* [RFC] [PATCH v2 1/4] padata: generic interface for parallel processing
2009-04-24 10:24 [RFC] [PATCH v2 0/4] Parallel IPsec Steffen Klassert
@ 2009-04-24 10:26 ` Steffen Klassert
2009-04-27 8:56 ` Herbert Xu
2009-04-24 10:27 ` [RFC] [PATCH v2 2/4] pcrypt: Add pcrypt crypto parallelization engine Steffen Klassert
` (3 subsequent siblings)
4 siblings, 1 reply; 16+ messages in thread
From: Steffen Klassert @ 2009-04-24 10:26 UTC (permalink / raw)
To: Herbert Xu; +Cc: David Miller, linux-crypto
This patch introduces an interface to process data objects
in parallel. On request it is possible to serialize again.
The parallelized objects return after serialization in the
same order as they were before the parallelization.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
---
include/linux/interrupt.h | 3 +-
include/linux/padata.h | 116 +++++++++++
kernel/Makefile | 2 +-
kernel/padata.c | 490 +++++++++++++++++++++++++++++++++++++++++++++
kernel/softirq.c | 2 +-
5 files changed, 610 insertions(+), 3 deletions(-)
create mode 100644 include/linux/padata.h
create mode 100644 kernel/padata.c
diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
index 91bb76f..a17679c 100644
--- a/include/linux/interrupt.h
+++ b/include/linux/interrupt.h
@@ -338,7 +338,8 @@ enum
TASKLET_SOFTIRQ,
SCHED_SOFTIRQ,
HRTIMER_SOFTIRQ,
- RCU_SOFTIRQ, /* Preferable RCU should always be the last softirq */
+ PADATA_SOFTIRQ,
+ RCU_SOFTIRQ, /* Preferable RCU should always be the last softirq */
NR_SOFTIRQS
};
diff --git a/include/linux/padata.h b/include/linux/padata.h
new file mode 100644
index 0000000..469359f
--- /dev/null
+++ b/include/linux/padata.h
@@ -0,0 +1,116 @@
+/*
+ * padata.h - header for the padata parallelization interface
+ *
+ * Copyright (C) 2008, 2009 secunet Security Networks AG
+ * Copyright (C) 2008, 2009 Steffen Klassert <steffen.klassert@secunet.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#ifndef PADATA_H
+#define PADATA_H
+
+#include <linux/interrupt.h>
+#include <linux/smp.h>
+#include <linux/list.h>
+
+enum
+{
+ NO_PADATA=0,
+ AEAD_ENC_PADATA,
+ AEAD_DEC_PADATA,
+ NR_PADATA
+};
+
+struct padata_priv {
+ struct list_head list;
+ struct call_single_data csd;
+ int cb_cpu;
+ int seq_nr;
+ unsigned int nr;
+ int info;
+ void (*parallel)(struct padata_priv *padata);
+ void (*serial)(struct padata_priv *padata);
+};
+
+struct padata_queue {
+ struct list_head list;
+ atomic_t num_obj;
+ int cpu_index;
+ spinlock_t lock;
+};
+
+struct parallel_data {
+ struct work_struct work;
+ struct padata_queue *queue;
+ atomic_t seq_nr;
+ atomic_t queued_objects;
+ cpumask_t cpu_map;
+ cpumask_t new_cpu_map;
+ u8 flags;
+#define PADATA_INIT 1
+#define PADATA_FLUSH_HARD 2
+#define PADATA_RESET_IN_PROGRESS 4
+ spinlock_t lock;
+};
+
+#ifdef CONFIG_USE_GENERIC_SMP_HELPERS
+extern void __init padata_init(unsigned int nr, cpumask_t cpu_map);
+extern void padata_dont_wait(unsigned int nr, struct padata_priv *padata);
+extern int padata_do_parallel(unsigned int softirq_nr, unsigned int nr,
+ struct padata_priv *padata, int cb_cpu);
+extern int padata_do_serial(unsigned int nr, struct padata_priv *padata);
+extern cpumask_t padata_get_cpumap(unsigned int nr);
+extern void padata_set_cpumap(unsigned int nr, cpumask_t cpu_map);
+extern void padata_add_cpu(unsigned int nr, int cpu);
+extern void padata_remove_cpu(unsigned int nr, int cpu);
+extern void padata_start(unsigned int nr);
+extern void padata_stop(unsigned int nr);
+#else
+static inline void padata_init(unsigned int nr,cpumask_t cpu_map)
+{
+}
+static inline void padata_dont_wait(unsigned int nr, struct padata_priv *padata)
+{
+}
+static inline int padata_do_parallel(unsigned int softirq_nr, unsigned int nr,
+ struct padata_priv *padata, int cb_cpu)
+{
+ return 0;
+}
