From: Alexander Shishkin <alexander.shishkin@linux.intel.com>
To: Greg KH <greg@kroah.com>
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>,
Chunyan Zhang <zhang.chunyan@linaro.org>,
laurent.fert@intel.com, yann.fouassier@intel.com,
linux-kernel@vger.kernel.org,
Alexander Shishkin <alexander.shishkin@linux.intel.com>
Subject: [PATCH 18/20] stm class: Fix a race in unlinking
Date: Mon, 15 Feb 2016 19:12:08 +0200 [thread overview]
Message-ID: <1455556330-12473-19-git-send-email-alexander.shishkin@linux.intel.com> (raw)
In-Reply-To: <1455556330-12473-1-git-send-email-alexander.shishkin@linux.intel.com>
There is a window in stm_source_link_drop(), during which the source's
link may change before locks are acquired. When this happens, it throws
a warning, since this is not an expected scenario.
This patch handles the race in such a way that if the link appears to
have changed by the time we took the locks, it will release them and
repeat the whole unlinking procedure from the beginning, unless the
other contender beat us to it.
Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
---
drivers/hwtracing/stm/core.c | 54 ++++++++++++++++++++++++++++++++++----------
1 file changed, 42 insertions(+), 12 deletions(-)
diff --git a/drivers/hwtracing/stm/core.c b/drivers/hwtracing/stm/core.c
index f86f56d853..30181821d9 100644
--- a/drivers/hwtracing/stm/core.c
+++ b/drivers/hwtracing/stm/core.c
@@ -694,18 +694,26 @@ err_free:
}
EXPORT_SYMBOL_GPL(stm_register_device);
-static void __stm_source_link_drop(struct stm_source_device *src,
- struct stm_device *stm);
+static int __stm_source_link_drop(struct stm_source_device *src,
+ struct stm_device *stm);
void stm_unregister_device(struct stm_data *stm_data)
{
struct stm_device *stm = stm_data->stm;
struct stm_source_device *src, *iter;
- int i;
+ int i, ret;
mutex_lock(&stm->link_mutex);
list_for_each_entry_safe(src, iter, &stm->link_list, link_entry) {
- __stm_source_link_drop(src, stm);
+ ret = __stm_source_link_drop(src, stm);
+ /*
+ * src <-> stm link must not change under the same
+ * stm::link_mutex, so complain loudly if it has;
+ * also in this situation ret!=0 means this src is
+ * not connected to this stm and it should be otherwise
+ * safe to proceed with the tear-down of stm.
+ */
+ WARN_ON_ONCE(ret);
}
mutex_unlock(&stm->link_mutex);
@@ -824,22 +832,28 @@ fail_detach:
*
* Caller must hold stm::link_mutex.
*/
-static void __stm_source_link_drop(struct stm_source_device *src,
- struct stm_device *stm)
+static int __stm_source_link_drop(struct stm_source_device *src,
+ struct stm_device *stm)
{
struct stm_device *link;
+ int ret = 0;
lockdep_assert_held(&stm->link_mutex);
- if (src->data->unlink)
- src->data->unlink(src->data);
-
/* for stm::link_list modification, we hold both mutex and spinlock */
spin_lock(&stm->link_lock);
spin_lock(&src->link_lock);
link = srcu_dereference_check(src->link, &stm_source_srcu, 1);
- if (WARN_ON_ONCE(link != stm))
+
+ /*
+ * The linked device may have changed since we last looked, because
+ * we weren't holding the src::link_lock back then; if this is the
+ * case, tell the caller to retry.
+ */
+ if (link != stm) {
+ ret = -EAGAIN;
goto unlock;
+ }
stm_output_free(link, &src->output);
list_del_init(&src->link_entry);
@@ -850,6 +864,11 @@ static void __stm_source_link_drop(struct stm_source_device *src,
unlock:
spin_unlock(&src->link_lock);
spin_unlock(&stm->link_lock);
+
+ if (!ret && src->data->unlink)
+ src->data->unlink(src->data);
+
+ return ret;
}
/**
@@ -865,18 +884,29 @@ unlock:
static void stm_source_link_drop(struct stm_source_device *src)
{
struct stm_device *stm;
- int idx;
+ int idx, ret;
+retry:
idx = srcu_read_lock(&stm_source_srcu);
+ /*
+ * The stm device will be valid for the duration of this
+ * read section, but the link may change before we grab
+ * the src::link_lock in __stm_source_link_drop().
