From 5dfc7a929f5ed80d5ce2905636579e4eda036f6f Mon Sep 17 00:00:00 2001 From: Cheng Jian Date: Sat, 26 Jun 2021 18:28:19 +0800 Subject: [PATCH 1/2] description/bpf: support dump kernel data and struct --- study/kernel/00-DESCRIPTION/BPF.md | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/study/kernel/00-DESCRIPTION/BPF.md b/study/kernel/00-DESCRIPTION/BPF.md index 5ee2963..61856aa 100644 --- a/study/kernel/00-DESCRIPTION/BPF.md +++ b/study/kernel/00-DESCRIPTION/BPF.md @@ -106,7 +106,7 @@ raw_tracepoint 相比 tracepoint -# libbpf-tools +# 3 libbpf-tools ------- @@ -118,6 +118,21 @@ raw_tracepoint 相比 tracepoint [How to use libbpf-tools with ubuntu 18.04 4.15 kernel](https://www.gitmemory.com/issue/iovisor/bcc/3232/761744003) + +# 4 debuging +------- + +[Dumping kernel data structures with BPF](https://lwn.net/Articles/818714) + +| 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | +|:----:|:----:|:---:|:----:|:---------:|:----:| +| 2018/03/01 | Jan H. Schönherr | [bpf: implement bpf based dumping of kernel data structures](http://patches.linaro.org/cover/221148) | BPF 支持 raw tracepoint 跟踪, 提供了对 tracepoint 参数的访问. | v1 ☐ | [PatchWork v2](http://patches.linaro.org/cover/221148) | +| 2020/06/23 | Alan Maguire | [bpf, printk: add BTF-based type printing](https://lore.kernel.org/patchwork/cover/1261591/) | 在 BPF 程序中通过 printk 打印内核结构体 | v3 ☐ | [PatchWork v3,bpf-next,0/8](https://lore.kernel.org/patchwork/cover/1261591/) | +| 2020/08/06 | Alan Maguire | [bpf: add bpf-based bpf_trace_printk()-like support](https://lore.kernel.org/patchwork/cover/1285027) | 在 BPF 程序中通过 trace_printk 打印信息 | RFC ☐ | [PatchWork RFC,bpf-next,0/4](https://lore.kernel.org/patchwork/cover/1285027) | +| 2020/09/28 | Alan Maguire | [bpf: add helpers to support BTF-based kernel data display](https://lore.kernel.org/patchwork/cover/1312835) | 在 BPF 程序中通过 printk 打印内核结构体 | v7 ☐ | [PatchWork v7,bpf-next,0/8](https://lore.kernel.org/patchwork/cover/1312835) | +| 2021/06/19 | Alan Maguire | [libbpf: BTF dumper support for typed data](https://lore.kernel.org/patchwork/cover/1448989) | 引入 btf_dump__dump_type_data, 可以用来输出结构体的信息. | v5 ☐ | [PatchWork v5,bpf-next,0/3](https://lore.kernel.org/patchwork/cover/1448989) | + +
* 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. From 157afba59f3a2223aa25eaff3156a4d2ba424480 Mon Sep 17 00:00:00 2001 From: Cheng Jian Date: Sat, 26 Jun 2021 18:28:43 +0800 Subject: [PATCH 2/2] description/memory: fairness zone allocation --- study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md | 32 ++++++++++++++++++- 1 file changed, 31 insertions(+), 1 deletion(-) diff --git a/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md b/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md index 66db493..8435588 100644 --- a/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md +++ b/study/kernel/00-DESCRIPTION/MEMORY_MANAGER.md @@ -230,6 +230,36 @@ 图中, 如果15对应的页是空闲的, 那么伙伴系统可以分配出连续的16个页框, 而由于15这一个页框被分配出去了, 导致最多只能分配出8个连续的页框. 假如这个页还会被回收回伙伴系统, 那么至少在这段时间内产生了碎片, 而如果更糟的, 如果这个页用来存储一些内核永久性的数据而不会被回收回来, 那么碎片将永远无法消除, 这意味着15这个页所处的最大内存块永远无法被连续的分配出去了. 假如上图中被分配出去的页都是不可移动的页, 那么就可以拿出一段内存, 专门用于分配不可移动页, 虽然在这段内存中有碎片, 但是避免了碎片散布到其他类型的内存中. 在系统中所有的内存都被标识为可移动的！