-
Notifications
You must be signed in to change notification settings - Fork 60
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix a bunch of warnings, add initial Github CI #167
Conversation
Rebased, pulling out the unrelated av1 patch. |
913fa2a
to
76bfcf9
Compare
The correct incantation of detecting |
Thanks for the patch, that's a lot of clean-up work! Looks like it's conflicting with your other PRs now that they've been merged. Can you get that cleared up? Then I can get it merged in. |
Signed-off-by: Emil Velikov <[email protected]>
The compiler throws a bunch of -Wchar-subscripts, since the default sign of char is architecture specific. Adjust the declaration to fix that. Signed-off-by: Emil Velikov <[email protected]>
Compiler throws a lovely warning about it and none of the existing drivers handle/support it - neither Mesa, i965 nor iHD Signed-off-by: Emil Velikov <[email protected]>
Copy the unreachable() macro from Mesa and use it to silence the compiler warnings. Signed-off-by: Emil Velikov <[email protected]>
There are a two disabled codecs, which are partially commented out. Thus the compiler rightfully complains about a bunch of unused-functions. Just throw these into another segment and drop the used/retain attributes. Thus if the compiler can decide to nuke them, yet it won't flag (m)any warnings. Plus in case they're retained (building without optimisations), it'll be impossible to pick them by mistake. Signed-off-by: Emil Velikov <[email protected]>
This enables some basic compiler warnings, all of which have been addressed with previous commits. Signed-off-by: Emil Velikov <[email protected]>
lastSliceParamsCount is unsigned, yet we've been comparing it against signed ints, while also initializing with signed int (buffer), which in itself should be unsigned ... just copy unsigned a few times until the compiler is (rightfully) happy. Signed-off-by: Emil Velikov <[email protected]>
Bear in mind there are a ton of unused-argument warnings, many of which noise so the warning is disabled. Signed-off-by: Emil Velikov <[email protected]>
Signed-off-by: Emil Velikov <[email protected]>
Annotate the local functions as static and enable the respective warnings. Signed-off-by: Emil Velikov <[email protected]>
... and drop initialization to O/NULL - it's the default already. Signed-off-by: Emil Velikov <[email protected]>
Enable the warning and the respective attribute. Thus the compiler can properly detect Wformat issues. Signed-off-by: Emil Velikov <[email protected]>
... and enable the warnings in meson. Signed-off-by: Emil Velikov <[email protected]>
Signed-off-by: Emil Velikov <[email protected]>
Pointer arithmetic isn't portable nor compliant C. Signed-off-by: Emil Velikov <[email protected]>
Quite noisy and not particularly useful warning - disable it. Signed-off-by: Emil Velikov <[email protected]>
Signed-off-by: Emil Velikov <[email protected]>
The latter ensures our actions are up-to date, while the former checks that we build successfully without any errors on Ubuntu 22.04 with gcc and clang-15. Heavily copied from libva. Note: the libffmpeg-nvenc-dev && libgstreamer-plugins-bad1.0-dev dependencies have been removed, since they pull the standalone libunwind-dev which conflicts with the pre-installed llvm one - libunwind-14-dev. As result the VP9 code path isn't built. Signed-off-by: Emil Velikov <[email protected]>
It was a fairly minimal conflict - rebased. Thanks for the prompt replies o/ |
Thanks for that. |
Greetings everyone,
I was looking at some unrelated libva/nvidia-vaapi-driver related issues (regressions/crashes caused by my patches) and noticed that the build is fairly noisy, since I've got bunch of warnings enabled in
CFLAGS
.Here is a handy MR that fixes them all, alongside a CI file on top so that new don't creep-in.