|
Yes, I can see that . the question was - why ?
|
|
|
|
|
You asked:
Vaclav_ wrote: In this example I gather the "-L" is linked to "lib" four times.
Do not get it .
-L/usr/lib/gcc/x86_64-linux-gnu/5/../../../../lib
Speaking for myself, the question asked suggests that you think the syntax somehow creates four links to "lib", and are unaware of the relationship between an absolute path and a relative path. Both Eddy Vluggen and I tried to answer the question we thought you had asked, and set you straight about what was going on here.
As I stated in my answer, I think this has come from gcc -v output. As I am not a gcc developer, I do not know why the compiler would generate such odd search paths. It could be something to do with hoops needed to generate correct path names on non unix-like operating systems, like say OpenVMS. The thing to do would be to ask the gcc developers mailing list. As it stands, it works fine, and I'm not aware of any significant performance penalty that accrues from doing things this way. Unless you're interested for interests sake, why worry about it? Essentially, it happens "under the hood" - without the -v option, gcc/g++/as/ld/etc can be considered a "black box" that takes in source code and produces either intermediate code [assembler (.s files) and object (.o files) being two possibilities], or an executable program. How it goes about it is, normally, uninteresting to the average developer. Sometimes, yes, you do need to get that information, particularly if you think you've found a bug and are working with the gcc development team to document the bug. Otherwise, we just loop through the "edit, compile, test" loop until we, or our customer, are happy with the results, not worry about how the compiler, or editor, or link-loader, etc actually do their work.
|
|
|
|
|
The only reason I can come up with, is to ensure that the "gcc/x86_64-linux-gnu/5" folder exists as it would fail if it doesn't.
Bastard Programmer from Hell 
If you can't read my code, try converting it here[^]
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.
|
|
|
|
|
Paths like that usually come from the config when you build the package on a local machine. Config scripts try to be all things to all people so often create what looks like illogical paths, but are necessary because of the way the scripts analyse a local system.
|
|
|
|
|
Are you saying that it came from the way Eclipse was installed and how IDE relates to the "tool chain"?
I need better understanding of this "install" process.
Especially how piece parts of the "application install " are spread into different Linux folders.
I do not see any point to "hit the books" until I run into issues like this.
Pretty munch tired of not knowing these "under the hood" processes - from the ground up.
This thread answered some of my questions, and I appreciate that.
I think will apply my newly acquired knowledge to get better understanding how Linux "packages" are processed / installed.
Thanks
|
|
|
|
|
See title.
Here is lscpu output on x86.
jim@jim-desktop:~$ lscpu
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 4
On-line CPU(s) list: 0-3
Thread(s) per core: 1
Core(s) per socket: 4
Socket(s): 1
NUMA node(s): 1
Vendor ID: GenuineIntel
CPU family: 6
Model: 55
Model name: Intel(R) Celeron(R) CPU J1900 @ 1.99GHz
Stepping: 3
CPU MHz: 2415.718
CPU max MHz: 2415.7000
CPU min MHz: 1332.8000
BogoMIPS: 3998.40
Virtualization: VT-x
L1d cache: 24K
L1i cache: 32K
L2 cache: 1024K
NUMA node0 CPU(s): 0-3
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology tsc_reliable nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 movbe popcnt tsc_deadline_timer rdrand lahf_lm 3dnowprefetch epb pti ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid tsc_adjust smep erms dtherm ida arat
Ana here is what armv7l saiz about bits - nothing.
pi@pi:~ $ lscpu
Architecture: armv7l
Byte Order: Little Endian
CPU(s): 4
On-line CPU(s) list: 0-3
Thread(s) per core: 1
Core(s) per socket: 4
Socket(s): 1
Model: 4
Model name: ARMv7 Processor rev 4 (v7l)
CPU max MHz: 1200.0000
CPU min MHz: 600.0000
BogoMIPS: 38.40
Flags: half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt vfpd32 lpae evtstrm crc32
Is this Linux distro depended ? Ubuntu versus Rasbpian?
Just asking.
|
|
|
|
|
The Architecture type x86_64 tells you the hardware is 64 bit Intel architecture. The operating modes tell you it can run 32-bit or 64-bit applications.
|
|
|
|
|
|
I have given up on building bluez library from source.
Now my question is
how to option / instruct (gcc) compiler / linker to use libbluetooth-dev on native OS?
I have downloaded correct version but have no clue how to or where to option the compiler / linker.
I also like to use correct terminology for the "libbluetoooth-dev" .
Is is a libray or package or what ?
