Growth statistics for the 7.2 kernel [LWN.net]

This article delivered to you by LWN subscribers

Subscribers to LWN.web made this text — and the whole lot that
surrounds it — potential. If you respect our content material, please
buy a subscription and make the subsequent
set of articles potential.

By Jonathan Corbet
August 17, 2026

Linus Torvalds released
the 7.2 kernel on August 17, after noting that the variety of fixes
coming in was nonetheless “larger than I’d have wished for“. In reality,
7.2 was one of many busiest improvement cycles within the kernel’s historical past,
including practically 600,000 traces of code. It’s time to take a look at some statistics
to get a deal with on how the kernel’s improvement neighborhood is altering.

There have been 16,418 non-merge commits introduced into the mainline for
the 7.2 launch, exceeding the (already busy) 7.1 launch; it’s, the truth is,
the second-busiest launch within the kernel’s historical past. Only the 6.7 release introduced in additional commits however, whereas
6.7 was dominated by the merging of the bcachefs improvement historical past, the
quantity for 7.2 is extra broad-based.

These commits have been contributed by 2,652 builders, a giant improve
over the earlier report (2,479 in 7.1). If one seems on the historical past of
developer involvement because the starting of the Git period, the result’s:

(This plot, like a lot of the data right here, comes from the subscriber-only LWN Kernel Source Database). There has been a transparent
pattern over that complete historical past: the variety of builders contributing to
every launch will increase over time. But the slope of that line has clearly
modified.

The most energetic builders engaged on 7.2 have been:

Most energetic 7.2 builders

By changesets Uwe Kleine-König 151 0.9% Rosen Penev 146 0.9% Ian Rogers 138 0.8% Johan Hovold 133 0.8% Arnaldo Carvalho de Melo 131 0.8% Chuck Lever 129 0.8% Eric Dumazet 121 0.7% Ville Syrjälä 115 0.7% Johannes Berg 112 0.7% Krzysztof Kozlowski 110 0.7% Thorsten Blum 108 0.7% Namjae Jeon 106 0.6% Christoph Hellwig 104 0.6% Sean Christopherson 100 0.6% Jakub Kicinski 96 0.6% SJ Park 94 0.6% Mike Rapoport 91 0.6% Dmitry Baryshkov 87 0.5% Eric Biggers 86 0.5% Rafael J. Wysocki 78 0.5%

By modified traces Matthew Stewart 83587 8.4% Hawking Zhang 62893 6.3% Miguel Ojeda 38979 3.9% Ian Rogers 31949 3.2% Darren Ye 20843 2.1% Ethan Nelson-Moore 16893 1.7% Harry Wentland 15233 1.5% Uwe Kleine-König 13803 1.4% Ingo Molnar 13649 1.4% Chuck Lever 10172 1.0% Eric Biggers 10067 1.0% Jakub Kicinski 9449 0.9% Rodrigo Siqueira 7749 0.8% Johannes Berg 7110 0.7% Arnd Bergmann 7085 0.7% Emil Tsalapatis 6810 0.7% Vivek Aknurwar 6544 0.7% Rob Clark 6490 0.6% Sabrina Dubroca 6401 0.6% Svyatoslav Ryhel 6072 0.6%

Uwe Kleine-König [KSDB]
was the most important contributor of commits to the 7.2 launch with cleanups
by way of the driving force subsystem. Rosen Penev [KSDB] contributed an extended
checklist of LLM-authored bug fixes all through the architecture-specific,
driver, networking, and crypto subsystems, amongst others. Ian Rogers [KSDB] labored extensively on
the user-space perf instrument, as did Arnaldo Carvalho de Melo [KSDB]. Johan Hovold [KSDB] made quite a few
enhancements, largely within the SPI and USB subsystems.

Matthew Stewart [KSDB]
reached the highest of the lines-changed column by following the time-honored
custom of including extra large, machine-generated amdgpu header recordsdata;
Hawking Zhang [KSDB] did
likewise. Miguel Ojeda [KSDB] introduced the Rust zerocopy
crate into the kernel supply. Darren Ye [KSDB] added the MediaTek mt8196
sound driver.

