public inbox for linux-s390@vger.kernel.org
 help / color / mirror / Atom feed
From: Harald Freudenberger <freude@linux.ibm.com>
To: dengler@linux.ibm.com, ifranzki@linux.ibm.com,
	fcallies@linux.ibm.com, hca@linux.ibm.com, gor@linux.ibm.com,
	agordeev@linux.ibm.com, seiden@linux.ibm.com,
	borntraeger@linux.ibm.com, frankja@linux.ibm.com,
	imbrenda@linux.ibm.com
Cc: linux-s390@vger.kernel.org, herbert@gondor.apana.org.au
Subject: [PATCH v4 03/24] s390/ap: Introduce ap message buffer pool
Date: Wed,  9 Apr 2025 16:02:44 +0200	[thread overview]
Message-ID: <20250409140305.58900-4-freude@linux.ibm.com> (raw)
In-Reply-To: <20250409140305.58900-1-freude@linux.ibm.com>

There is a need for a do-not-allocate-memory path through the AP bus
layer. The pkey layer may be triggered via the in-kernel interface
from a protected key crypto algorithm (namely PAES) to convert a
secure key into a protected key. This happens in a workqueue context,
so sleeping is allowed but memory allocations causing IO operations
are not permitted.

To accomplish this, an AP message memory pool with pre-allocated space
is established. When ap_init_apmsg() with use_mempool set to true is
called, instead of kmalloc() the ap message buffer is allocated from
the ap_msg_pool. This pool only holds a limited amount of buffers:
ap_msg_pool_min_items with the item size AP_DEFAULT_MAX_MSG_SIZE and
exactly one of these items (if available) is returned if
ap_init_apmsg() with the use_mempool arg set to true is called. When
this pool is exhausted and use_mempool is set true, ap_init_apmsg()
returns -ENOMEM without any attempt to allocate memory and the caller
has to deal with that.

Default values for this mempool of ap messages is:
 * Each buffer is 12KB (that is the default AP bus size
   and all the urgent messages should fit into this space).
 * Minimum items held in the pool is 8. This value is adjustable
   via module parameter ap.msgpool_min_items.

The zcrypt layer may use this flag to indicate to the ap bus that the
processing path for this message should not allocate memory but should
use pre-allocated memory buffer instead.  This is to prevent deadlocks
with crypto and io for example with encrypted swap volumes.

Signed-off-by: Harald Freudenberger <freude@linux.ibm.com>
Reviewed-by: Holger Dengler <dengler@linux.ibm.com>
---
 drivers/s390/crypto/ap_bus.c     | 54 ++++++++++++++++++++++++++++++--
 drivers/s390/crypto/ap_bus.h     |  3 +-
 drivers/s390/crypto/zcrypt_api.c | 10 +++---
 3 files changed, 58 insertions(+), 9 deletions(-)

diff --git a/drivers/s390/crypto/ap_bus.c b/drivers/s390/crypto/ap_bus.c
index 69f2d873748a..aac227397137 100644
--- a/drivers/s390/crypto/ap_bus.c
+++ b/drivers/s390/crypto/ap_bus.c
@@ -40,6 +40,7 @@
 #include <linux/module.h>
 #include <asm/uv.h>
 #include <asm/chsc.h>
+#include <linux/mempool.h>
 
 #include "ap_bus.h"
 #include "ap_debug.h"
@@ -101,6 +102,27 @@ static struct ap_config_info *const ap_qci_info_old = &qci[1];
  */
 debug_info_t *ap_dbf_info;
 