+static inline int padata_do_serial(unsigned int nr, struct padata_priv *padata)
+{
+ return 0;
+}
+static inline cpumask_t padata_get_cpumap(unsigned int nr)
+{
+ return cpu_online_map;
+}
+static inline void padata_set_cpumap(unsigned int nr, cpumask_t cpu_map)
+{
+}
+static inline padata_add_cpu(unsigned int nr, int cpu)
+{
+}
+static inline padata_remove_cpu(unsigned int nr, int cpu)
+{
+}
+static inline padata_start(unsigned int nr)
+{
+}
+static inline padata_stop(unsigned int nr)
+{
+}
+#endif
+#endif
diff --git a/kernel/Makefile b/kernel/Makefile
index 4242366..f2d7ea7 100644
--- a/kernel/Makefile
+++ b/kernel/Makefile
@@ -40,7 +40,7 @@ obj-$(CONFIG_RT_MUTEXES) += rtmutex.o
obj-$(CONFIG_DEBUG_RT_MUTEXES) += rtmutex-debug.o
obj-$(CONFIG_RT_MUTEX_TESTER) += rtmutex-tester.o
obj-$(CONFIG_GENERIC_ISA_DMA) += dma.o
-obj-$(CONFIG_USE_GENERIC_SMP_HELPERS) += smp.o
+obj-$(CONFIG_USE_GENERIC_SMP_HELPERS) += smp.o padata.o
ifneq ($(CONFIG_SMP),y)
obj-y += up.o
endif
diff --git a/kernel/padata.c b/kernel/padata.c
new file mode 100644
index 0000000..192c9a6
--- /dev/null
+++ b/kernel/padata.c
@@ -0,0 +1,490 @@
+/*
+ * padata.c - generic interface to process data streams in parallel
+ *
+ * Copyright (C) 2008, 2009 secunet Security Networks AG
+ * Copyright (C) 2008, 2009 Steffen Klassert <steffen.klassert@secunet.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#include <linux/module.h>
+#include <linux/cpumask.h>
+#include <linux/err.h>
+#include <linux/padata.h>
+
+#define MAX_SEQ_NR 1000000000
+
+static struct parallel_data padata_vec[NR_PADATA];
+static struct padata_priv *padata_get_next(struct parallel_data *par_data);
+
+static void padata_flush_hard(struct parallel_data *par_data)
+{
+ int cpu;
+ struct padata_priv *padata;
+ struct padata_queue *queue;
+
+ for_each_cpu_mask(cpu, par_data->cpu_map) {
+ queue = per_cpu_ptr(par_data->queue, cpu);
+
+ while(!list_empty(&queue->list)) {
+ padata = list_entry(queue->list.next, struct padata_priv, list);
+
+ spin_lock(&queue->lock);
+ list_del_init(&padata->list);
+ spin_unlock(&queue->lock);
+
+ atomic_dec(&par_data->queued_objects);
+ padata->serial(padata);
+ }
+ }
+}
+
+static void padata_flush_order(struct parallel_data *par_data)
+{
+ struct padata_priv *padata;
+
+ while (1) {
+ padata = padata_get_next(par_data);
+
+ if (padata && !IS_ERR(padata))
+ padata->serial(padata);
+ else
+ break;
+ }
+
+ padata_flush_hard(par_data);
+}
+
+static void padata_reset_work(struct work_struct *work)
+{
+ int cpu, cpu_index;
+ struct padata_queue *queue;
+ struct parallel_data *par_data;
+
+ par_data = container_of(work, struct parallel_data, work);
+
+ if (par_data->flags & (PADATA_INIT|PADATA_RESET_IN_PROGRESS))
+ return;
+
+ spin_lock_bh(&par_data->lock);
+ par_data->flags |= PADATA_RESET_IN_PROGRESS;
+
+ if (!(par_data->flags & PADATA_FLUSH_HARD))
+ padata_flush_order(par_data);
+ else
+ padata_flush_hard(par_data);
+
+ cpu_index = 0;
+
+ par_data->cpu_map = par_data->new_cpu_map;
+
+ for_each_cpu_mask(cpu, par_data->cpu_map) {
+ queue = per_cpu_ptr(par_data->queue, cpu);
+
+ atomic_set(&queue->num_obj, 0);
+ queue->cpu_index = cpu_index;
+ cpu_index++;
+ }
+ spin_unlock_bh(&par_data->lock);
+
+ atomic_set(&par_data->seq_nr, -1);
+ par_data->flags &= ~PADATA_RESET_IN_PROGRESS;
+ par_data->flags |= PADATA_INIT;
+}
+
+static struct padata_priv *padata_get_next(struct parallel_data *par_data)
+{
+ int cpu, num_cpus, empty;
+ int seq_nr, calc_seq_nr, next_nr;
+ struct padata_queue *queue, *next_queue;
+ struct padata_priv *padata;
+
+ empty = 0;
+ next_nr = -1;
+ next_queue = NULL;
+
+ num_cpus = cpus_weight(par_data->cpu_map);
+
+ for_each_cpu_mask(cpu, par_data->cpu_map) {
+ queue = per_cpu_ptr(par_data->queue, cpu);
+
+ /*
+ * Calculate the seq_nr of the object that should be
+ * next in this queue.
