[PATCH net-next 0/2] drivers/s390/net: Enable CONTEXT_ANALYSIS

[PATCH net-next 0/2] drivers/s390/net: Enable CONTEXT_ANALYSIS

[Heiko Carstens]

Enable CONTEXT_ANALYSYS for drivers/s390/net.

Static code checking for acquiring and releasing locks used to be done
with sparse. That was removed with [1] and replaced with a clang based
approach [2]. The new approach requires that each subsystem needs to be
explicitly enabled for checking.

Do that for drivers/s390/net. In order to avoid false positives due to
conditional locking, disable context analysis for specific functions of
the ctcm driver. All comments in ctcm indicate that the code should stay
as it is, and that such warnings should be ignored.

Disabling context analysis for specific functions keeps analysis enabled
for the rest of a file.

[1] 5b63d0ae94cc ("compiler-context-analysis: Remove Sparse support")
[2] 3269701cb256 ("compiler-context-analysis: Add infrastructure for Context Analysis with Clang")

Heiko Carstens (2):
  s390/ctcm: Add __context_unsafe() attribute to various functions
  drivers/s390/net: Enable CONTEXT_ANALYSIS

 drivers/s390/net/Makefile    |  2 ++
 drivers/s390/net/ctcm_fsms.c | 20 +++++++-------------
 drivers/s390/net/ctcm_mpc.c  |  6 ++----
 3 files changed, 11 insertions(+), 17 deletions(-)

-- 
2.53.0

[PATCH net-next 1/2] s390/ctcm: Add __context_unsafe() attribute to various functions

[Heiko Carstens]

Disable context analysis for various functions to get rid of context
analysis compile time warnings using clang caused by conditional
locking like e.g.:

drivers/s390/net/ctcm_fsms.c:1457:8:
  warning: spinlock 'arg->cdev->ccwlock' is not held on every path through here
drivers/s390/net/ctcm_fsms.c:1459:4:
  warning: releasing spinlock 'arg->cdev->ccwlock' that was not held

Use __context_unsafe() to provide a short comment why context analysis is
disabled for each function. Each of those functions already contains a
comment that the (previous) sparse context analysis warnings due to
conditional locking should be ignored.

Remove those comments everywhere and use the __context_unsafe() attribute
instead.

Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
---
 drivers/s390/net/ctcm_fsms.c | 20 +++++++-------------
 drivers/s390/net/ctcm_mpc.c  |  6 ++----
 2 files changed, 9 insertions(+), 17 deletions(-)

diff --git a/drivers/s390/net/ctcm_fsms.c b/drivers/s390/net/ctcm_fsms.c
index bf917f426453..84fd394d3525 100644
--- a/drivers/s390/net/ctcm_fsms.c
+++ b/drivers/s390/net/ctcm_fsms.c
@@ -545,6 +545,7 @@ static void chx_rxidle(fsm_instance *fi, int event, void *arg)
  * arg		Generic pointer, casted from channel * upon call.
  */
 static void ctcm_chx_setmode(fsm_instance *fi, int event, void *arg)
+__context_unsafe(/* Conditional locking */)
 {
 	struct channel *ch = arg;
 	int rc;
@@ -563,8 +564,6 @@ static void ctcm_chx_setmode(fsm_instance *fi, int event, void *arg)
 
 	if (event == CTC_EVENT_TIMER)	/* only for timer not yet locked */
 		spin_lock_irqsave(get_ccwdev_lock(ch->cdev), saveflags);
-			/* Such conditional locking is undeterministic in
-			 * static view. => ignore sparse warnings here. */
 
 	rc = ccw_device_start(ch->cdev, &ch->ccw[6], 0, 0xff, 0);
 	if (event == CTC_EVENT_TIMER)	/* see above comments */
@@ -648,6 +647,7 @@ static void ctcm_chx_start(fsm_instance *fi, int event, void *arg)
  * arg		Generic pointer, casted from channel * upon call.
  */
 static void ctcm_chx_haltio(fsm_instance *fi, int event, void *arg)
+__context_unsafe(/* Conditional locking */)
 {
 	struct channel *ch = arg;
 	unsigned long saveflags = 0;
@@ -662,15 +662,12 @@ static void ctcm_chx_haltio(fsm_instance *fi, int event, void *arg)
 
 	if (event == CTC_EVENT_STOP)	/* only for STOP not yet locked */
 		spin_lock_irqsave(get_ccwdev_lock(ch->cdev), saveflags);
-			/* Such conditional locking is undeterministic in
-			 * static view. => ignore sparse warnings here. */
 	oldstate = fsm_getstate(fi);
 	fsm_newstate(fi, CTC_STATE_TERM);
 	rc = ccw_device_halt(ch->cdev, 0);
 
