| Message ID | 20250108010728.207536-1-houtao@huaweicloud.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | bpf: Reduce the use of migrate_{disable|enable}() | expand |
Hello: This series was applied to bpf/bpf-next.git (master) by Alexei Starovoitov <ast@kernel.org>: On Wed, 8 Jan 2025 09:07:12 +0800 you wrote: > From: Hou Tao <houtao1@huawei.com> > > Hi, > > The use of migrate_{disable|enable} pair in BPF is mainly due to the > introduction of bpf memory allocator and the use of per-CPU data struct > in its internal implementation. The caller needs to disable migration > before invoking the alloc or free APIs of bpf memory allocator, and > enable migration after the invocation. > > [...] Here is the summary with links: - [bpf-next,v2,01/16] bpf: Remove migrate_{disable|enable} from LPM trie https://git.kernel.org/bpf/bpf-next/c/1b1a01db17af - [bpf-next,v2,02/16] bpf: Remove migrate_{disable|enable} in ->map_for_each_callback https://git.kernel.org/bpf/bpf-next/c/ea5b229630a6 - [bpf-next,v2,03/16] bpf: Remove migrate_{disable|enable} in htab_elem_free https://git.kernel.org/bpf/bpf-next/c/53f2ba0b1cc0 - [bpf-next,v2,04/16] bpf: Remove migrate_{disable|enable} from bpf_cgrp_storage_lock helpers https://git.kernel.org/bpf/bpf-next/c/25dc65f75b08 - [bpf-next,v2,05/16] bpf: Remove migrate_{disable|enable} from bpf_task_storage_lock helpers https://git.kernel.org/bpf/bpf-next/c/9e6c958b5466 - [bpf-next,v2,06/16] bpf: Disable migration when destroying inode storage https://git.kernel.org/bpf/bpf-next/c/e319cdc89566 - [bpf-next,v2,07/16] bpf: Disable migration when destroying sock storage https://git.kernel.org/bpf/bpf-next/c/7d1032d1e303 - [bpf-next,v2,08/16] bpf: Disable migration when cloning sock storage https://git.kernel.org/bpf/bpf-next/c/dfccfc47bde5 - [bpf-next,v2,09/16] bpf: Disable migration in bpf_selem_free_rcu https://git.kernel.org/bpf/bpf-next/c/090d7f2e640b - [bpf-next,v2,10/16] bpf: Disable migration before calling ops->map_free() https://git.kernel.org/bpf/bpf-next/c/4b7e7cd1c105 - [bpf-next,v2,11/16] bpf: Remove migrate_{disable|enable} in bpf_obj_free_fields() https://git.kernel.org/bpf/bpf-next/c/1d2dbe7120e8 - [bpf-next,v2,12/16] bpf: Remove migrate_{disable,enable} in bpf_cpumask_release() https://git.kernel.org/bpf/bpf-next/c/6a52b965ab6f - [bpf-next,v2,13/16] bpf: Remove migrate_{disable|enable} from bpf_selem_alloc() https://git.kernel.org/bpf/bpf-next/c/2269b32ab00e - [bpf-next,v2,14/16] bpf: Remove migrate_{disable|enable} from bpf_local_storage_alloc() https://git.kernel.org/bpf/bpf-next/c/4855a75ebf48 - [bpf-next,v2,15/16] bpf: Remove migrate_{disable|enable} from bpf_local_storage_free() https://git.kernel.org/bpf/bpf-next/c/7b984359e097 - [bpf-next,v2,16/16] bpf: Remove migrate_{disable|enable} from bpf_selem_free() https://git.kernel.org/bpf/bpf-next/c/d86088e2c35d You are awesome, thank you!
From: Hou Tao <houtao1@huawei.com> Hi, The use of migrate_{disable|enable} pair in BPF is mainly due to the introduction of bpf memory allocator and the use of per-CPU data struct in its internal implementation. The caller needs to disable migration before invoking the alloc or free APIs of bpf memory allocator, and enable migration after the invocation. The main users of bpf memory allocator are various kind of bpf maps in which the map values or the special fields in the map values are allocated by using bpf memory allocator. At present, the running context for bpf program has already disabled migration explictly or implictly, therefore, when these maps are manipulated in bpf program, it is OK to not invoke migrate_disable() and migrate_enable() pair. Howevers, it is not always the case when these maps are manipulated through bpf syscall, therefore many migrate_{disable|enable} pairs are added when the map can either be manipulated by BPF program or BPF syscall. The initial idea of reducing the use of migrate_{disable|enable} comes from Alexei [1]. I turned it into a patch set that archives the goals through the following three methods: 1. remove unnecessary migrate_{disable|enable} pair when the BPF syscall path also disables migration, it is OK to remove the pair. Patch #1~#3 fall into this category, while patch #4~#5 are partially included. 2. move the migrate_{disable|enable} pair from inner callee to outer caller Instead of invoking migrate_disable() in the inner callee, invoking migrate_disable() in the outer caller to simplify reasoning about when migrate_disable() is needed. Patch #4~#5 and patch #6~#19 belongs to this category. 3. add cant_migrate() check in the inner callee Add cant_migrate() check in the inner callee to ensure the guarantee that migration is disabled is not broken. Patch #1~#5, #13, #16~#19 also belong to this category. Please check the individual patches for more details. Comments are always welcome. Change Log: v2: * sqaush the ->map_free related patches (#10~#12, #15) into one patch * remove unnecessary cant_migrate() checks. v1: https://lore.kernel.org/bpf/20250106081900.1665573-1-houtao@huaweicloud.com Hou Tao (16): bpf: Remove migrate_{disable|enable} from LPM trie bpf: Remove migrate_{disable|enable} in ->map_for_each_callback bpf: Remove migrate_{disable|enable} in htab_elem_free bpf: Remove migrate_{disable|enable} from bpf_cgrp_storage_lock helpers bpf: Remove migrate_{disable|enable} from bpf_task_storage_lock helpers bpf: Disable migration when destroying inode storage bpf: Disable migration when destroying sock storage bpf: Disable migration when cloning sock storage bpf: Disable migration in bpf_selem_free_rcu bpf: Disable migration before calling ops->map_free() bpf: Remove migrate_{disable|enable} in bpf_obj_free_fields() bpf: Remove migrate_{disable,enable} in bpf_cpumask_release() bpf: Remove migrate_{disable|enable} from bpf_selem_alloc() bpf: Remove migrate_{disable|enable} from bpf_local_storage_alloc() bpf: Remove migrate_{disable|enable} from bpf_local_storage_free() bpf: Remove migrate_{disable|enable} from bpf_selem_free() kernel/bpf/arraymap.c | 6 ++---- kernel/bpf/bpf_cgrp_storage.c | 15 +++++++-------- kernel/bpf/bpf_inode_storage.c | 9 +++++---- kernel/bpf/bpf_local_storage.c | 30 +++++++++--------------------- kernel/bpf/bpf_task_storage.c | 15 +++++++-------- kernel/bpf/cpumask.c | 2 -- kernel/bpf/hashtab.c | 19 +++++++------------ kernel/bpf/helpers.c | 4 ---- kernel/bpf/lpm_trie.c | 20 ++++---------------- kernel/bpf/range_tree.c | 2 -- kernel/bpf/syscall.c | 10 +++++++--- net/core/bpf_sk_storage.c | 11 +++++++---- 12 files changed, 55 insertions(+), 88 deletions(-)