+ */
+ calc_seq_nr = (atomic_read(&queue->num_obj) * num_cpus)
+ + queue->cpu_index;
+
+ if (!list_empty(&queue->list)) {
+ padata = list_entry(queue->list.next,
+ struct padata_priv, list);
+
+ seq_nr = padata->seq_nr;
+
+ if (unlikely(calc_seq_nr != seq_nr)) {
+ par_data->flags &= ~PADATA_INIT;
+ par_data->flags |= PADATA_FLUSH_HARD;
+ padata = NULL;
+ goto out;
+ }
+ } else {
+ seq_nr = calc_seq_nr;
+ empty++;
+ }
+
+ if (next_nr < 0 || seq_nr < next_nr) {
+ next_nr = seq_nr;
+ next_queue = queue;
+ }
+ }
+
+ padata = NULL;
+
+ if (empty == num_cpus)
+ goto out;
+
+ if (!list_empty(&next_queue->list)) {
+ padata = list_entry(next_queue->list.next,
+ struct padata_priv, list);
+
+ spin_lock(&next_queue->lock);
+ list_del_init(&padata->list);
+ spin_unlock(&next_queue->lock);
+
+ atomic_dec(&par_data->queued_objects);
+ atomic_inc(&next_queue->num_obj);
+
+ goto out;
+ }
+
+ if (next_nr % num_cpus == next_queue->cpu_index) {
+ padata = ERR_PTR(-ENODATA);
+ goto out;
+ }
+
+ padata = ERR_PTR(-EINPROGRESS);
+out:
+ return padata;
+}
+
+static void padata_action(struct softirq_action *h)
+{
+ struct list_head *cpu_list, local_list;
+
+ cpu_list = &__get_cpu_var(softirq_work_list[PADATA_SOFTIRQ]);
+
+ local_irq_disable();
+ list_replace_init(cpu_list, &local_list);
+ local_irq_enable();
+
+ while (!list_empty(&local_list)) {
+ struct padata_priv *padata;
+
+ padata = list_entry(local_list.next,
+ struct padata_priv, csd.list);
+
+ list_del_init(&padata->csd.list);
+
+ padata->serial(padata);
+ }
+}
+
+static int padata_cpu_hash(unsigned int nr, struct padata_priv *padata)
+{
+ int cpu, target_cpu, this_cpu, cpu_index;
+
+ this_cpu = smp_processor_id();
+
+ if (padata->nr != 0)
+ return this_cpu;
+
+ if (!(padata_vec[nr].flags & PADATA_INIT))
+ return this_cpu;
+
+ padata->seq_nr = atomic_inc_return(&padata_vec[nr].seq_nr);
+
+ if (padata->seq_nr > MAX_SEQ_NR) {
+ padata_vec[nr].flags &= ~PADATA_INIT;
+ padata->seq_nr = 0;
+ schedule_work(&padata_vec[nr].work);
+ return this_cpu;
+ }
+
+ padata->nr = nr;
+
+ /*
+ * Hash the sequence numbers to the cpus by taking
+ * seq_nr mod. number of cpus in use.
+ */
+ cpu_index = padata->seq_nr % cpus_weight(padata_vec[nr].cpu_map);
+
+ target_cpu = first_cpu(padata_vec[nr].cpu_map);
+ for (cpu = 0; cpu < cpu_index; cpu++)
+ target_cpu = next_cpu(target_cpu, padata_vec[nr].cpu_map);
+
+ return target_cpu;
+}
+
+/*
+ * padata_dont_wait - must be called if an object that runs in parallel will
+ * not be serialized with padata_do_serial
+ *
+ * @nr: number of the padata instance
+ * @padata: object that will not be seen by padata_do_serial
+ */
+void padata_dont_wait(unsigned int nr, struct padata_priv *padata)
+{
+ struct padata_queue *queue;
+
+ if (!(padata_vec[nr].flags & PADATA_INIT))
+ return;
+
+ if (padata->nr == 0 || padata->nr != nr)
+ return;
+
+ queue = per_cpu_ptr(padata_vec[nr].queue, smp_processor_id());
+ atomic_inc(&queue->num_obj);
+
+ padata->nr = 0;
+ padata->seq_nr = 0;
+}
+EXPORT_SYMBOL(padata_dont_wait);
+
+/*
+ * padata_do_parallel - padata parallelization function
+ *
+ * @softirq_nr: number of the softirq that will do the parallelization
+ * @nr: number of the padata instance
+ * @padata: object to be parallelized
+ * @cb_cpu: cpu number on which the serialization callback function will run
+ */
+int padata_do_parallel(unsigned int softirq_nr, unsigned int nr,
+ struct padata_priv *padata, int cb_cpu)
+{
+ int target_cpu;
+
+ padata->cb_cpu = cb_cpu;
+
+ local_bh_disable();
+ target_cpu = padata_cpu_hash(nr, padata);
+ local_bh_enable();
+
+ send_remote_softirq(&padata->csd, target_cpu, softirq_nr);
+
+ return 1;
+}
+EXPORT_SYMBOL(padata_do_parallel);
+
+/*
+ * padata_do_serial - padata serialization function
+ *
