If i try to emerge valgrind, I get the following error: gcc -DHAVE_CONFIG_H -I. -I. -I.. -I./demangle -I../include -DVG_LIBDIR="\"/usr/ lib"\" -Winline -Wall -Wshadow -O -fomit-frame-pointer -mpreferred-stack-boundary=2 -g -mpreferred-stack-boundary=2 -c `test -f 'vg_scheduler.c' || echo './'`vg_scheduler.c vg_scheduler.c: In function `release_one_thread_waiting_on_mutex': vg_scheduler.c:2340: error: union has no member named `__m_owner' vg_scheduler.c:2345: error: union has no member named `__m_count' vg_scheduler.c:2346: error: union has no member named `__m_owner' vg_scheduler.c:2352: error: union has no member named `__m_owner' vg_scheduler.c:2352: error: `_pthread_descr' undeclared (first use in this function) vg_scheduler.c:2352: error: (Each undeclared identifier is reported only once vg_scheduler.c:2352: error: for each function it appears in.) vg_scheduler.c:2352: error: syntax error before "i" vg_scheduler.c: In function `do_pthread_mutex_lock': vg_scheduler.c:2396: error: union has no member named `__m_kind' vg_scheduler.c:2406: error: union has no member named `__m_count' vg_scheduler.c:2415: error: union has no member named `__m_count' vg_scheduler.c:2417: error: union has no member named `__m_owner' vg_scheduler.c:2420: error: union has no member named `__m_owner' vg_scheduler.c:2422: error: union has no member named `__m_kind' vg_scheduler.c:2424: error: union has no member named `__m_count' vg_scheduler.c:2428: error: union has no member named `__m_count' vg_scheduler.c:2461: error: union has no member named `__m_owner' vg_scheduler.c:2466: error: union has no member named `__m_count' vg_scheduler.c:2467: error: union has no member named `__m_owner' vg_scheduler.c:2467: error: `_pthread_descr' undeclared (first use in this function) vg_scheduler.c:2467: error: syntax error before "tid" vg_scheduler.c: In function `do_pthread_mutex_unlock': vg_scheduler.c:2501: error: union has no member named `__m_kind' vg_scheduler.c:2502: error: union has no member named `__m_kind' vg_scheduler.c:2503: error: union has no member named `__m_owner' vg_scheduler.c:2504: error: union has no member named `__m_owner' vg_scheduler.c:2508: error: union has no member named `__m_kind' vg_scheduler.c:2518: error: union has no member named `__m_count' vg_scheduler.c:2528: error: union has no member named `__m_count' vg_scheduler.c:2536: error: union has no member named `__m_owner' vg_scheduler.c:2546: error: union has no member named `__m_count' vg_scheduler.c:2547: error: union has no member named `__m_kind' vg_scheduler.c:2548: error: union has no member named `__m_count' vg_scheduler.c:2555: error: union has no member named `__m_count' vg_scheduler.c:2556: error: union has no member named `__m_owner' vg_scheduler.c: In function `do_pthread_cond_timedwait_TIMEOUT': vg_scheduler.c:2610: error: union has no member named `__m_owner' vg_scheduler.c:2612: error: union has no member named `__m_count' vg_scheduler.c:2617: error: union has no member named `__m_owner' vg_scheduler.c:2617: error: `_pthread_descr' undeclared (first use in this function) vg_scheduler.c:2617: error: syntax error before "tid" vg_scheduler.c:2618: error: union has no member named `__m_count' vg_scheduler.c:2630: error: union has no member named `__m_count' vg_scheduler.c: In function `release_N_threads_waiting_on_cond': vg_scheduler.c:2681: error: union has no member named `__m_owner' vg_scheduler.c:2683: error: union has no member named `__m_count' vg_scheduler.c:2687: error: union has no member named `__m_owner' vg_scheduler.c:2687: error: `_pthread_descr' undeclared (first use in this function) vg_scheduler.c:2687: error: syntax error before "i" vg_scheduler.c:2688: error: union has no member named `__m_count' vg_scheduler.c:2701: error: union has no member named `__m_count' vg_scheduler.c: In function `do_pthread_cond_wait': vg_scheduler.c:2750: error: union has no member named `__m_kind' vg_scheduler.c:2760: error: union has no member named `__m_count' vg_scheduler.c:2770: error: union has no member named `__m_count' vg_scheduler.c:2771: error: union has no member named `__m_owner' vg_scheduler.c: In function `scheduler_sanity': vg_scheduler.c:3570: error: union has no member named `__m_count' vg_scheduler.c:3571: error: union has no member named `__m_owner' vg_scheduler.c:3572: error: union has no member named `__m_owner' make[3]: *** [vg_scheduler.o] Error 1 