也就是说一开始其他类型都没有属于自己的内存, 而当要分配这些类型的内存时, 就要从可移动类型的内存中夺取一部分过来, 这样可以根据实际情况来分配其他类型的内存. + +### 2.1.1.3 fair allocation +------- + + +| 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | +|:----:|:----:|:---:|:----:|:---------:|:----:| +| 2013/08/02 | Mel Gorman | [mm: improve page aging fairness between zones/nodes](https://lore.kernel.org/patchwork/cover/397316) | NA | v2 ☑ 3.12-rc1 | [PatchWork v2,0/3](https://lore.kernel.org/patchwork/cover/397316) | +| 2013/12/18 | Mel Gorman | [Configurable fair allocation zone policy v4](https://lore.kernel.org/patchwork/cover/397316) | NA | v4 ☐ | [PatchWork RFC v4,0/6](https://lore.kernel.org/patchwork/cover/397316) | +| 2016/04/15 | Mel Gorman | [mm, page_alloc: Reduce cost of fair zone allocation policy retry](https://lore.kernel.org/patchwork/patch/668985) | [Optimise page alloc/free fast paths v3](https://lore.kernel.org/patchwork/cover/668967) 系列中的一个补丁. 降低了 fair zone 分配器的开销. | v3 ☑ 4.7-rc1 | [PatchWork v6 00/28](https://lore.kernel.org/patchwork/cover/668967) | +| 2016/07/08 | Mel Gorman | | [Move LRU page reclaim from zones to nodes v9](https://lore.kernel.org/patchwork/cover/696437)系列中的一个补丁. 公平区域分配策略在区域之间交叉分配请求, 以避免年龄倒置问题, 即回收新页面来平衡区域. LRU 回收现在是基于节点的, 所以这应该不再是一个问题, 公平区域分配策略本身开销也不小, 因此这个补丁移除了它. | v9 ☑ [4.8-rc1](https://kernelnewbies.org/Linux_4.8#Memory_management) | [PatchWork v21](https://lore.kernel.org/patchwork/cover/696437) | + + + +### 2.1.1.5 优化 +------- + + +| 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | +|:----:|:----:|:---:|:----:|:---------:|:----:| +| 2016/04/15 | Mel Gorman | [Optimise page alloc/free fast paths v3](https://lore.kernel.org/patchwork/cover/668967) | 优化 page 申请和释放的快速路径. 优化后
1. 在 free 路径中, 调试检查和页面区域/页面块仍然查找占主导地位, 目前仍没有明显的解决方案. 在 alloc 路径中, 主要的耗时操作是处理 zonelist、新页面准备和 fair zone 分配以及无数的统计更新. | v3 ☑ 4.7-rc1 | [PatchWork v6 00/28](https://lore.kernel.org/patchwork/cover/668967) | + +### 2.1.1.5 重构 +------- + +| 时间 | 作者 | 特性 | 描述 | 是否合入主线 | 链接 | +|:----:|:----:|:---:|:----:|:---------:|:----:| +| 2016/04/15 | Mel Gorman | [Rationalise `__alloc_pages` wrappers](https://patchwork.kernel.org/project/linux-mm/cover/20210225150642.2582252-1-willy@infradead.org) | NA | v3 ☑ 4.7-rc1 | [PatchWork v3,0/7](https://patchwork.kernel.org/project/linux-mm/cover/20210225150642.2582252-1-willy@infradead.org) | + + ### 2.1.1.3 内存水线 ------- @@ -591,7 +621,7 @@ CMA 的做法也是启动时预留, 但不同的是, 它允许这部分内存被 | 2010/11/19 | Mel Gorman | [big chunk memory allocator v4](https://lore.kernel.org/patchwork/cover/224757) | 大块内存分配器 | v4 ☐ | [PatchWork v4](https://lore.kernel.org/patchwork/cover/224757) | | 2012/04/03 | Michal Nazarewicz
*-*-*-*-*-*-*-*
Marek Szyprowski | [Contiguous Memory Allocator](https://lwn.net/Articles/486301) | 实现 CMA | v24 ☑ [3.5-rc1](https://kernelnewbies.org/Linux_3.5#Memory_Management) | [PatchWork v7](https://lore.kernel.org/patchwork/cover/229177)
*-*-*-*-*-*-*-*
[PatchWork v24](https://lore.kernel.org/patchwork/cover/295656) | | 2015/02/12 | Joonsoo Kim | [mm/compaction: enhance compaction finish condition](https://lore.kernel.org/patchwork/patch/542063) | 同样的, 之前 NULL 指针和错误指针的输出也很混乱, 进行了归一化. | v1 ☑ 4.1-rc1 | [PatchWork](https://lore.kernel.org/patchwork/cover/542063)
*-*-*-*-*-*-*-*
[关键 commit 2149cdaef6c0](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=2149cdaef6c0eb59a9edf3b152027392cd66b41f) | -| 2015/02/23 | SeongJae Park | [introduce gcma](https://lore.kernel.org/patchwork/patch/544555) | | RFC v2 ☐ | [PatchWork](https://lore.kernel.org/patchwork/cover/544555) | +| 2015/02/23 | SeongJae Park | [introduce gcma](https://lore.kernel.org/patchwork/patch/544555) | [GCMA(Guaranteed Contiguous Memory Allocator)方案](http://ceur-ws.org/Vol-1464/ewili15_12.pdf), 倾向于使用 writeback 的page cache 和 完成 swap out 的 anonymous pages 来做 seconday client, 进行迁移. 从而确保 primary client 的分配. | RFC v2 ☐ | [PatchWork](https://lore.kernel.org/patchwork/cover/544555), [GitHub](https://github.com/sjp38/linux.gcma/releases/tag/gcma/rfc/v2) |