+ * @nr: number of the padata instance
+ * @padata: object to be serialized
+ *
+ * returns 1 if the serialization callback function will be called
+ * from padata, 0 else
+ */
+int padata_do_serial(unsigned int nr, struct padata_priv *padata)
+{
+ int cpu;
+ struct padata_queue *reorder_queue;
+
+ if (!(padata_vec[nr].flags & PADATA_INIT))
+ return 0;
+
+ if (padata->nr != nr || padata->nr == 0) {
+ padata->serial(padata);
+ return 1;
+ }
+
+ cpu = smp_processor_id();
+
+ reorder_queue = per_cpu_ptr(padata_vec[nr].queue, cpu);
+
+ spin_lock(&reorder_queue->lock);
+ list_add_tail(&padata->list, &reorder_queue->list);
+ spin_unlock(&reorder_queue->lock);
+
+ atomic_inc(&padata_vec[nr].queued_objects);
+
+try_again:
+ if (!spin_trylock(&padata_vec[nr].lock))
+ goto out;
+
+ while(1) {
+ padata = padata_get_next(&padata_vec[nr]);
+
+ if (!padata || PTR_ERR(padata) == -EINPROGRESS)
+ break;
+ if (PTR_ERR(padata) == -ENODATA) {
+ spin_unlock(&padata_vec[nr].lock);
+ goto out;
+ }
+
+ send_remote_softirq(&padata->csd, padata->cb_cpu,
+ PADATA_SOFTIRQ);
+ }
+
+ if (unlikely(!(padata_vec[nr].flags & PADATA_INIT))) {
+ spin_unlock(&padata_vec[nr].lock);
+ goto reset_out;
+ }
+
+ spin_unlock(&padata_vec[nr].lock);
+
+ if (atomic_read(&padata_vec[nr].queued_objects))
+ goto try_again;
+
+out:
+ return 1;
+reset_out:
+ schedule_work(&padata_vec[nr].work);
+ return 1;
+}
+EXPORT_SYMBOL(padata_do_serial);
+
+/*
+ * padata_get_cpumap - get the cpu map that is actually in use
+ *
+ * @nr: number of the padata instance
+ */
+cpumask_t padata_get_cpumap(unsigned int nr)
+{
+ return padata_vec[nr].cpu_map;
+}
+EXPORT_SYMBOL(padata_get_cpumap);
+
+/*
+ * padata_set_cpumap - set the cpu map that padata uses
+ *
+ * @nr: number of the padata instance
+ * @cpu_map: the cpu map to use
+ */
+void padata_set_cpumap(unsigned int nr, cpumask_t cpu_map)
+{
+ padata_vec[nr].new_cpu_map = cpu_map;
+ padata_vec[nr].flags &= ~PADATA_INIT;
+ padata_vec[nr].flags |= PADATA_FLUSH_HARD;
+
+ schedule_work(&padata_vec[nr].work);
+}
+EXPORT_SYMBOL(padata_set_cpumap);
+
+/*
+ * padata_add_cpu - add a cpu to the padata cpu map
+ *
+ * @nr: number of the padata instance
+ * @cpu: cpu to remove
+ */
+void padata_add_cpu(unsigned int nr, int cpu)
+{
+ cpumask_t cpu_map = padata_vec[nr].cpu_map;
+
+ cpu_set(cpu, cpu_map);
+ padata_set_cpumap(nr, cpu_map);
+}
+EXPORT_SYMBOL(padata_add_cpu);
+
+/*
+ * padata_remove_cpu - remove a cpu from the padata cpu map
+ *
+ * @nr: number of the padata instance
+ * @cpu: cpu to remove
+ */
+void padata_remove_cpu(unsigned int nr, int cpu)
+{
+ cpumask_t cpu_map = padata_vec[nr].cpu_map;
+
+ cpu_clear(cpu, cpu_map);
+ padata_set_cpumap(nr, cpu_map);
+}
+EXPORT_SYMBOL(padata_remove_cpu);
+
+/*
+ * padata_start - start the parallel processing
+ *
+ * @nr: number of the padata instance
+ */
+void padata_start(unsigned int nr)
+{
+ if (padata_vec[nr].flags & PADATA_INIT)
+ return;
+
+ schedule_work(&padata_vec[nr].work);
+}
+EXPORT_SYMBOL(padata_start);
+
+/*
+ * padata_stop - stop the parallel processing
+ *
+ * @nr: number of the padata instance
+ */
+void padata_stop(unsigned int nr)
+{
+ padata_vec[nr].flags &= ~PADATA_INIT;
+}
+EXPORT_SYMBOL(padata_stop);
+
+/*
+ * padata_init - initialize a padata instance
+ *
+ * @nr: number of the padata instance
+ * @cpu_map: map of the cpu set that padata uses for parallelization
+ */
+void __init padata_init(unsigned int nr, cpumask_t cpu_map)
+{
+ int cpu, cpu_index;
+ struct padata_queue *percpu_queue, *queue;
+
+ percpu_queue = alloc_percpu(struct padata_queue);
+
+ if (!percpu_queue) {
+ printk("padata_init: Failed to alloc the serialization"
+ "queues for padata nr %d, exiting!\n", nr);
+ return;
+ }
+
+ cpu_index = 0;
+
+ for_each_possible_cpu(cpu) {