When get the above answered I should be able to apply it to different architecture.
I have also run into some kind of "foregin package " options, but not sure I need it. Maybe later.
Help is much appreciated.
Cheers
|
|
|
|
|
What is "libbluetoooth-dev"? Unix/Linux libraries should have a .a (archive) or .so (shared object) extension to identify the type of library. I have explained a number of times that to include a library named libxxx.a or libxxx.so, you should add an option of the form -lxxx to your build (either gcc or ld). If the name of the library does not follow the standard conventions, then use -l:filename.
For all ld opions see ld(1): GNU linker - Linux man page[^].
|
|
|
|
|
Partially answered my own question - libbluetooth-dev is a "package".
Using it / linking to it is non issue.
|
|
|
|
|
I have made some progress in tracing the "configure" script, all the way to
****
@ line 6764 debug START converting (?)
@ line 6765 --build = x86_64-pc-linux-gnu (?)
@ line 6766 --host = armv7l-unknown-linux-gnueabihf (?)
****
checking how to convert x86_64-pc-linux-gnu file names to armv7l-unknown-linux-gnueabihf format... ****
@ line 6831 unhandled host =armv7l-unknown-linux-gnueabihf (and normal native builds) (?)
****
****
@ line 6853 (?)
****
func_convert_file_noop
Now I do not get why "they" are converting build NAME to host FORMAT, but since the triplet "case " switch does not check for anything near "linux" and falls thru to "unhandled " I am beginning to wonder it the source I got from www.bluez.org is for Windowze only.
Is there a away to very that?
I am also stuck as far as what does "func_convert_file_noop" do.
Addendum
Here is a part of "configure" script I believe is responsible in instructing the compiler to build a libray of --host architecture.
As far as I can tell it does not evaluate for "Linux" as OS in optioned triplet , but falls thru to "unhandled" and then the conpiler buidls "native " AKA --build architecture libraries.
{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to convert $build file names to $host format" >&5
$as_echo_n "checking how to convert $build file names to $host format... " >&6; }
if ${lt_cv_to_host_file_cmd+:} false; then :
$as_echo_n "(cached) " >&6
else
case $host in
*-*-mingw* )
case $build in
*-*-mingw* ) # actually msys
lt_cv_to_host_file_cmd=func_convert_file_msys_to_w32
;;
*-*-cygwin* )
lt_cv_to_host_file_cmd=func_convert_file_cygwin_to_w32
;;
* ) # otherwise, assume *nix
lt_cv_to_host_file_cmd=func_convert_file_nix_to_w32
;;
esac
;;
*-*-cygwin* )
case $build in
*-*-mingw* ) # actually msys
lt_cv_to_host_file_cmd=func_convert_file_msys_to_cygwin
;;
*-*-cygwin* )
lt_cv_to_host_file_cmd=func_convert_file_noop
;;
* ) # otherwise, assume *nix
lt_cv_to_host_file_cmd=func_convert_file_nix_to_cygwin
;;
esac
;;
* ) # unhandled hosts (and "normal" native builds)
lt_cv_to_host_file_cmd=func_convert_file_noop
;;
esac
fi
to_host_file_cmd=$lt_cv_to_host_file_cmd
{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $lt_cv_to_host_file_cmd" >&5
$as_echo "$lt_cv_to_host_file_cmd" >&6; }
{ $as_echo "$as_me:${as_lineno-$LINENO}: checking how to convert $build file names to toolchain format" >&5
$as_echo_n "checking how to convert $build file names to toolchain format... " >&6; }
if ${lt_cv_to_tool_file_cmd+:} false; then :
$as_echo_n "(cached) " >&6
else
#assume ordinary cross tools, or native build.
lt_cv_to_tool_file_cmd=func_convert_file_noop
case $host in
*-*-mingw* )
case $build in
*-*-mingw* ) # actually msys
lt_cv_to_tool_file_cmd=func_convert_file_msys_to_w32
;;
esac
;;
esac
fi
to_tool_file_cmd=$lt_cv_to_tool_file_cmd
{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $lt_cv_to_tool_file_cmd" >&5
$as_echo "$lt_cv_to_tool_file_cmd" >&6; }
modified 28-Mar-19 16:24pm.
|
|
|
|
|
1) the configure script is AUTOGENTERATED by the autoconf tool set. There's generally no good that can come from messing about in it.
2) I have told you several times use --host=arm-linux-gnueabihf.