make[3]: Leaving directory `/var/tmp/portage/valgrind-2.0_pre20031012/work/ valgrind-20031012/coregrind' make[2]: *** [all-recursive] Error 1 make[2]: Leaving directory `/var/tmp/portage/valgrind-2.0_pre20031012/work/ valgrind-20031012/coregrind' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/var/tmp/portage/valgrind-2.0_pre20031012/work/ valgrind-20031012' make: *** [all] Error 2 !!! ERROR: dev-util/valgrind-2.0_pre20031012 failed. !!! Function src_compile, Line 37, Exitcode 2 !!! (no error message) Reproducible: Always Steps to Reproduce: 1. emerge valgrind Portage 2.0.49-r15 (default-x86-1.4, gcc-3.3.2, glibc-2.3.2-r8, 2.6.0-test9-mm2) ================================================================= System uname: 2.6.0-test9-mm2 i686 AMD Athlon(tm) XP 1600+ Gentoo Base System version 1.4.3.11 distcc 2.11.2 i686-pc-linux-gnu (protocols 1 and 2) (default port 3632) [disabled] ccache version 2.3 [enabled] ACCEPT_KEYWORDS="x86 ~x86" AUTOCLEAN="yes" CFLAGS="-march=athlon-xp -O3 -ffast-math -funroll-loops -pipe -fomit-frame-pointer -msse -mfpmath=sse,387 -falign-functions=16 -falign-labels=1 -falign-loops=16 -falign-jumps=16 -ftracer" CHOST="i686-pc-linux-gnu" COMPILER="gcc3" CONFIG_PROTECT="/etc /var/qmail/control /usr/kde/2/share/config /usr/kde/3/ share/config /usr/X11R6/lib/X11/xkb /usr/kde/3.2/share/config /usr/share/texmf/ tex/generic/config/ /usr/share/texmf/tex/platex/config/ /usr/share/config" CONFIG_PROTECT_MASK="/etc/gconf /etc/env.d" CXXFLAGS="-march=athlon-xp -O3 -ffast-math -funroll-loops -pipe -fomit-frame-pointer -msse -mfpmath=sse,387 -falign-functions=16 -falign-labels=1 -falign-loops=16 -falign-jumps=16 -ftracer" DISTDIR="/usr/portage-distfiles" FEATURES="autoaddcvs sandbox ccache" GENTOO_MIRRORS="http://mirrors.sec.informatik.tu-darmstadt.de/gentoo http:// linux.rz.ruhr-uni-bochum.de/download/gentoo-mirror http://www.ibiblio.org/pub/ Linux/distributions/gentoo http://gentoo.oregonstate.edu" MAKEOPTS="-j2" PKGDIR="/usr/portage-packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="x86 oss avi gif jpeg libg++ mad mikmod png spell truetype xmms xv zlib gdbm berkdb slang readline tetex svga sdl tcpd pam libwww ssl perl python esd imlib gtk qt motif opengl ldap X gtk2 3dnow -apm kde cdr dga fbcon mozilla mpeg ncurses pdflib quicktime samba sse acpi aalib crypt ipv6 -nls -gnome alsa gpm nptl oggvorbis xml2 -java arts encode -debug cups usb foomaticdb prelink snmp mysql mmx imap"
please try out the 2.0.0 release and see if it fixes your problem
No, the bug is not fixed.
I saw the same bug has already been submitted (by a Gentooer, so I'm guessing you submitted it upstream). I suppose we'll just have wait for a response from them.
the same output is recived with 2.0.0, granted I had to change the ebuild because the 2.6 config patch is not needed for a 2.6 kernel anymore. Portage 2.0.49-r15 (default-x86-1.4, gcc-3.3.2, glibc-2.3.2-r8, 2.6.0-test9-mm2) ================================================================= System uname: 2.6.0-test9-mm2 i686 AMD Athlon(tm) MP 1900+ Gentoo Base System version 1.4.3.12 ACCEPT_KEYWORDS="x86 ~x86" AUTOCLEAN="yes" CFLAGS="-march=athlon-xp -O3 -fomit-frame-pointer -pipe -falign-functions=64 -mfpmath=sse -m3dnow -msse -mmmx -ffast-math" CHOST="i686-pc-linux-gnu" COMPILER="gcc3" CONFIG_PROTECT="/etc /var/qmail/control /usr/kde/2/share/config /usr/kde/3/share/config /usr/X11R6/lib/X11/xkb /usr/kde/3.2/share/config /usr/kde/3.1/share/config /usr/share/texmf/tex/generic/config/ /usr/share/texmf/tex/platex/config/ /usr/share/texmf/dvips/config/ /usr/share/texmf/dvipdfm/config/ /usr/share/texmf/xdvi/ /usr/share/config" CONFIG_PROTECT_MASK="/etc/gconf /etc/env.d" CXXFLAGS="-march=athlon-xp -O3 -fomit-frame-pointer -pipe -falign-functions=64 -mfpmath=sse -m3dnow -msse -mmmx -ffast-math" DISTDIR="/usr/portage/distfiles" FEATURES="autoaddcvs -sandbox ccache buildpkg" GENTOO_MIRRORS="http://gentoo.oregonstate.edu http://distro.ibiblio.org/pub/Linux/distributions/gentoo" MAKEOPTS="-j3" PKGDIR="/usr/portage/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage" SYNC="rsync://rsync.gentoo.org/gentoo-portage"
Caleb, in comment #3, you said that you saw the same bug submitted upstream. I looked at bugs.kde.org for all valgrind bugs. I didn't see this bug there. I saw http://bugs.kde.org/show_bug.cgi?id=68360 , but that isn't exactly what's happening here. Am I looking in the wrong place?
