update to ffmpeg 7.1 by elv-peter · Pull Request #141 · eluv-io/avpipe

elv-peter · 2025-11-08T03:26:52Z

https://github.com/eluv-io/avpipe/commits/peter/ffmpeg-7.1 is too messy due to merges from master

For this branch, keep all commits rebased on top of master.

… error transcoding from MPEG-TS

…quick fix for mp4e tests

Copilot

Pull Request Overview

This PR migrates the codebase to use newer FFmpeg APIs, removing deprecated functions and updating to modern channel layout handling. The changes address deprecation warnings and prepare the code for compatibility with recent FFmpeg versions.

Key Changes

Removed libavresample dependency in favor of libswresample
Migrated from deprecated channel layout APIs (channel_layout, channels) to the new AVChannelLayout struct with ch_layout
Updated I/O callback signatures to use io_close2 instead of io_close and added const qualifiers
Replaced deprecated struct fields: frame_number → frame_num, pkt_duration → duration, coded_picture_number removed
Removed deprecated avcodec_close() calls (now handled by avcodec_free_context())
Improved error handling in MP4 validation to allow partial parsing
Disabled flaky HLS tests and updated test data paths

Reviewed Changes

Copilot reviewed 19 out of 19 changed files in this pull request and generated 12 comments.

Show a summary per file

File	Description
rules.make	Removed libavresample from pkg-config dependencies
mp4e/mp4_test.go	Updated test data paths to use cmd/mez-validator/testdata
mp4e/mp4.go	Improved error handling to support partial MP4 parsing
live/lhr_tool_test.go	Disabled three flaky HLS tests due to unavailable manifest URLs
libavpipe/src/avpipe_xc.c	Migrated to new channel layout API, updated io_close2, replaced deprecated frame fields, added delay_moov flag
libavpipe/src/avpipe_utils.c	Updated dump functions to use ch_layout and frame_num
libavpipe/src/avpipe_mux.c	Changed io_close signature to return int and use io_close2
libavpipe/src/avpipe_io.{h,c}	Updated elv_io_close signature to return int for io_close2 compatibility
libavpipe/src/avpipe_format.c	Replaced deprecated fields: channels → ch_layout.nb_channels, pkt_duration → duration
libavpipe/src/avpipe_filters.c	Migrated audio filter channel layout handling to new API
libavpipe/src/avpipe_copy_mpegts.c	Updated MPEG-TS encoder channel layout handling
libavpipe/include/avpipe_xc.h	Changed AVCodec pointer to const, updated writer callback signature
exc/elv_xc.c	Added const qualifier to writer functions, migrated channel layout parsing
exc/elv_mux.c	Added const qualifier to writer functions
avpipe_test.go	Added profile and level parameters to TestABRMuxing
avpipe_errors.go	Removed libavresample cgo directive
avpipe.go	Removed libavresample, added channel layout helper, implemented ChannelLayout with proper memory management
avpipe.c	Updated writer callback signatures with const qualifiers

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-11-08T03:31:07Z

 func TestParseSegment(t *testing.T) {
-	f := "testdata/vfsegment.mp4"
+	//f := "testdata/vfsegment.mp4" // something wrong with this file
+	f := "../cmd/mez-validator/testdata/video-mez-segment-1.mp4"


Brittle test data path dependency. The test now depends on files in ../cmd/mez-validator/testdata/ which creates a fragile cross-package dependency. If the cmd/mez-validator package is moved or its testdata reorganized, this test will break. Consider either: 1) copying necessary test files to mp4e/testdata/, 2) using a relative path that's less likely to break, or 3) documenting this dependency clearly if it's intentional.

Copilot · 2025-11-08T03:31:07Z

 // TODO validate the segments together as one file
 func TestCheckSegments(t *testing.T) {
-	directory := "testdata"
+	directory := "../cmd/mez-validator/testdata"


Brittle test data path dependency. Similar to TestParseSegment, this test now depends on ../cmd/mez-validator/testdata/ which creates a fragile cross-package dependency. Consider the same alternatives: copying test files locally, using a more stable path, or documenting the dependency.

