macos: copy rows of another stride one by one (Kitedesk T-4204)

FlutterRgbaRendererPluginOnRgba gets rows padded to stride_align, and the
pixel buffer has its own bytes per row ((width * 4 + 63) & ~63 on Apple
Silicon). The frame was copied in one memcpy of len bytes, so rows came out
shifted whenever the two differ: Kitedesk sends stride_align 1 for a session
drawn in software, whose "all displays" view still uses textures, and any
width that is not a multiple of 16 then shears (1366, 1001, 1279, 3838). A
stride above the pixel buffer's wrote past its data.

Rows are now copied one by one, width * 4 bytes each, when the strides
differ, and in one memcpy of the rows otherwise. A frame shorter than its
rows promise is dropped instead of being read past its end.

Tests: rows_follow_the_frame_stride and short_frames_are_dropped pass; the
other eight that failed still do.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
yunchan 2026-11-15 09:44:38 +09:00
parent 9a377b6155
commit 2c334da6f9

View file

@ -4,6 +4,9 @@
//
// Created by kingtous on 2023/2/17.
//
// Modified by the Kitedesk fork (2026, T-4204): rows are copied one by one
// when their stride differs from the pixel buffer's.
//
import Foundation
import FlutterMacOS
@ -34,6 +37,12 @@ import CoreVideo
}
private func _markFrameAvaliable(buffer: UnsafePointer<UInt8>, len: Int, width: Int, height: Int, stride_align: Int) -> Bool {
// The rows of buffer are padded to stride_align.
let row = width * 4
let stride = stride_align > 1 ? (row + stride_align - 1) / stride_align * stride_align : row
guard width > 0 && height > 0 && len >= stride * height else {
return false
}
var pixelBufferCopy: CVPixelBuffer?
// macOS only support 32BGRA currently.
let dict: [String: Any] = [
@ -50,7 +59,14 @@ import CoreVideo
CVPixelBufferLockBaseAddress(pixelBufferCopy!, [])
let ptr = CVPixelBufferGetBaseAddress(pixelBufferCopy!)!
memcpy(ptr, buffer, len)
let bytesPerRow = CVPixelBufferGetBytesPerRow(pixelBufferCopy!)
if (bytesPerRow == stride) {
memcpy(ptr, buffer, stride * height)
} else {
for y in 0..<height {
memcpy(ptr + y * bytesPerRow, buffer + y * stride, row)
}
}
CVPixelBufferUnlockBaseAddress(pixelBufferCopy!, [])
self.data = pixelBufferCopy
self.width = width