The carrier_lock spinlock protects the carrier detection. While it is
hold, framer_get_status() is called witch in turn takes a mutex.
This is not correct and can lead to a deadlock.

A run with PROVE_LOCKING enabled detected the issue:
  [ BUG: Invalid wait context ]
  ...
  c204ddbc (&framer->mutex){+.+.}-{3:3}, at: framer_get_status+0x40/0x78
  other info that might help us debug this:
  context-{4:4}
  2 locks held by ifconfig/146:
  #0: c0926a38 (rtnl_mutex){+.+.}-{3:3}, at: devinet_ioctl+0x12c/0x664
  #1: c2006a40 (&qmc_hdlc->carrier_lock){....}-{2:2}, at: qmc_hdlc_framer_set_carrier+0x30/0x98

Avoid the spinlock usage and convert carrier_lock to a mutex.

Fixes: 54762918ca85 ("net: wan: fsl_qmc_hdlc: Add framer support")
Cc: stable@vger.kernel.org
Signed-off-by: Herve Codina <herve.codina@bootlin.com>
---
 drivers/net/wan/fsl_qmc_hdlc.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/net/wan/fsl_qmc_hdlc.c b/drivers/net/wan/fsl_qmc_hdlc.c
index c5e7ca793c43..64b4bfa6fea7 100644
--- a/drivers/net/wan/fsl_qmc_hdlc.c
+++ b/drivers/net/wan/fsl_qmc_hdlc.c
@@ -18,6 +18,7 @@
 #include <linux/hdlc.h>
 #include <linux/mod_devicetable.h>
 #include <linux/module.h>
+#include <linux/mutex.h>
 #include <linux/platform_device.h>
 #include <linux/slab.h>
 #include <linux/spinlock.h>
@@ -37,7 +38,7 @@ struct qmc_hdlc {
 	struct qmc_chan *qmc_chan;
 	struct net_device *netdev;
 	struct framer *framer;
-	spinlock_t carrier_lock; /* Protect carrier detection */
+	struct mutex carrier_lock; /* Protect carrier detection */
 	struct notifier_block nb;
 	bool is_crc32;
 	spinlock_t tx_lock; /* Protect tx descriptors */
@@ -60,7 +61,7 @@ static int qmc_hdlc_framer_set_carrier(struct qmc_hdlc *qmc_hdlc)
 	if (!qmc_hdlc->framer)
 		return 0;
 
-	guard(spinlock_irqsave)(&qmc_hdlc->carrier_lock);
+	guard(mutex)(&qmc_hdlc->carrier_lock);
 
 	ret = framer_get_status(qmc_hdlc->framer, &framer_status);
 	if (ret) {
@@ -706,7 +707,7 @@ static int qmc_hdlc_probe(struct platform_device *pdev)
 
 	qmc_hdlc->dev = dev;
 	spin_lock_init(&qmc_hdlc->tx_lock);
-	spin_lock_init(&qmc_hdlc->carrier_lock);
+	mutex_init(&qmc_hdlc->carrier_lock);
 
 	qmc_hdlc->qmc_chan = devm_qmc_chan_get_bychild(dev, dev->of_node);
 	if (IS_ERR(qmc_hdlc->qmc_chan))
-- 
2.45.0

…
> hold, framer_get_status() is called witch in turn takes a mutex.
…

                                      which?

Regards,
Markus

On Tue, Jul 30, 2024 at 08:31:04AM +0200, Herve Codina wrote:
> The carrier_lock spinlock protects the carrier detection. While it is
> hold, framer_get_status() is called witch in turn takes a mutex.
> This is not correct and can lead to a deadlock.
> 
> A run with PROVE_LOCKING enabled detected the issue:
>   [ BUG: Invalid wait context ]
>   ...
>   c204ddbc (&framer->mutex){+.+.}-{3:3}, at: framer_get_status+0x40/0x78
>   other info that might help us debug this:
>   context-{4:4}
>   2 locks held by ifconfig/146:
>   #0: c0926a38 (rtnl_mutex){+.+.}-{3:3}, at: devinet_ioctl+0x12c/0x664
>   #1: c2006a40 (&qmc_hdlc->carrier_lock){....}-{2:2}, at: qmc_hdlc_framer_set_carrier+0x30/0x98
> 
> Avoid the spinlock usage and convert carrier_lock to a mutex.
> 
> Fixes: 54762918ca85 ("net: wan: fsl_qmc_hdlc: Add framer support")
> Cc: stable@vger.kernel.org
> Signed-off-by: Herve Codina <herve.codina@bootlin.com>

Reviewed-by: Simon Horman <horms@kernel.org>