 	if (event == CTC_EVENT_STOP)
 		spin_unlock_irqrestore(get_ccwdev_lock(ch->cdev), saveflags);
-			/* see remark above about conditional locking */
 
 	if (rc != 0 && rc != -EBUSY) {
 		fsm_deltimer(&ch->timer);
@@ -824,6 +821,7 @@ static void ctcm_chx_setuperr(fsm_instance *fi, int event, void *arg)
  * arg		Generic pointer, casted from channel * upon call.
  */
 static void ctcm_chx_restart(fsm_instance *fi, int event, void *arg)
+__context_unsafe(/* Conditional locking */)
 {
 	struct channel *ch = arg;
 	struct net_device *dev = ch->netdev;
@@ -842,9 +840,6 @@ static void ctcm_chx_restart(fsm_instance *fi, int event, void *arg)
 	fsm_newstate(fi, CTC_STATE_STARTWAIT);
 	if (event == CTC_EVENT_TIMER)	/* only for timer not yet locked */
 		spin_lock_irqsave(get_ccwdev_lock(ch->cdev), saveflags);
-			/* Such conditional locking is a known problem for
-			 * sparse because its undeterministic in static view.
-			 * Warnings should be ignored here. */
 	rc = ccw_device_halt(ch->cdev, 0);
 	if (event == CTC_EVENT_TIMER)
 		spin_unlock_irqrestore(get_ccwdev_lock(ch->cdev), saveflags);
@@ -999,6 +994,7 @@ static void ctcm_chx_txiniterr(fsm_instance *fi, int event, void *arg)
  * arg		Generic pointer, casted from channel * upon call.
  */
 static void ctcm_chx_txretry(fsm_instance *fi, int event, void *arg)
+__context_unsafe(/* Conditional locking */)
 {
 	struct channel *ch = arg;
 	struct net_device *dev = ch->netdev;
@@ -1042,9 +1038,6 @@ static void ctcm_chx_txretry(fsm_instance *fi, int event, void *arg)
 		fsm_addtimer(&ch->timer, 1000, CTC_EVENT_TIMER, ch);
 		if (event == CTC_EVENT_TIMER) /* for TIMER not yet locked */
 			spin_lock_irqsave(get_ccwdev_lock(ch->cdev), saveflags);
-			/* Such conditional locking is a known problem for
-			 * sparse because its undeterministic in static view.
-			 * Warnings should be ignored here. */
 		if (do_debug_ccw)
 			ctcmpc_dumpit((char *)&ch->ccw[3],
 					sizeof(struct ccw1) * 3);
@@ -1383,6 +1376,7 @@ static void ctcmpc_chx_txdone(fsm_instance *fi, int event, void *arg)
  * arg		Generic pointer, casted from channel * upon call.
  */
 static void ctcmpc_chx_rx(fsm_instance *fi, int event, void *arg)
+__context_unsafe(/* Conditional locking */)
 {
 	struct channel		*ch = arg;
 	struct net_device	*dev = ch->netdev;
@@ -1462,7 +1456,7 @@ static void ctcmpc_chx_rx(fsm_instance *fi, int event, void *arg)
 			spin_lock_irqsave(
 				get_ccwdev_lock(ch->cdev), saveflags);
 		rc = ccw_device_start(ch->cdev, &ch->ccw[0], 0, 0xff, 0);
-		if (dolock) /* see remark about conditional locking */
+		if (dolock)
 			spin_unlock_irqrestore(
 				get_ccwdev_lock(ch->cdev), saveflags);
 		if (rc != 0)
@@ -1539,6 +1533,7 @@ static void ctcmpc_chx_firstio(fsm_instance *fi, int event, void *arg)
  * arg		Generic pointer, casted from channel * upon call.
  */
 void ctcmpc_chx_rxidle(fsm_instance *fi, int event, void *arg)
+__context_unsafe(/* Conditional locking */)
 {
 	struct channel *ch = arg;
 	struct net_device *dev = ch->netdev;
@@ -1566,7 +1561,6 @@ void ctcmpc_chx_rxidle(fsm_instance *fi, int event, void *arg)
 		ch->ccw[1].count = ch->max_bufsize;
 		CTCM_CCW_DUMP((char *)&ch->ccw[0], sizeof(struct ccw1) * 3);
 		if (event == CTC_EVENT_START)
-			/* see remark about conditional locking */
 			spin_lock_irqsave(get_ccwdev_lock(ch->cdev), saveflags);
 		rc = ccw_device_start(ch->cdev, &ch->ccw[0], 0, 0xff, 0);
 		if (event == CTC_EVENT_START)
diff --git a/drivers/s390/net/ctcm_mpc.c b/drivers/s390/net/ctcm_mpc.c
index aeb102537e7f..08e36685e578 100644
--- a/drivers/s390/net/ctcm_mpc.c
+++ b/drivers/s390/net/ctcm_mpc.c
@@ -1647,6 +1647,7 @@ static int mpc_validate_xid(struct mpcg_info *mpcginfo)
  * CTCM_PROTO_MPC only
  */
 static void mpc_action_side_xid(fsm_instance *fsm, void *arg, int side)
+__context_unsafe(/* Conditional locking */)
 {
 	struct channel *ch = arg;
 	int rc = 0;
@@ -1774,9 +1775,6 @@ static void mpc_action_side_xid(fsm_instance *fsm, void *arg, int side)
 	CTCM_D3_DUMP((char *)ch->xid_id, 4);
 