+ queue = per_cpu_ptr(percpu_queue, cpu);
+
+ if (cpu_isset(cpu, cpu_map)) {
+ queue->cpu_index = cpu_index;
+ cpu_index++;
+ }
+
+ INIT_LIST_HEAD(&queue->list);
+ spin_lock_init(&queue->lock);
+ atomic_set(&queue->num_obj, 0);
+ }
+
+ INIT_WORK(&padata_vec[nr].work, padata_reset_work);
+
+ atomic_set(&padata_vec[nr].seq_nr, -1);
+ atomic_set(&padata_vec[nr].queued_objects, 0);
+ padata_vec[nr].cpu_map = cpu_map;
+ padata_vec[nr].new_cpu_map = cpu_map;
+ padata_vec[nr].queue = percpu_queue;
+ padata_vec[nr].flags = 0;
+ spin_lock_init(&padata_vec[nr].lock);
+}
+
+static int __init padata_initcall(void)
+{
+ open_softirq(PADATA_SOFTIRQ, padata_action);
+
+ return 0;
+}
+subsys_initcall(padata_initcall);
diff --git a/kernel/softirq.c b/kernel/softirq.c
index b525dd3..12c9b64 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -56,7 +56,7 @@ static DEFINE_PER_CPU(struct task_struct *, ksoftirqd);
char *softirq_to_name[NR_SOFTIRQS] = {
"HI", "TIMER", "NET_TX", "NET_RX", "BLOCK",
- "TASKLET", "SCHED", "HRTIMER", "RCU"
+ "TASKLET", "SCHED", "HRTIMER", "PADATA", "RCU"
};
/*
--
1.5.4.2
^ permalink raw reply related [flat|nested] 16+ messages in thread* [RFC] [PATCH v2 2/4] pcrypt: Add pcrypt crypto parallelization engine
2009-04-24 10:24 [RFC] [PATCH v2 0/4] Parallel IPsec Steffen Klassert
2009-04-24 10:26 ` [RFC] [PATCH v2 1/4] padata: generic interface for parallel processing Steffen Klassert
@ 2009-04-24 10:27 ` Steffen Klassert
2009-04-27 8:56 ` Herbert Xu
2009-04-24 10:28 ` [RFC] [PATCH v2 3/4] xfrm: Add a netlink attribute for software crypto accelerators Steffen Klassert
` (2 subsequent siblings)
4 siblings, 1 reply; 16+ messages in thread
From: Steffen Klassert @ 2009-04-24 10:27 UTC (permalink / raw)
To: Herbert Xu; +Cc: David Miller, linux-crypto
This patch adds a parallel crypto template that takes a crypto
algorithm and converts it to process the crypto transforms in
parallel. For the moment only aead is supported.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
---
crypto/Kconfig | 13 ++
crypto/Makefile | 2 +
crypto/pcrypt.c | 415 +++++++++++++++++++++++++++++++++++++++++++++
crypto/pcrypt_core.c | 106 ++++++++++++
include/crypto/pcrypt.h | 56 ++++++
include/linux/crypto.h | 2 +
include/linux/interrupt.h | 2 +
kernel/softirq.c | 3 +-
8 files changed, 598 insertions(+), 1 deletions(-)
create mode 100644 crypto/pcrypt.c
create mode 100644 crypto/pcrypt_core.c
create mode 100644 include/crypto/pcrypt.h
diff --git a/crypto/Kconfig b/crypto/Kconfig
index 74d0e62..b05fc95 100644
--- a/crypto/Kconfig
+++ b/crypto/Kconfig
@@ -112,6 +112,19 @@ config CRYPTO_NULL
help
These are 'Null' algorithms, used by IPsec, which do nothing.
+config CRYPTO_PCRYPT_CORE
+ bool
+ select CRYPTO_AEAD
+
+config CRYPTO_PCRYPT
+ tristate "Parallel crypto engine (EXPERIMENTAL)"
+ depends on USE_GENERIC_SMP_HELPERS && EXPERIMENTAL
+ select CRYPTO_MANAGER
+ select CRYPTO_PCRYPT_CORE
+ help
+ This converts an arbitrary crypto algorithm into a parallel
+ algorithm that is executed in a softirq.
+
config CRYPTO_WORKQUEUE
tristate
diff --git a/crypto/Makefile b/crypto/Makefile
index 673d9f7..84b9d17 100644
--- a/crypto/Makefile
+++ b/crypto/Makefile
@@ -56,6 +56,8 @@ obj-$(CONFIG_CRYPTO_XTS) += xts.o
obj-$(CONFIG_CRYPTO_CTR) += ctr.o
obj-$(CONFIG_CRYPTO_GCM) += gcm.o
obj-$(CONFIG_CRYPTO_CCM) += ccm.o
+obj-$(CONFIG_CRYPTO_PCRYPT_CORE) += pcrypt_core.o
+obj-$(CONFIG_CRYPTO_PCRYPT) += pcrypt.o
obj-$(CONFIG_CRYPTO_CRYPTD) += cryptd.o
obj-$(CONFIG_CRYPTO_DES) += des_generic.o
obj-$(CONFIG_CRYPTO_FCRYPT) += fcrypt.o
diff --git a/crypto/pcrypt.c b/crypto/pcrypt.c
new file mode 100644
index 0000000..23f7eb7
--- /dev/null
+++ b/crypto/pcrypt.c
@@ -0,0 +1,415 @@
+/*
+ * pcrypt - Parallel crypto wrapper.