Upstream released 2.1.0, the NEWS-entry "68360: Valgrind does not compile against 2.6.0-testX kernels" looked quite hopeful, but I see the exact same error with these sources.
if we run this sed script we get a lot less errors: sed -i \ -e 's:__m_owner:__data.__owner:g' \ -e 's:__m_kind:__data.__kind:g' \ -e 's:__m_count:__data.__count:g' \ coregrind/vg_scheduler.c \ coregrind/vg_libpthread.c \ include/vg_skin.h
then it seems that a _pthread_descr no longer exists so if we look at bits/pthreadtypes.h we now find that owner is of type int and not _pthread_descr ... so if we put '#define _pthread_descr int' inside of vg_include.h it makes more things happy and lets vg_scheduler.c compile next up is vg_libpthread.c ... at this point we see that it gets pretty pissed at not knowing anything about the innerards of pthread_attr_t ... and looking at bits/pthreadtypes.h, i guess you're not supposed to :) anyways, it seems the interface has changed from using members of this struct to function calls ... pthread.h has gained a whole bunch of get/set functions: old way: myg = (pthread_attr_t*)attr->__guardsize new way: pthread_attr_setguardsize(attr, &myg) so on and so forth for
err i'm an idiot ;) the entire point of vg_libpthread.c is provide a wrapper around all the pthread functions which means knowing the inner structure of the pthread stuff ... so scratch that last comment about vg_libpthread.c heh
you wrote: > pthread_attr_setguardsize(attr, &myg) pthread_attr_GETguardsize ??
There is a bug report for this here http://bugs.kde.org/show_bug.cgi?id=69616 The patch applies against CVS and compiles and runs properly with NPTL. A few chunks fail against 2.1.0 though.
Created attachment 25410 [details, diff] Diff to make valgrind-2.1.0 compile against nptl version of pthreadtypes.h This diff is based on the work by Nicholas Nethercote. See http://bugs.kde.org/show_bug.cgi?id=69616 for more info.
Should that be applied across the board, or only for nptl folks?
Compiles, but when run, gives: $ valgrind ./hello grep: error while loading shared libraries: libc.so.6: cannot open shared object file: No such file or directory /usr/bin/valgrind: './hello' not found in $PATH, aborting. (where hello is a "hello world" program) Should I open another bug report for this?
I had manualy patched 2.1.0 as well, and ran into the libc.so.6 problem... cvs on the other hand seemes to be no problem. Caleb.. this should be safe for non nptl people as well because it is simply re impliments parts of the old pthreadtyps within valgrind. thus taking care of the changes made in the new version of the file without relying on the new version of the file. err yea
btw... I tried this patch against the 2.1.0 archive, and i run into the libc.6.so error again.
Created attachment 25450 [details, diff] 1 of 3 part patch first of 2 patches to bring 2.1.0 up to cvs
Created attachment 25451 [details, diff] part 2 of 3 this gets us to cvs as of 2/11/2004 (still runs as 2.1)
Created attachment 25452 [details, diff] part 3 of 3 this patches to work with nptl. I have tested and it all seems to work for me Thanks again
btw the new version of valgrind will requiere you to specify the tool to use, as they have removed valgrinds defaulting. ie valgrind ./hello will out put the same as just valgrind so you will need to do something like valgrind --tool=<tool> ./hello <tool> can be helgrind, addrcheck, cachegrind, memcheck, corecheck, lackey, none. of course you can setup a config file that will use a default action
i can verify that the 3 part patch set worked for me ... now if only valgrind would stop lying to me and telling me my program leaks memory
*** Bug 43305 has been marked as a duplicate of this bug. ***
BTW this issue is already fixed in valgrind's CVS, which means this issue will be nonexistent with the next release. Here is the fix's mention in the CVS log: date: 2004/02/28 15:40:36; author: nethercote; state: Exp; lines: +80 -108 Some newer systems have a new version of pthreadtypes.h that comes from glibc 2.3.3, which is binary-compatible, but not source compatible with the old one, which came from LinuxThreads. We were using the types defined in the old one, which caused compilation errors on systems using the new one. This commit introduces our own versions of these types. Our versions are laid out identically to the LinuxThreads ones, but the field names are different. We convert all pthread types to our versions before using them, so we don't rely on the pthreadtypes.h types any more. Hopefully this will fix the problem; I have three reports that it does. Let's see...
Oops... Next time I should read _all_ comments in a bug report. ;-)
Valgrind 2.1.1 compiles with the new NPTL headers. Please include 2.1.1 in the portage tree. You should also include calltree 0.9.7 that depends on >=valgrind-2.0.0 I had to put both of them in my overlay tree because digikam depends on them.
2.1.1 is now in portage
*** Bug 56622 has been marked as a duplicate of this bug. ***
