All of lore.kernel.org
 help / color / mirror / Atom feed
From: Caleb Connolly <caleb.connolly@linaro.org>
To: Alex Elder <elder@kernel.org>,
	"David S. Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>
Cc: Alex Elder <elder@linaro.org>, netdev@vger.kernel.org
Subject: [PATCH net] net: ipa: disable ipa interrupt during suspend
Date: Sun, 15 Jan 2023 17:59:24 +0000	[thread overview]
Message-ID: <20230115175925.465918-1-caleb.connolly@linaro.org> (raw)

The IPA interrupt can fire when pm_runtime is disabled due to it racing
with the PM suspend/resume code. This causes a splat in the interrupt
handler when it tries to call pm_runtime_get().

Explicitly disable the interrupt in our ->suspend callback, and
re-enable it in ->resume to avoid this. If there is an interrupt pending
it will be handled after resuming. The interrupt is a wake_irq, as a
result even when disabled if it fires it will cause the system to wake
from suspend as well as cancel any suspend transition that may be in
progress. If there is an interrupt pending, the ipa_isr_thread handler
will be called after resuming.

Fixes: 1aac309d3207 ("net: ipa: use autosuspend")
Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
---
 drivers/net/ipa/ipa_interrupt.c | 10 ++++++++++
 drivers/net/ipa/ipa_interrupt.h | 16 ++++++++++++++++
 drivers/net/ipa/ipa_power.c     | 17 +++++++++++++++++
 3 files changed, 43 insertions(+)

diff --git a/drivers/net/ipa/ipa_interrupt.c b/drivers/net/ipa/ipa_interrupt.c
index d458a35839cc..c1b3977e1ae4 100644
--- a/drivers/net/ipa/ipa_interrupt.c
+++ b/drivers/net/ipa/ipa_interrupt.c
@@ -127,6 +127,16 @@ static irqreturn_t ipa_isr_thread(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
+void ipa_interrupt_irq_disable(struct ipa *ipa)
+{
+	disable_irq(ipa->interrupt->irq);
+}
+
+void ipa_interrupt_irq_enable(struct ipa *ipa)
+{
+	enable_irq(ipa->interrupt->irq);
+}
+
 /* Common function used to enable/disable TX_SUSPEND for an endpoint */
 static void ipa_interrupt_suspend_control(struct ipa_interrupt *interrupt,
 					  u32 endpoint_id, bool enable)
diff --git a/drivers/net/ipa/ipa_interrupt.h b/drivers/net/ipa/ipa_interrupt.h
index f31fd9965fdc..8a1bd5b89393 100644
--- a/drivers/net/ipa/ipa_interrupt.h
+++ b/drivers/net/ipa/ipa_interrupt.h
@@ -85,6 +85,22 @@ void ipa_interrupt_suspend_clear_all(struct ipa_interrupt *interrupt);
  */
 void ipa_interrupt_simulate_suspend(struct ipa_interrupt *interrupt);
 
+/**
+ * ipa_interrupt_irq_enable() - Enable IPA interrupts
+ * @ipa:	IPA pointer
+ *
+ * This enables the IPA interrupt line
+ */
+void ipa_interrupt_irq_enable(struct ipa *ipa);
+
+/**
+ * ipa_interrupt_irq_disable() - Disable IPA interrupts
+ * @ipa:	IPA pointer
+ *
+ * This disables the IPA interrupt line
+ */
+void ipa_interrupt_irq_disable(struct ipa *ipa);
+
 /**
  * ipa_interrupt_config() - Configure the IPA interrupt framework
  * @ipa:	IPA pointer
diff --git a/drivers/net/ipa/ipa_power.c b/drivers/net/ipa/ipa_power.c
index 8420f93128a2..8057be8cda80 100644
--- a/drivers/net/ipa/ipa_power.c
+++ b/drivers/net/ipa/ipa_power.c
@@ -181,6 +181,17 @@ static int ipa_suspend(struct device *dev)
 
 	__set_bit(IPA_POWER_FLAG_SYSTEM, ipa->power->flags);
 
+	/* Increment the disable depth to ensure that the IRQ won't
+	 * be re-enabled until the matching _enable call in
+	 * ipa_resume(). We do this to ensure that the interrupt
+	 * handler won't run whilst PM runtime is disabled.
+	 *
+	 * Note that disabling the IRQ is NOT the same as disabling
+	 * irq wake. If wakeup is enabled for the IPA then the IRQ
+	 * will still cause the system to wake up, see irq_set_irq_wake().
+	 */
+	ipa_interrupt_irq_disable(ipa);
+
 	return pm_runtime_force_suspend(dev);
 }
 
@@ -193,6 +204,12 @@ static int ipa_resume(struct device *dev)
 
 	__clear_bit(IPA_POWER_FLAG_SYSTEM, ipa->power->flags);
 
+	/* Now that PM runtime is enabled again it's safe
+	 * to turn the IRQ back on and process any data
+	 * that was received during suspend.
+	 */
+	ipa_interrupt_irq_enable(ipa);
+
 	return ret;
 }
 
-- 
2.39.0


             reply	other threads:[~2023-01-15 18:00 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-15 17:59 Caleb Connolly [this message]
2023-01-17 18:29 ` [PATCH net] net: ipa: disable ipa interrupt during suspend Alex Elder
2023-01-18  3:10 ` patchwork-bot+netdevbpf
2023-01-18 10:19 ` [PATCH net] net: ipa: disable ipa interrupt during suspend: manual merge Matthieu Baerts

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=20230115175925.465918-1-caleb.connolly@linaro.org \
    --to=caleb.connolly@linaro.org \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=elder@kernel.org \
    --cc=elder@linaro.org \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.