JakeHillion/scx
1
0
Fork 0
mirror of https://github.com/JakeHillion/scx.git synced 2024-11-30 04:50:24 +00:00
Code Issues Packages Projects Releases Wiki Activity

scx_userland: improve BPF logging

Browse Source 
Always report task comm, nr_queued and nr_scheduled in the log messages.
Moreover, report also task name (comm) and cpu when possible.

All these extra information can be really helpful to trace and debug
scheduling issues.

Signed-off-by: Andrea Righi <andrea.righi@canonical.com>
This commit is contained in:
Andrea Righi 2024-01-07 11:25:11 +01:00
parent 295873ac41
commit 8ccbbdadee
1 changed files with 27 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

42
scheds/rust/scx_rustland/src/bpf/main.bpf.c
View File

@ -86,9 +86,10 @@ volatile u64 nr_failed_dispatches, nr_sched_congested;
const volatile bool debug;
/* Allow to use bpf_printk() only when @debug is set */
#define dbg_msg(_fmt, ...) do { \
if (debug) \
bpf_printk(_fmt, ##__VA_ARGS__); \
#define dbg_msg(_fmt, ...) do { \
if (debug) \
bpf_printk(_fmt " nr_queued=%lu nr_scheduled=%llu", \
##__VA_ARGS__, nr_queued, nr_scheduled); \
} while(0)
/*
@ -281,7 +282,7 @@ static bool is_usersched_needed(void)
*/
static void dispatch_local(struct task_struct *p, u64 enq_flags)
{
dbg_msg("%s: pid=%d", __func__, p->pid);
dbg_msg("dispatch: pid=%d (%s) cpu=local", p->pid, p->comm);
scx_bpf_dispatch(p, SCX_DSQ_LOCAL, slice_ns, enq_flags | SCX_ENQ_LOCAL);
}
@ -340,12 +341,12 @@ static void dispatch_task(struct task_struct *p, s32 cpu, u64 enq_flags)
* simply report the event as a "dispatch failure" and keep
* going.
*/
dbg_msg("%s: pid=%d (fail)", __func__, p->pid);
dbg_msg("dispatch: pid=%d (%s) cpu=any", p->pid, p->comm);
__sync_fetch_and_add(&nr_failed_dispatches, 1);
scx_bpf_dispatch(p, SCX_DSQ_GLOBAL, slice_ns, enq_flags);
return;
}
dbg_msg("%s: pid=%d cpu=%ld", __func__, p->pid, cpu);
dbg_msg("dispatch: pid=%d (%s) cpu=%ld", p->pid, p->comm, cpu);
scx_bpf_dispatch(p, SCX_DSQ_LOCAL_ON | cpu, slice_ns,
enq_flags | SCX_ENQ_LOCAL);
}
@ -450,6 +451,15 @@ static void get_task_info(struct queued_task_ctx *task,
task->cpu = scx_bpf_task_cpu(p);
}
/*
* User-space scheduler is congested: log that and increment congested counter.
*/
static void sched_congested(struct task_struct *p)
{
dbg_msg("congested: pid=%d (%s)", p->pid, p->comm);
__sync_fetch_and_add(&nr_sched_congested, 1);
}
/*
* Task @p becomes ready to run. We can dispatch the task directly here if the
* user-space scheduler is not required, or enqueue it to be processed by the
@ -485,10 +495,9 @@ void BPF_STRUCT_OPS(rustland_enqueue, struct task_struct *p, u64 enq_flags)
* previously used CPU in this case).
*/
get_task_info(&task, p, false);
dbg_msg("enqueue: pid=%d", task.pid);
dbg_msg("enqueue: pid=%d (%s)", p->pid, p->comm);
if (bpf_map_push_elem(&queued, &task, 0)) {
dbg_msg("scheduler congested: pid=%d", task.pid);
__sync_fetch_and_add(&nr_sched_congested, 1);
sched_congested(p);
dispatch_task(p, -1, enq_flags);
__sync_fetch_and_add(&nr_kernel_dispatches, 1);
return;
@ -572,13 +581,15 @@ void BPF_STRUCT_OPS(rustland_dispatch, s32 cpu, struct task_struct *prev)
*/
void BPF_STRUCT_OPS(rustland_running, struct task_struct *p)
{
dbg_msg("start: pid=%d (%s)", p->pid, p->comm);
s32 cpu = scx_bpf_task_cpu(p);
dbg_msg("start: pid=%d (%s) cpu=%ld", p->pid, p->comm, cpu);
/*
* Mark the CPU as busy by setting the pid as owner (ignoring the
* user-space scheduler).
*/
if (!is_usersched_task(p))
set_cpu_owner(scx_bpf_task_cpu(p), p->pid);
set_cpu_owner(cpu, p->pid);
}
/*
@ -586,7 +597,9 @@ void BPF_STRUCT_OPS(rustland_running, struct task_struct *p)
*/
void BPF_STRUCT_OPS(rustland_stopping, struct task_struct *p, bool runnable)
{
dbg_msg("stop: pid=%d (%s)", p->pid, p->comm);
s32 cpu = scx_bpf_task_cpu(p);
dbg_msg("stop: pid=%d (%s) cpu=%ld", p->pid, p->comm, cpu);
/*
* Mark the CPU as idle by setting the owner to 0.
*/
@ -654,7 +667,7 @@ void BPF_STRUCT_OPS(rustland_disable, struct task_struct *p)
{
struct queued_task_ctx task = {};
dbg_msg("exiting: pid=%d", task.pid);
dbg_msg("exit: pid=%d (%s)", p->pid, p->comm);
get_task_info(&task, p, true);
if (bpf_map_push_elem(&queued, &task, 0)) {
/*
@ -668,8 +681,7 @@ void BPF_STRUCT_OPS(rustland_disable, struct task_struct *p)
* thing to avoid introducing extra overhead to free up
* resources).
*/
dbg_msg("scheduler congested: pid=%d", task.pid);
__sync_fetch_and_add(&nr_sched_congested, 1);
sched_congested(p);
return;
}
__sync_fetch_and_add(&nr_queued, 1);