Copilot · 2025-11-08T03:31:08Z

    int64_t h = *((int64_t *)(inctx->opaque));
    int64_t fd = *(int64_t *)outctx->opaque;
-    int bwritten = AVPipeWriteOutput(h, fd, buf, buf_size);
+    int bwritten = AVPipeWriteOutput(h, fd, (uint8_t *)buf, buf_size);


Type casting away const qualifier. The function signature was changed to accept const uint8_t *buf but here it's cast back to uint8_t *. This defeats the purpose of the const correctness improvement and could lead to undefined behavior if AVPipeWriteOutput modifies the buffer. Either change AVPipeWriteOutput to accept const uint8_t * or reconsider the const qualifier in the signature.

Suggested change

int bwritten = AVPipeWriteOutput(h, fd, (uint8_t *)buf, buf_size);

int bwritten = AVPipeWriteOutput(h, fd, buf, buf_size);

Copilot · 2025-11-08T03:31:08Z

    xcparams_t *xcparams = (inctx != NULL) ? inctx->params : NULL;
    int64_t fd = *(int64_t *)outctx->opaque;
-    int bwritten = AVPipeWriteMuxOutput(fd, buf, buf_size);
+    int bwritten = AVPipeWriteMuxOutput(fd, (uint8_t *)buf, buf_size);


Type casting away const qualifier. Similar to line 647, the const qualifier is being cast away. This defeats the purpose of the const correctness improvement in the function signature. Either AVPipeWriteMuxOutput should accept const uint8_t * or reconsider the const qualifier.

Suggested change

int bwritten = AVPipeWriteMuxOutput(fd, (uint8_t *)buf, buf_size);

int bwritten = AVPipeWriteMuxOutput(fd, buf, buf_size);

Copilot

Pull Request Overview

Copilot reviewed 19 out of 19 changed files in this pull request and generated 4 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-11-12T03:42:18Z

+		goavpipe.Log.Error("InSeeker SEEK_END", "offset", offset, "whence", whence, "input_size", h.input.Size(), "about_to_call_seek", true)
+	}
+
+	// FFmpeg 7.1: Handle AVSEEK_SIZE (0x10000 = 65536) - return file size directly
+	if int(whence) == 65536 { // AVSEEK_SIZE
+		size := h.input.Size()
+		goavpipe.Log.Error("InSeeker AVSEEK_SIZE", "offset", offset, "whence", whence, "returning_size", size)
+		return size, nil
+	}
+
 	n, err := h.input.Seek(int64(offset), int(whence))
+	if int(whence) == 2 { // io.SeekEnd
+		goavpipe.Log.Error("InSeeker SEEK_END result", "offset", offset, "whence", whence, "returned_pos", n, "error", err)


Using Log.Error() for debug/trace messages is incorrect. Lines 269, 275, and 281 log normal control flow information at ERROR level. These should use Log.Debug() or Log.Trace() instead, as they're informational messages for debugging FFmpeg 7.1 behavior, not actual errors.

Suggested change

goavpipe.Log.Error("InSeeker SEEK_END", "offset", offset, "whence", whence, "input_size", h.input.Size(), "about_to_call_seek", true)

}

// FFmpeg 7.1: Handle AVSEEK_SIZE (0x10000 = 65536) - return file size directly

if int(whence) == 65536 { // AVSEEK_SIZE

size := h.input.Size()

goavpipe.Log.Error("InSeeker AVSEEK_SIZE", "offset", offset, "whence", whence, "returning_size", size)

return size, nil

}

n, err := h.input.Seek(int64(offset), int(whence))

if int(whence) == 2 { // io.SeekEnd

goavpipe.Log.Error("InSeeker SEEK_END result", "offset", offset, "whence", whence, "returned_pos", n, "error", err)

goavpipe.Log.Debug("InSeeker SEEK_END", "offset", offset, "whence", whence, "input_size", h.input.Size(), "about_to_call_seek", true)

}

// FFmpeg 7.1: Handle AVSEEK_SIZE (0x10000 = 65536) - return file size directly

if int(whence) == 65536 { // AVSEEK_SIZE

size := h.input.Size()

goavpipe.Log.Debug("InSeeker AVSEEK_SIZE", "offset", offset, "whence", whence, "returning_size", size)

return size, nil

}

n, err := h.input.Seek(int64(offset), int(whence))

if int(whence) == 2 { // io.SeekEnd

goavpipe.Log.Debug("InSeeker SEEK_END result", "offset", offset, "whence", whence, "returned_pos", n, "error", err)

Copilot · 2025-11-12T03:42:19Z

+    // FFmpeg 7.1: Use size callback to fake stream size instead of deprecated 'written' field
+    // This tells FFmpeg the total size of the stream for avio_size() calls
+    if (inctx->sz > 0) {
+        avioctx->seek = in_handlers->avpipe_seeker;  // Ensure seeker handles SEEK_END for size
+    }


The comment says 'Use size callback to fake stream size' but the code is setting avioctx->seek, not implementing a size callback. The avioctx->seek is likely already set earlier, and this doesn't replace the deprecated avioctx->written behavior. The proper approach would be to ensure the seeker callback handles AVSEEK_SIZE (whence=0x10000) as done in avpipe.go:273.