 	if (!in_hardirq()) {
-			 /* Such conditional locking is a known problem for
-			  * sparse because its static undeterministic.
-			  * Warnings should be ignored here. */
 		spin_lock_irqsave(get_ccwdev_lock(ch->cdev), saveflags);
 		gotlock = 1;
 	}
@@ -1784,7 +1782,7 @@ static void mpc_action_side_xid(fsm_instance *fsm, void *arg, int side)
 	fsm_addtimer(&ch->timer, 5000 , CTC_EVENT_TIMER, ch);
 	rc = ccw_device_start(ch->cdev, &ch->ccw[8], 0, 0xff, 0);
 
-	if (gotlock)	/* see remark above about conditional locking */
+	if (gotlock)
 		spin_unlock_irqrestore(get_ccwdev_lock(ch->cdev), saveflags);
 
 	if (rc != 0) {
-- 
2.53.0

[PATCH net-next 2/2] drivers/s390/net: Enable CONTEXT_ANALYSIS

[Heiko Carstens]

All drivers in drivers/s390/net pass clang's compile time context
analysis. Therefore enable CONTEXT_ANALYSIS.

Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
---
 drivers/s390/net/Makefile | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/s390/net/Makefile b/drivers/s390/net/Makefile
index 537514cc52fb..038ba3e4005d 100644
--- a/drivers/s390/net/Makefile
+++ b/drivers/s390/net/Makefile
@@ -3,6 +3,8 @@
 # S/390 network devices
 #
 
+CONTEXT_ANALYSIS := y
+
 ctcm-y += ctcm_main.o ctcm_fsms.o ctcm_mpc.o ctcm_sysfs.o ctcm_dbug.o
 obj-$(CONFIG_CTCM) += ctcm.o fsm.o
 obj-$(CONFIG_SMSGIUCV) += smsgiucv.o
-- 
2.53.0

Re: [PATCH net-next 0/2] drivers/s390/net: Enable CONTEXT_ANALYSIS

[Alexandra Winter]

On 05.08.26 16:50, Heiko Carstens wrote:
> Enable CONTEXT_ANALYSYS for drivers/s390/net.
> 
> Static code checking for acquiring and releasing locks used to be done
> with sparse. That was removed with [1] and replaced with a clang based
> approach [2]. The new approach requires that each subsystem needs to be
> explicitly enabled for checking.
> 
> Do that for drivers/s390/net. In order to avoid false positives due to
> conditional locking, disable context analysis for specific functions of
> the ctcm driver. All comments in ctcm indicate that the code should stay
> as it is, and that such warnings should be ignored.
> 
> Disabling context analysis for specific functions keeps analysis enabled
> for the rest of a file.
> 
> [1] 5b63d0ae94cc ("compiler-context-analysis: Remove Sparse support")
> [2] 3269701cb256 ("compiler-context-analysis: Add infrastructure for Context Analysis with Clang")
> 
> Heiko Carstens (2):
>   s390/ctcm: Add __context_unsafe() attribute to various functions
>   drivers/s390/net: Enable CONTEXT_ANALYSIS
> 
>  drivers/s390/net/Makefile    |  2 ++
>  drivers/s390/net/ctcm_fsms.c | 20 +++++++-------------
>  drivers/s390/net/ctcm_mpc.c  |  6 ++----
>  3 files changed, 11 insertions(+), 17 deletions(-)
> 

Acked-by: Alexandra Winter <wintera@linux.ibm.com>

Re: [PATCH net-next 1/2] s390/ctcm: Add __context_unsafe() attribute to various functions

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Unsafe conditional locking using `!in_hardirq()` leads to guaranteed self-deadlocks on PREEMPT_RT or when using forced threaded interrupts (`threadirqs`).
--

--- Patch [1]: [PATCH net-next 1/2] s390/ctcm: Add __context_unsafe() attribute to various functions ---
commit c3d2fbe266a854e35b040af4b1c68b39c513a193
Author: Heiko Carstens <hca@linux.ibm.com>
Subject: s390/ctcm: Add __context_unsafe() attribute to various functions

Disable context analysis for various functions to get rid of context
analysis compile time warnings using clang caused by conditional
locking.