From your post 3 days ago:
Quote: sudo ./configure --prefix=/usr/local/BLUEZ/bluez-5.50 --mandir=/usr/share/man --sysconfdir=/etc --localstatedir=/var --enable-library --disable-systemd --host=armv7l-unknown-linux-gnueabihf
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for armv7l-unknown-linux-gnueabihf-strip... no
checking for strip... strip
checking for a thread-safe mkdir -p... /bin/mkdir -p
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether make supports nested variables... yes
checking how to create a pax tar archive... gnutar
checking for style of include used by make... GNU
checking for armv7l-unknown-linux-gnueabihf-gcc... no
checking for gcc... gcc
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables...
checking whether we are cross compiling... no
Now, I would expect anyone paying any attention (and surely anyone who's as nit-picky as you seem to be) to think to themselves:
Hmm ... looks like the configure script is trying to use armv7l-unknown-linux-gnueabihf-gcc. That prefix is what I supplied as the value of the --host argument. I don't have arm7l-unkown-linux-gneabihf-gcc on my system. I do have, however arm-linux-gnueabihf-gcc - maybe I'll try --host=arm-linux-gneuabihf and see if that works.
And lo-and-behold configure completes, and you can start make. and wonder of wonders it uses arm-linux-gnueabihf-gcc, just like we want it too. Result!
BUT,BUT, and again BUT: until you've got the prerequisites (GLib, D-Bus) built for armhf make will fail anyway!
3) if you do some googling around for autoconf --target you should find information about the so called candian cross Cross-Compilation (automake). Take the time to read down to the bottom of the page. There it tells you that --target is used when compiling compiler tools i.e. ld, ar, as, etc (binutils) in a three-way cross-compilation. For example, when you are compiling binutils on an X86 to run on an armhf but will produce code for PowerPC. --target has NOTHING to do with cross compiling in the normal way. I have not tried it by suspect that since configure is AUTOGENERATED that the --target option always shows up in the script, whether its required or not.
Now, stop elephanting around.
|
|
|
|
|
pi@pi $ sudo bash config.guess
armv7l-unknown-linux-gnueabihf
|
|
|
|
|
Stop using sudo for everything. That's like using Run as Administrator all the time.
As it happens, I do know what I'm talking about. As an example I'll cross compile grep for you. grep does have an optional dependency on libpcre - perl regular expressions, which I won't complicate this by also compiling. Fortunately we can disable that by using --disable-perl-regex
$ ls
grep-3.0 grep-3.0.tar.xz
configure can build in a directory other than the source directory, so lets do that. We'll use --prefx=$HOME/apt so we don't need to use sudo to install.
$ mkdir build
$ cd build
$ ../grep-3.0/configure --host=arm-linux-gnueabihf --prefix=$HOME/arm --disable-perl-regexp > configure.out 2>&1 && echo Success || echo Failed
Success
did we configure for cross compile?
$ grep cross configure.out
checking whether we are cross compiling... yes
checking for working fcntl.h... cross-compiling
lets make and install ...
$ make > make.out 2>&1 && echo Success || echo Failure
Success
$ ls $HOME/arm
ls: cannot access '/home/k5054/arm': No such file or directory
$ make install > make_install.out 2>&1 && echo Success || echo Failure
Success
so far so good ... lets see what I have in $HOME/arm
$ cd $HOME/arm
$ ls
bin share
$ cd bin
$ ls
egrep fgrep grep
$ file *
egrep: POSIX shell script, ASCII text executable
fgrep: POSIX shell script, ASCII text executable
grep: ELF 32-bit LSB shared object, ARM, EABI5 version 1 (SYSV), dynamically linked, interpreter /lib/ld-, for GNU/Linux 3.2.0, BuildID[sha1]=b44e80774f0b6413d50f10c1bd7f57d10e88a6af, with debug_info, not stripped
so we've got an ARM EABI5 executable for $HOME/arm/bin/grep. how does that compare to /bin/grep?
$ file /bin/grep
/bin/grep: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 2.6.32, BuildID[sha1]=0b72a4b88afe6711c7154541e1f002331041c71f, stripped
so that's an X86_64 executable ... things are looking good, lets copy that over to the pi and see if it will work
$ scp -r -q arm pi3:
$ ssh pi3
Linux pi3 4.14.98-v7+ #1200 SMP Tue Feb 12 20:27:48 GMT 2019 armv7l
Last login: Thu Mar 28 00:33:06 2019 from 192.168.235.202
lets test arm/bin/grep
k5054@pi3:~ $ ./arm/bin/grep --version
grep (GNU grep) 3.0
Copyright (C) 2017 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>.
