[Coastrnaut]

Fixes #974

Texture data passed to projectM, e.g. from the texture load callback added in 4.2, is tightly packed, but GL_UNPACK_ALIGNMENT defaults to 4. For RGB (or 1/2 channel) images whose row size isn't a multiple of 4 bytes, the driver expects padded rows: the texture is skewed and glTexImage2D reads past the end of the application's buffer. That's the crash in Texture::Texture() → _mesa_TexImage2D → memcpy from #974, which only happens with some cover images, depending on their width.

Texture::Update(), the only place projectM uploads pixel data, now sets the alignment to 1 for the upload and restores the previous value afterwards, so the application's GL state is left alone.

Testing

Windows, RTX A6000, GL 3.3 core. A small test app registers a texture load callback that returns a tightly packed image placed directly before a PAGE_NOACCESS page, so any read past its end crashes, and a preset that draws sampler_cover.

Image	master	this PR
501×500, 3 channels	crash	OK
333×200, 3 channels	crash	OK
257×129, 1 channel	crash	OK
500×500, 3 channels	OK	OK
501×500 / 333×200, 4 channels	OK	OK

With this PR the 501×500 and 333×200 RGB textures render identically to the same images given as RGBA (max pixel difference 0).

Backwards compatibility: I loaded all 14,536 presets from cream-of-the-crop, en-d, milkdrop-original and projectm-classic, with the preset folders as texture search path (loading the textures changed the output of ~7,000 presets, so image loading was exercised), and rendered 30 frames each with fixed audio, frame times and random seeds (the seeds were a local patch only, identical on both sides). Shader compile/link results, failed loads and the rendered frames are identical to master for every preset. All textures projectM creates itself are RGBA, where the alignment doesn't matter.

Unit tests pass. There's no new unit test because the test suite doesn't create a GL context.

Retested on current master (2026-11-12)

• Unit tests: 164/164 libprojectM and 98/98 playlist tests pass.
• Presets: I loaded all 14,536 presets from cream-of-the-crop, en-d, milkdrop-original and projectm-classic one after another with this branch and with master (GL 3.3 core, RTX A6000, preset folders as texture search path, 30 frames each with fixed audio, frame times and random seeds; the seeds are a local patch only, identical on both sides). Shader compile/link results, failed loads and all rendered frames are identical to master.
• Together with my other PRs: All my open PRs (Fix calling convention of glGetStringi pointer in GLProbe #1059 to Build for Windows Store (UWP) without app-side workarounds #1065, Return int from the NaN-safe multiplication of two ints #1067 to Fix shaders that declare or name samplers through macros #1072, Build Apple frameworks for iOS, tvOS, watchOS and visionOS, and support EAGL contexts #1075, Don't shift the blur textures by Direct3D's half-pixel offset #1078 and Render with 10 bits per color channel if the target framebuffer has more than 8 #1079) merge with each other and with master without conflicts. Merged together (all except Render with 10 bits per color channel if the target framebuffer has more than 8 #1079, which I opened after this run), they build on desktop GL and GLES, all unit tests pass, and the corpus gives 0 failed loads, 0 shader compile/link failures and 0 GL errors in an OpenGL ES context, with and without textures.