linux-kernel - [RFC 11/16] padata: Cap helpers started to online CPUs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed,  5 Jan 2022 19:46:51 -0500
From:   Daniel Jordan <daniel.m.jordan@...cle.com>
To:     Alexander Duyck <alexanderduyck@...com>,
        Alex Williamson <alex.williamson@...hat.com>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Ben Segall <bsegall@...gle.com>,
        Cornelia Huck <cohuck@...hat.com>,
        Dan Williams <dan.j.williams@...el.com>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Dietmar Eggemann <dietmar.eggemann@....com>,
        Herbert Xu <herbert@...dor.apana.org.au>,
        Ingo Molnar <mingo@...hat.com>,
        Jason Gunthorpe <jgg@...dia.com>,
        Johannes Weiner <hannes@...xchg.org>,
        Josh Triplett <josh@...htriplett.org>,
        Michal Hocko <mhocko@...e.com>, Nico Pache <npache@...hat.com>,
        Pasha Tatashin <pasha.tatashin@...een.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Steffen Klassert <steffen.klassert@...unet.com>,
        Steve Sistare <steven.sistare@...cle.com>,
        Tejun Heo <tj@...nel.org>,
        Tim Chen <tim.c.chen@...ux.intel.com>,
        Vincent Guittot <vincent.guittot@...aro.org>
Cc:     linux-mm@...ck.org, kvm@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-crypto@...r.kernel.org,
        Daniel Jordan <daniel.m.jordan@...cle.com>
Subject: [RFC 11/16] padata: Cap helpers started to online CPUs

padata can start num_possible_cpus() helpers, but this is too many
considering that every job's main thread participates and there may
be fewer online than possible CPUs.

Limit overall concurrency, including main thread(s), to
num_online_cpus() with the padata_works_inuse counter to prevent
CPU-intensive threads flooding the system in case of concurrent jobs.

Signed-off-by: Daniel Jordan <daniel.m.jordan@...cle.com>
---
 kernel/padata.c | 24 ++++++++++++++++++------
 1 file changed, 18 insertions(+), 6 deletions(-)

diff --git a/kernel/padata.c b/kernel/padata.c
index 0f4002ed1518..e27988d3e9ed 100644
--- a/kernel/padata.c
+++ b/kernel/padata.c
@@ -50,6 +50,7 @@ struct padata_work {
 
 static DEFINE_SPINLOCK(padata_works_lock);
 static struct padata_work *padata_works;
+static unsigned int padata_works_inuse;
 static LIST_HEAD(padata_free_works);
 
 struct padata_mt_job_state {
@@ -98,11 +99,16 @@ static struct padata_work *padata_work_alloc(void)
 
 	lockdep_assert_held(&padata_works_lock);
 
-	if (list_empty(&padata_free_works))
-		return NULL;	/* No more work items allowed to be queued. */
+	/* Are more work items allowed to be queued? */
+	if (padata_works_inuse >= num_online_cpus())
+		return NULL;
+
+	if (WARN_ON_ONCE(list_empty(&padata_free_works)))
+		return NULL;
 
 	pw = list_first_entry(&padata_free_works, struct padata_work, pw_list);
 	list_del(&pw->pw_list);
+	++padata_works_inuse;
 	return pw;
 }
 
@@ -111,7 +117,11 @@ static int padata_work_alloc_mt(int nworks, struct list_head *head)
 	int i;
 
 	spin_lock(&padata_works_lock);
-	/* Start at 1 because the current task participates in the job. */
+	/*
+	 * Increment inuse and start iterating at 1 to account for the main
+	 * thread participating in the job with its stack-allocated work.
+	 */
+	++padata_works_inuse;
 	for (i = 1; i < nworks; ++i) {
 		struct padata_work *pw = padata_work_alloc();
 
@@ -128,20 +138,22 @@ static void padata_work_free(struct padata_work *pw)
 {
 	lockdep_assert_held(&padata_works_lock);
 	list_add(&pw->pw_list, &padata_free_works);
+	WARN_ON_ONCE(!padata_works_inuse);
+	--padata_works_inuse;
 }
 
 static void padata_works_free(struct list_head *works)
 {
 	struct padata_work *cur, *next;
 
-	if (list_empty(works))
-		return;
-
 	spin_lock(&padata_works_lock);
 	list_for_each_entry_safe(cur, next, works, pw_list) {
 		list_del(&cur->pw_list);
 		padata_work_free(cur);
 	}
+	/* To account for the main thread finishing its part of the job. */
+	WARN_ON_ONCE(!padata_works_inuse);
+	--padata_works_inuse;
 	spin_unlock(&padata_works_lock);
 }
 
-- 
2.34.1