+/*
+ * There is a need for a do-not-allocate-memory path through the AP bus
+ * layer. The pkey layer may be triggered via the in-kernel interface from
+ * a protected key crypto algorithm (namely PAES) to convert a secure key
+ * into a protected key. This happens in a workqueue context, so sleeping
+ * is allowed but memory allocations causing IO operations are not permitted.
+ * To accomplish this, an AP message memory pool with pre-allocated space
+ * is established. When ap_init_apmsg() with use_mempool set to true is
+ * called, instead of kmalloc() the ap message buffer is allocated from
+ * the ap_msg_pool. This pool only holds a limited amount of buffers:
+ * ap_msg_pool_min_items with the item size AP_DEFAULT_MAX_MSG_SIZE and
+ * exactly one of these items (if available) is returned if ap_init_apmsg()
+ * with the use_mempool arg set to true is called. When this pool is exhausted
+ * and use_mempool is set true, ap_init_apmsg() returns -ENOMEM without
+ * any attempt to allocate memory and the caller has to deal with that.
+ */
+static mempool_t *ap_msg_pool;
+static unsigned int ap_msg_pool_min_items = 8;
+module_param_named(msgpool_min_items, ap_msg_pool_min_items, uint, 0440);
+MODULE_PARM_DESC(msgpool_min_items, "AP message pool minimal items.");
+
 /*
  * AP bus rescan related things.
  */
@@ -549,11 +571,22 @@ static void ap_poll_thread_stop(void)
 /*
  * ap_init_apmsg() - Initialize ap_message.
  */
-int ap_init_apmsg(struct ap_message *ap_msg)
+int ap_init_apmsg(struct ap_message *ap_msg, bool use_mempool)
 {
-	unsigned int maxmsgsize = atomic_read(&ap_max_msg_size);
+	unsigned int maxmsgsize;
 
 	memset(ap_msg, 0, sizeof(*ap_msg));
+
+	if (use_mempool) {
+		ap_msg->msg = mempool_alloc_preallocated(ap_msg_pool);
+		if (!ap_msg->msg)
+			return -ENOMEM;
+		ap_msg->bufsize = AP_DEFAULT_MAX_MSG_SIZE;
+		ap_msg->flags |= AP_MSG_FLAG_MEMPOOL;
+		return 0;
+	}
+
+	maxmsgsize = atomic_read(&ap_max_msg_size);
 	ap_msg->msg = kmalloc(maxmsgsize, GFP_KERNEL);
 	if (!ap_msg->msg)
 		return -ENOMEM;
@@ -568,7 +601,12 @@ EXPORT_SYMBOL(ap_init_apmsg);
  */
 void ap_release_apmsg(struct ap_message *ap_msg)
 {
-	kfree_sensitive(ap_msg->msg);
+	if (ap_msg->flags & AP_MSG_FLAG_MEMPOOL) {
+		memzero_explicit(ap_msg->msg, ap_msg->bufsize);
+		mempool_free(ap_msg->msg, ap_msg_pool);
+	} else {
+		kfree_sensitive(ap_msg->msg);
+	}
 }
 EXPORT_SYMBOL(ap_release_apmsg);
 
@@ -2457,6 +2495,14 @@ static int __init ap_module_init(void)
 	/* init ap_queue hashtable */
 	hash_init(ap_queues);
 
+	/* create ap msg buffer memory pool */
+	ap_msg_pool = mempool_create_kmalloc_pool(ap_msg_pool_min_items,
+						  AP_DEFAULT_MAX_MSG_SIZE);
+	if (!ap_msg_pool) {
+		rc = -ENOMEM;
+		goto out;
+	}
+
 	/* set up the AP permissions (ioctls, ap and aq masks) */
 	ap_perms_init();
 
@@ -2503,6 +2549,7 @@ static int __init ap_module_init(void)
 out_bus:
 	bus_unregister(&ap_bus_type);
 out:
+	mempool_destroy(ap_msg_pool);
 	ap_debug_exit();
 	return rc;
 }
@@ -2513,6 +2560,7 @@ static void __exit ap_module_exit(void)
 	ap_irq_exit();
 	root_device_unregister(ap_root_device);
 	bus_unregister(&ap_bus_type);
+	mempool_destroy(ap_msg_pool);
 	ap_debug_exit();
 }
 
diff --git a/drivers/s390/crypto/ap_bus.h b/drivers/s390/crypto/ap_bus.h
index 483231bcdea6..60c41d8116ad 100644
--- a/drivers/s390/crypto/ap_bus.h
+++ b/drivers/s390/crypto/ap_bus.h
@@ -236,8 +236,9 @@ struct ap_message {
 #define AP_MSG_FLAG_SPECIAL  0x0001	/* flag msg as 'special' with NQAP */
 #define AP_MSG_FLAG_USAGE    0x0002	/* CCA, EP11: usage (no admin) msg */
 #define AP_MSG_FLAG_ADMIN    0x0004	/* CCA, EP11: admin (=control) msg */
+#define AP_MSG_FLAG_MEMPOOL  0x0008 /* ap msg buffer allocated via mempool */
 
-int ap_init_apmsg(struct ap_message *ap_msg);
+int ap_init_apmsg(struct ap_message *ap_msg, bool use_mempool);
 void ap_release_apmsg(struct ap_message *ap_msg);
 
 enum ap_sm_wait ap_sm_event(struct ap_queue *aq, enum ap_sm_event event);
diff --git a/drivers/s390/crypto/zcrypt_api.c b/drivers/s390/crypto/zcrypt_api.c
index 78df25da0b95..ad09c5306e28 100644
--- a/drivers/s390/crypto/zcrypt_api.c
+++ b/drivers/s390/crypto/zcrypt_api.c
@@ -648,7 +648,7 @@ static long zcrypt_rsa_modexpo(struct ap_perms *perms,
 
 	trace_s390_zcrypt_req(mex, TP_ICARSAMODEXPO);
 
-	rc = ap_init_apmsg(&ap_msg);
+	rc = ap_init_apmsg(&ap_msg, false);
 	if (rc)
 		goto out;
 
@@ -753,7 +753,7 @@ static long zcrypt_rsa_crt(struct ap_perms *perms,
 
 	trace_s390_zcrypt_req(crt, TP_ICARSACRT);
 
-	rc = ap_init_apmsg(&ap_msg);
+	rc = ap_init_apmsg(&ap_msg, false);
 	if (rc)
 		goto out;
 
@@ -861,7 +861,7 @@ static long _zcrypt_send_cprb(bool userspace, struct ap_perms *perms,
 
 	xcrb->status = 0;
 
-	rc = ap_init_apmsg(&ap_msg);
+	rc = ap_init_apmsg(&ap_msg, false);
 	if (rc)
 		goto out;
 
@@ -1045,7 +1045,7 @@ static long _zcrypt_send_ep11_cprb(bool userspace, struct ap_perms *perms,
 
 	trace_s390_zcrypt_req(xcrb, TP_ZSENDEP11CPRB);
 
-	rc = ap_init_apmsg(&ap_msg);
+	rc = ap_init_apmsg(&ap_msg, false);
 	if (rc)
 		goto out;
 
@@ -1219,7 +1219,7 @@ static long zcrypt_rng(char *buffer)
 
 	trace_s390_zcrypt_req(buffer, TP_HWRNGCPRB);
 
-	rc = ap_init_apmsg(&ap_msg);
+	rc = ap_init_apmsg(&ap_msg, false);
 	if (rc)
 		goto out;
 	rc = prep_rng_ap_msg(&ap_msg, &func_code, &domain);
-- 
2.43.0


  parent reply	other threads:[~2025-04-09 14:03 UTC|newest]

Thread overview: 50+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-04-09 14:02 [PATCH v4 00/24] AP bus/zcrypt/pkey/paes no-mem-alloc patches Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 01/24] s390/ap: Move response_type struct into ap_msg struct Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 02/24] s390/ap/zcrypt: Rework AP message buffer allocation Harald Freudenberger
2025-04-09 14:02 ` Harald Freudenberger [this message]
2025-04-09 14:02 ` [PATCH v4 04/24] s390/zcrypt: Avoid alloc and copy of ep11 targets if kernelspace cprb Harald Freudenberger
2025-04-09 16:19   ` Holger Dengler
2025-04-09 14:02 ` [PATCH v4 05/24] s390/ap/zcrypt: New xflag parameter Harald Freudenberger
2025-04-09 16:25   ` Holger Dengler
2025-04-09 14:02 ` [PATCH v4 06/24] s390/zcrypt: Introduce cprb mempool for cca misc functions Harald Freudenberger
2025-04-11 12:40   ` Holger Dengler
2025-04-14 14:17     ` Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 07/24] s390/zcrypt: Introduce cprb mempool for ep11 " Harald Freudenberger
2025-04-11 12:58   ` Holger Dengler
2025-04-14 14:21     ` Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 08/24] s390/zcrypt: Rework zcrypt function zcrypt_device_status_mask_ext Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 09/24] s390/zcrypt: Introduce pre-allocated device status array for cca misc Harald Freudenberger
2025-04-11 13:06   ` Holger Dengler
2025-04-09 14:02 ` [PATCH v4 10/24] s390/zcrypt: Introduce pre-allocated device status array for ep11 misc Harald Freudenberger
2025-04-11 13:08   ` Holger Dengler
2025-04-09 14:02 ` [PATCH v4 11/24] s390/zcrypt: Remove unused functions from cca misc Harald Freudenberger
2025-04-11 13:10   ` Holger Dengler
2025-04-09 14:02 ` [PATCH v4 12/24] s390/zcrypt: Remove CCA and EP11 card and domain info caches Harald Freudenberger
2025-04-11 13:25   ` Holger Dengler
2025-04-14 14:39     ` Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 13/24] s390/zcrypt/pkey: Rework cca findcard() implementation and callers Harald Freudenberger
2025-04-11 14:16   ` Holger Dengler
2025-04-14 14:41     ` Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 14/24] s390/zcrypt/pkey: Rework ep11 " Harald Freudenberger
2025-04-11 14:18   ` Holger Dengler
2025-04-14 14:42     ` Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 15/24] s390/zcrypt: Rework cca misc functions kmallocs to use the cprb mempool Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 16/24] s390/zcrypt: Propagate xflags argument with cca_get_info() Harald Freudenberger
2025-04-11 14:25   ` Holger Dengler
2025-04-14 14:48     ` Harald Freudenberger
2025-04-14 15:04     ` Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 17/24] s390/zcrypt: Locate ep11_domain_query_info onto the stack instead of kmalloc Harald Freudenberger
2025-04-09 14:02 ` [PATCH v4 18/24] s390/zcrypt: Rework ep11 misc functions to use cprb mempool Harald Freudenberger
2025-04-09 14:03 ` [PATCH v4 19/24] s390/pkey: Rework CCA pkey handler to use stack for small memory allocs Harald Freudenberger
2025-04-09 14:03 ` [PATCH v4 20/24] s390/pkey: Rework EP11 " Harald Freudenberger
2025-04-09 14:03 ` [PATCH v4 21/24] s390/uv: Rename find_secret() to uv_find_secret() and publish Harald Freudenberger
2025-04-11 14:53   ` Holger Dengler
2025-04-14  8:08   ` Steffen Eiden
2025-04-14 15:15     ` Harald Freudenberger
2025-04-09 14:03 ` [PATCH v4 22/24] s390/pkey: Use preallocated memory for retrieve of UV secret metadata Harald Freudenberger
2025-04-11 15:24   ` Holger Dengler
2025-04-14  8:02   ` Steffen Eiden
2025-04-09 14:03 ` [PATCH v4 23/24] s390/zcrypt/pkey: Provide and pass xflags within pkey and zcrypt layers Harald Freudenberger
2025-04-11 14:36   ` Holger Dengler
2025-04-09 14:03 ` [PATCH v4 24/24] s390/pkey/crypto: Introduce xflags param for pkey in-kernel API Harald Freudenberger
2025-04-14 13:34   ` Holger Dengler

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=20250409140305.58900-4-freude@linux.ibm.com \
    --to=freude@linux.ibm.com \
    --cc=agordeev@linux.ibm.com \
    --cc=borntraeger@linux.ibm.com \
    --cc=dengler@linux.ibm.com \
    --cc=fcallies@linux.ibm.com \
    --cc=frankja@linux.ibm.com \
    --cc=gor@linux.ibm.com \
    --cc=hca@linux.ibm.com \
    --cc=herbert@gondor.apana.org.au \
    --cc=ifranzki@linux.ibm.com \
    --cc=imbrenda@linux.ibm.com \
    --cc=linux-s390@vger.kernel.org \
    --cc=seiden@linux.ibm.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox