Linus Torvalds shipped Linux 7.3-rc5 today, and Phoronix flagged it with a headline anyone tracking the Oct 18 stable target should notice: "Another Week, Another Large RC" (https://www.phoronix.com/news/Linux-7.3-rc5-Released). If you're staging test builds or penciling in an upgrade window, that phrase raises an obvious question: does a large rc5 mean the release is about to slip, or is this just normal kernel-development noise that resolves itself by rc7?
RC size alone rarely determines the final ship date. What matters more is whether Torvalds signals concern about stability as the cycle heads into its back half. Nothing in the current reporting on 7.3-rc5 indicates he's flagged a delay — the Oct 18 target still stands as of this week's release (https://www.phoronix.com/news/Linux-7.3-rc5-Released). But "large" is doing real work in that headline, and it's worth unpacking what it signals for anyone deciding when to start testing.
What a "Large" RC Actually Means
When a kernel maintainer calls an RC large, they're talking about diff size — the volume of changed lines and touched files relative to a typical RC at that point in the cycle. Early RCs (rc1, rc2) are expected to be big; they absorb the merge window's backlog. By rc5, the diff should be shrinking as the tree settles toward stability, so a large rc5 breaks that expected taper. For more on this, see read about audit ubuntu cloud images for missing packages.
That doesn't automatically mean the kernel is unstable. Large late-cycle diffs commonly trace back to one of three causes: a subsystem maintainer sending a bigger-than-usual batch of queued fixes, a regression that required a broader fix touching multiple drivers, or ordinary driver churn — especially in graphics, networking, or architecture-specific code — landing all at once. None of these are red flags by themselves.
The real red flag is Torvalds explicitly saying he's uneasy about merging so much this late in the cycle. That kind of comment is what has historically preceded an extra RC week, not the diff size in isolation.
The practical distinction for sysadmins is size versus risk. A large diff full of driver updates for hardware you don't run is noise. A large diff touching core scheduler, memory management, or filesystem code is worth reading the actual rc5 announcement for, not just the headline.
What This Means for the Oct 18 Target
The evidence available here doesn't include confirmed RC counts or release dates for the 7.0 through 7.2 cycles, so a literal side-by-side comparison isn't possible from this reporting. What is broadly true of how Torvalds runs a cycle is the shape of the decision: he tends to default to seven release candidates and adds an eighth only when he judges, usually in the rc6 or rc7 announcement, that the tree isn't calm enough to tag stable. That call typically isn't made or confirmed at rc5.
Treat the table below as a planning heuristic drawn from that general pattern, not as verified data tied to specific named 7.x releases:
| Cycle pattern | RC count | What it usually signals | |---|---|---| | Standard cycle | 7 RCs | Stable tags on the Sunday after rc7, as originally planned | | Extended cycle | 8 RCs | Torvalds adds a week, typically citing late regressions or a diff he's not comfortable shipping | | Compressed cycle | 6 RCs | Happens only when the tree is unusually calm; uncommon |
Applied to 7.3, a large rc5 is worth watching, not worth acting on yet. If rc6 returns to a normal size and Torvalds' notes read calm, Oct 18 is very likely to hold. If rc6 is also unusually large, or if Torvalds mentions lingering regressions, that's the point where an rc8 becomes a real possibility and Oct 18 should be treated as tentative.
What to Do This Week
- Keep testing rc5 now. Diff size isn't a reason to wait — testing is what surfaces the regressions that would justify an extra RC in the first place.
- Prioritize hardware-specific configs. Graphics and storage stacks are where late-cycle churn tends to concentrate; that's where problems show up first.
- Watch the rc6 announcement, not the rc5 headline. A return to normal diff size all but confirms Oct 18. Another large or messy rc6, paired with any hesitation from Torvalds, is the real signal the calendar is about to move.
- Hold production upgrades for the tagged stable release. Track rc builds for testing, not deployment, regardless of how calm or chaotic a given RC looks.
Also read: a closer look at wine vs wine-staging: which one should gamers use?
