Tarantool development patches archive
 help / color / mirror / Atom feed
From: Alexander Turenko <alexander.turenko@tarantool.org>
To: "Alexander V. Tikhonov" <avtikhon@tarantool.org>
Cc: tarantool-patches@dev.tarantool.org
Subject: Re: [Tarantool-patches] [PATCH v1] build: fix unit tests build with lrt
Date: Thu, 28 Nov 2019 19:41:47 +0300	[thread overview]
Message-ID: <20191128164146.7kmvbyd24fmuifei@tkn_work_nb> (raw)
In-Reply-To: <2697bfced492ce88fc91d8d9dc1e2f1a68c7f147.1574430214.git.avtikhon@tarantool.org>

fiber.cc now uses clock_getttime(), which requires -lrt with glibc prior
2.1. Let's add librt dependency for core library for glibc of those
versions.

We should check for glibc version (using __GLIBC__ and __GLIBC_MINOR__
macros) or perform a check whether we able to use clock_gettime() w/o
librt.

So the patch should look like this:

 | diff --git a/CMakeLists.txt b/CMakeLists.txt
 | index 7d6c084..ca968eb 100644
 | --- a/CMakeLists.txt
 | +++ b/CMakeLists.txt
 | @@ -116,6 +116,12 @@ else ()
 |      set(HAVE_CLOCK_GETTIME ${HAVE_CLOCK_GETTIME_DECL})
 |  endif ()
 |  set(CMAKE_REQUIRED_LIBRARIES "")
 | +# According to `man 2 clockgettime` the glibc wrapper requires
 | +# -lrt on glibc versions before 2.17. We need to check whether
 | +# the function is available without -lrt to set this linker option
 | +# conditionally.
 | +check_function_exists(clock_gettime HAVE_CLOCK_GETTIME_WITHOUT_RT)
 | +
 |  check_symbol_exists(__get_cpuid cpuid.h HAVE_CPUID)
 |  
 |  # Checks for libev
 | diff --git a/src/lib/core/CMakeLists.txt b/src/lib/core/CMakeLists.txt
 | index e60b519..72e631b 100644
 | --- a/src/lib/core/CMakeLists.txt
 | +++ b/src/lib/core/CMakeLists.txt
 | @@ -46,3 +46,11 @@ target_link_libraries(core salad small uri decNumber ${LIBEV_LIBRARIES}
 |  if (ENABLE_BACKTRACE AND NOT TARGET_OS_DARWIN)
 |      target_link_libraries(core gcc_s ${UNWIND_LIBRARIES})
 |  endif()
 | +
 | +# Since fiber.top() introduction, fiber.cc, which is part of core
 | +# library, depends on clock_gettime() syscall, so we should set
 | +# -lrt when it is appropriate. See a comment for
 | +# HAVE_CLOCK_GETTIME_WITHOUT_RT in ${REPO}/CMakeLists.txt.
 | +if ("${HAVE_CLOCK_GETTIME}" AND NOT "${HAVE_CLOCK_GETTIME_WITHOUT_RT}")
 | +    target_link_libraries(core rt)
 | +endif()

CCed Serge as author of the patch that introduces the regression. Serge,
it seems we should disable fiber.top() or give explicit error when
HAVE_CLOCK_GETTIME is false.

WBR, Alexander Turenko.

On Fri, Nov 22, 2019 at 04:51:38PM +0300, Alexander V. Tikhonov wrote:
> After the commit 77fa45bd05f8cdd4c0f9bad85185ef5b61528d49

Please, use format commit_hash ('commit message header').

> the unit tests builds failed like:
> 
> /opt/rh/devtoolset-6/root/usr/libexec/gcc/x86_64-redhat-linux/6.3.1/ld:
>   ../../src/lib/core/libcore.a(fiber.c.o): undefined reference to symbol
>   'clock_gettime@@GLIBC_2.2.5'
> //lib64/librt.so.1: error adding symbols: DSO missing from command line
> collect2: error: ld returned 1 exit status
> test/unit/CMakeFiles/cbus.test.dir/build.make:108: recipe for target
>   'test/unit/cbus.test' failed
> make[2]: *** [test/unit/cbus.test] Error 1
> 
> To fix the issue the 'rt' library was added to test unit cmake file only
> for linux and freebsd targets.

It is glibc flaw, no need to change something on FreeBSD.

> 
> Close #4639
> ---
> 
> Github: https://github.com/tarantool/tarantool/tree/avtikhon/gh-4639-lrt-unit-tests-full-ci
> Issue: https://github.com/tarantool/tarantool/issues/4639

  reply	other threads:[~2019-11-28 16:41 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-22 13:51 Alexander V. Tikhonov
2019-11-28 16:41 ` Alexander Turenko [this message]
2019-11-29  7:12   ` Alexander Tikhonov
  -- strict thread matches above, loose matches on Subject: below --
2019-11-29  7:05 Alexander V. Tikhonov
2019-11-22 11:39 Alexander V. Tikhonov

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=20191128164146.7kmvbyd24fmuifei@tkn_work_nb \
    --to=alexander.turenko@tarantool.org \
    --cc=avtikhon@tarantool.org \
    --cc=tarantool-patches@dev.tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH v1] build: fix unit tests build with lrt' \
    /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

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox