* pkgtemp: double lock
@ 2010-09-04 11:51 Jiri Slaby
2010-09-04 17:08 ` Fenghua Yu
0 siblings, 1 reply; 2+ messages in thread
From: Jiri Slaby @ 2010-09-04 11:51 UTC (permalink / raw)
To: fenghua.yu; +Cc: Brown, Len, LKML
Hi,
stanse found that you double lock pdev_list_mutex:
-> pkgtemp_device_remove
-> mutex_lock(&pdev_list_mutex);
-> pkgtemp_device_add
-> mutex_lock(&pdev_list_mutex);
Could you fix that?
regards,
--
js
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: pkgtemp: double lock
2010-09-04 11:51 pkgtemp: double lock Jiri Slaby
@ 2010-09-04 17:08 ` Fenghua Yu
0 siblings, 0 replies; 2+ messages in thread
From: Fenghua Yu @ 2010-09-04 17:08 UTC (permalink / raw)
To: Jiri Slaby; +Cc: Yu, Fenghua, Brown, Len, LKML
On Sat, Sep 04, 2010 at 04:51:21AM -0700, Jiri Slaby wrote:
> Hi,
>
> stanse found that you double lock pdev_list_mutex:
> -> pkgtemp_device_remove
> -> mutex_lock(&pdev_list_mutex);
> -> pkgtemp_device_add
> -> mutex_lock(&pdev_list_mutex);
>
> Could you fix that?
>
> regards,
> --
> js
I have already sent a patch to lkml to fix the issue. It's not in upstream yet.
I'll will push it to Linus.
>From fenghua.yu@intel.com Wed Aug 18 16:03:04 2010
From: Fenghua Yu <fenghua.yu@intel.com>
When a sibling is added to dev_list after a cpu is hot-removed, pdev_list_m=
utex
has been locked already. But pkgtemp_device_add() tries to lock pdev_list_m=
utex
again. This is incorrect. The patch fixes this issue.
The patch also removes __cpuinit for pkgtemp_device_add() to avoid section
mismatch warning.
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
---
drivers/hwmon/pkgtemp.c | 18 +++++++++++++-----
1 files changed, 13 insertions(+), 5 deletions(-)
diff --git a/drivers/hwmon/pkgtemp.c b/drivers/hwmon/pkgtemp.c
index 74157fc..928a016 100644
--- a/drivers/hwmon/pkgtemp.c
+++ b/drivers/hwmon/pkgtemp.c
@@ -276,7 +276,7 @@ struct pdev_entry {
static LIST_HEAD(pdev_list);
static DEFINE_MUTEX(pdev_list_mutex);
=20
-static int __cpuinit pkgtemp_device_add(unsigned int cpu)
+static int pkgtemp_device_add(unsigned int cpu)
{
int err;
struct platform_device *pdev;
@@ -341,26 +341,34 @@ static void pkgtemp_device_remove(unsigned int cpu)
{
struct pdev_entry *p, *n;
unsigned int i;
- int err;
=20
- mutex_lock(&pdev_list_mutex);
list_for_each_entry_safe(p, n, &pdev_list, list) {
if (p->cpu !=3D cpu)
continue;
=20
+ mutex_lock(&pdev_list_mutex);
platform_device_unregister(p->pdev);
list_del(&p->list);
kfree(p);
+ mutex_unlock(&pdev_list_mutex);
+ /*
+ * Select one of removed cpu's siblings to represent sensor
+ * for this package.
+ * If there is no more running sibling in a package, the
+ * package sensor for this package is not available to user
+ * space any more.
+ */
for_each_cpu(i, cpu_core_mask(cpu)) {
+ int err;
+
if (i !=3D cpu) {
err =3D pkgtemp_device_add(i);
if (!err)
break;
}
}
- break;
+ return;
}
- mutex_unlock(&pdev_list_mutex);
}
=20
static int __cpuinit pkgtemp_cpu_callback(struct notifier_block *nfb,
--=20
1.6.0.3
^ permalink raw reply related [flat|nested] 2+ messages in thread
end of thread, other threads:[~2010-09-04 17:18 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-09-04 11:51 pkgtemp: double lock Jiri Slaby
2010-09-04 17:08 ` Fenghua Yu
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox