The patch below does not apply to the 5.4-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.4.y
git checkout FETCH_HEAD
git cherry-pick -x c7f49dadfcdf27e1f747442e874e9baa52ab7674
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025082105-partridge-unboxed-5a0e@gregkh' --subject-prefix 'PATCH 5.4.y' HEAD^..
Possible dependencies:
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From c7f49dadfcdf27e1f747442e874e9baa52ab7674 Mon Sep 17 00:00:00 2001
From: Eric Biggers <ebiggers(a)kernel.org>
Date: Tue, 8 Jul 2025 12:38:28 -0700
Subject: [PATCH] crypto: x86/aegis - Fix sleeping when disallowed on
PREEMPT_RT
skcipher_walk_done() can call kfree(), which takes a spinlock, which
makes it incorrect to call while preemption is disabled on PREEMPT_RT.
Therefore, end the kernel-mode FPU section before calling
skcipher_walk_done(), and restart it afterwards.
Moreover, pass atomic=false to skcipher_walk_aead_encrypt() instead of
atomic=true. The point of atomic=true was to make skcipher_walk_done()
safe to call while in a kernel-mode FPU section, but that does not
actually work. So just use the usual atomic=false.
Fixes: 1d373d4e8e15 ("crypto: x86 - Add optimized AEGIS implementations")
Cc: stable(a)vger.kernel.org
Signed-off-by: Eric Biggers <ebiggers(a)kernel.org>
Signed-off-by: Herbert Xu <herbert(a)gondor.apana.org.au>
diff --git a/arch/x86/crypto/aegis128-aesni-glue.c b/arch/x86/crypto/aegis128-aesni-glue.c
index f1b6d40154e3..3cb5c193038b 100644
--- a/arch/x86/crypto/aegis128-aesni-glue.c
+++ b/arch/x86/crypto/aegis128-aesni-glue.c
@@ -119,7 +119,9 @@ crypto_aegis128_aesni_process_crypt(struct aegis_state *state,
walk->dst.virt.addr,
round_down(walk->nbytes,
AEGIS128_BLOCK_SIZE));
+ kernel_fpu_end();
skcipher_walk_done(walk, walk->nbytes % AEGIS128_BLOCK_SIZE);
+ kernel_fpu_begin();
}
if (walk->nbytes) {
@@ -131,7 +133,9 @@ crypto_aegis128_aesni_process_crypt(struct aegis_state *state,
aegis128_aesni_dec_tail(state, walk->src.virt.addr,
walk->dst.virt.addr,
walk->nbytes);
+ kernel_fpu_end();
skcipher_walk_done(walk, 0);
+ kernel_fpu_begin();
}
}
@@ -176,9 +180,9 @@ crypto_aegis128_aesni_crypt(struct aead_request *req,
struct aegis_state state;
if (enc)
- skcipher_walk_aead_encrypt(&walk, req, true);
+ skcipher_walk_aead_encrypt(&walk, req, false);
else
- skcipher_walk_aead_decrypt(&walk, req, true);
+ skcipher_walk_aead_decrypt(&walk, req, false);
kernel_fpu_begin();
The patch below does not apply to the 5.10-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.10.y
git checkout FETCH_HEAD
git cherry-pick -x c7f49dadfcdf27e1f747442e874e9baa52ab7674
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025082104-syrup-acquire-04a4@gregkh' --subject-prefix 'PATCH 5.10.y' HEAD^..
Possible dependencies:
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From c7f49dadfcdf27e1f747442e874e9baa52ab7674 Mon Sep 17 00:00:00 2001
From: Eric Biggers <ebiggers(a)kernel.org>
Date: Tue, 8 Jul 2025 12:38:28 -0700
Subject: [PATCH] crypto: x86/aegis - Fix sleeping when disallowed on
PREEMPT_RT
skcipher_walk_done() can call kfree(), which takes a spinlock, which
makes it incorrect to call while preemption is disabled on PREEMPT_RT.
Therefore, end the kernel-mode FPU section before calling
skcipher_walk_done(), and restart it afterwards.
Moreover, pass atomic=false to skcipher_walk_aead_encrypt() instead of
atomic=true. The point of atomic=true was to make skcipher_walk_done()
safe to call while in a kernel-mode FPU section, but that does not
actually work. So just use the usual atomic=false.
