An EFI variable in a pull is not a tarball
Phoronix and a kernel diary say Linux 7.4's EFI updates merged, so a direct firmware boot can publish the loader's disk id. A merged pull is not a 7.4 release tarball.
The short version
No. Phoronix on October 26, 2026, and a kernel diary the day before, say the EFI updates merged. The patch, posted in September, has the EFI stub set a volatile loader disk identifier when firmware boots the kernel directly, so a systemd generator can find the disk. This page did not verify a tagged 7.4 tarball. Merge reports are not that tag.
Merged is what the diaries say
Phoronix reported on October 26, 2026 that the EFI updates for the Linux 7.4 cycle had been merged. Kernel Dispatch's October 25 mainline note says the same pull landed. This page did not clone a tag and did not see a 7.4 tarball. The short-slice change already filed from this cycle is a different pull, about how long a short scheduling slice lasts. Two merges in one window are two changes. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.The short slice is a different pull.
September 6 is the patch
Vincent Mailhol's third version of the patch, dated September 6, 2026 on the kernel mailing list, is the explanation this page uses. Direct firmware boot is the case. The stub publishes the partition identifier the generator needs, and it refuses to overwrite a loader variable that is already present. The parser that walks the device path stays in the patch. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.
The specification names the variables
systemd's Boot Loader Interface is the document that defines LoaderDevicePartUUID and the neighboring loader variables. The kernel patch is writing a variable that document already named. The kernel is not inventing a private side channel. Asked directly, What do the October accounts say happened? That the EFI update pull was merged. Phoronix says so on October 26. Kernel Dispatch says the same of the October 25 mainline window. Neither account is a tarball this page downloaded. The document named in the sources is the record this page will keep.
A sound pull is not this variable
The same 7.4 window includes a sound subsystem pull already filed, and a memblock search change that was a patch rather than a release. None of those is the EFI stub's variable. Readers who stack the headlines into one feature will mis-count the cycle. A Rust program at the edge of systemd is about which language a leaf uses, not about which identifier the stub writes. The document named in the sources is the record this page will keep. The document named in the sources is the record this page will keep.The sound pull is another file.
Memblock was the cautionary twin
An earlier brief in this cycle had to say that a memblock search patch was not itself a release. The EFI news is further along, because two diaries say it merged, and it is still not a tarball. The same caution has two strengths. Asked directly, Is this the same 7.4 change as the short slice or the sound pull? No. Those are other pulls in the same release window. A disk identifier for the generator is a third change. The document named in the sources is the record this page will keep.A search patch was not a release.
Who the variable is for
Machines whose firmware starts the kernel directly, without a boot loader that would have set the identifier itself. A machine that already boots through a loader which sets the variable is the case the sentinel is there to leave alone. Asked directly, Why does the patch exist? So systemd-gpt-auto-generator can find the disk when no boot loader sits between the firmware and the kernel. The patch's explanation is that sentence. The device-path parser is not reprinted here. October 27's KVM pull, page-modification logging on AMD SVM, is another file in that 7.4 window and is still not a tarball. The document named in the sources is the record this page will keep.The KVM pull is a different file.
What this page will not paste
The device-path walk, the variable flags, and the generator's matching rules are the patch and the specification. Copying them would turn a news brief into a port. Memory-safety language in a release note is a different exam and does not set an EFI variable. Asked directly, Where is the specification? The Boot Loader Interface on systemd's site names the variables. A Rust leaf inside systemd is a different fact about that project and does not set the variable. The document named in the sources is the record this page will keep.Safety language is not a variable.
systemd's role is the consumer
The generator that reads the identifier is systemd's. A decision to write some systemd leaves in Rust does not change which variable the stub is asked to publish. The consumer and the language of a leaf are separate sentences. Asked directly, What does the September patch say it sets? When firmware boots the kernel directly, the EFI stub sets a volatile LoaderDevicePartUUID, and it uses LoaderInfo as a sentinel that bails out if the variable is already set. The systemd specification is the document that names those variables.A leaf language is not the variable.
What would make it a 7.4 release
A tagged tarball or a release announcement that names the version. Until this page has seen that tag, the accurate sentence is that two diaries report a merge. The short answer this page will stand on is this: No. Phoronix on October 26, 2026, and a kernel diary the day before, say the EFI updates merged. The patch, posted in September, has the EFI stub set a volatile loader disk identifier when firmware boots the kernel directly, so a systemd generator can find the disk. This page did not verify a tagged 7.4 tarball. Merge reports are not that tag.
Sources
The reports this brief is filing. Futureweb did not republish them.
Questions
What do the October accounts say happened?
That the EFI update pull was merged. Phoronix says so on October 26. Kernel Dispatch says the same of the October 25 mainline window. Neither account is a tarball this page downloaded.
What does the September patch say it sets?
When firmware boots the kernel directly, the EFI stub sets a volatile LoaderDevicePartUUID, and it uses LoaderInfo as a sentinel that bails out if the variable is already set. The systemd specification is the document that names those variables.
Why does the patch exist?
So systemd-gpt-auto-generator can find the disk when no boot loader sits between the firmware and the kernel. The patch's explanation is that sentence. The device-path parser is not reprinted here.
Is this the same 7.4 change as the short slice or the sound pull?
No. Those are other pulls in the same release window. A disk identifier for the generator is a third change.
Where is the specification?
The Boot Loader Interface on systemd's site names the variables. A Rust leaf inside systemd is a different fact about that project and does not set the variable.