Linux Kernel 7.1.5 Drops With Over 2,000 Commits, Capping a "New Normal" for Stable Updates
Greg Kroah-Hartman lands the latest v7.1.x maintenance release, tackling critical UAF bugs, filesystem deadlocks, and a massive spike in patch volume that the community is already calling routine.
Linux Kernel 7.1.5 shipped today, bringing 2,070 commits and roughly 796 contributors to a stable maintenance release that should keep your system running a little longer. Greg Kroah-Hartman dropped the patch on July 24, 2026, and if you're running anything in the v7.1 series, it's already the latest stable point in that branch.
The v7.1 series is currently the active mainline branch, with 7.2-rc4 sitting in testing for about five days now. You're probably wondering why a "stable" update carries over two thousand commits. The Linux community has settled on a phrase for it: the "new normal." AI-assisted development, faster syzbot discovery cycles, and a broader security audit push have quietly turned massive stable releases into the default rather than the exception.
The Bugs That Actually Matter
The headline fix comes from Thomas Gleixner, who patched a critical use-after-free race in the POSIX CPU timer subsystem. A non-leader exec() call could race with timer deletion, leaving kernel memory in a state that looked fine until it didn't. The patch adds proper synchronization primitives and reshuffles the task lookup logic to keep the race from winning.
Next up is a high-severity Bluetooth L2CAP use-after-free tracked down by Siwei Zhang. The callback returned a channel pointer after releasing the parent socket lock, giving another thread enough time to accept and free the socket before anyone actually used it. The fix hands lifetime ownership of the child channel straight to the core kernel.
If you run an SMB server via ksmbd, two related patches from Davide Ornaghi and Namjae Jeon close a path traversal hole and a credential bypass. The traversal bug let authenticated clients step past share boundaries using .. components during file creation. The credential issue let FSCTL operations like SET_SPARSE and SET_ZERO_DATA run with worker credentials instead of the opener's, effectively skipping permission checks.
Chris Mason locked down a binder LSM context sizing bug where security_release_secctx() zeroed the context length before the bounds calculation ran. That inflated the end offset just enough to copy straight into the security context area. Pratyush Yadav from Google added validation to the liveupdate subsystem, making sure ioctl operations only run on compatible session types after a kexec.
Filesystems, Drivers, and the Rest of the Pile
Filesystem-wise, f2fs got a deadlock fix in the gc_merge path during garbage collection, plus better error handling for corrupted xattr entries. exfat added iomap infrastructure groundwork and swapped out unsafe arithmetic macros for type-safe inline functions. XFS finally marked newly added RTGs as free during growfs on zoned devices, which stops the filesystem from waiting for space that will never arrive.
The subsystem breakdown tells a story more than a feature list. netfilter contributed 58 patches, wifi added 48, ntfs brought 33, and bpf landed 31. apparmor quietly slipped 20 security module patches through the door, while pinctrl, crypto, riscv, and media each rounded out the top ten with 19 to 25 changes.
Driver fixes are just as scattered as the numbers suggest. ALSA patched an async firmware callback race in the TAS2781 HDA driver and updated the Scarlett 2i2 Gen 4 firmware interface. The ueagle-atm modem driver finally killed a concurrent firmware load bug during disconnect that syzbot had been reporting for years. AMD's DRM stack reverted a Color Manager regression in 3DLUT, Shaper, and Blend from the DCN42 sync. IIO forced explicit sample requests in the HID rotation driver after it started returning stale quaternions.
Core subsystems aren't idle either. iomap consolidated bio submission to share read-endio logic across filesystems, fixing XFS T10 protection offloading that a prior refactoring had quietly broken. The GPIB driver got a revert for an incorrect cb7210 fix that leaked memory when request_irq failed.
Keep in mind that the longterm branches remain untouched by this cycle. 6.18, 6.12, 6.6, 6.1, 5.15, and 5.10 are still getting their own maintenance updates, so if you're pinned to a longterm line, you'll want to watch the corresponding tree rather than tracking v7.1.x.
The sheer volume here isn't a bug. It's a symptom. Two thousand commits in a stable branch used to take a full release cycle. Now it happens on a Tuesday.
However, at the same time, the kernel community has clearly decided that massive stable updates are the baseline now rather than a rare event. Testing came from a broad cross-section: Debian developers, Fedora QA, CIP contributors, Broadcom, and Google. That spread means enterprise, distribution, and embedded workflows all saw validation before this hit the mirrors.
Linux kernel 7.1.5 released
Linux kernel version 7.1.5 is now available:
Full source: https://cdn.kernel.org/pub/linux/kernel/v7.x/linux-7.1.5.tar.xz
Patch: https://cdn.kernel.org/pub/linux/kernel/v7.x/patch-7.1.5.xz
PGP Signature: https://cdn.kernel.org/pub/linux/kernel/v7.x/linux-7.1.5.tar.sign
You can view the summary of the changes at the following URL:
https://git.kernel.org/stable/ds/v7.1.5/v7.1.4
