How to compile a Go application using gopacket for 32bit mips - go

I am trying to compile a little application using the gopacket library to linux on a 32bit mips cpu. Unfortunately I am getting loads of errors like this:
/home/cdutz/go/pkg/mod/github.com/google/gopacket#v1.1.19/pcap/pcap.go:30:22: undefined: pcapErrorNotActivated
On a "normal" linux system these seem to be defined in "pcap_unix.go" while on windows the values are coming from "defs_windows_amd64.go". I do have the libpcap in a 32bit mips version on my target system, which is good as my target system doesn't have the extra space to install all the tools needed to compile anything on it. I know that libpcap doesn't exist as a 1-to-1 version on windows, so this probably explains the "defs"-files. But I would generally expect it to have the same API as the one on my linux system.
[UPDATE]
So it seems number 1 of the things that needs to be ensured is that cgo is executed. This is done by setting the environment variable:
CGO_ENABLED=1
Next we need to ensure the mips compatible versions of the libpcap are installed (the header files are identical on any architecture). In order to do this on my Ubuntu 16.4 I first needed to enable the 'mips' archirecture:
dpkg --add-architecture mips
And add the debian repo to the /etc/apt/sources.list
deb [trusted=yes] http://ftp.de.debian.org/debian buster main
As soon as that's done, I could install the mips binaries:
apt install libpcap-dev:mips libpcap0.8-dev:mips libc6-dev:mips libdbus-1-dev:mips libpcap0.8:mips
In order to cross compile, it seems I need a gccgo version that can do that. For this I installed:
apt-get install gccgo-mips-linux-gnu
Now comes something I am not sure I did right, but when running the go build with compiler=gccgo it always picked the amd64 version and using anything else but 'gccgo' as compiler argument didn't work, so I updated the symlink in /usr/bin/gccgo to point to: 'mips-linux-gnu-gccgo-8' (in the same directory).
After all of these changes, I was able to almost build everything with this command:
go build -compiler=gccgo
If I enable the additional output with the '-x' option, I can see that cgo is now doing it's thing. It's also compiling all the other modules. But on the pcap one it now fails with:
cc1: error: command line option '-c' is valid for the driver but not for C
So this is where I'm currently stuck at.

Ok, after 3 days I think I managed to get things working and I'll summ up what I did.
In the end the gccgo path was a dead end, so instead of installing gccgo-mips-linux-gnu I installed gcc-mips-linux-gnu.
Next I set the CC environment variable to point to this:
export CC=/usr/bin/mips-linux-gnu-gcc-8
That was actually what was missing.
So summing it up on my Ubuntu 16.04 system:
dpkg --add-architecture mips
echo "deb [trusted=yes] http://ftp.de.debian.org/debian buster main" > /etc/apt/sources.list
apt update
apt install -y wget git build-essential mc
apt install -y gcc-mips-linux-gnu
apt install -y libpcap-dev:mips libpcap0.8-dev:mips libc6-dev:mips libdbus-1-dev:mips libpcap0.8:mips
export CC=/usr/bin/mips-linux-gnu-gcc-8
export GOOS=linux
export GOARCH=mips
export GOMIPS=softfloat
export CGO_ENABLED=1
go build
I hope this might help others.

Related

Ubuntu: cgo: C compiler "gcc-11" not found: exec: "gcc-11": executable file not found in $PATH

I am using Ubuntu 20.04, and I installed Go using Homebrew.
The version of Go is go 1.19.
When I run my application with go run . or go build ., this error comes up:
# github.com/mattn/go-sqlite3
cgo: C compiler "gcc-11" not found: exec: "gcc-11": executable file not found in $PATH
I have tried running
sudo apt install gcc
but the terminal tells me that gcc is already the newest version (4:9.3.0-1ubuntu2).
When I run gcc -v
the terminal tells me gcc version 9.4.0 (Ubuntu 9.4.0-1ubuntu1~20.04.1)
And I have already run apt update and apt upgrade already, and then reinstall gcc, but it's still version 9.4.0
My question is how do I install gcc-11? Or is it not supported in Ubuntu 20.04? Or should I not use go 1.19?
Thanks
The go developers provide a simple way to manage multiple go versions at once: https://go.dev/doc/manage-install
It should not be too hard to explore if this is a Homebrew artifact, or something generic to go, or a specific version of go on your system.
You might also look at the output of go env which should list configured defaults for various dependencies. You are looking for something like CC=gcc in the output.