Fixes: 1d373d4e8e15 ("crypto: x86 - Add optimized AEGIS implementations")
Cc: stable(a)vger.kernel.org
Signed-off-by: Eric Biggers <ebiggers(a)kernel.org>
Signed-off-by: Herbert Xu <herbert(a)gondor.apana.org.au>
diff --git a/arch/x86/crypto/aegis128-aesni-glue.c b/arch/x86/crypto/aegis128-aesni-glue.c
index f1b6d40154e3..3cb5c193038b 100644
--- a/arch/x86/crypto/aegis128-aesni-glue.c
+++ b/arch/x86/crypto/aegis128-aesni-glue.c
@@ -119,7 +119,9 @@ crypto_aegis128_aesni_process_crypt(struct aegis_state *state,
walk->dst.virt.addr,
round_down(walk->nbytes,
AEGIS128_BLOCK_SIZE));
+ kernel_fpu_end();
skcipher_walk_done(walk, walk->nbytes % AEGIS128_BLOCK_SIZE);
+ kernel_fpu_begin();
}
if (walk->nbytes) {
@@ -131,7 +133,9 @@ crypto_aegis128_aesni_process_crypt(struct aegis_state *state,
aegis128_aesni_dec_tail(state, walk->src.virt.addr,
walk->dst.virt.addr,
walk->nbytes);
+ kernel_fpu_end();
skcipher_walk_done(walk, 0);
+ kernel_fpu_begin();
}
}
@@ -176,9 +180,9 @@ crypto_aegis128_aesni_crypt(struct aead_request *req,
struct aegis_state state;
if (enc)
- skcipher_walk_aead_encrypt(&walk, req, true);
+ skcipher_walk_aead_encrypt(&walk, req, false);
else
- skcipher_walk_aead_decrypt(&walk, req, true);
+ skcipher_walk_aead_decrypt(&walk, req, false);
kernel_fpu_begin();
The patch below does not apply to the 5.15-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.15.y
git checkout FETCH_HEAD
git cherry-pick -x c7f49dadfcdf27e1f747442e874e9baa52ab7674
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025082104-whooping-armband-6944@gregkh' --subject-prefix 'PATCH 5.15.y' HEAD^..
Possible dependencies:
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From c7f49dadfcdf27e1f747442e874e9baa52ab7674 Mon Sep 17 00:00:00 2001
From: Eric Biggers <ebiggers(a)kernel.org>
Date: Tue, 8 Jul 2025 12:38:28 -0700
Subject: [PATCH] crypto: x86/aegis - Fix sleeping when disallowed on
PREEMPT_RT
skcipher_walk_done() can call kfree(), which takes a spinlock, which
makes it incorrect to call while preemption is disabled on PREEMPT_RT.
Therefore, end the kernel-mode FPU section before calling
skcipher_walk_done(), and restart it afterwards.
Moreover, pass atomic=false to skcipher_walk_aead_encrypt() instead of
atomic=true. The point of atomic=true was to make skcipher_walk_done()
safe to call while in a kernel-mode FPU section, but that does not
actually work. So just use the usual atomic=false.