+ *
+ * Copyright (C) 2009 secunet Security Networks AG
+ * Copyright (C) 2009 Steffen Klassert <steffen.klassert@secunet.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#include <crypto/algapi.h>
+#include <crypto/internal/aead.h>
+#include <linux/err.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <crypto/pcrypt.h>
+
+struct pcrypt_instance_ctx {
+ struct crypto_spawn spawn;
+ unsigned int tfm_count;
+};
+
+struct pcrypt_aead_ctx {
+ struct crypto_aead *child;
+ unsigned int tfm_nr;
+};
+
+static int pcrypt_do_parallel(struct padata_priv *padata, unsigned int tfm_nr,
+ unsigned int softirq, unsigned int padata_nr)
+{
+ unsigned int cpu, cpu_index, num_cpus, cb_cpu;
+ cpumask_t cpu_map;
+
+ cpu_map = padata_get_cpumap(padata_nr);
+ num_cpus = cpus_weight(cpu_map);
+
+ cpu_index = tfm_nr % num_cpus;
+
+ cb_cpu = first_cpu(cpu_map);
+ for (cpu = 0; cpu < cpu_index; cpu++)
+ cb_cpu = next_cpu(cb_cpu, cpu_map);
+
+ return padata_do_parallel(softirq, padata_nr, padata, cb_cpu);
+}
+
+static int pcrypt_aead_setkey(struct crypto_aead *parent,
+ const u8 *key, unsigned int keylen)
+{
+ struct pcrypt_aead_ctx *ctx = crypto_aead_ctx(parent);
+
+ return crypto_aead_setkey(ctx->child, key, keylen);
+}
+
+static int pcrypt_aead_setauthsize(struct crypto_aead *parent,
+ unsigned int authsize)
+{
+ struct pcrypt_aead_ctx *ctx = crypto_aead_ctx(parent);
+
+ return crypto_aead_setauthsize(ctx->child, authsize);
+}
+
+static void pcrypt_aead_serial(struct padata_priv *padata)
+{
+ struct pcrypt_request *preq = pcrypt_padata_request(padata);
+ struct aead_request *req = pcrypt_request_ctx(preq);
+
+ aead_request_complete(req->base.data, padata->info);
+}
+
+static void pcrypt_aead_giv_serial(struct padata_priv *padata)
+{
+ struct pcrypt_request *preq = pcrypt_padata_request(padata);
+ struct aead_givcrypt_request *req = pcrypt_request_ctx(preq);
+
+ aead_request_complete(req->areq.base.data, padata->info);
+}
+
+static void pcrypt_aead_done(struct crypto_async_request *areq, int err)
+{
+ struct aead_request *req = areq->data;
+ struct pcrypt_request *preq = aead_request_ctx(req);
+ struct padata_priv *padata = pcrypt_request_padata(preq);
+
+ padata->info = err;
+ req->base.flags &= ~CRYPTO_TFM_REQ_MAY_SLEEP;
+
+ local_bh_disable();
+ if (padata_do_serial(padata->nr, padata))
+ goto out;
+
+ aead_request_complete(req, padata->info);
+
+out:
+ local_bh_enable();
+}
+
+static void pcrypt_aead_enc(struct padata_priv *padata)
+{
+ struct pcrypt_request *preq = pcrypt_padata_request(padata);
+ struct aead_request *req = pcrypt_request_ctx(preq);
+
+ padata->info = crypto_aead_encrypt(req);
+
+ if (padata->info)
+ return;
+
+ if (padata_do_serial(AEAD_ENC_PADATA, padata))
+ return;
+
+ aead_request_complete(req->base.data, padata->info);
+}
+
+static int pcrypt_aead_encrypt(struct aead_request *req)
+{
+ int err;
+ struct pcrypt_request *preq = aead_request_ctx(req);
+ struct aead_request *creq = pcrypt_request_ctx(preq);
+ struct padata_priv *padata = pcrypt_request_padata(preq);
+ struct crypto_aead *aead = crypto_aead_reqtfm(req);
+ struct pcrypt_aead_ctx *ctx = crypto_aead_ctx(aead);
+ u32 flags = aead_request_flags(req);
+
+ memset(padata, 0, sizeof(struct padata_priv));
+
+ padata->parallel = pcrypt_aead_enc;
+ padata->serial = pcrypt_aead_serial;
+
+ aead_request_set_tfm(creq, ctx->child);
+ aead_request_set_callback(creq, flags & ~CRYPTO_TFM_REQ_MAY_SLEEP,
+ pcrypt_aead_done, req);
+ aead_request_set_crypt(creq, req->src, req->dst,
+ req->cryptlen, req->iv);
+ aead_request_set_assoc(creq, req->assoc, req->assoclen);
+
+ if (pcrypt_do_parallel(padata, ctx->tfm_nr, AEAD_ENC_SOFTIRQ,
+ AEAD_ENC_PADATA))
+ err = -EINPROGRESS;
+ else
+ err = crypto_aead_encrypt(creq);
+
+ return err;
+}
+
+static void pcrypt_aead_dec(struct padata_priv *padata)
+{
+ struct pcrypt_request *preq = pcrypt_padata_request(padata);
+ struct aead_request *req = pcrypt_request_ctx(preq);
+
+ padata->info = crypto_aead_decrypt(req);
+
+ if (padata->info)
