linux-kernel - Re: [PATCH] sched_groups are expected to be circular linked list, make it so right after allocation

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]

Message-ID: <1336566096.2527.30.camel@twins>
Date:	Wed, 09 May 2012 14:21:36 +0200
From:	Peter Zijlstra <a.p.zijlstra@...llo.nl>
To:	Igor Mammedov <imammedo@...hat.com>
Cc:	Jiang Liu <liuj97@...il.com>, linux-kernel@...r.kernel.org,
	mingo@...nel.org, pjt@...gle.com, tglx@...utronix.de,
	seto.hidetoshi@...fujitsu.com
Subject: Re: [PATCH] sched_groups are expected to be circular linked list,
 make it so right after allocation

On Wed, 2012-05-09 at 13:58 +0200, Igor Mammedov wrote:
> On 05/09/2012 01:52 PM, Peter Zijlstra wrote:
> > On Wed, 2012-05-09 at 13:44 +0200, Igor Mammedov wrote:
> >> This patch fixes only build_sched_groups path, but there is another fail path
> >> that results in below OOPS.
> >> build_overlap_sched_groups() may exit without setting groups and later it will crash
> >> init_sched_groups_power as well.
> >
> > if that allocation fails? Or is there another fail path?
> 
> build_overlap_sched_groups(struct sched_domain *sd, int cpu)
> ...
>                  if (cpumask_test_cpu(cpu, sg_span))
>                          groups = sg;
> ...
> 
> above test fails and leaves local var groups set to NULL
> and before exit there is:
> 
>    sd->groups = groups;
> 
> which resets sd->groups to NULL 

Cute! So we're building groups for @cpu, for a domain on the same @cpu,
but none of the groups actually span this @cpu. This would imply the
domain doesn't actually contain @cpu.

> and I'm not sure if it is correct at all to skip this
> assignment if groups == NULL. 

It would avoid exploding, but nothing in the above situation is anywhere
near correct.

Does something like the below give any clues as to how we got there?

---
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1043,6 +1043,15 @@ struct sched_domain {
 	unsigned long span[0];
 };
 
+static inline char *sched_domain_name(struct sched_domain *sd)
+{
+#ifdef CONFIG_SCHED_DEBUG
+	return sd->name;
+#else
+	return "";
+#endif
+}
+
 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
 {
 	return to_cpumask(sd->span);
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -5993,6 +5993,22 @@ build_overlap_sched_groups(struct sched_
 		last = sg;
 		last->next = first;
 	}
+	if (!groups) {
+		char str[256];
+
+		printk(KERN_ERR "sched: Topology is hosed for CPU-%d!!\n", cpu);
+		cpulist_scnprintf(str, sizeof(str), sched_domain_span(sd));
+		printk(KERN_ERR "sched: domain: %s %s\n", sched_domain_name(sd), str);
+		
+		sg = first;
+		if (sg) do {
+			cpulist_scnprintf(str, sizeof(str), sched_group_cpus(sg));
+			printk(KERN_ERR "sched:  group: %s\n", str);
+			sg = sg->next;
+		} while (sg != first);
+
+		BUG();
+	}
 	sd->groups = groups;
 
 	return 0;

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/