Just over 8% of the commits in 7.2 carried Tested-by tags, whereas 48%
included Reviewed-by tags. The overview tags, specifically, are exhibiting a
delicate downward pattern in current releases:

Whether that decline displays an precise lower within the quantity of overview
happening is unclear at finest. In any case essentially the most energetic testers and
critiques for this cycle have been:

Test and overview credit in 7.2
Tested-by Dan Wheeler 117 7.0% Luiz Capitulino 54 3.2% James Clark 44 2.6% Sarthak Sharma 44 2.6% Venkat Rao Bagalkote 32 1.9% Tommaso Merciai 27 1.6% David Riley 27 1.6% Jie Gan 27 1.6% Geert Uytterhoeven 25 1.5% Arthur Kiyanovski 25 1.5% Marek Szyprowski 25 1.5% Randy Dunlap 24 1.4% Shivaprasad G Bhat 24 1.4% Rinitha S 21 1.3% Vitaly Prosyak 20 1.2%	Reviewed-by Konrad Dybcio 231 2.2% Dmitry Baryshkov 225 2.1% Andy Shevchenko 175 1.7% Geert Uytterhoeven 171 1.6% Krzysztof Kozlowski 152 1.5% Frank Li 147 1.4% Ilpo Järvinen 145 1.4% Christoph Hellwig 143 1.4% Jan Kara 121 1.2% Christian König 120 1.1% Linus Walleij 116 1.1% Alex Deucher 105 1.0% Simon Horman 102 1.0% Jeff Layton 101 1.0% Ido Schimmel 94 0.9%

Dan Wheeler [KSDB] is, as
all the time, the highest tester by advantage of his work with AMD graphics-driver
patches. On
the overview facet, Konrad Dybcio [KSDB] added
his tag to 231 adjustments whereas Dmitry Baryshkov [KSDB] tagged
225; the identical two, who each are inclined to give attention to drivers and devicetree recordsdata
for Qualcomm units, have been on the prime of that checklist for the 7.1 cycle as properly.

Work on the 7.2 kernel was supported by 249 employers that we have been capable of
establish; essentially the most energetic of these have been:

Most energetic 7.2 employers

By changesets (Unknown) 2743 16.7% Intel 1512 9.2% Google 1191 7.3% Red Hat 873 5.3% AMD 813 5.0% Qualcomm 767 4.7% (None) 743 4.5% NVIDIA 535 3.3% Meta 441 2.7% (Consultant) 429 2.6% SUSE 358 2.2% Renesas Electronics 328 2.0% IBM 307 1.9% Kylin 278 1.7% NXP Semiconductors 264 1.6% Oracle 245 1.5% Arm 222 1.4% BayLibre 203 1.2% Microsoft 201 1.2% Bootlin 194 1.2%

By traces modified AMD 224121 22.4% (Unknown) 103861 10.4% Google 102565 10.3% Qualcomm 59021 5.9% Intel 53661 5.4% Red Hat 51559 5.2% Meta 38635 3.9% (None) 28218 2.8% NVIDIA 27417 2.7% MediaTek 22025 2.2% NXP Semiconductors 17692 1.8% BayLibre 15972 1.6% Oracle 12233 1.2% Microsoft 11786 1.2% Renesas Electronics 11103 1.1% Linaro 10209 1.0% (Consultant) 9362 0.9% IBM 9205 0.9% Texas Instruments 8398 0.8% SUSE 8350 0.8%

The most vital change this time round, arguably, is the numerous
improve in builders with unknown affiliation. This is partially, at
least, a mirrored image of the continued flood of latest builders getting into the
kernel neighborhood.

