diff --git a/study/kernel/00-DESCRIPTION/DEBUGGING.md b/study/kernel/00-DESCRIPTION/DEBUGGING.md index 251c230..52aa70c 100644 --- a/study/kernel/00-DESCRIPTION/DEBUGGING.md +++ b/study/kernel/00-DESCRIPTION/DEBUGGING.md @@ -412,7 +412,7 @@ facebook 在 LPC-2021 公布了其[最新基于 BOLT 优化 Linux 内核的进 BOLT 代码在 [github 开源](https://github.com/facebookincubator/BOLT). -## 13.2 Shrinking the kernel +## 13.3 Shrinking the kernel ------- @@ -522,6 +522,17 @@ Mold 是目前 Unix 链接器的现代替代品, 已经达到了 1.0 版本. 由 2022 年 6 月 Mold 1.3 继续对 LTO 进行了优化. 参见 phoronix 报道 [Mold 1.3 High Speed Linker Released With LTO Improvements](https://www.phoronix.com/scan.php?page=news_item&px=Mold-1.3-Released). + +## 13.9 Compiler Optimization +------- + +[Experimental -O3 Optimizing The Linux Kernel For Better Performance Brought Up Again](https://www.phoronix.com/scan.php?page=news_item&px=O3-Optimize-Kernel-2022-Patches) + +| 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | +|:---:|:----:|:---:|:----:|:---------:|:----:| +| 2022/06/21 | Miko Larsson | [Kconfig: -O3 enablement](https://lore.kernel.org/all/20220621133526.29662-1-mikoxyzzz@gmail.com) | 允许所有架构支持 -O3 编译. | v1 ☐☑✓ | [LORE v1,0/2](https://lore.kernel.org/all/20220621133526.29662-1-mikoxyzzz@gmail.com) | + + # 14 FTRACE ------- diff --git a/study/kernel/00-DESCRIPTION/LOCKING.md b/study/kernel/00-DESCRIPTION/LOCKING.md index 3c61ac6..517ceef 100644 --- a/study/kernel/00-DESCRIPTION/LOCKING.md +++ b/study/kernel/00-DESCRIPTION/LOCKING.md @@ -139,6 +139,16 @@ spinlock 的值出现变化时, 所有试图获取这个 spinlock 的 CPU 都需 | 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) | + +### 1.4.3 +------- + + +| 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | +|:----:|:----:|:---:|:----:|:---------:|:----:| +| 2022/06/21 | guoren@kernel.org | [riscv: Add qspinlock support](https://lore.kernel.org/all/20220621144920.2945595-1-guoren@kernel.org) | TODO | v6 ☐☑✓ | [LORE v5](https://lore.kernel.org/lkml/20220620155404.1968739-1-guoren@kernel.org)
*-*-*-*-*-*-*-*
[LORE v6,0/2](https://lore.kernel.org/all/20220621144920.2945595-1-guoren@kernel.org) | + + ## 1.5 PV_SPINLOCK ------- @@ -235,9 +245,12 @@ PV_SPINLOCKS 的合入引起了[性能问题 Performance overhead of paravirt_op | 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | |:----:|:----:|:---:|:----:|:---------:|:----:| +| 2015/06/22 | Peter Zijlstra | [percpu rwsem -v2](https://lore.kernel.org/all/20150622121623.291363374@infradead.org) | TODO | v2 ☐☑✓ | [LORE v2,0/13](https://lore.kernel.org/all/20150622121623.291363374@infradead.org) | | 2012/08/31 | Mikulas Patocka | [Fix a crash when block device is read and block size is changed at the same time](https://lore.kernel.org/patchwork/cover/323377) | NA | v2 ☑ 3.8-rc1 | [PatchWork 0/4](https://lore.kernel.org/patchwork/cover/323377) | | 2020/11/07 | Oleg Nesterov | [percpu_rw_semaphore: reimplement to not block the readers unnecessarily](https://lore.kernel.org/patchwork/cover/1342950) | NA | v2 ☑ 3.8-rc1 | [PatchWork v2,0/5](https://lore.kernel.org/patchwork/cover/339247), [PatchWork](https://lore.kernel.org/patchwork/cover/339702)
*-*-*-*-*-*-*-*
