Fix 32-bit overflow in AudioStreamInfo::frames_to_microseconds

[kahrendt]

Summary

frames_to_microseconds computed frames * US_PER_SECOND as a uint32 * uint32 product, which overflows above ~4295 frames (~97 ms at 44.1 kHz). Most callers dodged this by pre-reducing the frame count, but the hard-sync drop path in SyncTask passes a frame count bounded only by the decode buffer size, so a large drop could silently corrupt decoded_timestamp precisely during resync.

Changes

• Widen the intermediate multiply to 64-bit and return int64_t, making the conversion exact for any frame count.
• This removes the need for the two-step remainder decomposition that callers used purely to dodge the overflow:
  • Delete frames_to_milliseconds_with_remainder (and its gcd helper and the ms_sample_rate_gcd_ member).
  • Collapse the two call sites in SyncTask (track_sent_audio and the unplayed-frames calc in process_playback_progress) to a single frames_to_microseconds() call with identical numeric results: the decomposition was a pure overflow-dodge, not a precision trick.
  • The hard-sync drop site is now correct for large drops with no further change.

Performance note (32-bit targets)

The only added cost is a 64-bit/32-bit software divide (__udivdi3) in place of the previous all-32-bit arithmetic. This runs at audio-chunk granularity (~tens of Hz), not per-sample, so the absolute cost is negligible (well under 0.01% CPU). The genuinely hot hand-optimized path (the client/time formatter) is unaffected.

Testing

Verified the library and both host examples build cleanly. Behavioral test coverage, a large-frame case pinning this fix, will be a follow-up unit-test-infrastructure PR.

[Copilot]

Pull request overview

Fixes an overflow bug in AudioStreamInfo::frames_to_microseconds by widening the intermediate arithmetic, ensuring correct timestamp/playtime math when handling larger frame counts (notably in SyncTask’s hard-sync drop/resync paths).

Changes:

• Change frames_to_microseconds to use a 64-bit intermediate multiply and return int64_t.
• Remove the milliseconds + remainder decomposition helper and simplify SyncTask playtime computations to a single frames_to_microseconds() call.
• Remove now-unused gcd helper and related state.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
src/sync_task.cpp	Simplifies playtime accumulation to rely on the widened frames_to_microseconds() conversion.
src/audio_stream_info.h	Updates API signature/docs for frames_to_microseconds() and removes the milliseconds-with-remainder helper + state.
src/audio_stream_info.cpp	Implements the 64-bit-safe conversion and removes the old gcd-based decomposition logic.

---

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