I need to load and run a C module using dlopen() / LoadLibrary(), and be able to review extensive profiling & debugging information of the hosted module only (not the host program), via a live log displayed as part of the host:
How exactly would I go about extracting stack state and profiling info from the hosted module, at runtime? gprof and gdb? And doing so without slowing things down awfully?
Could any of the debugging and profiling information that comes back, be mixed up (i.e. between host and hosted), provided that I compile both the host and hosted modules with the correct, respective compiler switches? I need a completely clean separation of log data i.e. just the hosted module.
Related
I am using MoltenVK on MacBook Air with Apple Silicon. I have set my environment according to instructions on Vulkan Tutorial website but when I try to run my program in Xcode I get the following two error messages from validation layer:
ERROR (Callback: Validation Layer): Loading layer library /usr/local/share/vulkan/explicit_layer.d/../../../lib/libVkLayer_api_dump.dylib
ERROR (Callback: Validation Layer): Loading layer library /usr/local/share/vulkan/explicit_layer.d/../../../lib/libVkLayer_khronos_validation.dylib
I have played around with the settings in vkconfig and added environment variable VK_INSTANCE_LAYERS pointing to list of those two layers that failed to load and VK_LAYER_PATH that points to path /Users/jakub/VulkanSDK/1.3.224.1/macOS/share/vulkan/explicit_layer.d. None of that worked.
Those environment variables only work on macOS if you launch the apps from the command line. Apple does not allow environment variables to be set for GUI apps launched from the desktop, etc. A better way to use layers on macOS is to use VKConfig. You can create a configuration with the layers you want (and there's a nice GUI for tweaking layer parameters). This creates a "metalayer" that the loader will always see and make use without the need for environment variables. Also, make sure you do not link statically to MoltenVK, as this does not use the actual loader and layers will not work (although linking statically is perfectly fine for shipping your apps).
I am building a kiosk application using webkitgtk on the raspberry pi 4.
This application will not be connected to the internet and all the html,css, javascript for the UI are all located on the local filesystem.
I am using buildroot to setup the Linux system, starting with the pi 4 defconfig provided in buildroot.
I have enabled all the packages needed to get webkitgtk running.
Also, the kiosk application has been tested on my desktop, using the same software stack and it works
However, when i try to launch the application on the raspberry pi, a blank page pops up. I have played around with the WebKitWebSettings object associated with my WebKitWebView by enabling local file access. It still shows up a blank screen.
Also included in my pi4 application bundle is a simple gtk3+ application. This launches successfully!
I will really appreciate some pointers as to why this is happening as i have sort of reached a dead end
UPDATE
I enabled the MiniBrowser app that comes with the Webkitgtk package.
Entering the local url, The page does not load. It only gives me a message at the top saying "Successfully downloaded".
It seems to be treating my input as a download
UPDATE 2
After some more experimenting, i was finally able to get webkitgtk working on the pi 4.
The problem seems to originate from using the webkit_web_view_load_uri() api.
It does not seem to recognize my html document as a web page.
I got around it using the webkit_web_view_load_html() call. This included some hacks by first reading in the contents of the html doc into a character buffer, and passing it to webkit_web_view_load_html().
You also have to provide a base path to this function call to be able to resolve all the urls (scripts, css, images etc) in your html document.
Another problem i haven't been able to work around is, SVG images are not loading in webkitgtk. I have used jpg formats and they work. I suspect this my be due to a configuration switch in building webkigtk
It's hard for me to figure out what might be happening without having access to your environment and settings. My gut feeling is that pages are showing blank because perhaps some shared libraries are missing. You can check that with:
$ ldd WebKitBuild/GTK/Release/bin/MiniBrowser
I am using buildroot to setup the Linux system, starting with the pi 4 defconfig provided in buildroot.
There's a buildroot repository for building WPE for RPi. WPE (WebPlatform for Embeded) is like WebKitGTK but doesn't depend on GTK toolkit. Another important difference is that WPE runs natively on Wayland.
