]{linebreak} *:GNU Make 3.80 #f2dcdc 55 code new 2004 Nov anonymous al 2004 Nov mlelstv 1 3 SIGSEGV at end of make check#include + #include #define PTHREAD_EX_INTERNAL #include "pthread_ex.h" *************** static void *pthread_create_wrapper(void *** 62,69 **** EX_CTX_INITIALIZE(ex_ctx); pthread_setspecific(pthread_ex_ctx_key, ex_ctx); ! /* perform original operation */ ! return wrapper->entry(wrapper->arg); } /* pthread_create() wrapper */ --- 63,73 ---- EX_CTX_INITIALIZE(ex_ctx); pthread_setspecific(pthread_ex_ctx_key, ex_ctx); ! /* perform original operation, but free wrapper first */ ! void *(*wentry)(void *) = wrapper->entry; ! void *warg = wrapper->arg; ! free(wrapper); ! return wentry(warg); } /* pthread_create() wrapper */ *************** int pthread_create_ex(pthread_t *thread, *** 71,81 **** const pthread_attr_t *attr, void *(*entry)(void *), void *arg) { ! pthread_create_ex_t wrapper; /* spawn thread but execute start function through wrapper */ ! wrapper.entry = entry; ! wrapper.arg = arg; ! return pthread_create(thread, attr, pthread_create_wrapper, &wrapper); } --- 75,86 ---- const pthread_attr_t *attr, void *(*entry)(void *), void *arg) { ! pthread_create_ex_t* wrapper = (pthread_create_ex_t*)malloc(sizeof(pthread_create_ex_t)); ! if (wrapper == NULL) return ENOMEM; /* spawn thread but execute start function through wrapper */ ! wrapper->entry = entry; ! wrapper->arg = arg; ! return pthread_create(thread, attr, pthread_create_wrapper, wrapper); }
$ uname -a Linux bongo 2.4.27 #1 Fri Oct 29 18:20:52 CEST 2004 i686 AMD_Athlon(tm)_XP_1700+ unknown PLD Linux $ rpm -q gcc gcc-3.3.5-1 (gdb) run Starting program: /home/users/radek/rpm/BUILD/al-0.9.1/al_test Test Suite: OSSP al (Assembly Line) __________________________________________________________________ Test: assembly line data copying .............................. OK Test: assembly line splicing .................................. OK Test: assembly line labelling ................................. OK Test: assembly line buffer attach ............................. OK __________________________________________________________________ Test Summary: 4 tests (4 ok, 0 failed), 531 checks (531 ok, 0 failed) Test Suite: OK Program received signal SIGSEGV, Segmentation fault. 0x4f9ee587 in free () from /lib/libc.so.6 (gdb) bt #0 0x4f9ee587 in free () from /lib/libc.so.6 #1 0x0804afad in ts_suite_free (ts=0x804e8f8) at ts.c:379 #2 0x0804a72f in main (argc=1, argv=0xb41a2984) at al_test.c:412#f2dcdc 43 code new 2004 Apr anonymous al 2004 Apr rse 3 3 al does not compile on powerpc and amd64 In ts.c line 205 va_list pointers are assigned to each other. This does not work on powerpc and amd64. One should use va_copy from C99 for this. #f2dcdc 194 code new 2013 Jul anonymous 2013 Jul rse 1 1 Errors while compiling: libuuid.a could not read symbols: Bad value. Compiling on x86_64 Arch Linux results in the following error: /usr/bin/ld: src/perlDeps/uuid-1.6.2/perl/../build/.libs/libuuid.a(uuid.o): relocation R_X86_64_32 against `.rodata.str1.8' can not be used when making a shared object; recompile with -fPIC src/perlDeps/uuid-1.6.2/perl/../build/.libs/libuuid.a: could not read symbols: Bad value collect2: error: ld returned 1 exit status make[1]: *** [blib/arch/auto/OSSP/uuid/uuid.so] Error 1 make[1]: Leaving directory `src/perlDeps/uuid-1.6.2/perl' make: *** [all] Error 2 #f2dcdc 178 code new 2009 Oct anonymous 2009 Oct rse 2 1 Problem compiling uuid-1.6.2 # make make[1]: Entering directory `/tmp/uuid-1.6.2/php' ./configure: line 1623: PHP_INIT_BUILD_SYSTEM: command not found ./configure: line 1629: syntax error near unexpected token `config.nice' ./configure: line 1629: `PHP_CONFIG_NICE(config.nice)' make[1]: *** [config] Error 2 make[1]: Leaving directory `/tmp/uuid-1.6.2/php' make: *** [php/modules/uuid.so] Error 2