Copilot · 2025-11-12T03:42:19Z

    int64_t h = *((int64_t *)(inctx->opaque));
    int64_t fd = *(int64_t *)outctx->opaque;
-    int bwritten = AVPipeWriteOutput(h, fd, buf, buf_size);
+    int bwritten = AVPipeWriteOutput(h, fd, (uint8_t *)buf, buf_size);


Casting away const (from 'const uint8_t *' to 'uint8_t *') violates the const contract and could lead to undefined behavior if AVPipeWriteOutput modifies the buffer. Either update AVPipeWriteOutput's signature to accept 'const uint8_t *', or document why this cast is safe.

Suggested change

int bwritten = AVPipeWriteOutput(h, fd, (uint8_t *)buf, buf_size);

int bwritten = AVPipeWriteOutput(h, fd, buf, buf_size);

Copilot · 2025-11-12T03:42:19Z

 4. OutputHandler: is the output handler with Write/Seek/Close methods. An implementation of this
    interface is needed by ffmpeg to write encoded streams properly.
+
+TODO: Call C.free for every C.CString to not leak memory.


This TODO indicates a known memory leak issue where C.CString allocations are not being freed throughout the codebase. While line 983 correctly uses 'defer C.free(unsafe.Pointer(cName))', this suggests there may be other locations with leaks. This should be tracked and addressed systematically.

Suggested change

TODO: Call C.free for every C.CString to not leak memory.

// TODO [MEMORY LEAK]: For every allocation with C.CString, immediately defer C.free(unsafe.Pointer(...)) to avoid memory leaks.

elv-peter · 2026-04-01T22:05:06Z

Moving to ffmpeg 8 instead

elv-peter added 5 commits November 7, 2025 18:59

Fix build errors after updating FFmpeg from version 4.4 to 7.1

c34a3a2

Fix use of AV_CHANNEL_ORDER_UNSPEC

bf27221

fix setting of channel layout on audio filters; add delay_moov to fix…

8c5f7e6

… error transcoding from MPEG-TS

disable live unit tests failing due to a source stream that is down; …

5edb79b

…quick fix for mp4e tests

fix goavpipe move related error

9fb928e

elv-peter requested review from JanFalkin, Copilot and elv-serban November 8, 2025 03:26

Copilot AI reviewed Nov 8, 2025

View reviewed changes

elv-peter mentioned this pull request Nov 11, 2025

Fix build errors after updating FFmpeg from version 4.4 to 7.1 #122

Closed

JanFalkin and others added 2 commits November 11, 2025 16:43

fix in seeker getting incorrect size

5e66f2c

fixes for PR - mainly handling of delay_moov flag and encoder setup

c69966f

elv-peter requested a review from Copilot November 12, 2025 03:39

Copilot started reviewing on behalf of elv-peter November 12, 2025 03:40 View session

Copilot finished reviewing on behalf of elv-peter November 12, 2025 03:41

Copilot AI reviewed Nov 12, 2025

View reviewed changes

change ERROR logs to DEBUG

c0243d2

elv-peter closed this Apr 1, 2026

	int bwritten = AVPipeWriteOutput(h, fd, (uint8_t *)buf, buf_size);
	int bwritten = AVPipeWriteOutput(h, fd, buf, buf_size);

	int bwritten = AVPipeWriteMuxOutput(fd, (uint8_t *)buf, buf_size);
	int bwritten = AVPipeWriteMuxOutput(fd, buf, buf_size);

	TODO: Call C.free for every C.CString to not leak memory.
	// TODO [MEMORY LEAK]: For every allocation with C.CString, immediately defer C.free(unsafe.Pointer(...)) to avoid memory leaks.

Conversation

elv-peter commented Nov 8, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Key Changes

Reviewed Changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Uh oh!

Copilot AI Nov 12, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 12, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 12, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 12, 2025

Choose a reason for hiding this comment

Uh oh!

elv-peter commented Apr 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants