From cbbee68b1c8507e3ce52c955e4e240de6eee824f Mon Sep 17 00:00:00 2001 From: Cheng Jian Date: Tue, 18 Jan 2022 23:56:09 +0800 Subject: [PATCH] description: update 20210118 --- study/kernel/00-DESCRIPTION/DEBUGGING.md | 2 +- study/kernel/00-DESCRIPTION/LOCKING.md | 22 +++++++--- study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md | 41 +++++++++++++------ study/kernel/00-DESCRIPTION/SCHEDULER.md | 7 +++- 4 files changed, 51 insertions(+), 21 deletions(-) diff --git a/study/kernel/00-DESCRIPTION/DEBUGGING.md b/study/kernel/00-DESCRIPTION/DEBUGGING.md index 08c51f9..713de06 100644 --- a/study/kernel/00-DESCRIPTION/DEBUGGING.md +++ b/study/kernel/00-DESCRIPTION/DEBUGGING.md @@ -498,7 +498,7 @@ Mold 是目前 Unix 链接器的现代替代品, 已经达到了 1.0 版本. 由 |:----:|:----:|:---:|:----:|:---------:|:----:| | 2021/06/22 | Daniel Bristot de Oliveira | [hwlat improvements and osnoise/timerlat tracers](https://patchwork.kernel.org/project/linux-trace-devel/patch/162936876189.187130.17558311387542061930.stgit@devnote2) | NA | v5 ☑ 5.14-rc1 | [Patchwork v5,09/14](https://lore.kernel.org/all/e649467042d60e7b62714c9c6751a56299d15119.1624372313.git.bristot@redhat.com), [COMMIT](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=bce29ac9ce0bb0b0b146b687ab978378c21e9078) | | 2021/11/29 | Daniel Bristot de Oliveira | [osnoise: Support multiple instances (for RTLA)](https://patchwork.kernel.org/project/linux-trace-devel/patch/162936876189.187130.17558311387542061930.stgit@devnote2) | 为 osnoise/timerlat 跟踪器启用多个实例. 目前, osnoise 和 timerlat 仅在单个实例上运行. 为了减少这种限制, 本系列增加了对同一跟踪程序的并行实例的支持. 也就是说, 可以使用不同的跟踪配置运行 osnoise tracer 的两个实例. 例如, 一个仅用于跟踪器输出, 另一个用于跟踪器和一组跟踪点. 这个补丁集是 RTLA 的内核依赖项. 此修补程序集与 [RTLA](https://patchwork.kernel.org/project/linux-trace-devel/cover/cover.1638182284.git.bristot@kernel.org) 一起发送, 但我们将内核和用户空间修补程序集分开. | v9 ☑ 5.16-rc1 | [Patchwork V9,0/9](https://patchwork.kernel.org/project/linux-trace-devel/cover/cover.1635702894.git.bristot@kernel.org), [COMMIT](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2fac8d6486d5c34e2ec7028580142b8209da3f92) | -| 2021/11/29 | Daniel Bristot de Oliveira | [RTLA: An interface for osnoise/timerlat tracers](https://patchwork.kernel.org/project/linux-trace-devel/patch/162936876189.187130.17558311387542061930.stgit@devnote2) | [rtla](https://github.com/bristot/rtsl) 是一个旨在分析 Linux 实时属性的工具集. rtla 并没有将 Linux 作为一个黑盒来测试, 而是利用内核跟踪功能来提供关于属性和意外结果的根本原因的精确信息. 首先, 它提供了一个 osnoise 和 timerlat 示踪器的接口. 在未来, 它还将作为 rtsl 和其他延迟 / 噪声跟踪器的基础. V6 版本前五个补丁是 [osnoise: Support multiple instances](https://lore.kernel.org/lkml/cover.1628775552.git.bristot@kernel.org) 的重新发送, 它为 osnoise/timerlat 跟踪器启用多个实例. 接下来的七个补丁是 rtla, rtla osnoise 和 rtla timerlat. 后面的版本都将内核态和用户态拆开来提交. | v5 ☐ 5.14-rc1 | [2021/10/27 Patchwork v6,00/20](https://lore.kernel.org/lkml/cover.1635284863.git.bristot@kernel.org)
*-*-*-*-*-*-*-*
[2021/11/29 Patchwork V8,00/14](https://patchwork.kernel.org/project/linux-trace-devel/cover/cover.1638182284.git.bristot@kernel.org) | +| 2021/12/10 | Daniel Bristot de Oliveira | [RTLA: An interface for osnoise/timerlat tracers](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?id=5dce5904e3b9bc11d9635dd3458c7ff32caebb59) | [rtla](https://github.com/bristot/rtsl) 是一个旨在分析 Linux 实时属性的工具集. rtla 并没有将 Linux 作为一个黑盒来测试, 而是利用内核跟踪功能来提供关于属性和意外结果的根本原因的精确信息. 首先, 它提供了一个 osnoise 和 timerlat 示踪器的接口. 在未来, 它还将作为 rtsl 和其他延迟 / 噪声跟踪器的基础. V6 版本前五个补丁是 [osnoise: Support multiple instances](https://lore.kernel.org/lkml/cover.1628775552.git.bristot@kernel.org) 的重新发送, 它为 osnoise/timerlat 跟踪器启用多个实例. 接下来的七个补丁是 rtla, rtla osnoise 和 rtla timerlat. 后面的版本都将内核态和用户态拆开来提交. 参见 [Linux 5.17 Picks Up A Real-Time Analysis Tool](https://www.phoronix.com/scan.php?page=news_item&px=Linux-5.17-RTLA) | v5 ☐ 5.17-rc1 | [2021/10/27 Patchwork v6,00/20](https://lore.kernel.org/lkml/cover.1635284863.git.bristot@kernel.org)
*-*-*-*-*-*-*-*
[2021/11/29 Patchwork v8,00/14](https://patchwork.kernel.org/project/linux-trace-devel/cover/cover.1638182284.git.bristot@kernel.org)
*-*-*-*-*-*-*-*
[LORE v9,00/14](https://lore.kernel.org/all/cover.1639158831.git.bristot@kernel.org) | https://patchwork.kernel.org/project/linux-trace-devel/list/?submitter=200911&state=*&archive=both¶m=1&page=2 diff --git a/study/kernel/00-DESCRIPTION/LOCKING.md b/study/kernel/00-DESCRIPTION/LOCKING.md index 05cbbc3..53a4b42 100644 --- a/study/kernel/00-DESCRIPTION/LOCKING.md +++ b/study/kernel/00-DESCRIPTION/LOCKING.md @@ -84,17 +84,21 @@ CAS 的原理是, 将旧值与一个期望值进行比较, 如果相等, 则更 ## 1.2 ticket LOCK ------- -https://lwn.net/Articles/267968 +[Ticket spinlocks](https://lwn.net/Articles/267968) | 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | |:----:|:----:|:---:|:----:|:---------:|:----:| | 2007/11/01 | Nick Piggin | [ticket spinlocks for x86](https://lore.kernel.org/patchwork/cover/95892) | X86 架构 ticket spinlocks 的实现. | v1 ☑ 2.6.25-rc1(部分合入) | [PatchWork RFC](https://lore.kernel.org/patchwork/cover/85789)
*-*-*-*-*-*-*-*
[PatchWork](https://lore.kernel.org/patchwork/cover/95892), [PatchWork](https://lore.kernel.org/patchwork/cover/95894) | | 2021/09/19 | Guo Ren | [riscv: locks: introduce ticket-based spinlock implementation](https://patchwork.kernel.org/project/linux-riscv/patch/20210919165331.224664-1-guoren@kernel.org) | riscv 架构 ticket spinlocks 的实现. | v1 ☐ |[PatchWork](https://patchwork.kernel.org/project/linux-riscv/patch/20210919165331.224664-1-guoren@kernel.org) | +| 2013/10/09 | Nick Piggin | [arm64: locks: introduce ticket-based spinlock implementation](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=52ea2a560a9dba57fe5fd6b4726b1089751accf2) | ARM64 架构 ticket spinlocks 的实现. | v1 ☑ 3.13-rc1 | [COMMIT](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=52ea2a560a9dba57fe5fd6b4726b1089751accf2) | +| 2015/02/10 | Nick Piggin | [arm64: locks: patch in lse instructions when supported by the CPU](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=81bb5c6420635dfd058c210bd342c29c95ccd145) | ARM64 ticket spinlocks 使用 LSE 进行优化. | v1 ☑ 4.3-rc1 | [COMMIT](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=81bb5c6420635dfd058c210bd342c29c95ccd145) | [Linux中的spinlock机制[一] - CAS和ticket spinlock](https://zhuanlan.zhihu.com/p/80727111) +[BAKERY ALGORITHM](https://remonstrate.wordpress.com/tag/bakery-algorithm) +[Lamport 面包店算法](https://blog.csdn.net/pizi0475/article/details/17649949) ## 1.3 MCS lock ------- @@ -110,10 +114,10 @@ spinlock 的值出现变化时, 所有试图获取这个 spinlock 的 CPU 都需 | 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | |:----:|:----:|:---:|:----:|:---------:|:----:| -| 2015/04/24 | Waiman Long | [qspinlock: a 4-byte queue spinlock with PV support](https://lore.kernel.org/patchwork/cover/127444) | X86 架构 qspinlocks 的实现. | v16 ☑ 4.2-rc1 | [PatchWork RFC](https://lore.kernel.org/lkml/20140310154236.038181843@infradead.org)