Packages apt-get vs brew?

Im trying to get something running in a lab, and I need to install a whole bunch of packages to compile the code:
sudo apt-get install build-essential libgtk2.0-dev libwxgtk3.0-dev libwx-perl libmodule-build-perl git cpanminus libextutils-cppguess-perl libboost-all-dev libxmu-dev liblocal-lib-perl wx-common libopengl-perl libwx-glcanvas-perl libtbb-dev libxmu-dev freeglut3-dev libwxgtk-media3.0-dev libboost-thread-dev libboost-system-dev libboost-filesystem-dev libcurl4-openssl-dev libextutils-makemaker-cpanfile-perl
The problem is, I'm working on a Mac; So I only have brew available - and the package names are not equivalent.
Am I trying to do something dumb and impossible, or is there a way to cross reference those apt-get packages to ones available in brew and install those so I can build the app?
For added context, its a modified "slicer" application for generating files that a 3d printer uses to make parts. I could fire up an ubuntu VM and use apt-get, but I'm on an M1 mac at the moment and recompiling experimental QEMU code just so I can fire up an ubuntu VM takes me way more out of my depth than I already am... I'm running brew on a duplicated, rosetta emulation forced terminal.
There is no straightforward way to automatically figure out whether a corresponding Homebrew package exists for each of these packages, no. But you can probably guess the majority, and manually figure out the rest. Each Debian package has a link to the upstream sources, and a home page if one exists, from which you can often find links to packages for other architectures, etc.
From the Debian package search page you can search e.g. for the libxmu-dev package, and discover the corresponding package page for Buster (the current stable Debian release), which in turn has links to the upstream repo, etc. But this is an X11 package, so it's not straightforwardly compatible with macOS, which uses an entirely different GUI architecture.

