Running WinDbg in headless mode - c++

Is there a way to WinDbg in pure headless mode ?
My use case is that I should be able to imitate "!analyze -v" command for a minidump on command line without launching WinDbg GUI.

With the Debugging Tools for Windows, there is also cdb, the console debugger.
To debug a crash dump, use the -z "<dump>" option. To run a command immediately, use -c "<command>". To output everything into a file you can redirect the output with a usual DOS > <file> or open a log file using .logopen <file>. To exit CDB after the analysis, use q.
You want to make sure that you have symbols set up correctly, so include a .symfix <path>;.reload. I prefer using .symfix <symbolpath> in the command rather than -y <symbolpath>, so I can always check in the log that it has been set correctly.
Full command line (using redirect)
"<windbgpath>\cdb.exe" -z "<dumpfile>" -c ".symfix <symbolpath>;.reload;!analyze -v;q" > "<logfile>"
Disadvantage of redirect: line ending is LF only. And you get a lot of noise when the debugger starts.
Full command line (using log file)
"<windbgpath>\cdb.exe" -z "<dumpfile>" -c ".logopen <logfile>;.symfix <symbolpath>;.reload;!analyze -v;.logclose;q"

Related

Script for Notepad++ NppExec for C++ in ubuntu

I just switched to ubuntu and I wanted to setup notepad++ for CPP.
So I used the NppExec plugin to compile within notepad++,
My script was :
npp_save
g++ "$(FULL_CURRENT_PATH)" -o "$(CURRENT_DIRECTORY)\$(NAME_PART)obj"
./"$(NAME_PART)obj"
Here the "obj" I used is to just save the file with an "obj" keyword nothing else.
The last line ./"$(NAME_PART)obj" is to run the program.
But it looks not working in ubuntu, it produces this error:
NPP_SAVE: Z:\home\username\cpp\test.cpp
g++ "Z:\home\username\cpp\test.cpp" -o "Z:\home\username\cpp\testobj"
; about to start a child process: "g++ "Z:\home\username\cpp\test.cpp" -o "Z:\home\username\cpp\testobj"
CreatProcess() failed with error code 2:
File not found.
./"testobj"
; about to start a child process: "./"testobj""
CreatProcess() failed with error code 2:
File not found.
I have investigated some of what I think is the problem, so I think is the usage of / and \ in changing the directory.
I don't know how to fix that, so I can not be sure.
Any ideas? :) I am using vim btw in the same machine and it is working perfectly.
In theory it might be possible (see below), in practice it is rather convoluted and works only for simple compiles (like single file hello world type).
I would suggest you try a linux program, e.g.
an editor like
scite (same editing engine as notepad++) or
kate
or a real IDE like
kdeveloper or
qtcreator.
The problems with Notepad++ inside wine and g++ outside wine (from the linux install ) are this:
notepad++ inside wine under linux is still a windows program
NppExec can only do, what a cmd inside wine can do.
starting g++ directly inside cmd is an error due to g++ being a linux binary and not a windows binary
that is your CreatProcess() failed with error code 2, it means: you are trying to execute a linux program inside wine.
That does not work! (At least not so easy.)
Though you can start linux program inside cmd inside wine using start /unix ...
started this way, g++ wants linux paths and NppExec through its variables will provide only windows paths (whatever wine has set up as drives like Z:\home\username\src\hello.cpp)
though you can convert wine paths to linux paths via the winepath -u command.
g++ started through 'start /unix ... ' inside a cmd inside wine has no proper terminal to report errors to you
though you can start an xterm for g++ and have g++ reports its messages to the xterm
the downside is that g++ will report errors using the linux paths in the xterm, so you cannot double click on an error message an get to the corresponding filename and line.
You get the idea: its complicated not comfortable.
What worked for me for a helloword.cpp was this NppExec script:
NPP_SAVE
npp_run cmd /c start /unix /usr/bin/xterm -e "/usr/bin/winepath -u '$(FULL_CURRENT_PATH)' | xargs g++ -o /tmp/a.out && /tmp/a.out ; echo 'Press return'; read"
The second line
uses an xterm,
let winepath convert the Z:\home\... path to /home/... and
have that send to g++ for compilation using /tmp/a.out as binary
if compile is successfull, /tmp/a.out is executed
the echo and read are for keeping the xterm open so that you can read the output.
If you really want to use Notepad++ inside wine, one option might be using Gnu Make outside of wine and have NppExec run make all or make run similar to the g++ in my script example. That would work for more complicated compiles.