*-*-*-*-*-*-*-*
[LORE v16 00/14](https://lore.kernel.org/all/1429901803-29771-1-git-send-email-Waiman.Long@hp.com/) | | 2014/01/21 | Tim Chen | [MCS Lock: MCS lock code cleanup and optimizations](https://lore.kernel.org/patchwork/cover/435770) | MCS LOCK 重构, 增加了新的文件 `include/linux/mcs_spinlock.h` | v9 ☑ 3.15-rc1 | [LKML v6 0/6](https://lkml.org/lkml/2013/9/25/532)
*-*-*-*-*-*-*-*
[PatchWork v9 0/6](https://lore.kernel.org/patchwork/cover/435770), [关键 commit](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e72246748ff006ab928bc774e276e6ef5542f9c5) | | 2014/02/10 | Peter Zijlstra | [locking/core patches](https://lore.kernel.org/patchwork/cover/440565) | PV SPINLOCK | v1 ☑ 4.2-rc1 | [PatchWork](https://lore.kernelorg/lkml/20140210195820.834693028@infradead.org) | + ## 1.4 qspinlock ------- @@ -130,10 +134,9 @@ spinlock 的值出现变化时, 所有试图获取这个 spinlock 的 CPU 都需 | 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | |:----:|:----:|:---:|:----:|:---------:|:----:| -| 2017/04/10 | Yury Norov | [arm64: queued spinlocks and rw-locks](http://patches.linaro.org/cover/98492) | X86 架构 qspinlocks 的实现. | RFC ☐ | [PatchWork RFC](https://patchwork.kernel.org/project/linux-arm-kernel/patch/1491860104-4103-4-git-send-email-ynorov@caviumnetworks.com) | -| 2018/04/26 | Will Deacon | [kernel/locking: qspinlock improvements](https://patchwork.kernel.org/project/linux-arm-kernel/patch/1524738868-31318-2-git-send-email-will.deacon@arm.com/) | qspinlocks 优化. | v3 ☑ 4.18-rc1 | [LWN](https://lwn.net/Articles/751105))
*-*-*-*-*-*-*-*
[LKML v3 00/14](https://lkml.org/lkml/2018/4/26/340) | -| 2018/06/16 | Will Deacon | [arm64: locking: Replace ticket lock implementation with qspinlock](http://patches.linaro.org/cover/98492) | ARM64 架构 qspinlocks 的实现. | RFC ☑ 4.19-rc1 | [PatchWork RFC](https://patchwork.kernel.org/project/linux-arm-kernel/patch/1530010812-17161-3-git-send-email-will.deacon@arm.com) | - +| 2017/04/10 | Yury Norov | [arm64: queued spinlocks and rw-locks](http://patches.linaro.org/cover/98492) | ARM64 架构 qspinlocks 的实现. | RFC ☐ | [PatchWork RFC](https://patchwork.kernel.org/project/linux-arm-kernel/patch/1491860104-4103-4-git-send-email-ynorov@caviumnetworks.com)
*-*-*-*-*-*-*-*
[LORE 0/3](https://lore.kernel.org/lkml/20170503145141.4966-1-ynorov@caviumnetworks.com) | +| 2018/04/26 | Will Deacon | [kernel/locking: qspinlock improvements](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?id=baa8c6ddf7be33f2b0ddeb68906d668caf646baa) | 优化并实现了 qspinlocks 的通用框架. | v3 ☑ 4.18-rc1 | [LWN](https://lwn.net/Articles/751105), [LORE v3,00/14](https://lore.kernel.org/all/1524738868-31318-1-git-send-email-will.deacon@arm.com) | +| 2018/06/26 | Will Deacon | [Hook up qspinlock for arm64](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?id=5d168964aece0b4a41269839c613683c5d7e0fb2) | ARM64 架构 qspinlocks 的实现. | v1 ☑ 4.19-rc1 | [LORE 0/3](https://lore.kernel.org/linux-arm-kernel/1530010812-17161-1-git-send-email-will.deacon@arm.com) | ## 1.5 PV_SPINLOCK ------- @@ -288,6 +291,13 @@ PV_SPINLOCKS 的合入引起了[性能问题 Performance overhead of paravirt_op [An introduction to lockless algorithms](https://lwn.net/Articles/844224) +# 9 ATOMIC +------- + +| 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | +|:----:|:----:|:---:|:----:|:---------:|:----:| +| 2015/08/06 | Will Deacon | [Add generic support for relaxed atomics](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?id=0ca326de7aa9cb253db9c1a3eb3f0487c8dbf912) | ARM64 引入 relaxed atomics. | v5 ☑ 4.3-rc1 | [LORE 0/5](https://lore.kernel.org/lkml/1436790687-11984-1-git-send-email-will.deacon@arm.com)