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Written by Mike Haertel and others, see <http://git.sv.gnu.org/cgit/grep.git/tree/AUTHORS>.
that worked! how does that compare to /bin/grep?
k5054@pi3:~ $ /bin/grep --version
grep (GNU grep) 2.27
Copyright (C) 2016 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>.
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Written by Mike Haertel and others, see <http://git.sv.gnu.org/cgit/grep.git/tree/AUTHORS>.
it's a different version, very good! - now lets try a grep pattern, and count the number of times the word "is" appears in the version text of /bin/grep
k5054@pi3:~ $ /bin/grep --version | arm/bin/grep -c '\<is\>'
2
looks like my cross-compiled grep is working fine
k5054@pi3:~ $ logout
well, looks like that all worked -- seems --host=arm-linux-gnueabihf is the right thing after all
|
|
|
|
|
I hope this is my last post on bluez package.
I made wrong choice.
END OF STORY
FYI update
guess what I DID NOT find in "configure"?
System types:
--build=BUILD configure for building on BUILD [guessed]
--host=HOST cross-compile to build programs to run on HOST [BUILD]
#CCC ADD
# looks as there is no NEED for target
_ACEOF
Will update - pending actual test of the bluez build and C++ link.
This is a SPECIFIC follow-up on the original post.
If you have an issue with this - please ignore me.
I have managed to recompile Linux "stack " - bluez for armhf
sudo dpkg -l "bluez"
[sudo] password for jim:
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Architecture Description
+++-==============-============-============-=================================
ii bluez:armhf 5.50-1 armhf bluez-5.50
I have verified that I do have "libbluetooth.so" file linked via -L option and linker did same.
I am still getting this error telling me the file format is wrong.
Is it possible that gcc crosscompile prefix arm-linux-gnueabihf and "bluez" package architecture armhf is causing this?
I did try to checkinstall for arm-linux-gnueabihf architecture but it did not go.
I also build same package for armv7l architecture with same results.
-L/usr/local/BLUEZ/bluez-5.50/lib -L/usr/lib/gcc-cross/arm-linux-gnueabihf/5 -L/usr/lib/gcc-cross/arm-linux-gnueabihf/5/../../../../arm-linux-gnueabihf/lib/../lib -L/lib/../lib -L/usr/lib/../lib -L/usr/lib/gcc-cross/arm-linux-gnueabihf/5/../../../../arm-linux-gnueabihf/lib ./src/BLUE_CROSS.o -lbluetooth -lstdc++ -lm -lgcc_s -lgcc -lc -lgcc_s -lgcc /usr/lib/gcc-cross/arm-linux-gnueabihf/5/crtend.o /usr/lib/gcc-cross/arm-linux-gnueabihf/5/../../../../arm-linux-gnueabihf/lib/../lib/crtn.o
/usr/local/BLUEZ/bluez-5.50/lib/libbluetooth.so: file not recognized: File format not recognized
collect2: error: ld returned 1 exit status
makefile:45: recipe for target 'BLUE_CROSS_193210' failed
modified 1-Apr-19 11:52am.
|
|
|
|
|
what does file /usr/local/BLUEZ/bluez-5.50/lib/libbluetooth.so return? If it is not ELF 32-bit LSB shared object, ARM, EABI5 version 1 (SYSV), then you need to review your installation of the bluez library for armhf.
|
|
|
|
|
"return"?
Forgive me my ignorance , but how do I do that?
jim@jim-desktop:/usr/local/BLUEZ/bluez-5.50/lib$ ls
a2mp.h cmtp.h libbluetooth.la sdp.h
amp.h cups libbluetooth.so sdp_lib.h
bluetooth hci.c libbluetooth.so.3 sdp.lo
bluetooth.c hci.h libbluetooth.so.3.18.16 uuid.c
bluetooth.h hci_lib.h mgmt.h uuid.h
bluetooth.lo hci.lo pkgconfig uuid.lo
bluez.pc hidp.h rfcomm.h
bluez.pc.in l2cap.h sco.h
bnep.h libbluetooth-internal.la sdp.c
jim@jim-desktop:/usr/local/BLUEZ/bluez-5.50/lib$ file libbluetooth.so
libbluetooth.so: symbolic link to libbluetooth.so.3.18.16
<pre lang="text">
jim@jim-desktop:/usr/local/BLUEZ/bluez-5.50/lib$ stat libbluetooth.so
File: 'libbluetooth.so' -> 'libbluetooth.so.3.18.16'
Size: 23 Blocks: 0 IO Block: 4096 symbolic link
Device: 812h/2066d Inode: 919828 Links: 1
Access: (0777/lrwxrwxrwx) Uid: ( 0/ root) Gid: ( 0/ root)
Access: 2019-03-22 11:41:14.976991502 -0500
Modify: 2019-03-22 11:09:42.000000000 -0500
Change: 2019-03-22 11:29:48.319241561 -0500
Birth: -
modified 22-Mar-19 19:37pm.
|
|
|
|
|
|
OK, I missed the "symbolic link".