Fixes: 1d373d4e8e15 ("crypto: x86 - Add optimized AEGIS implementations")
Cc: stable(a)vger.kernel.org
Signed-off-by: Eric Biggers <ebiggers(a)kernel.org>
Signed-off-by: Herbert Xu <herbert(a)gondor.apana.org.au>
diff --git a/arch/x86/crypto/aegis128-aesni-glue.c b/arch/x86/crypto/aegis128-aesni-glue.c
index f1b6d40154e3..3cb5c193038b 100644
--- a/arch/x86/crypto/aegis128-aesni-glue.c
+++ b/arch/x86/crypto/aegis128-aesni-glue.c
@@ -119,7 +119,9 @@ crypto_aegis128_aesni_process_crypt(struct aegis_state *state,
walk->dst.virt.addr,
round_down(walk->nbytes,
AEGIS128_BLOCK_SIZE));
+ kernel_fpu_end();
skcipher_walk_done(walk, walk->nbytes % AEGIS128_BLOCK_SIZE);
+ kernel_fpu_begin();
}
if (walk->nbytes) {
@@ -131,7 +133,9 @@ crypto_aegis128_aesni_process_crypt(struct aegis_state *state,
aegis128_aesni_dec_tail(state, walk->src.virt.addr,
walk->dst.virt.addr,
walk->nbytes);
+ kernel_fpu_end();
skcipher_walk_done(walk, 0);
+ kernel_fpu_begin();
}
}
@@ -176,9 +180,9 @@ crypto_aegis128_aesni_crypt(struct aead_request *req,
struct aegis_state state;
if (enc)
- skcipher_walk_aead_encrypt(&walk, req, true);
+ skcipher_walk_aead_encrypt(&walk, req, false);
else
- skcipher_walk_aead_decrypt(&walk, req, true);
+ skcipher_walk_aead_decrypt(&walk, req, false);
kernel_fpu_begin();
The patch below does not apply to the 6.1-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.1.y
git checkout FETCH_HEAD
git cherry-pick -x c7f49dadfcdf27e1f747442e874e9baa52ab7674
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025082103-chamomile-hesitant-52e3@gregkh' --subject-prefix 'PATCH 6.1.y' HEAD^..
Possible dependencies:
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From c7f49dadfcdf27e1f747442e874e9baa52ab7674 Mon Sep 17 00:00:00 2001
From: Eric Biggers <ebiggers(a)kernel.org>
Date: Tue, 8 Jul 2025 12:38:28 -0700
Subject: [PATCH] crypto: x86/aegis - Fix sleeping when disallowed on
PREEMPT_RT
skcipher_walk_done() can call kfree(), which takes a spinlock, which
makes it incorrect to call while preemption is disabled on PREEMPT_RT.
Therefore, end the kernel-mode FPU section before calling
skcipher_walk_done(), and restart it afterwards.
Moreover, pass atomic=false to skcipher_walk_aead_encrypt() instead of
atomic=true. The point of atomic=true was to make skcipher_walk_done()
safe to call while in a kernel-mode FPU section, but that does not
actually work. So just use the usual atomic=false.
Fixes: 1d373d4e8e15 ("crypto: x86 - Add optimized AEGIS implementations")
Cc: stable(a)vger.kernel.org
Signed-off-by: Eric Biggers <ebiggers(a)kernel.org>
Signed-off-by: Herbert Xu <herbert(a)gondor.apana.org.au>
diff --git a/arch/x86/crypto/aegis128-aesni-glue.c b/arch/x86/crypto/aegis128-aesni-glue.c
index f1b6d40154e3..3cb5c193038b 100644
--- a/arch/x86/crypto/aegis128-aesni-glue.c
+++ b/arch/x86/crypto/aegis128-aesni-glue.c
@@ -119,7 +119,9 @@ crypto_aegis128_aesni_process_crypt(struct aegis_state *state,
walk->dst.virt.addr,
round_down(walk->nbytes,
AEGIS128_BLOCK_SIZE));
+ kernel_fpu_end();
skcipher_walk_done(walk, walk->nbytes % AEGIS128_BLOCK_SIZE);
+ kernel_fpu_begin();
}
if (walk->nbytes) {
@@ -131,7 +133,9 @@ crypto_aegis128_aesni_process_crypt(struct aegis_state *state,
aegis128_aesni_dec_tail(state, walk->src.virt.addr,
walk->dst.virt.addr,
walk->nbytes);
+ kernel_fpu_end();
skcipher_walk_done(walk, 0);
+ kernel_fpu_begin();
}
}
@@ -176,9 +180,9 @@ crypto_aegis128_aesni_crypt(struct aead_request *req,
struct aegis_state state;
if (enc)
- skcipher_walk_aead_encrypt(&walk, req, true);
+ skcipher_walk_aead_encrypt(&walk, req, false);
else
- skcipher_walk_aead_decrypt(&walk, req, true);
+ skcipher_walk_aead_decrypt(&walk, req, false);
kernel_fpu_begin();
The patch below does not apply to the 5.4-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable(a)vger.kernel.org>.
To reproduce the conflict and resubmit, you may use the following commands:
git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.4.y
git checkout FETCH_HEAD
git cherry-pick -x 51888393cc64dd0462d0b96c13ab94873abbc030
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025082130-duchess-reflux-c692@gregkh' --subject-prefix 'PATCH 5.4.y' HEAD^..
Possible dependencies:
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From 51888393cc64dd0462d0b96c13ab94873abbc030 Mon Sep 17 00:00:00 2001
From: "Rafael J. Wysocki" <rafael.j.wysocki(a)intel.com>
Date: Wed, 9 Jul 2025 12:41:45 +0200
Subject: [PATCH] PM: runtime: Take active children into account in
pm_runtime_get_if_in_use()
For all practical purposes, there is no difference between the situation
in which a given device is not ignoring children and its active child
count is nonzero and the situation in which its runtime PM usage counter
is nonzero. However, pm_runtime_get_if_in_use() will only increment the
device's usage counter and return 1 in the latter case.
For consistency, make it do so in the former case either by adjusting
pm_runtime_get_conditional() and update the related kerneldoc comments
accordingly.
Fixes: c111566bea7c ("PM: runtime: Add pm_runtime_get_if_active()")
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki(a)intel.com>
Reviewed-by: Ulf Hansson <ulf.hansson(a)linaro.org>
Reviewed-by: Sakari Ailus <sakari.ailus(a)linux.intel.com>
Cc: 5.10+ <stable(a)vger.kernel.org> # 5.10+: c0ef3df8dbae: PM: runtime: Simplify pm_runtime_get_if_active() usage
Cc: 5.10+ <stable(a)vger.kernel.org> # 5.10+
Link: https://patch.msgid.link/12700973.O9o76ZdvQC@rjwysocki.net
diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c
index c55a7c70bc1a..2ba0dfd1de5a 100644
--- a/drivers/base/power/runtime.c
+++ b/drivers/base/power/runtime.c
@@ -1191,10 +1191,12 @@ EXPORT_SYMBOL_GPL(__pm_runtime_resume);
*
* Return -EINVAL if runtime PM is disabled for @dev.
*
- * Otherwise, if the runtime PM status of @dev is %RPM_ACTIVE and either
- * @ign_usage_count is %true or the runtime PM usage counter of @dev is not
- * zero, increment the usage counter of @dev and return 1. Otherwise, return 0
- * without changing the usage counter.
+ * Otherwise, if its runtime PM status is %RPM_ACTIVE and (1) @ign_usage_count
+ * is set, or (2) @dev is not ignoring children and its active child count is
+ * nonero, or (3) the runtime PM usage counter of @dev is not zero, increment
+ * the usage counter of @dev and return 1.
+ *
+ * Otherwise, return 0 without changing the usage counter.
*
* If @ign_usage_count is %true, this function can be used to prevent suspending
* the device when its runtime PM status is %RPM_ACTIVE.
@@ -1216,7 +1218,8 @@ static int pm_runtime_get_conditional(struct device *dev, bool ign_usage_count)
retval = -EINVAL;
} else if (dev->power.runtime_status != RPM_ACTIVE) {
retval = 0;
- } else if (ign_usage_count) {
+ } else if (ign_usage_count || (!dev->power.ignore_children &&
+ atomic_read(&dev->power.child_count) > 0)) {
retval = 1;
atomic_inc(&dev->power.usage_count);
} else {
@@ -1249,10 +1252,16 @@ EXPORT_SYMBOL_GPL(pm_runtime_get_if_active);
* @dev: Target device.
*
* Increment the runtime PM usage counter of @dev if its runtime PM status is
- * %RPM_ACTIVE and its runtime PM usage counter is greater than 0, in which case
- * it returns 1. If the device is in a different state or its usage_count is 0,
- * 0 is returned. -EINVAL is returned if runtime PM is disabled for the device,
- * in which case also the usage_count will remain unmodified.
+ * %RPM_ACTIVE and its runtime PM usage counter is greater than 0 or it is not
+ * ignoring children and its active child count is nonzero. 1 is returned in
+ * this case.
+ *
+ * If @dev is in a different state or it is not in use (that is, its usage
+ * counter is 0, or it is ignoring children, or its active child count is 0),
+ * 0 is returned.
+ *
+ * -EINVAL is returned if runtime PM is disabled for the device, in which case
+ * also the usage counter of @dev is not updated.
*/
int pm_runtime_get_if_in_use(struct device *dev)
{