*-*-*-*-*-*-*-*
[LORE v2,0/7](https://lore.kernel.org/lkml/1437060758-10381-1-git-send-email-will.deacon@arm.com)
*-*-*-*-*-*-*-*
[LORE v5,0/8](https://lore.kernel.org/all/1438880084-18856-1-git-send-email-will.deacon@arm.com) | +
* 本作品/博文 ( [AderStep-紫夜阑珊-青伶巷草 Copyright ©2013-2017](http://blog.csdn.net/gatieme) ), 由 [成坚(gatieme)](http://blog.csdn.net/gatieme) 创作. diff --git a/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md b/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md index 6099d0e..3b90ad1 100644 --- a/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md +++ b/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md @@ -98,8 +98,10 @@ Andrew Morton 一般以一封 [incoming](https://lore.kernel.org/linux-mm/20211105133408.cccbb98b71a77d5e8430aba1@linux-foundation.org) 的邮件向 Linus 发起 push 请求, Linus pull 之后的 Merge 标题为 Merge branch 'akpm' (patches from Andrew). +- [x] Merge branch 'akpm' (patches from Andrew) + ```cpp -git log --oneline | grep -E "Merge |Linux" | grep -E "akpm|Linux" | less +git log --oneline v5.15...v5.16 | grep -E "Merge branch | Linux " | grep -E "akpm| Linux " | less ``` | 版本 | 发布时间 | 合并链接 | @@ -107,7 +109,7 @@ git log --oneline | grep -E "Merge |Linux" | grep -E "akpm|Linux" | less | 5.13 | [2021/06/02](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=62fb9874f5da) | [2021/04/30, 5.13-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=d42f323a7df0b298c07313db00b44b78555ca8e6)
[2021/05/05, 5.13-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8404c9fbc84b741f66cff7d4934a25dd2c344452)
[2021/05/15, 5.13-rc2](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a4147415bdf152748416e391dd5d6958ad0a96da)
[2021/05/22, 5.13-rc3](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=34c5c89890d6295621b6f09b18e7ead9046634bc)
[2021/06/05, 5.13-rc5](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e5220dd16778fe21d234a64e36cf50b54110025f)
[2021/06/16, 5.13-rc7](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=70585216fe7730d9fb5453d3e2804e149d0fe201)
[2021/06/25, 5.13](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=7ce32ac6fb2fc73584b567c73ae0c47528954ec6) | | 5.14 | [2021/08/29](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=73f3af7b4611) | [2021/06/29, 5.14-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=65090f30ab791810a3dc840317e57df05018559c)
[2021/07/02, 5.14-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=71bd9341011f)
[2021/07/09, 5.14-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=bd9c35060329)
[2021/07/15, 5.14-rc2](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=dd9c7df94c1b)
[2021/07/24, 5.14-rc3](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=bca1d4de3981)
[2021/07/30, 5.14-rc4](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=ad6ec09d9622)
[2021/08/13, 5.14-rc6](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=dfa377c35d70)
[2021/08/20, 5.14-rc7](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=ed3bad2e4fd7)