How to work in batch mode

I have inherited an ANSI C++ program that: has no GUIs and is supposed to run in batch mode, generating lots of data (we are talking 100,000+ ASCII files). We are thinking that in long term we’ll run it under UNIX. For now, I have a MacBook Air running OS X 10.9.4 and I loaded Xcode 5.1.1. It compiles without errors or warnings.
I need to test a program as follows:
<prompt> myprogram datain dataout1 datout2
Where is the compiled program? In which directory? Can I copy my datain file in that directory?
For repeated execution under Windows (Command Prompt window) I normally would have a batch file of the type:
myprogram datain1 dataout11 datout12
myprogram datain2 dataout21 datout22
myprogram datain3 dataout31 datout32
........
myprogram datainn dataoutn1 datoutn2
Can I do something similar with OS X? Where can I find the applicable documentation?
You will want to look for your terminal emulation program. See http://en.wikipedia.org/wiki/Terminal_(OS_X) for how to use it, and it should be the bash shell which is one of the unix shells
You can also do a shell script see
http://tldp.org/HOWTO/Bash-Prog-Intro-HOWTO.html for some bash shell scripting info
For such a simple operation, you can write a shell script that will look almost exactly the same as the batch file you use on Windows. The key difference between Windows' cmd.exe and *nix shells here is that the current directory is not part of the search path for executables (the way it is on Windows), so if you put the shell script in the same folder as the compiled executable, you will need to prefix the program name with ./ (to mean "look in the current directory"). For example:
#!/bin/sh
./myprogram datain1 dataout11 datout12
./myprogram datain2 dataout21 datout22
./myprogram datain3 dataout31 datout32
........
./myprogram datainn dataoutn1 datoutn2
If the shell script and executable are not in the same folder, you can use either an absolute path or an appropriate relative path.
Also, to run the script you will either need to make it executable:
$ chmod +x myscript.sh
$ ./myscript.sh
or invoke the shell with the script as an argument:
$ sh myscript.sh

How do redirect STDOUT out of XCode compiled Command Line executable?

I have a C++ program that writes to stdout that I am trying to compile into a command line utility in XCode. I need the program compiled into a release build (which it seems, XCode calls an Archive - go figure), not run it as debug.
I basically want to pipe the output of my program into a textfile like this
MyProgram > TextFile.txt
I can compile the program into a command line and run it like this via a Bash shell like this:
Open MyProgram
This spawns a new process (not what I want). However, I could live with that if this worked:
Open MyProgram > TextFile.txt
...But it doesnt :-(. If I try to just run MyProgram directly from Bash, I get the error: -bash: MyProgram: command not found.
What am I doing wrong? How can I compile my command line tool to NOT require the Open command under Mac OSX?
Thanks for any help you can provide. I am picking up C++ on the Mac platform and I am beginning to find it quite it a bit more troublesome than Visual Studio. Does it ever get less painful to work with? :-)
Make your project a Command Line Tool. This will make it so you can run it from the command line directly.
Run it by typing ./MyProgram or ./MyProgram > TextFile.txt not open MyProgram.

What needs to be done to get a distributable program from Eclipse?

I’ve produced a C++ program in Eclipse running on Redhat, which compiles and runs fine through Eclipse.
I thought that to run it separately to Eclipse you use the build artifact which is in the directory set via the project’s properties.
However this executable doesn’t run (I know it’s an executable as I’ve set it to be an executable via the project’s properties and it shows up as such via the ls command and the file explorer).
When attempting to run it using the executable’s name, I get the error:
bash: <filename>: command not found
When attempting to run it as a bash file:
<filename>: <filename>: cannot execute binary file
And when running it with "./" before the file name, nothing happens. Nothing new appears in the running processes and the terminal just goes to the next line as though I’d just pressed enter with no command.
Any help?
You've more or less figure out the first error yourself. when you just run <filename> , it is not in your PATH environment variable, so you get "command not found". You have to give a full or relative path when to the program in order to run it, even if you're in the same directory as the program - you run it with ./<filename>
When you do run your program, it appears to just exit as soon as you start it - we can't help much with that without knowing what the program does or see some code.
You can do some debugging, e.g. after the program just exits run echo $? to see if it exited with a particular exit value, or run your program using the strace tool to see what it does (or do it the usual way, insert printf debugging, or debug it with gdb)

How to dump the entire GDB session to a file, including commands I type and their output?

In bash, I can use the script command, which dumps everything that shows on the shell to a file, including:
commands typed
PS1 line
stdout and stderr of commands
What is the equivalent in gdb?
I tried to run shell script from inside GDB, but after I hit return, I was in the shell and lost the shell prompt and could not run command any more. Moreover I could not use ctrl+c or ctrl+\ to exit. I needed to force kill the /bin/login tty2 to exit.
If you want to log GDB's output, you can use the GDB logging output commands, eg.
set logging file mylog.txt
set logging on
If you want to redirect your program's output to a file, you can use a redirect, eg.
run myprog > mylog.txt
see the chapter on program IO in the GDB manual for more information
Create a text file, i.e. gdbCommands.txt, with the following commands
set logging on my_log_file\nbt 10\nq
bt 10, indicates the number of lines (function calls) we need from the backtrace, in our example is 10 lines.
Execute gdb using the following command, assuming a core dump file core.2345
gdb -x gdbCommands.txt myApp core.2345
Open my_log_file and inspect backtrace!
howto-redirect-gdb-backtrace-output-to-a-text-file
I have logging enabled using:
set trace-commands on
set pagination off
set logging file $log
and show logging reports (to both to terminal and file):
+show logging
Currently logging to mylog.
Logs will be appended to the log file.
Output will be logged and displayed
If I print the value of a variable that also gets logged (to both to terminal and file):
+p myvar
$2 = 0
But if I do command like where or “info b” all I get logged to the file is:
+where
+info b
Anyone know why or how to fix it?
Have a look at the GDB documentation. Search for "Canned Sequences of Commands". There is a way to save GDB commands in a file and run them with the source command and you can use some GDB commands in these scripts to print information available to GDB (like echo, output and printf).
If you want that output in a file, use set logging file FILE.