exec: "gcc": executable file not found in %PATH% when trying go build

I am using Windows 10. When I tried to build Chaincode it reported this error
# github.com/hyperledger/fabric/vendor/github.com/miekg/pkcs11
exec: "gcc": executable file not found in %PATH%
My chaincode imports:
import (
"fmt"
"strconv"
"github.com/hyperledger/fabric/core/chaincode/shim"
pb "github.com/hyperledger/fabric/protos/peer"
)
It's running fine in Docker.
gcc (the GNU Compiler Collection) provides a C compiler. On Windows, install TDM-GCC. The github.com/miekg/pkcs11 package uses cgo. Cgo enables the creation of Go packages that call C code.
If you are running Ubuntu do:
apt-get install build-essential
This solved the problem. It installs the gcc/g++ compilers and libraries.
I also encountered this message, but in my case, it was missing gcc.exe. I used choco and installed mingw, and then it worked.
details:
download choco
choco install mingw -y
check: gcc -v
1) Install .exe from > https://sourceforge.net/projects/mingw-w64/
1.2) ! use x86_64 architecture
2) Add C:\Program Files\mingw-w64\x86_64-8.1.0-posix-seh-rt_v6-rev0\mingw64\bin to PATH in User Variables and in System Variables. For me it works.
! To edit Path variable press Windows key, type 'path', choose 'Edit the system environment variables', click 'Environment Variables', find Path variable in System variables and in User variables then edit.
On Windows install http://tdm-gcc.tdragon.net/download, that is all.
If you are using an alpine based image with your Dockerfile
Install build-base which will be met with your requirements.
apk add build-base
$ go env
check CGO_ENABLED if its 1 change it to 0 by
$export CGO_ENABLED=0
For my case :
os: windows 10
command:
choco install mingw
install choco if not installed:
Link: https://www.liquidweb.com/kb/how-to-install-chocolatey-on-windows/
worked for me.
The proper explanations why go build does not work for hyperledger in Windows environment are given as other answers.
For your compilation purposes, just to make it work without installing anything extra, you can try the following
go build --tags nopkcs11
It worked for me. I hope same works for you too.
You can try - this is not a solution but a temp workaround
cgo_enabled=0 go build
Once you install gcc - and make sure %PATH has a way to find it (gcc.exe) - this should go away.
Also running this one will ensure the cgo_enabled variable will stay this way as long as terminal is open. That way you don't have to prefix it each time you do a build.
export cgo_enabled=0 go build
just followed instructions from following and it solve my issue
https://code.visualstudio.com/docs/cpp/config-mingw
it ask to install Mingw-w64 via MSYS2
important command is pacman -S --needed base-devel mingw-w64-x86_64-toolchain
then add C:\msys64\mingw64\bin to PATH
thanks
For Ubuntu, what worked for me was to simply run:
sudo apt install gcc
On Amazon Linux 2:
Install go
wget https://go.dev/dl/go1.18.1.linux-amd64.tar.gz
rm -rf /usr/local/go && tar -C /usr/local -xzf go1.18.1.linux-amd64.tar.gz
export PATH=$PATH:/usr/local/go/bin
Install gcc
sudo yum groupinstall "Development Tools"
I recommend using the package group, even though it can be done without it, because groupinstall gives you the necessary packages to compile software on Amazon Linux and Redhat, CentOS for that matter.
on Ubuntu its very easy but on windows need to do it:
download MinGW on http://www.mingw.org/
install on basic package Gcc-g++ (see this image)
add on environment Patch of windows variables.
restart and continue with "go get ..."
If you are running Ubuntu do:
sudo apt-get update
sudo apt-get install build-essential.
If the above commands do not work do:
sudo add-apt-repository "deb http://archive.ubuntu.com/ubuntu $(lsb_release -sc) main universe"
The main component contains applications that are free software, can be freely redistributed and are fully supported by the Ubuntu team. & The universe component is a snapshot of the free, open-source, and Linux world.
Then install package by following command in terminal:
sudo apt-get update
sudo apt-get install build-essential.
For more info click here: https://itectec.com/ubuntu/ubuntu-problem-installing-build-essential-on-14-04-1-lts-duplicate/
Just add this to your Dockerfile
RUN apk add alpine-sdk
gcc should not be necessary, unless you are cross compiling for a non-windows platform, or use cgo.
If you still need gcc, however, you should install MinGW, which provides a gcc port for Windows (Cygwin and msys should also work, although I have never actually tested this).
Edit: I see from your error message now, that it is a dependency that requires gcc. If you didn't already know this, gcc is a c/c++ compiler, and in this case it is probably needed to compile c source files included by a dependency or sub-dependency.
Instruction to fix the "exec: “gcc”: executable file not found in %PATH%" error with MSYS2:
Download MSYS2.
Put MSYS2 folder into your $PATH.
Start the MSYS2 command line program.
Run this command: pacman -S gcc.
Kindly install the MINGW after GUI will automatically take.
http://mingw.org/wiki/Getting_Started
On Windows, you can install gcc by Scoop:
scoop install gcc
you need to download MingGW64
put MingGW64 folder into your $PATH
run go build xxx.go (with cgo library)
Hi jaswanth the main problem is that you haven't register your %GO_HOME%\pkg\tool\windows_amd64 to yuour Environment Path.
%GO_HOME% is the repository where you install your go at the first time.
same as other, just install tdm-gcc, but you can use its terminal, "MinGW", you can access it from start menu folder tdm-gcc, after start, browse to your project, and run it again
I'm a Windows user and I downloaded tdm-gcc (MinGW-w64 based) from the link below:
https://jmeubank.github.io/tdm-gcc/
After installation, it made a folder named "TDM-GCC-64".
I added "C:\TDM-GCC-64\bin" to my PATH, And it fixed my problem.