> diff --git a/drivers/s390/net/ctcm_fsms.c b/drivers/s390/net/ctcm_fsms.c
> index bf917f4264532..84fd394d35250 100644
> --- a/drivers/s390/net/ctcm_fsms.c
> +++ b/drivers/s390/net/ctcm_fsms.c

[ ... ]

> @@ -1462,7 +1456,7 @@ static void ctcmpc_chx_rx(fsm_instance *fi, int event, void *arg)
>  			spin_lock_irqsave(
>  				get_ccwdev_lock(ch->cdev), saveflags);
>  		rc = ccw_device_start(ch->cdev, &ch->ccw[0], 0, 0xff, 0);
> -		if (dolock) /* see remark about conditional locking */
> +		if (dolock)
>  			spin_unlock_irqrestore(
>  				get_ccwdev_lock(ch->cdev), saveflags);

[Severity: High]
This is a pre-existing issue, but does this conditional locking using
!in_hardirq() (which sets dolock earlier in ctcmpc_chx_rx) cause
self-deadlocks on PREEMPT_RT or when using forced threaded interrupts?

Since the CIO interrupt handler already holds the CCW device lock, when
threaded interrupts are enabled, the hardware interrupt handler runs in
process context and in_hardirq() evaluates to false.

Would this cause ctcmpc_chx_rx() to mistakenly attempt to re-acquire the lock
it already holds, leading to an immediate self-deadlock?

[ ... ]

> diff --git a/drivers/s390/net/ctcm_mpc.c b/drivers/s390/net/ctcm_mpc.c
> index aeb102537e7f5..08e36685e5788 100644
> --- a/drivers/s390/net/ctcm_mpc.c
> +++ b/drivers/s390/net/ctcm_mpc.c

[ ... ]

> @@ -1774,9 +1775,6 @@ static void mpc_action_side_xid(fsm_instance *fsm, void *arg, int side)
>  	CTCM_D3_DUMP((char *)ch->xid_id, 4);
>  
>  	if (!in_hardirq()) {
> -			 /* Such conditional locking is a known problem for
> -			  * sparse because its static undeterministic.
> -			  * Warnings should be ignored here. */
>  		spin_lock_irqsave(get_ccwdev_lock(ch->cdev), saveflags);
>  		gotlock = 1;
>  	}

[Severity: High]
This is a pre-existing issue, but does checking !in_hardirq() for conditional
locking lead to the same self-deadlock here in mpc_action_side_xid()?

If this FSM action is executed from the interrupt handler while threaded
interrupts are active, !in_hardirq() will evaluate to true, causing an attempt
to acquire the already-held get_ccwdev_lock().

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260805145032.1409325-1-hca@linux.ibm.com?part=1

Re: [PATCH net-next 2/2] drivers/s390/net: Enable CONTEXT_ANALYSIS

[sashiko-bot]

> All drivers in drivers/s390/net pass clang's compile time context
> analysis. Therefore enable CONTEXT_ANALYSIS.
> 
> Signed-off-by: Heiko Carstens <hca@linux.ibm.com>

Sashiko has reviewed this patch and found no issues. It looks great!

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260805145032.1409325-1-hca@linux.ibm.com?part=2

Re: [PATCH net-next 0/2] drivers/s390/net: Enable CONTEXT_ANALYSIS

[Simon Horman]

On Wed, Aug 05, 2026 at 04:50:30PM +0200, Heiko Carstens wrote:
> Enable CONTEXT_ANALYSYS for drivers/s390/net.
> 
> Static code checking for acquiring and releasing locks used to be done
> with sparse. That was removed with [1] and replaced with a clang based
> approach [2]. The new approach requires that each subsystem needs to be
> explicitly enabled for checking.
> 
> Do that for drivers/s390/net. In order to avoid false positives due to
> conditional locking, disable context analysis for specific functions of
> the ctcm driver. All comments in ctcm indicate that the code should stay
> as it is, and that such warnings should be ignored.
> 
> Disabling context analysis for specific functions keeps analysis enabled
> for the rest of a file.
> 
> [1] 5b63d0ae94cc ("compiler-context-analysis: Remove Sparse support")
> [2] 3269701cb256 ("compiler-context-analysis: Add infrastructure for Context Analysis with Clang")
> 
> Heiko Carstens (2):
>   s390/ctcm: Add __context_unsafe() attribute to various functions
>   drivers/s390/net: Enable CONTEXT_ANALYSIS

For the series:

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