Thread by @embeddedgus

Taking a look into every fall-through warning is not the most glamorous and sexy thing. But the ongoing efforts to enabling -Wimplicit-fallthrough
keep paying off. :) lore.kernel.org/patchwork/patc…

This bug has been out there for almost two years.

Roses are red,
Violets are blue,
I'm almost done,
With the -Wimplicit-fallthrough 💻🔍

Bugs, bugs, ancient bugs!
Another years-old bug found while working on the -Wimplicit-fallthrough stuff: lore.kernel.org/patchwork/patc…

This one was introduced in January 2012.

Happy to see we continue placing breaks where they should go 😀:
cgit.freedesktop.org/~agd5f/linux/c…

Yay! Another missing break makes it to multiple stable trees:
lwn.net/Articles/78271…

Two more of these bugfixes landed in stable during the weekend 🐧 lwn.net/Articles/78385…

Stephen Rothwell is now building linux-next with -Wimplicit-fallthrough. Newly introduced code that is buggy is detected, and now a bugfix takes hours instead of months or years:
lore.kernel.org/lkml/201904021…

Once -Wimplicit-fallthrough finally makes it to mainline, a bugfix should take seconds or minutes before the code is even submitted; therefore completely eliminating this bug class. 🐧

Another missing break years-old bug. This time in wireless-drivers:
lore.kernel.org/patchwork/patc…

It was introduced in July 2016. 🐧

This is certainly not the most glamorous and sexy thing... but it's worth it:
lore.kernel.org/lkml/8df20a3a-… 🐧

Another missing break bug:
lore.kernel.org/patchwork/patc… 🐧

This one was introduced just a week ago.

I've just sent my first pull-request to Linus. Fingers crossed! 🤞
lore.kernel.org/lkml/201905061… 🐧

"The pull request you sent on Mon, 6 May 2019 ... has been merged into torvalds/linux.git:"
git.kernel.org/torvalds/c/b4b… 🐧😃!!!

And today, we woke up with only 23 of these warnings left in linux-next. This means the work is 99% complete.

A bugfix for a 12-year-old bug has been finally pulled and will be merged into mainline soon:
git.kernel.org/pub/scm/linux/… 🐧

Yep; this bug has been out there since 2007. Briefly explained: the problem is that the code always returns "success" even on failure.

I'm happy that this bugfix made it to the last official 3.18.y release:
lwn.net/Articles/78868… 🐧

Two more of these bugfixes have landed in multiple stable trees:
lwn.net/Articles/78877… 🐧

Quick update:

There are only 5 of 2311 of these warnings left in linux-next. 🐧

All of them have a patch now.
Work progress: 99.78%

Stop introducing fall-through warnings, please. There are many cases in which it's much clearer (hence better for code maintenance) to add breaks instead of letting the code to fall through:
lore.kernel.org/patchwork/patc…

Last week we'd only 2 warnings left in linux-next. Today we've 4

Fingers crossed! 🤞

lore.kernel.org/lkml/858bd3e0-… 🐧

Stress is waiting for Linus to merge your final changes to complete a task you've been working on since the fall of 2017, hours before the merge window closes. So, here's a #Caturday pic for stress relief.

Two more of these bugfixes will land in -stable, soon:

lore.kernel.org/patchwork/patc…
lore.kernel.org/patchwork/patc… 🐧

After almost two years of work, -Wimplicit-fallthrough will be finally globally enabled in Linux v5.3. I'll go grab a beer. Have a great weekend everybody. 🐧
git.kernel.org/pub/scm/linux/…

"The C language allows switch cases to "fall through" when a "break" statement is missing at the end of a case. This, however, introduces ambiguity in the code, as it's not always clear if the missing break is intentional or a bug..." 🐧
kernel.org/doc/html/lates…

"One source of kernel bugs has been closed, a switch that can only be seen as a best-case scenario.". 🐧
lwn.net/SubscriberLink…

Friday night, casually building ARCH=sh for the first time, I found a 10-year-old bug thanks to -Wimplicit-fallthrough:
lore.kernel.org/patchwork/patc… 🐧

This is the perfect way to officially start my weekend. Cheers! 🍻

Happy to see other people finding more bugs (this time in ARM64) thanks to -Wimplicit-fallthrough:
git.kernel.org/pub/scm/linux/… 🐧

This bug was introduced in December 2017. It's fixed in mainline now and will be applied to 4.17+ stable kernels, soon.

This made my day:
lore.kernel.org/linux-iio/2019… 🐧

Mission accomplished. :)

Applied to mainline now:
git.kernel.org/pub/scm/linux/… 🐧

Kernel CI build for yesterday's linux-next (next-20190820):
lore.kernel.org/linux-next/5d5…
70 fall-through warnings.

Today's linux-next (next-20190821):
lore.kernel.org/linux-next/5d5…
0 fall-through warnings on x86_64, i386, arm, arm64, mips, riscv, arc and sh. 😀🐧

And yet another years-old bug is still out there in the wild. Hopefully, the bugfix will make it to mainline and land in stable, soon:
lore.kernel.org/lkml/201909090… 🐧

This bug was introduced in 2015.
ARCH=powerpc
mpc83xx_defconfig

Happy to see this bugfix finally landing in mainline. It should be applied to multiple stable trees, soon:
git.kernel.org/pub/scm/linux/… 🐧

This bugfix has now been applied to stable trees 5.3.2, 5.2.18 and 4.19.76:
lwn.net/Articles/80101… 🐧

As commented at @KernelRecipes a couple of weeks ago, 'fallthrough' pseudo-keyword will replace all /* fallthrough */ markings. So, yeah, we are now forking C to make the kernel more secure.😃

lore.kernel.org/lkml/CAHk-=whO… 🐧

u-boot is about to globally enable -Wimplicit-fallthrough:

lists.denx.de/pipermail/u-bo…

and that's great news! 😃

I don't think I've documented this fix for a "missing continue" bug in this thread before, so here I go:
git.kernel.org/linus/2c930e3d… 🐧

The bug was introduced in August 2014 by commit
65c7843c942d0 ("usbtest: Add interrupt EP testcases").
The bugfix was applied to -stable in 2017.

The bugfix mentioned above was part of a small series:
mail-archive.com/linux-kernel@v… 🐧
mail-archive.com/linux-kernel@v… 🐧

At the time I was new to kernel development and I wasn't into the good practice of adding the "Fixes:" and "stable" tags to the changelog text.

Another fix (I haven't documented before) for a bug that was found thanks to the efforts to enable -Wimplicit-fallthrough:

"net: sched: cls_api: fix dead code in switch"
git.kernel.org/linus/2ed9db30… 🐧