+ */
stm = srcu_dereference(src->link, &stm_source_srcu);
+ ret = 0;
if (stm) {
mutex_lock(&stm->link_mutex);
- __stm_source_link_drop(src, stm);
+ ret = __stm_source_link_drop(src, stm);
mutex_unlock(&stm->link_mutex);
}
srcu_read_unlock(&stm_source_srcu, idx);
+
+ /* if it did change, retry */
+ if (ret == -EAGAIN)
+ goto retry;
}
static ssize_t stm_source_link_show(struct device *dev,
--
2.7.0
next prev parent reply other threads:[~2016-02-15 17:13 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-02-15 16:08 [git pull] stm class/intel_th: Updates for char-misc-next Alexander Shishkin
2016-02-15 16:44 ` Greg KH
2016-02-15 17:11 ` [PATCH 00/20] " Alexander Shishkin
2016-02-15 17:11 ` [PATCH 01/20] stm class: Use a signed return type for stm_find_master_chan Alexander Shishkin
2016-02-15 17:11 ` [PATCH 02/20] stm class: Fix master deallocation in device unregistering Alexander Shishkin
2016-02-15 17:11 ` [PATCH 03/20] intel_th: Depend on HAS_IOMEM Alexander Shishkin
2016-02-15 17:11 ` [PATCH 04/20] intel_th: gth: Remove commented-out code Alexander Shishkin
2016-02-15 17:11 ` [PATCH 05/20] intel_th: Update scratchpad bits according to enabled output activity Alexander Shishkin
2016-02-15 17:11 ` [PATCH 06/20] intel_th: msu: Fix offset for wrapped block Alexander Shishkin
2016-02-15 17:11 ` [PATCH 07/20] intel_th: msu: Release resources on read error Alexander Shishkin
2016-02-15 17:11 ` [PATCH 08/20] intel_th: sth: Sanitize packet callback's return values Alexander Shishkin
2016-02-15 17:11 ` [PATCH 09/20] intel_th: Set root device's drvdata early Alexander Shishkin
2016-02-15 17:12 ` [PATCH 10/20] intel_th: Use real device index in the node names Alexander Shishkin
2016-02-15 17:12 ` [PATCH 11/20] stm class: Use driver's packet callback return value Alexander Shishkin
2016-02-15 17:12 ` [PATCH 12/20] stm class: Support devices with multiple instances Alexander Shishkin
2016-02-15 17:12 ` [PATCH 13/20] stm class: dummy_stm: Create multiple devices Alexander Shishkin
2016-02-15 17:12 ` [PATCH 14/20] stm class: Add heartbeat stm source device Alexander Shishkin
2016-02-15 17:12 ` [PATCH 15/20] stm class: Fix unlocking braino in the error path Alexander Shishkin
2016-02-15 17:12 ` [PATCH 16/20] stm class: Guard output assignment against concurrency Alexander Shishkin
2016-02-15 17:12 ` [PATCH 17/20] stm class: Fix unbalanced module/device refcounting Alexander Shishkin
2016-02-15 17:12 ` Alexander Shishkin [this message]
2016-02-15 17:12 ` [PATCH 19/20] stm class: Plug stm device's unlink callback Alexander Shishkin
2016-02-15 17:12 ` [PATCH 20/20] stm class: dummy_stm: Add link callback for fault injection Alexander Shishkin
2016-02-16 12:29 ` [git pull] stm class/intel_th: Updates for char-misc-next Alexander Shishkin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1455556330-12473-19-git-send-email-alexander.shishkin@linux.intel.com \
--to=alexander.shishkin@linux.intel.com \
--cc=greg@kroah.com \
--cc=laurent.fert@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mathieu.poirier@linaro.org \
--cc=yann.fouassier@intel.com \
--cc=zhang.chunyan@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
Powered by JetHome