From: "Dmitry V. Levin" <ldv@altlinux.org> To: ALT Devel discussion list <devel@lists.altlinux.org> Subject: Re: [devel] I: illegal RPATH: relative paths Date: Mon, 12 Dec 2011 18:12:01 +0400 Message-ID: <20111212141201.GC24413@altlinux.org> (raw) In-Reply-To: <20111212134733.GB24413@altlinux.org> [-- Attachment #1: Type: text/plain, Size: 4301 bytes --] Hi, On Mon, Dec 12, 2011 at 05:47:33PM +0400, Dmitry V. Levin wrote: > Анализ логов тестовой пересборки показывает, что в Sisyphus-x86_64 очень много > пакетов с ошибочными и/или некорректными RPATH. [...] > - относительные пути вместо абсолютных; Относительные пути вместо абсолютных в RPATH - это беда. Вот как это выглядит на простом примере. Обычная простая программа без RPATH: $ echo 'int main(void){return 0;}' |gcc -xc - -O2 -Wall -o a.out && strace -eopen ./a.out open("/etc/ld.so.cache", O_RDONLY) = 3 open("/lib64/libc.so.6", O_RDONLY) = 3 +++ exited with 0 +++ Та же самая программа, но с RPATH, содержащим относительный путь: $ echo 'int main(void){return 0;}' |gcc -xc - -O2 -Wall -o a.out -Wl,-rpath,foo && strace -eopen ./a.out open("foo/tls/x86_64/libc.so.6", O_RDONLY) = -1 ENOENT (No such file or directory) open("foo/tls/libc.so.6", O_RDONLY) = -1 ENOENT (No such file or directory) open("foo/x86_64/libc.so.6", O_RDONLY) = -1 ENOENT (No such file or directory) open("foo/libc.so.6", O_RDONLY) = -1 ENOENT (No such file or directory) open("/etc/ld.so.cache", O_RDONLY) = 3 open("/lib64/libc.so.6", O_RDONLY) = 3 +++ exited with 0 +++ Как видно, относительный путь в RPATH приводит к тому, что ld.so во время запуска программы ищет все слинкованные библиотеки относительно текущего каталога, т.е. в совершенно непредсказуемых местах. Как правило, относительные пути в RPATH - это просто ошибки сборки. Вот, например, самая распространенная из таких ошибок в Sisyphus-x86_64: $ lftp -c 'cat http://git.altlinux.org/beehive/logs/Sisyphus-x86_64/latest/success/CoinSYMPHONY-5.4.1-alt1.svn20110903' | grep RPATH verify-elf: WARNING: ./usr/bin/symphony: RPATH entry found: path=/usr/lib/openmpi/lib verify-elf: WARNING: ./usr/lib64/libSymAppl.so.0.0.0: RPATH entry found: path=/usr/lib/openmpi/lib verify-elf: WARNING: ./usr/lib64/libOsiSym.so.0.0.0: RPATH entry found: path=/usr/lib/openmpi/lib verify-elf: WARNING: ./usr/lib64/libSym.so.0.0.0: RPATH entry found: path=/usr/lib/openmpi/lib Скорее всего, мейнтейнер имел в виду "/usr/lib/openmpi/lib", а вовсе не "path=/usr/lib/openmpi/lib", который у него получился. Вот список исходных пакетов с относительными путями в RPATH на x86-64, в формате ACL: CoinBcp real @everybody CoinBcps real @everybody CoinBlis real @everybody CoinBonmin real @everybody CoinCgl real @everybody CoinClp real @everybody CoinDip real @everybody CoinFlopCpp real @everybody CoinMP real @everybody CoinOS real @everybody CoinSYMPHONY real @everybody CoinSmi real @everybody CoinVol real @everybody GASNet real @qa @everybody adios real @everybody afepack-complex real @everybody afepack-real real @everybody arpack real @qa @everybody blopex-complex real @everybody blopex-real real @everybody blzpack real @qa @everybody ccaffeine real sin @python @qa @everybody cheetah real @qa @everybody chicken voins oddity dsdp real @qa @everybody exodusii real @everybody gromacs mike @qa @everybody h5part-mpi real @everybody ibutils stanv inger real @qa katrin zver @qa libblacs real @qa @everybody libhdf5-7-mpi real @everybody libnetcdf7-mpi real @everybody libproxy shaba aris @gnome libscalapack real @qa @everybody mISDN mithraen mct real @qa @everybody mpe2 real @qa @everybody mpip real sin @python @qa @everybody mpitests real @qa @everybody mumps real @qa @everybody p4est real @everybody parmetis0 real @everybody pastix real @qa @everybody petscfem-complex real @everybody petscfem-real real @everybody primme real @qa @everybody pswarm real @python @qa @everybody pympi real @qa @python @everybody python-module-boostmpi real @everybody python-module-mpi4py real sin @python @qa @everybody salsa-complex real @qa @everybody salsa-real real @qa @everybody scalasca real @qa @everybody scotch real @qa @everybody scribus manowar real @everybody @qa silo real sin @python @qa @everybody spooles real @qa @everybody sprng real @qa @everybody sprng1 real @everybody squid-kerberos-ldap-helper manowar @qa sundials real @qa @everybody trlan real @qa @everybody Я предлагаю считать относительные пути в RPATH грубыми ошибками, приводящими к ошибке сборки даже в режиме проверки rpath=relaxed. -- ldv [-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]
next prev parent reply other threads:[~2011-12-12 14:12 UTC|newest] Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top 2011-12-12 13:47 [devel] I: illegal RPATH, introduction Dmitry V. Levin 2011-12-12 14:12 ` Dmitry V. Levin [this message] 2011-12-12 18:01 ` [devel] I: illegal RPATH: relative paths Igor Vlasenko 2011-12-13 3:40 ` REAL 2011-12-13 4:18 ` Damir Shayhutdinov 2011-12-13 5:52 ` REAL 2011-12-13 6:50 ` Damir Shayhutdinov 2011-12-13 7:56 ` REAL 2011-12-13 11:06 ` Dmitry V. Levin 2011-12-13 12:11 ` REAL 2011-12-13 17:24 ` Michael Shigorin 2011-12-13 11:30 ` Dmitry V. Levin 2011-12-13 12:33 ` REAL 2011-12-13 16:01 ` Dmitry V. Levin 2011-12-13 16:11 ` Денис Смирнов 2011-12-13 16:46 ` Damir Shayhutdinov 2011-12-14 6:44 ` Денис Смирнов 2011-12-14 8:53 ` Sergey Y. Afonin 2011-12-14 10:07 ` REAL 2011-12-14 11:09 ` Michael Shigorin 2011-12-14 12:11 ` REAL 2011-12-13 17:25 ` Michael Shigorin 2011-12-13 17:32 ` Dmitry V. Levin 2011-12-13 19:17 ` Michael Shigorin 2011-12-13 21:29 ` Dmitry V. Levin 2011-12-13 3:37 ` REAL 2011-12-21 10:55 ` Michael Shigorin 2011-12-31 15:43 ` Yuri N. Sedunov 2011-12-12 14:33 ` [devel] I: illegal RPATH: out of standard library tree paths Dmitry V. Levin 2011-12-13 3:43 ` REAL 2011-12-13 11:17 ` Dmitry V. Levin 2011-12-13 12:25 ` REAL 2011-12-12 14:57 ` [devel] I: illegal RPATH: paths to standard libraries Dmitry V. Levin
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20111212141201.GC24413@altlinux.org \ --to=ldv@altlinux.org \ --cc=devel@lists.altlinux.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: link
ALT Linux Team development discussions This inbox may be cloned and mirrored by anyone: git clone --mirror http://lore.altlinux.org/devel/0 devel/git/0.git # If you have public-inbox 1.1+ installed, you may # initialize and index your mirror using the following commands: public-inbox-init -V2 devel devel/ http://lore.altlinux.org/devel \ devel@altlinux.org devel@altlinux.ru devel@lists.altlinux.org devel@lists.altlinux.ru devel@linux.iplabs.ru mandrake-russian@linuxteam.iplabs.ru sisyphus@linuxteam.iplabs.ru public-inbox-index devel Example config snippet for mirrors. Newsgroup available over NNTP: nntp://lore.altlinux.org/org.altlinux.lists.devel AGPL code for this site: git clone https://public-inbox.org/public-inbox.git