[2021/08/25, 5.14](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=73f3af7b4611) | | 5.15 | [2021/10/31](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8bb7eca972ad) | [2021/09/08, 5.15-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2d338201d531)
[2021/09/08, 5.15-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a3fa7a101dcf)
[2021/09/25, 5.15-rc3](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a3b397b4fffb)
[2021/10/19, 5.15-rc6](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=d9abdee5fd5a)
[2021/10/29, 5.15](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2c04d67ec1eb) | -| 5.16 | [2021/10/31](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8bb7eca972ad) | [2021/11/06, 5.16-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=512b7931ad05)
[2021/11/09, 5.16-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=59a2ceeef6d6)
[2021/11/11, 5.16-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=dbf49896187f)
[2021/11/20, 5.16-rc2](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=923dcc5eb0c1)
[2021/11/20, 5.16-rc5](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=df442a4ec740) +| 5.16 | [2021/10/31](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8bb7eca972ad) | [2021/11/06, 5.16-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=512b7931ad05)
[2021/11/09, 5.16-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=59a2ceeef6d6)
[2021/11/11, 5.16-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=dbf49896187f)
[2021/11/20, 5.16-rc2](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=923dcc5eb0c1)
[2021/11/20, 5.16-rc5](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=df442a4ec740)
[2021/12/25, 5.16-rc7](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=d0cc67b2781654ac71c73d303e0347e5e9b10ad3)
[2021/12/25, 5.16-rc8](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=f87bcc88f3028af584b0820bdf6e0e4cdc759d26) | cgit 上查看 MM 所有的 log 信息 : @@ -264,6 +266,8 @@ Linux 一开始是在一台i386上的机器开发的, i386 的硬件页表是2 ## 1.4 [Clarifying memory management with page folios](https://lwn.net/Articles/849538) ------- +### 1.4.1 page folios +------- [LWN: 利用page folio来明确内存操作！](https://blog.csdn.net/Linux_Everything/article/details/115388078) @@ -288,11 +292,19 @@ Linux 一开始是在一台i386上的机器开发的, i386 的硬件页表是2 | 2022/01/02 | "Matthew Wilcox (Oracle)" | [MConvert GUP to folios](https://patchwork.kernel.org/project/linux-mm/cover/20220102215729.2943705-1-willy@infradead.org) | NA | v1 ☐ | [PatchWork 00/17](https://patchwork.kernel.org/project/linux-mm/cover/20220102215729.2943705-1-willy@infradead.org) | +### 1.4.2 Pulling slabs out of struct page +------- + +[Pulling slabs out of struct page](https://lwn.net/Articles/871982) + +[Struct slab comes to 5.17](https://lwn.net/Articles/881039/) + + | 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | |:----:|:----:|:---:|:----:|:---------:|:----:| -| 2022/01/04 | "Matthew Wilcox (Oracle)" | [Separate struct slab from struct page](https://patchwork.kernel.org/project/linux-mm/cover/20211004134650.4031813-1-willy@infradead.org) | struct page 结构定义中比较复杂的部分之一是 slab 分配器所使用的部分. 一般来说, 如果将 slab 的数据类型从 page 结构体中分离是有好处的, 而且它还有助于防止尾页滑落到任何地方. | v2 ☐ | [2021/07/15 PatchWork 00/62](https://patchwork.kernel.org/project/linux-mm/cover/20211004134650.4031813-1-willy@infradead.org)
*-*-*-*-*-*-*-*
[2021/12/01 PatchWork v2,00/33](https://patchwork.kernel.org/project/linux-mm/cover/20211201181510.18784-1-vbabka@suse.cz)
*-*-*-*-*-*-*-*
