SDL_BlitSurface segfault

[Starbuck5]

I've been chewing on the SDL 2.29.1 pre-release for a few days now trying to figure out some failures and segfaults in the pygame-ce test suite.

I found that one of our image saving routines was segfaulting, and I tracked it down to SDL_BlitSurface and this commit bb969ac

Here is a minimum reproducible example (this segfaults for me):

SDL_Surface* surface = SDL_CreateRGBSurfaceWithFormat(0, 1, 1, 32, SDL_PIXELFORMAT_ARGB8888);
SDL_Surface* linebuf = SDL_CreateRGBSurfaceWithFormat(0, 1, 1, 32, SDL_PIXELFORMAT_ARGB8888);
SDL_Rect srcrect = {0, 0, 1, 1};
SDL_BlitSurface(surface, &srcrect, linebuf, NULL);

This code lacks a dest rect, and SDL_UpperBlit now sends that NULL through into SDL_LowerBlit, where it didn't previously.

[1bsyl]

I see, this should be
return SDL_LowerBlit(src, &r_src, dst, &r_dst);
I'll fix that

[1bsyl]

done.
for SDL3, that was actually ok 3639743
issue, was when doing the merge probably.

[1bsyl]

@Starbuck5 please give a try

[oddbookworm]

@1bsyl I'm not @Starbuck5, but I just tried with the SDL 2.29.2 prerelease, and it seems to be working correctly now. The tests that were segfaulting before no longer do so with that prerelease. I've got a draft pull request up that's running through the entire test suite on all of our platforms, so when those all finish I can update this comment with any findings

Update: looks fine so far

Pull Request for reference

Update to the update:
New pull request that has the permissions fixed
All of our unit tests pass, so it looks good!

[Starbuck5]

Thanks for the quick action @1bsyl, I've tested your patch and confirmed it works. (no longer segfaults)

Thanks @oddbookworm for checking up on this too :)