Yosemite and Valgrind

Can you tell me how to install valgrind on yosemite? When I try to install it i get " checking for the kernel version... unsupported (14.0.0)
configure: error: Valgrind works on Darwin 10.x, 11.x, 12.x and 13.x (Mac OS X 10.6/7/8/9) "
There is no official path or update, and I didn't found anything (except http://comments.gmane.org/gmane.comp.kde.devel.bugs/1553705 , but they didn't resolve that problem).
As there's no stable release that supports Yosemite, you can install the latest development version with
brew install --HEAD valgrind
Whilst it may have been the case in past OS X release cycles that Valgrind took a period of time before achieving reasonable feature support, basic OS X 10.10 support is already available in Valgrind trunk due to significant work on pre-release Yosemite.
From the mailing list:
There has been some effort recently to improve Valgrind's support for
Yosemite. If you develop on Mac OS, you might like to try out the
trunk (svn co svn://svn.valgrind.org/valgrind/trunk) and report any
breakage you get. Support for Yosemite is good enough that at least
one large graphical application (Firefox) runs OK. Support for the
previous release, 10.9 (Mavericks), is also substantially improved.
Note that the work has targetted 64 bit processes only. 32 bit might
work, and probably better on Mavericks, but I suspect it will be
increasingly problematic on Yosemite due to Valgrind's 32 bit x86
instruction set support not having progressed passed SSSE3.
Julian Seward
http://sourceforge.net/p/valgrind/mailman/message/33047840/
Full disclosure: I'm one of the new Valgrind developers who contributed patches to support OS X 10.10
Valerio's svn workflow will download every branch which is time and resource consuming. A better procedure is to download just the trunk:
svn co svn://svn.valgrind.org/valgrind/trunk valgrind
cd valgrind
./autogen.sh
./configure
make
make install
Here is my take on it. I more or less had a clean mac with xcode installed.
Got it compiling and running with the following:
# build/install autoconf/automake/libtool so that 'autogen' works
curl -OL http://ftpmirror.gnu.org/autoconf/autoconf-2.69.tar.gz
tar -xzf autoconf-2.69.tar.gz
cd autoconf-2.69
./configure && make && sudo make install
cd ..
curl -OL http://ftpmirror.gnu.org/automake/automake-1.14.tar.gz
tar -xzf automake-1.14.tar.gz
cd automake-1.14
./configure && make && sudo make install
cd..
curl -OL http://ftpmirror.gnu.org/libtool/libtool-2.4.2.tar.gz
tar -xzf libtool-2.4.2.tar.gz
cd libtool-2.4.2
./configure && make && sudo make install
cd ..
svn co svn://svn.valgrind.org/valgrind/trunk valgrind
cd valgrind
./autogen.sh
# important: configure-params, otherwise make ends in errors
./configure -disable-tls --enable-only64bit --build=amd64-darwin
make
# sudo, otherwise it fails due to permissions
sudo make install
Note that callgrind_control (from valgrind-3.11.0 SVN) doesn't appear to work on OS X, looks like a perl-script and the commandline tool which it runs (vgdb -l) prints something 'unexpected' which that script doesn't correctly parse ( so it won't be able to find the other process running with valgrind ).
Alternatively, the perl script just calls vgdb, we can also directly do that ( just figure out your process-id manually):
vgdb --pid=2858 instrumentation on
Worked for me on 10.10.1 :
svn co svn://svn.valgrind.org/valgrind
cd valgrind
./autogen.sh
./configure
make
make install
All of the solutions listed here failed for me. What finally ended up working was to use mac ports.
sudo port install valgrind-devel
Here's how to install it using alternative sources besides the official svn (because it seems to be intermittently available).
https://crispyappstudiosblog.wordpress.com/2015/07/07/installing-valgrind-on-osx-yosemite/
1) Navigate to this git mirror of the svn and download the latest
available version: http://repo.or.cz/w/valgrind.git
2) You need VEX as well, so grab the latest version here:
http://repo.or.cz/w/vex.git
3) Extract both of them. Put the entire contents of the VEX folder
into a folder called VEX in the top level of the valgrind directory.
cd to the valgrind directory, and execute the following:
Run ./autogen.sh
Run ./configure
Run make
Run sudo make install
Test it out by running valgrind --version You should be running at
least 3.11.0 SVN for it work on Yosemite.
I installed it on my mac by installing homebrew and then running this 3 commands in the terminal.
brew update
brew doctor
brew install --HEAD valgrind
PS: I have Os X El Capitan (10.11) but this should work with previous versions too.
I finally got Valgrind to work on my OSX El Capitan 10.11.12.
User Kalmiya's answer worked for me first after I installed Xcode commandline tools.
Type this in the terminal:
xcode-select --install
Now follow Kalmiya's post, step by step.
https://stackoverflow.com/a/30366798/3633475
Here is another take on the svn install. The previous ones did not work for me, since I needed to have automake and autoconf installed, which I did not, even though I had the latest version of the Xcode command line tools installed.
I got the following from this site. I also had to link automake and autoconf after doing brew install automake and brew install autoconf by doing brew link automake and brew link autoconf for this to work.
# Check out their repo...
$ svn co svn://svn.valgrind.org/valgrind/trunk valgrind-trunk
# and hop into it.
$ cd valgrind-trunk
# You need to have autoconf and automake installed to build Valgrind
# This example uses Homebrew to install these dependencies
# (MacPorts should also work)
# (Permission error? add sudo!)
$ brew install automake
$ brew install autoconf
# run autogen.sh in valgrind-trunk
$ ./autogen.sh
# Tricky, there are some hard wired paths in the Valgrind sources.
# You need to symlink the mach folder in your XCode SDK to /usr/include/mach
# Be sure to use the proper Xcode SDK "MacOSX10.10.sdk" in the path!
$ ln -sv /Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk/usr/include/mach /usr/include/mach
# Run configure + set install paths in valgrind-trunk
$ ./configure --prefix=/usr/local
# Run Make and make install (permission error? add sudo!) in valgrind-trunk
$ make
$ make install
# Check it works
$ valgrind --version
valgrind-3.11.0.SVN
I have used kalmiya's instructions to build valgrind as a conda package for OSX Yosemite. For those who work with anaconda/conda, just do
conda install -c https://conda.binstar.org/groakat valgrind
Side-note:
I needed to install the command line tools as described below to get valgrind compiled.
https://stackoverflow.com/a/30471647/2156909
I got valgrind on Yosemite compiled, but had to use a hack to do so. While I think you should be using xcode-select install to get all command line tools (after which valgrind should make properly), but if you don't want to do this (eg. size of Xcode tools too big), you can also get the Darwin OSX code and copy the following files to /usr/include/mach
mach_vm.defs
task.defs
thread_act.defs
vm_map.defs
This allowed a clean compile and install, although note it is a rather slack hack.

Problem with gcc 4.6 installation on ubuntu

I am trying to install gcc 4.6 (mainly for having C++0x better supported) in my ubuntu 9.10 (via virtualbox). I referred to previous questions, but I am getting a different error.
I am referring this link for the installation. Now, I have done till the ./gcc-xx/configure ... step. Though it was giving some flex package related error. Mostly due to that make is also failing with below errors:
build/gengtype.o: In function
adjust_field_rtx_def':
/home/milind/ubuntu_shared/GCC/build/gcc/../../gcc-4.6-20110610/gcc/gengtype.c:978:
undefined reference tolexer_line'
/home/milind/ubuntu_shared/GCC/build/gcc/../../gcc-4.6-20110610/gcc/gengtype.c:1032:
undefined reference to lexer_line'
/home/milind/ubuntu_shared/GCC/build/gcc/../../gcc-4.6-20110610/gcc/gengtype.c:1042:
undefined reference tolexer_line' ...............
Now this is giving me a hard time figuring it out because I have already flex/bison latest versions installed. I searched over internet for 2 days almost but no luck. Any help would be really appreciated. Also note that, I already have gcc 4.4 installed in /usr/bin/gcc and I have unzipped the gcc 4.6 tar in my home directory local folder.
[Note: I am also ok with installing ubuntu 11.10 too (which has gcc 4.6) as last resort. But I don't know if its .iso image is available.]
I got this fixed. I followed following procedure:
[Note: run all the commands with sudo, if you are not login as root. e.g. sudo ls -ltr; sudo make install;
As mentioned in the link in my
question, download the gcc4.6...tar
file in a temporary place
Now find the place where current
gcc is stored. e.g. My earlier
gcc4.4 was stored in
/usr/lib/gcc/i486-linux-gnu. Which
has a folder called 4.4, 4.4.1
Create a folder named 4.6 (or
4.6.1/2/3 etc.) and put that
.tar file inside it. Untar the
file as shown in link.
Follow all the procedure as per the
link. Use nohup <command> & to
track the logs. i.e. nohup make
clean all & followed by tail -f
nohup.out
If some error comes, it means some
package is missing. Mostly those
package will be present in your
current gcc version. You can
install them there itself. For
example, in my case zlib was
missing. I ran sudo apt-get install
zlib1g-dev libssl-dev and it worked
fine. Otherwise download from internet and install it.
Once your gcc is installed, you
can simply check it using type
gcc-4.6. In my case it showed that
it's stored as
/usr/local/bin/g++-4.6.
Either you can use the same path to
compile or you can put an alias in
your bash/tcsh/ksh. e.g.
/usr/local/bin/g++-4.6 -std=c++0x
-Wall test.cpp
FWIW Debian testing and unstable have gcc-4.6 as a standard package. So you can simply install that distro inside of virtualbox or, as I've done on my Ubuntu 11.04 server at home, via kvm. In the past, I also used to use dchroot build environments.
There may also be prepackaged gcc-4.6 binaries at launchpad.

Resources