[2022/01/04 PatchWork v4,00/32](https://patchwork.kernel.org/project/linux-mm/cover/20220104001046.12263-1-vbabka@suse.cz) | +| 2022/01/04 | "Matthew Wilcox (Oracle)" | [Separate struct slab from struct page](https://patchwork.kernel.org/project/linux-mm/cover/20211004134650.4031813-1-willy@infradead.org) | struct page 结构定义中比较复杂的部分之一是 slab 分配器所使用的部分. 一般来说, 如果将 slab 的数据类型从 page 结构体中分离是有好处的, 而且它还有助于防止尾页滑落到任何地方. | v2 ☐ | [2021/07/15 PatchWork RFC,00/62](https://patchwork.kernel.org/project/linux-mm/cover/20211004134650.4031813-1-willy@infradead.org)
*-*-*-*-*-*-*-*
[2021/12/01 PatchWork v2,00/33](https://patchwork.kernel.org/project/linux-mm/cover/20211201181510.18784-1-vbabka@suse.cz)
*-*-*-*-*-*-*-*
[2022/01/04 PatchWork v4,00/32](https://patchwork.kernel.org/project/linux-mm/cover/20220104001046.12263-1-vbabka@suse.cz) | | 2021/10/12 | Johannes Weiner | [PageSlab: eliminate unnecessary compound_head() calls](https://patchwork.kernel.org/project/linux-mm/cover/20211012180148.1669685-1-hannes@cmpxchg.org) | 重构代码, 消除二义性, 使得代码更加简洁. PageSlab() 目前对所有调用站点施加一个 compound_head() 调用, 即使只有极少数情况会遇到尾页. 这组补丁气泡尾分辨率到少数需要它的网站, 并消除它在其他地方. 这个改动很独立, 它的灵感来自于 Willy 的补丁 Separate struct slab from struct page](https://patchwork.kernel.org/project/linux-mm/cover/20210715200030.899216-1-willy@infradead.org). 为了让逻辑更清晰, 代码更简洁. PageSlab() 的调用应该完全从对 compound_head() 的无限制调用中分离出来, 因为它们本身就有不必要的开销. | v1 ☐ | [PatchWork 00/11](https://patchwork.kernel.org/project/linux-mm/cover/20211012180148.1669685-1-hannes@cmpxchg.org), [LKML](https://lkml.org/lkml/2021/10/12/820) | -| 2021/11/16 | Vlastimil Babka | [Separate struct slab from struct page](https://patchwork.kernel.org/project/linux-mm/cover/20211116001628.24216-1-vbabka@suse.cz) | NA | RFC ☐ | [PatchWork RFC,00/32](https://patchwork.kernel.org/project/linux-mm/cover/20211116001628.24216-1-vbabka@suse.cz) | +| 2022/01/04 | Vlastimil Babka | [Separate struct slab from struct page](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?id=9d6c59c1c0d62a314a2b46839699b200cccd2d08) | NA | RFC ☑ 5.17-rc1 | [PatchWork RFC,00/32](https://patchwork.kernel.org/project/linux-mm/cover/20211116001628.24216-1-vbabka@suse.cz)
*-*-*-*-*-*-*-*
[PatchWork v4,00/32](https://patchwork.kernel.org/project/linux-mm/cover/20220104001046.12263-1-vbabka@suse.cz) | ## 1.5 页面初始化 @@ -2340,18 +2352,18 @@ LWN 上 Mel 写的关于 Huge Page 的连载. -## 7.1 标准大页 HUGETLB 支持 +## 7.1 标准大页 HugeTLB 支持 ------- -### 7.1.1 引入 HUGETLB +### 7.1.1 引入 HugeTLB ------- 如果能使用更大的页面, 则能很好地解决上述问题. 试想如果使用2MB的页(一个页相当于512个连续的4KB 页面), 则所需的 TLB 表项由原来的 512个变成1个, 这将大大提高 TLB hit 的机率; 缺页中断也由原来的512次变为1次, 这对性能的提升是不言而喻的. -然而, 前面也说了 Linux 对4KB大小的页面认知是根植其中的, 想要支持更大的页面, 需要对非常多的核心的代码进行大改动, 这是不现实的. 于是, 为了支持大页面, 有了一个所谓 HUGETLB 的实现. +然而, 前面也说了 Linux 对4KB大小的页面认知是根植其中的, 想要支持更大的页面, 需要对非常多的核心的代码进行大改动, 这是不现实的. 于是, 为了支持大页面, 有了一个所谓 HugeTLB 的实现. 它的实现是在系统启动时, 按照用户指定需求的最大大页的大小和个数, 预留足够的物理内存空间. 用户在程序中可以使用 **mmap()** 系统调用或共享内存的方式映射和访问这些大页, 参考官方文档: [HugeTLB Pages](https://www.kernel.org/doc/html/latest/admin-guide/mm/hugetlbpage.html). @@ -2402,7 +2414,7 @@ LWN 上 Mel 写的关于 Huge Page 的连载. huge page 最开始只支持 PMD 级别(基础页 4K, 则 PMD 级别为 2MB)的大页, 自 3.8 版本加入这个 [commit 42d7395feb56 ("mm: support more pagesizes for MAP_HUGETLB/SHM_HUGETLB")](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=42d7395feb56f0655cd8b68e06fc6063823449f8) 之后, 利用 shmget()/mmap() 的 flag 参数中未使用的 bits, 可以支持其他的 huge page 大小(比如 1GB). -进一步的, ARM64 MMU 支持一个连续位(Contiguous bit), 该位表示 TTE 是可缓存在单个 TLB 条目中的一组连续条目之一. 