getrusage: move thread_group_cputime_adjusted() outside of lock_task_sighand()
authorOleg Nesterov <oleg@redhat.com>
Mon, 22 Jan 2024 15:50:50 +0000 (16:50 +0100)
committerAndrew Morton <akpm@linux-foundation.org>
Thu, 8 Feb 2024 05:20:32 +0000 (21:20 -0800)
Patch series "getrusage: use sig->stats_lock", v2.

This patch (of 2):

thread_group_cputime() does its own locking, we can safely shift
thread_group_cputime_adjusted() which does another for_each_thread loop
outside of ->siglock protected section.

This is also preparation for the next patch which changes getrusage() to
use stats_lock instead of siglock, thread_group_cputime() takes the same
lock.  With the current implementation recursive read_seqbegin_or_lock()
is fine, thread_group_cputime() can't enter the slow mode if the caller
holds stats_lock, yet this looks more safe and better performance-wise.

Link: https://lkml.kernel.org/r/20240122155023.GA26169@redhat.com
Link: https://lkml.kernel.org/r/20240122155050.GA26205@redhat.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Dylan Hatch <dylanbhatch@google.com>
Tested-by: Dylan Hatch <dylanbhatch@google.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
kernel/sys.c

index e219fcfa112d863eeef58381d04fd4bab16a1e32..70ad06ad852e5940bc46fc25225e701366cd86c7 100644 (file)
@@ -1785,17 +1785,19 @@ void getrusage(struct task_struct *p, int who, struct rusage *r)
        struct task_struct *t;
        unsigned long flags;
        u64 tgutime, tgstime, utime, stime;
-       unsigned long maxrss = 0;
+       unsigned long maxrss;
+       struct mm_struct *mm;
        struct signal_struct *sig = p->signal;
 
-       memset((char *)r, 0, sizeof (*r));
+       memset(r, 0, sizeof(*r));
        utime = stime = 0;
+       maxrss = 0;
 
        if (who == RUSAGE_THREAD) {
                task_cputime_adjusted(current, &utime, &stime);
                accumulate_thread_rusage(p, r);
                maxrss = sig->maxrss;
-               goto out;
+               goto out_thread;
        }
 
        if (!lock_task_sighand(p, &flags))
@@ -1819,9 +1821,6 @@ void getrusage(struct task_struct *p, int who, struct rusage *r)
                fallthrough;
 
        case RUSAGE_SELF:
-               thread_group_cputime_adjusted(p, &tgutime, &tgstime);
-               utime += tgutime;
-               stime += tgstime;
                r->ru_nvcsw += sig->nvcsw;
                r->ru_nivcsw += sig->nivcsw;
                r->ru_minflt += sig->min_flt;
@@ -1839,19 +1838,24 @@ void getrusage(struct task_struct *p, int who, struct rusage *r)
        }
        unlock_task_sighand(p, &flags);
 
-out:
-       r->ru_utime = ns_to_kernel_old_timeval(utime);
-       r->ru_stime = ns_to_kernel_old_timeval(stime);
+       if (who == RUSAGE_CHILDREN)
+               goto out_children;
 
-       if (who != RUSAGE_CHILDREN) {
-               struct mm_struct *mm = get_task_mm(p);
+       thread_group_cputime_adjusted(p, &tgutime, &tgstime);
+       utime += tgutime;
+       stime += tgstime;
 
-               if (mm) {
-                       setmax_mm_hiwater_rss(&maxrss, mm);
-                       mmput(mm);
-               }
+out_thread:
+       mm = get_task_mm(p);
+       if (mm) {
+               setmax_mm_hiwater_rss(&maxrss, mm);
+               mmput(mm);
        }
+
+out_children:
        r->ru_maxrss = maxrss * (PAGE_SIZE / 1024); /* convert pages to KBs */
+       r->ru_utime = ns_to_kernel_old_timeval(utime);
+       r->ru_stime = ns_to_kernel_old_timeval(stime);
 }
 
 SYSCALL_DEFINE2(getrusage, int, who, struct rusage __user *, ru)