Now I need to go back to "configure" to find out why the library is build for host and
not the target.
Thanks very much for your help, appreciate it,
PS
It would have been helpful if the linker told me this.
Maybe there is an option somewhere to do that.
jim@jim-desktop:/usr/local/BLUEZ/bluez-5.50/lib$ ls
cups libbluetooth.la libbluetooth.so libbluetooth.so.3 libbluetooth.so.3.18.16 pkgconfig
jim@jim-desktop:/usr/local/BLUEZ/bluez-5.50/lib$ file libbluetooth.so.3.18.16
libbluetooth.so.3.18.16: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, BuildID[sha1]=55d60353934d3672e2c88fbc64ceaba37c4d0442, not stripped
jim@jim-desktop:/usr/local/BLUEZ/bluez-5.50/lib$ file libbluetooth.so
libbluetooth.so: symbolic link to libbluetooth.so.3.18.16
jim@jim-desktop:/usr/local/BLUEZ/bluez-5.50/lib$ file libbluetooth.so.3
|
|
|
|
|
Vaclav_ wrote: It would have been helpful if the linker told me this.
The linker's role is to combine different object code files into an executable. It expects all objects to be in the correct format, so if they are not then it can only reject them. It cannot be expected to analyse the content of every 'foreign' object to tell the user what it is.
|
|
|
|
|
You can check where your apt installed your package using
$ dpkg -L bluez
/.
/bin
/bin/hciconfig
/etc
/etc/bluetooth
/etc/bluetooth/input.conf
/etc/bluetooth/main.conf
...
/usr/share/man/man1/rctest.1.gz
/usr/share/man/man1/rfcomm.1.gz
/usr/share/man/man1/sdptool.1.gz
/usr/share/man/man8
/usr/share/man/man8/bluetoothd.8.gz
/usr/sbin/bluetoothd
$
Note that if you're querying packages, you don't need to sudo to execute dpkg. Also note that the normal package for bluez does not include libbluetooth. For that you need packages libbluetooth3 and libbluetooth-dev (for headers, etc).
$ dpkg -L bluez | grep libbluetooth.so
$ dpkg -L libbluetooth3 | grep libbluetooth.so
/usr/lib/arm-linux-gnueabihf/libbluetooth.so.3.18.14
/usr/lib/arm-linux-gnueabihf/libbluetooth.so.3
$ dpkg -L libbluetooth-dev | grep libbluetooth.so
/usr/lib/arm-linux-gnueabihf/libbluetooth.so
$
I expect if you've built these packages for armhf and installed them on your X86 sytem, then they may not be in the locations listed here.
|
|
|
|
|
Thanks for reply.
After Richard's post I can correctly identify the file wrong architecture - as already suspected and reported by the linker.
Upon preliminary investigation it looks as "bluez" package "configure" is flaky.
It's sketchy documentation clearly states that it is smart enough to sort-of self option itself.
I can prove otherwise as far as --build, --host and --target options go.
I really do not care how it "automatically " determines them, but it totally ignores --target when given as an option!
Not only that - it also selectively ignores other architecture options.
I can provide "configure" file outputs if anybody cares.
Here is a snippet
checking build system type... x86_64-pc-linux-gnu
checking host system type... x86_64-pc-linux-gnu
Yes, I did search for "target" - nothing , nada
The real question is - how to resolve this?
www.bluez.org is useless
Ubuntu forum probably does not care
other forums are off limits and I really do not want to be just statistic anyway
If it is bug - where do I go next?
I guess learn how to edit "configure " is my real option.
But I need to verify if the target architecture I am using is valid , since I cannot rely on "configure" to tell me !
For example - this passed --target=armv7l, but was ignored.
I am open to suggestions,
|
|
|
|
|
Try opening the configure script with your favourite editor and do a search for the keywords mentioned above. You should then be able to see what options it is able to handle.
|
|
|
|
|
Did that already.
There is no "checking ..target..." similar to "host" and "build".
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