通过对该位的支持可以增加新的中间大页的大小. 可用的巨大页面大小取决于基本页面大小 PAGE_SIZE. +进一步的, ARM64 MMU 支持一个连续位(Contiguous bit), 该位表示 TTE 是可缓存在单个 TLB 条目中的一组连续条目之一. 通过对该位的支持可以增加新的中间大页的大小. 可用的巨大页面大小取决于基本页面大小 PAGE_SIZE. 参见 [arm64: Add support for PTE contiguous bit.](https://lore.kernel.org/lkml/1450380686-20911-1-git-send-email-dwoods@ezchip.com) 在不使用连续页面的情况下, 大页面大小如下所示: @@ -2427,9 +2439,7 @@ huge page 最开始只支持 PMD 级别(基础页 4K, 则 PMD 级别为 2MB)的 --------------------------------------------------- - - -如果基本页面大小设置为 64KB，则默认情况下会启用 2MB 页面。在将来，4KB 和 64KB 的页面都可以使用 2MB 作为默认的巨大页面大小。 +如果基本页面大小设置为 64KB, 则默认情况下会启用 2MB 页面. 在将来, 4KB 和 64KB 的页面都可以使用 2MB 作为默认的巨大页面大小. | 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | |:----:|:----:|:---:|:----:|:---------:|:----:| @@ -2498,7 +2508,6 @@ hugepages_treat_as_movable 的目的是减少内存碎片, 而 hugetlb 页面的 | 2018/01/31 | Michal Hocko | [mm, hugetlb: remove hugepages_treat_as_movable sysctl](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?id=944d9fec8d7aee3f2e16573e9b6a16634b33f403) | 移除 hugepages_treat_as_movable, 不再允许 HugeTLB 从 ZONE_MOVABLE 中分配. 而是[使用 hugepage_migration_support(ed) 来做控制](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=83467efbdb7948146581a56cbd683a22a0684bbb). | RFC ☑ 4.16-rc1 | [LORE RFC](https://lore.kernel.org/all/20171003072619.8654-1-mhocko@kernel.org), [COMMIT](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=d6cb41cc44c63492702281b1d329955ca767d399) | - 但是, HugeTLB 的分配仍然有两个严重的问题: 1. 分配不是 NUMA 感知的. 在 NUMA 机器上, 内核在节点之间平均分配引导时分配的大量页面. 例如, 假设您有一个四节点 NUMA 机器, 并希望在引导时分配四个 1G 的巨大页面. 内核将为每个节点分配一个巨大的页面. 另一方面, 有些用户希望能够指定应该从哪个 NUMA 节点分配巨大的页面. 这样他们就可以在特定的 NUMA 节点上放置虚拟机. @@ -2535,7 +2544,13 @@ HugeTLB 的分配和使用多数情况下是静态的. x86_64 支持 2M 和 1G #### 7.1.7.3 Numa Aware HugeTLB reserve ------- -HugeTLB 预留空间时可以精细化控制不同 NUMA NODE 上预留的空间. +HugeTLB CMA 在设计的时候, 已经考虑了 NUMA 的存在. + +1. 在 [mm: using CMA for 1 GB hugepages allocation](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?id=cf11e85fc08cc6a4fe3ac2ba2e610c962bf20bc3) 引入 hugetlb_cma 参数的过程中, Roman Gushchin 发现 CMA 并没有一个在指定 NUMA 节点上预留和分配内存的能力, 因此 [mm: cma: NUMA node interface](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8676af1ff2d28e64e5636147821bda7524cf007d) 增加了尝试在特定节点上[预留](https://elixir.bootlin.com/linux/v5.7/source/include/linux/cma.h#L33)和分配连续内存的能力. 注意如果指定的节点无法分配, 它将回退到其他节点进行分配. + +2. 此外 [hugetlb: add support for gigantic page allocation at runtime](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=944d9fec8d7aee3f2e16573e9b6a16634b33f403) 在运行时动态分配 HugeTLB 内存时, 可以在指定节点下 `/sys/devices/system/node/node1/hugepages/hugepages-1048576kB/nr_hugepages` 的 HugeTLB 接口中进行动态分配. + +随后内核开发者考虑预留 HugeTLB 空间时也可以精细化控制不同 NUMA NODE 上预留的空间, 通过扩展 HugeTLB 的启动参数解析工作, 增加各个节点上的精细化预留. | 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | |:----:|:----:|:---:|:----:|:---------:|:----:| diff --git a/study/kernel/00-DESCRIPTION/SCHEDULER.md b/study/kernel/00-DESCRIPTION/SCHEDULER.md index 345d151..4cf700e 100644 --- a/study/kernel/00-DESCRIPTION/SCHEDULER.md +++ b/study/kernel/00-DESCRIPTION/SCHEDULER.md @@ -51,12 +51,17 @@ Mainline Merge Window - [x] Merge tag 'sched-urgent-date' + +```cpp +git log --oneline v5.15...v5.16 | grep -E "Merge tag | Linux " | grep -E "sched| Linux " +``` + | 版本 | 发布时间 | 合并链接 | |:---:|:-------:|:-------:| | 5.13 | 2021/06/28 | [Merge tag 'sched-core-2021-04-28', 5.13-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=16b3d0cf5bad844daaf436ad2e9061de0fe36e5c)