New builders and their instruments

Each of the earlier two kernel releases had set a report for the variety of
first-time builders contributing adjustments; 7.2 has damaged that report
once more, with 613 contributors [KSDB] who had not been seen
right here earlier than. Here, as soon as once more, the pattern is obvious:

The most energetic new developer was Bryam Vargas [KSDB], who contributed
43 commits fixing issues within the enter, NVMe, F2FS, AppArmor,
Landlock, SELinux, NTFS, WiFi, ATA, SCSI, networking, Ceph, and Thunderbolt
subsystems, amongst others. As has been noticed right here earlier than, it could be
troublesome for a seasoned kernel developer to succeed in deeply into that many
subsystems over the course of a single improvement cycle — at the very least, if
they wished to know what they’re doing. For a beforehand unknown
developer to do that is shocking at finest. Vargas’s patches don’t carry
Assisted-by tags, although, so we’re left to imagine that this work was accomplished
with out LLM help.

Overall, 1,111 adjustments in 7.2 (underneath 7% of the whole) carried
Assisted-by tags — a big improve over the 301 seen in 7.1. The
actual rely of LLM-assisted patches in 7.2 should certainly be far greater than
that, although. Many builders don’t provide these tags (usually — however not
all the time — out of ignorance of the foundations), maintainers hardly ever ask about LLM
use, and at the very least one maintainer actively strips those
tags when making use of patches. Accurately monitoring commits developed with
LLM help might prove to not be potential, but it surely typically looks like
the kernel neighborhood doesn’t care sufficient to essentially attempt.

The different method during which LLM utilization is making itself felt in kernel
improvement is within the space of patch overview. There are 252 patches
in 7.2 with Reported-by tags crediting the Sashiko review system [KSDB];
568 commits point out Sashiko of their changelogs. Other LLMs are
additionally clearly getting used to seek out kernel bugs, although, with actual results.

The 7.2 launch accommodates 4,830 commits with Fixes tags, indicating
that they repair a bug launched by a earlier commit. To see simply how
vital a rise is, contemplate this plot, an up to date model of 1
that has been seen right here earlier than (most lately for the 6.17 development cycle):

The heavy inexperienced line plots what number of commits in every launch contained Fixes
tags, whereas the heavy brown line is the variety of commits that have been
recognized by Fixes tags in subsequent commits. The crossover level is at
5.9, suggesting that, as of that launch, the kernel neighborhood began
fixing extra bugs than it launched. That conclusion won’t maintain,
although; it can take years for all the post-5.9 bugs to be recognized
and glued, and that line will shift because it occurs.

The noteworthy a part of the plot, although, is on the right-hand facet, the place
the variety of commits carrying Fixes tags will increase dramatically, after
having been roughly degree for fairly a number of releases. That, it appears, is
what occurs as soon as giant numbers of individuals begin turning LLMs towards the
activity of discovering bugs in present software program. This goes to be an
attention-grabbing pattern to look at. The variety of bugs can’t be infinite — even
within the kernel — so the speed of bug detection and fixing should finally
fall off. The present bugs could have been fastened, and LLM overview could have
lowered the variety of new bugs launched. At least, that’s the hope.

If that drop-off does not occur, there can be some attention-grabbing
inquiries to reply. It appears fairly clear that the unfold of LLMs has
enabled folks to create kernel patches who wouldn’t have been able to
doing so earlier than. If that’s true, it’s not a lot of a stretch to conclude
that, in lots of instances, the submitters of those patches don’t absolutely
perceive what these patches do. Perhaps the LLMs are adequate that
this understanding is now not crucial, but it surely may be that we’re
making a pile of code that no one has ever actually understood, however which
will have to be maintained within the coming years.

The Linux kernel is aimed toward manufacturing use, not analysis or
experimentation. But there’s presently an enormous experiment being run on
the kernel’s improvement processes and neighborhood. This neighborhood has
absorbed new processes and instruments many instances over its 35-year historical past, and
will nearly actually discover a method to try this this time as properly. But that
absorption has not all the time been with out turbulence alongside the way in which.

Meanwhile, there are practically 14,000 non-merge changesets sitting in
the linux-next repository and poised to be pulled for the 7.3 launch; over
1,100 of them comprise Assisted-by tags. It would appear that the excessive degree
of exercise that has characterised current releases won’t be ending
anytime quickly. As all the time, regulate LWN as we observe the subsequent
improvement cycle (and those who come after).

---