JakeHillion/scx
1
0
Fork 0
mirror of https://github.com/JakeHillion/scx.git synced 2024-11-25 11:00:24 +00:00
Code Issues Packages Projects Releases Wiki Activity

scx_layered: Work around verification failure in antistall_set() on old kernels

Browse Source 
In earlier kernels, the iterator variable wasn't trusted making the verifier
choke on calling kfuncs on its dereferences. Work around by re-looking up
the task by PID.
This commit is contained in:
Tejun Heo 2024-11-18 21:37:36 -10:00
parent 61f378c1cd
commit dbcd233f17
1 changed files with 9 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
scheds/rust/scx_layered/src/bpf/main.bpf.c
View File

@ -2497,7 +2497,7 @@ static bool layered_monitor(void)
*/
u64 antistall_set(u64 dsq_id, u64 jiffies_now)
{
struct task_struct *p;
struct task_struct *__p, *p = NULL;
struct task_ctx *tctx;
s32 cpu;
u64 *antistall_dsq, *delay, cur_delay;
@ -2511,7 +2511,12 @@ u64 antistall_set(u64 dsq_id, u64 jiffies_now)
// verifier
bpf_rcu_read_lock();
bpf_for_each(scx_dsq, p, dsq_id, 0) {
bpf_for_each(scx_dsq, __p, dsq_id, 0) {
/* XXX verifier workaround: drop the following block later */
if (p)
bpf_task_release(p);
if (!(p = bpf_task_from_pid(__p->pid)))
continue;
if (!(tctx = lookup_task_ctx(p)))
goto unlock;
@ -2562,6 +2567,8 @@ look_for_cpu:
}
unlock:
if (p)
bpf_task_release(p);
bpf_rcu_read_unlock();
return 0;
}