+ return;
+
+ if (padata_do_serial(AEAD_DEC_PADATA, padata))
+ return;
+
+ aead_request_complete(req->base.data, padata->info);
+}
+
+static int pcrypt_aead_decrypt(struct aead_request *req)
+{
+ int err;
+ struct pcrypt_request *preq = aead_request_ctx(req);
+ struct aead_request *creq = pcrypt_request_ctx(preq);
+ struct padata_priv *padata = pcrypt_request_padata(preq);
+ struct crypto_aead *aead = crypto_aead_reqtfm(req);
+ struct pcrypt_aead_ctx *ctx = crypto_aead_ctx(aead);
+ u32 flags = aead_request_flags(req);
+
+ memset(padata, 0, sizeof(struct padata_priv));
+
+ padata->parallel = pcrypt_aead_dec;
+ padata->serial = pcrypt_aead_serial;
+
+ aead_request_set_tfm(creq, ctx->child);
+ aead_request_set_callback(creq, flags & ~CRYPTO_TFM_REQ_MAY_SLEEP,
+ pcrypt_aead_done, req);
+ aead_request_set_crypt(creq, req->src, req->dst,
+ req->cryptlen, req->iv);
+ aead_request_set_assoc(creq, req->assoc, req->assoclen);
+
+ if (pcrypt_do_parallel(padata, ctx->tfm_nr, AEAD_DEC_SOFTIRQ,
+ AEAD_DEC_PADATA))
+ err = -EINPROGRESS;
+ else
+ err = crypto_aead_decrypt(creq);
+
+ return err;
+}
+
+static void pcrypt_aead_givenc(struct padata_priv *padata)
+{
+ struct pcrypt_request *preq = pcrypt_padata_request(padata);
+ struct aead_givcrypt_request *req = pcrypt_request_ctx(preq);
+
+ padata->info = crypto_aead_givencrypt(req);
+
+ if (padata->info)
+ return;
+
+ if (padata_do_serial(AEAD_ENC_PADATA, padata))
+ return;
+
+ aead_request_complete(req->areq.base.data, padata->info);
+}
+
+static int pcrypt_aead_givencrypt(struct aead_givcrypt_request *req)
+{
+ int err;
+ struct aead_request *areq = &req->areq;
+ struct pcrypt_request *preq = aead_request_ctx(areq);
+ struct aead_givcrypt_request *creq = pcrypt_request_ctx(preq);
+ struct padata_priv *padata = pcrypt_request_padata(preq);
+ struct crypto_aead *aead = aead_givcrypt_reqtfm(req);
+ struct pcrypt_aead_ctx *ctx = crypto_aead_ctx(aead);
+ u32 flags = aead_request_flags(areq);
+
+ memset(padata, 0, sizeof(struct padata_priv));
+
+ padata->parallel = pcrypt_aead_givenc;
+ padata->serial = pcrypt_aead_giv_serial;
+
+ aead_givcrypt_set_tfm(creq, ctx->child);
+ aead_givcrypt_set_callback(creq, flags & ~CRYPTO_TFM_REQ_MAY_SLEEP,
+ pcrypt_aead_done, areq);
+ aead_givcrypt_set_crypt(creq, areq->src, areq->dst,
+ areq->cryptlen, areq->iv);
+ aead_givcrypt_set_assoc(creq, areq->assoc, areq->assoclen);
+ aead_givcrypt_set_giv(creq, req->giv, req->seq);
+
+
+ if (pcrypt_do_parallel(padata, ctx->tfm_nr, AEAD_ENC_SOFTIRQ,
+ AEAD_ENC_PADATA))
+ err = -EINPROGRESS;
+ else
+ err = crypto_aead_givencrypt(creq);
+
+ return err;
+}
+
+static int pcrypt_aead_init_tfm(struct crypto_tfm *tfm)
+{
+ struct crypto_instance *inst = crypto_tfm_alg_instance(tfm);
+ struct pcrypt_instance_ctx *ictx = crypto_instance_ctx(inst);
+ struct pcrypt_aead_ctx *ctx = crypto_tfm_ctx(tfm);
+ struct crypto_aead *cipher;
+
+ ictx->tfm_count++;
+ ctx->tfm_nr = ictx->tfm_count;
+
+ cipher = crypto_spawn_aead(crypto_instance_ctx(inst));
+
+ if (IS_ERR(cipher))
+ return PTR_ERR(cipher);
+
+ ctx->child = cipher;
+ tfm->crt_aead.reqsize = sizeof(struct pcrypt_request)
+ + sizeof(struct aead_givcrypt_request)
+ + crypto_aead_reqsize(cipher);
+
+ return 0;
+}
+
+static void pcrypt_aead_exit_tfm(struct crypto_tfm *tfm)
+{
+ struct pcrypt_aead_ctx *ctx = crypto_tfm_ctx(tfm);
+
+ crypto_free_aead(ctx->child);
+}
+
+static struct crypto_instance *pcrypt_alloc_instance(struct crypto_alg *alg)
+{
+ struct crypto_instance *inst;
+ struct pcrypt_instance_ctx *ctx;
+ int err;
+
+ inst = kzalloc(sizeof(*inst) + sizeof(*ctx), GFP_KERNEL);
+ if (!inst) {
+ inst = ERR_PTR(-ENOMEM);
+ goto out;
+ }
+
+ err = -ENAMETOOLONG;
+ if (snprintf(inst->alg.cra_driver_name, CRYPTO_MAX_ALG_NAME,
+ "pcrypt(%s)", alg->cra_driver_name) >= CRYPTO_MAX_ALG_NAME)
+ goto out_free_inst;
+
+ if (snprintf(inst->alg.cra_name, CRYPTO_MAX_ALG_NAME,
+ "pcrypt(%s)", alg->cra_name) >= CRYPTO_MAX_ALG_NAME)
+ goto out_free_inst;
+
+ ctx = crypto_instance_ctx(inst);