[Merge tag 'sched-urgent-2021-05-09', 5.13-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9819f682e48c6a7055c5d7a6746411dd3969b0e5)
[Merge tag 'sched-urgent-2021-05-15', 5.13-rc2](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c12a29ed9094b4b9cde8965c12850460b9a79d7c)
[Merge tag 'sched-urgent-2021-06-12', 5.13-rc6](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=99f925947ab0fd5c17b74460d8b32f1aa1c86e3a)
[Merge tag 'sched_urgent_for_v5.13_rc6', 5.13-rc7, 2021/06/20](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=cba5e97280f53ec7feb656fcdf0ec00a5c6dd539)
[Merge tag 'sched-urgent-2021-06-24', 5.13](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=666751701b6e4b6b6ebc82186434806fa8a09cf3)
| | 5.14 | 2021/08/29 | [Merge tag 'sched-core-2021-06-28', 5.14-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=54a728dc5e4feb0a9278ad62b19f34ad21ed0ee4)
[Merge tag 'sched-urgent-2021-06-30', 5.14-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a6eaf3850cb171c328a8b0db6d3c79286a1eba9d)
[Merge tag 'sched-urgent-2021-07-11', 5.14-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=877029d9216dcc842f50d37571f318cd17a30a2d)
[Merge tag 'sched-urgent-2021-08-08', 5.14-rc5](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=713f0f37e8128e8a0190a98f5a4be71fb32a671a)
[Merge tag 'sched_urgent_for_v5.14', 5.14](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=537b57bd5a202af145c266d4773971c2c9f90cd9) | | 5.15 | 2021/11/1 | NA | [Merge tag 'sched-core-2021-08-30'](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5d3c0db4598c5de511824649df2aa976259cf10a)
[Merge remote-tracking branch 'tip/sched/arm64'](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=65266a7c6abf)
[Merge tag 'sched_urgent_for_v5.15_rc1'](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=56c244382fdb)
[Merge tag 'sched_urgent_for_v5.15_rc4'](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=777feabaea77)
[Merge tag 'sched_urgent_for_v5.15_rc7'](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=6c62666d8879) -| 5.16 | NA | [Merge tag 'sched-core-2021-11-01', 5.16-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9a7e0a90a454a7826ecbca055a6ec9271b70c686)
[Merge tag 'sched_urgent_for_v5.16_rc1', 5.16-rc3](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=fc661f2dcb7e)
[Merge tag 'sched-urgent-2021-11-28', 5.16-rc3](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=97891bbf38f7)
[Merge tag 'sched_urgent_for_v5.16_rc4', 5.16-rc4](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1d213767dc6f) | +| 5.16 | 2022/01/09 | [Merge tag 'sched-core-2021-11-01', 5.16-rc1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9a7e0a90a454a7826ecbca055a6ec9271b70c686)
[Merge tag 'sched_urgent_for_v5.16_rc1', 5.16-rc3](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=fc661f2dcb7e)
[Merge tag 'sched-urgent-2021-11-28', 5.16-rc3](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=97891bbf38f7)
[Merge tag 'sched_urgent_for_v5.16_rc4', 5.16-rc4](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1d213767dc6f)
[Merge tag 'sched-urgent-2021-12-12', 5.16-rc5](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=773602256a2ca73455b0baeae5737c4a9ed6ef49) | cgit 上查看 sched 所有的 log 信息 :