[PatchWork](https://lore.kernel.org/patchwork/patch/339064) | | 2020/11/18 | Oleg Nesterov | [percpu_rw_semaphore: lockdep + config](https://lore.kernel.org/patchwork/cover/1342950) | NA | v1 ☑ 3.8-rc1 | [PatchWork 0/3](https://lore.kernel.org/patchwork/cover/341521) | +| 2020/01/31 | Peter Zijlstra | [locking: Percpu-rwsem rewrite](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?id=41f0e29190ac9e38099a37abd1a8a4cb1dc21233) | TODO | v1 ☐☑✓ | [LORE v1,0/7](https://lore.kernel.org/all/20200131150703.194229898@infradead.org) | + # 3 MUTEX ------- diff --git a/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md b/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md index 14615c3..ae20dfd 100644 --- a/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md +++ b/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md @@ -3841,6 +3841,11 @@ v2.5 的时候引入了 shrink 机制, 并提供了 API 统一了各个模块的 Facebook 指出他们也面临过同样的问题, 所有的 workload 都需要放到 container 里去执行, 用户需要明确申明需要使用多少内存, 不过其实没人知道自己真的会用到多少内存, 因此用户申请的内存数量都太多了, 也就有了类似的overcommit和reclaim问题. Facebook的方案是采用 [PSI(pressure-stall information)](https://lwn.net/Articles/759781), 根据这个来了解内存是否变得很紧张了, 相应的会把LRU list里最久未用的page砍掉. 假如这个举动导致更多的 refault 发生. 不过通过调整内存的回收就调整的激进程度可以缓和 refault. 从而达到较合理的结果, 同时占用的CPU时间也会小得多. 描述其设计方案的论文在 ASPLOS’22 中发表, 并成为四篇 Best Paper 之一, 参见 [TMO: Transparent Memory Offloading in Datacenters](https://dl.acm.org/doi/10.1145/3503222.3507731). 网络上的分析 [知乎, 大荒落 TMO: Transparent Memory Offloading in Datacenters](https://zhuanlan.zhihu.com/p/477786756). +参见: + +phoronix 报道 [Meta's Transparent Memory Offloading Saves Them 20~32% Of Memory Per Linux Server](https://www.phoronix.com/scan.php?page=news_item&px=Meta-Transparent-TMO). + +Meta(原 Facebook) 博客 [Transparent memory offloading: more memory at a fraction of the cost and power](https://engineering.fb.com/2022/06/20/data-infrastructure/transparent-memory-offloading-more-memory-at-a-fraction-of-the-cost-and-power). | 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 18fd1ba..85661ca 100644 --- a/study/kernel/00-DESCRIPTION/SCHEDULER.md +++ b/study/kernel/00-DESCRIPTION/SCHEDULER.md @@ -214,11 +214,19 @@ CFS的算法和实现都相当简单, 众多的测试表明其性能也非常优 [[Announce] [patch] Modular Scheduler Core and Completely Fair Scheduler [CFS]](https://lore.kernel.org/lkml/20070413202100.GA9957@elte.hu/) +#### 1.1.3.1 CFS 调度器 +------- + | 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | |:-----:|:----:|:----:|:----:|:------------:|:----:| | 2007/07/09 | Ingo Molnar | [CFS Scheduler](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c31f2e8a42c41efa46397732656ddf48cc77593e) | 实现 CFS 调度器. | v1 ☑✓ 2.6.23-rc1 | [LORE](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c31f2e8a42c41efa46397732656ddf48cc77593e) | | 2007/10/15 | Ingo Molnar | [CFS Group Scheduler](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9c63d9c021f375a2708ad79043d6f4dd1291a085) | CFS 支持组调度 | v1 ☑✓ 2.6.24-rc1| [LORE](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=9c63d9c021f375a2708ad79043d6f4dd1291a085) | +#### 1.1.3.2 load_weight +------- + +[Linux CFS 调度器之负荷权重 load_weight--Linux 进程的管理与调度 (二十五）](https://blog.csdn.net/gatieme/article/details/52067665) + ### 1.1.4 CK 的 BFS 和 MuQSS -------