Em Fri, May 05, 2023 at 01:48:52PM -0700, Namhyung Kim escreveu: > On Fri, May 5, 2023 at 1:46 PM Ian Rogers <irogers@xxxxxxxxxx> wrote: > > > > On Fri, May 5, 2023 at 1:43 PM Jiri Olsa <olsajiri@xxxxxxxxx> wrote: > > > > > > On Fri, May 05, 2023 at 10:04:47AM -0700, Ian Rogers wrote: > > > > On Fri, May 5, 2023 at 9:56 AM Arnaldo Carvalho de Melo <acme@xxxxxxxxxx> wrote: > > > > > > > > > > Em Fri, May 05, 2023 at 10:33:15AM -0300, Arnaldo Carvalho de Melo escreveu: > > > > > > Em Fri, May 05, 2023 at 01:03:14AM +0200, Jiri Olsa escreveu: > > > > > > That with the preserve_access_index isn't needed, we need just the > > > > > > fields that we access in the tools, right? > > > > > > > > > > I'm now doing build test this in many distro containers, without the two > > > > > reverts, i.e. BPF skels continue as opt-out as in my pull request, to > > > > > test build and also for the functionality tests on the tools using such > > > > > bpf skels, see below, no touching of vmlinux nor BTF data during the > > > > > build. > > > > > > > > > > - Arnaldo > > > > > > > > > > From 882adaee50bc27f85374aeb2fbaa5b76bef60d05 Mon Sep 17 00:00:00 2001 > > > > > From: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx> > > > > > Date: Thu, 4 May 2023 19:03:51 -0300 > > > > > Subject: [PATCH 1/1] perf bpf skels: Stop using vmlinux.h generated from BTF, > > > > > use subset of used structs + CO-RE > > > > > > > > > > Linus reported a build break due to using a vmlinux without a BTF elf > > > > > section to generate the vmlinux.h header with bpftool for use in the BPF > > > > > tools in tools/perf/util/bpf_skel/*.bpf.c. > > > > > > > > > > Instead add a vmlinux.h file with the structs needed with the fields the > > > > > tools need, marking the structs with __attribute__((preserve_access_index)), > > > > > so that libbpf's CO-RE code can fixup the struct field offsets. > > > > > > > > > > In some cases the vmlinux.h file that was being generated by bpftool > > > > > from the kernel BTF information was not needed at all, just including > > > > > linux/bpf.h, sometimes linux/perf_event.h was enough as non-UAPI > > > > > types were not being used. > > > > > > > > > > To keep te patch small, include those UAPI headers from the trimmed down > > > > > vmlinux.h file, that then provides the tools with just the structs and > > > > > the subset of its fields needed for them. > > > > > > > > > > Testing it: > > > > > > > > > > # perf lock contention -b find / > /dev/null > > > > > > I tested perf lock con -abv -L rcu_state sleep 1 > > > and needed fix below > > > > > > jirka > > > > I thought this was fixed by: > > https://lore.kernel.org/lkml/20230427234833.1576130-1-namhyung@xxxxxxxxxx/ Those are upstream already: ⬢[acme@toolbox perf-tools]$ git log --oneline torvalds/master | grep -m1 -B1 "perf lock contention: Fix struct rq lock access" b9f82b5c63bf5390 perf lock contention: Rework offset calculation with BPF CO-RE e53de7b65a3ca59a perf lock contention: Fix struct rq lock access ⬢[acme@toolbox perf-tools]$ > > but I think that is just in perf-tools-next. > Right, but we might still need the empty rq definition. Yeah, without the empty struct diff libbpf complains about a mismatch of just a forward declaration as the type for 'runqueues' on the lock_contention.bpf.c file while the kernel has a the type as 'struct rq': [root@quaco ~]# perf lock con -ab sleep 1 libbpf: extern (var ksym) 'runqueues': incompatible types, expected [95] fwd rq, but kernel has [55509] struct rq libbpf: failed to load object 'lock_contention_bpf' libbpf: failed to load BPF skeleton 'lock_contention_bpf': -22 Failed to load lock-contention BPF skeleton lock contention BPF setup failed [root@quaco ~]# Adding: struct rq {}; libbpf is happy: [root@quaco ~]# perf lock con -ab sleep 1 contended total wait max wait avg wait type caller 2 50.64 us 25.38 us 25.32 us spinlock tick_do_update_jiffies64+0x25 1 26.18 us 26.18 us 26.18 us spinlock tick_do_update_jiffies64+0x25 [root@quaco ~]# - Arnaldo