If you're interested in having a webapp embedded in a browser running in a device with limited capabilities, WPE is a better choice than WebKitGTK. The buildroot repo for building WPE for RPi is here:
https://github.com/WebPlatformForEmbedded/buildroot
There's is also this very interesting step-by-step guide on how to build WPE for RPi3:
https://samdecrock.medium.com/building-wpe-webkit-for-raspberry-pi-3-cdbd7b5cb362
I'm not sure whether the buildroot recipe would work for RPi4. It seems to work for all previous versions, so you might be stepping in new land if you try to build WPE on RPi4.
If you have an RPi3 available I'd try to build WPE for RPi3 first, and make sure that works. Then try for RPi4.
I have a client server program on C. The program has used libraries such as ffmpeg, libfftw3 etc, now I want to run it on my webpage. I have object files as well as executable files compiled from the C programs. I was searching for some platform which will allow me to run the client-server, just like the way a shell script on terminal allows.
Can anyone please tell me how can I do it.
You can run the program "from your web page" if you rebuild your program using Google's Native Client (NaCl) libraries. https://code.google.com/p/nativeclient/
This builds an executable that is run from Chrome, on the users machine, inside a sandboxed environment. The executable is built and distributed in an intermediate, machine independent format that Chrome coverts into a runtime format when it downloads and runs the app.
The performance is good (a little slower than full speed). The downside is that you are limited to the NaCl libraries. And, you have to run the app with Chrome.
If you want to run this in a client-server configuration, you can have the server run on the same machine or on a different machine.
It's easier to write the server program on your own server machine once and communicate via sockets, otherwise you have to port it to all the machines you want it to run on.
If you are interested in running the server on your local machine, you can use WebSockets to connect simple web pages with HTML and JavaScript to the server.
The right approach depends on what your needs are.
I've installed gdb and gdbserver on an angstrom linux ARM board (with external access), and am trying to get source level debugging of a shared library working from my local machine. Currently, if I ssh into the device, I can run gdb and I am able to get everything working, including setting a breakpoint, hitting it and doing a backtrace.
My problem comes when I try and do the same using gdbserver and running gdb on my host machine in order to accomplish the same thing (eventually I'd like to get this working in eclipse, but in gdb is good enough for the moment).
I notice that when I just use gdb on the server and run "info shared", it correctly loads symbol files (syms read : yes for all), which I'm then able to debug. I've had no such luck doing so remotely, using "symbol-file" or "directory" or "shared". Its obviously seeing the files, but I can't get it load any symbols, even when I specify remote files directly. Any advice on what I can try next?
There are a few different ways for this to fail, but the typical one is for gdb to pick up local files rather than files from the server.
There are also a few different ways to fix this, but the simplest by far is to do this before invoking target remote:
(gdb) set sysroot remote:
This tells gdb to fetch files from the remote system. If you have debug info there (which I gather from your post that you do), then it will all work fine.
The typical problem with this approach is that it requires copying data from the remote. This can be a pain if you have a bad link. In this case you can keep a copy of the data locally and point sysroot at the copy. However, this requires some attention to keeping things in sync.
First run up to main, and then set solib-search-path . Otherwise, gdbserver stops in the dynamic loader, before libraries can be loaded. More details at: Debugging shared libraries with gdbserver
Is it possible to load your WPF application when the machine starts up? Our WPF application is taking 30 seconds on our existing client hardware, and we'd like to "hide" that by having our application startup when the machine is booted. But the fact that our application boots on startup should be invisible to the user.
How would we go about building such a thing? Is there a term to google that will provide a tutorial? It would be nice if there's a WPF specific-one we can grab.
Note: In case there's any moral qualms to this (since I agree that having 20 programs startup when your computer does is frustrating!), we're building software for a specialized industry who will be using their computer to run primarily our applications.
The easiest approach would be
make sure your big libraries are installed in the GAC
write a dummy app that uses (most of) those libs. Actually use them so they get loaded.
run the dummy on startup
make the dummy stop by itself, or keep it alive hidden
This would in no way interfere with the starting of the normal application. The benefit is that the libraries get loaded and jitted. The drawback that your app still needs to load & initialize.
Sasha had a few more tips not listed here, rebasing your modules, leveraging the prefetch facility and compression (UPX, yuk!).