Linus Torvalds announced the release of an unusually large version of the Linux 4 kernel. 12, which was added more than 1 million lines of new code. Most of them are supported by AMD Radeon RX Vega, driver for Intel Atom IPU and other drivers.
The Linux 4 kernel. 12.
The Linux OS creator, Linus Torvalds, announced the release of a new version of the Linux kernel under the number 4. 12. Compared to 4. 11, a large amount of new code was added to the new version - more than 1 million lines, as calculated by the resource Phoronix. The size of the patch reaches 89 MB, it contains about 13 thousand. Edit from 1,5 thousand. Developers.
"Historically 4. 12 is one of the biggest releases, and I think that, in the final analysis, only 4. 9 has more additions. But 4. 9 was great at least partially because Greg declared it to be a core with a long support period. A 4. 12 is simply big in itself, "Torvalds writes.. "Greg" should be understood as one of the leading Linux developers Greg Kroah-Hartman (Greg Kroah-Hartman).
Reasons for "weighting".
Version 4. 12 turned out great for several reasons. Part of the new code is supported by new Radeon RX Vega video cards from AMD. According to Torvalds, it took up almost half of the total additional volume, resulting in more than 85% of the code written down had to driver. Another "heavy" addition was the driver for Intel Atom IPU.
In addition, updates have been added for the POWER9, ARM and Nvidia chips, as well as the USB Type-C port manager. In addition, the developers performed some stabilization of the kernel in the form of randomization of the allocation of the kernel address space (KASLR), which by default is used for x86 systems.
Other changes.
Development of version 4. 12 began in mid-May 2017. In the new release appeared I / O planners Budget Fair Queueing (BFQ) and Kyber. Their task is to provide access to data in several threads for multi-core systems. Thus, the construction of a multilevel queuing system was completed.
In the structure of the LivePatch API, which allows patches to be applied to a running kernel, a hybrid consistency model was added.
In this consistency monitoring model, the stack analysis of kPatch is combined with the mechanism for evaluating individual tasks of kGraft. Now the kernel without delays can sew more complex patches than before, even if they relate to changes in functions or semantics of data.
In addition, for Device Mapper, the dm-integrity module was created, which is responsible for emulating the block device. Also appeared the Trusted Execution Environment (TEE) framework, with which you can create protected environments on ARM TrustZone chips.