+ err = crypto_init_spawn(&ctx->spawn, alg, inst,
+ CRYPTO_ALG_TYPE_MASK);
+ if (err)
+ goto out_free_inst;
+
+ inst->alg.cra_priority = alg->cra_priority + 100;
+ inst->alg.cra_blocksize = alg->cra_blocksize;
+ inst->alg.cra_alignmask = alg->cra_alignmask;
+
+out:
+ return inst;
+
+out_free_inst:
+ kfree(inst);
+ inst = ERR_PTR(err);
+ goto out;
+}
+
+static struct crypto_instance *pcrypt_alloc_aead(struct rtattr **tb)
+{
+ struct crypto_instance *inst;
+ struct crypto_alg *alg;
+ struct crypto_attr_type *algt;
+
+ algt = crypto_get_attr_type(tb);
+
+ alg = crypto_get_attr_alg(tb, algt->type,
+ (algt->mask & CRYPTO_ALG_TYPE_MASK)
+ | CRYPTO_ALG_PCRYPT);
+ if (IS_ERR(alg))
+ return ERR_CAST(alg);
+
+ inst = pcrypt_alloc_instance(alg);
+ if (IS_ERR(inst))
+ goto out_put_alg;
+
+ inst->alg.cra_flags = CRYPTO_ALG_TYPE_AEAD | CRYPTO_ALG_ASYNC
+ | CRYPTO_ALG_PCRYPT;
+ inst->alg.cra_type = &crypto_aead_type;
+
+ inst->alg.cra_aead.ivsize = alg->cra_aead.ivsize;
+ inst->alg.cra_aead.geniv = alg->cra_aead.geniv;
+ inst->alg.cra_aead.maxauthsize = alg->cra_aead.maxauthsize;
+
+ inst->alg.cra_ctxsize = sizeof(struct pcrypt_aead_ctx);
+
+ inst->alg.cra_init = pcrypt_aead_init_tfm;
+ inst->alg.cra_exit = pcrypt_aead_exit_tfm;
+
+ inst->alg.cra_aead.setkey = pcrypt_aead_setkey;
+ inst->alg.cra_aead.setauthsize = pcrypt_aead_setauthsize;
+ inst->alg.cra_aead.encrypt = pcrypt_aead_encrypt;
+ inst->alg.cra_aead.decrypt = pcrypt_aead_decrypt;
+ inst->alg.cra_aead.givencrypt = pcrypt_aead_givencrypt;
+
+out_put_alg:
+ crypto_mod_put(alg);
+ return inst;
+}
+
+static struct crypto_instance *pcrypt_alloc(struct rtattr **tb)
+{
+ struct crypto_attr_type *algt;
+
+ algt = crypto_get_attr_type(tb);
+ if (IS_ERR(algt))
+ return ERR_CAST(algt);
+
+ switch (algt->type & algt->mask & CRYPTO_ALG_TYPE_MASK) {
+ case CRYPTO_ALG_TYPE_AEAD:
+ return pcrypt_alloc_aead(tb);
+ }
+
+ return ERR_PTR(-EINVAL);
+}
+
+static void pcrypt_free(struct crypto_instance *inst)
+{
+ struct pcrypt_instance_ctx *ctx = crypto_instance_ctx(inst);
+
+ crypto_drop_spawn(&ctx->spawn);
+ kfree(inst);
+}
+
+static struct crypto_template pcrypt_tmpl = {
+ .name = "pcrypt",
+ .alloc = pcrypt_alloc,
+ .free = pcrypt_free,
+ .module = THIS_MODULE,
+};
+
+static int __init pcrypt_init(void)
+{
+ padata_start(AEAD_ENC_PADATA);
+ padata_start(AEAD_DEC_PADATA);
+
+ return crypto_register_template(&pcrypt_tmpl);
+}
+
+static void __exit pcrypt_exit(void)
+{
+ padata_stop(AEAD_ENC_PADATA);
+ padata_stop(AEAD_DEC_PADATA);
+
+ crypto_unregister_template(&pcrypt_tmpl);
+}
+
+module_init(pcrypt_init);
+module_exit(pcrypt_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Parallel crypto engine");
diff --git a/crypto/pcrypt_core.c b/crypto/pcrypt_core.c
new file mode 100644
index 0000000..61c0411
--- /dev/null
+++ b/crypto/pcrypt_core.c
@@ -0,0 +1,106 @@
+/*
+ * pcrypt_core.c - Core functions for the pcrypt crypto parallelization
+ *
+ * Copyright (C) 2009 secunet Security Networks AG
+ * Copyright (C) 2009 Steffen Klassert <steffen.klassert@secunet.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#include <linux/interrupt.h>
+#include <linux/cpu.h>
+#include <linux/err.h>
+#include <linux/module.h>
+#include <crypto/pcrypt.h>
+
+static void aead_enc_action(struct softirq_action *h)
+{
+ struct list_head *cpu_list, local_list;
+
+ cpu_list = &__get_cpu_var(softirq_work_list[AEAD_ENC_SOFTIRQ]);
+
+ local_irq_disable();
+ list_replace_init(cpu_list, &local_list);
+ local_irq_enable();
+
+ while (!list_empty(&local_list)) {
+ struct padata_priv *padata;
+
+ padata = list_entry(local_list.next, struct padata_priv,
+ csd.list);
+
+ list_del_init(&padata->csd.list);
+
+ padata->parallel(padata);
+ }
+}
+
+static void aead_dec_action(struct softirq_action *h)
+{
+ struct list_head *cpu_list, local_list;
+
+ cpu_list = &__get_cpu_var(softirq_work_list[AEAD_DEC_SOFTIRQ]);
+
+ local_irq_disable();
+ list_replace_init(cpu_list, &local_list);
+ local_irq_enable();
+
+ while (!list_empty(&local_list)) {
+ struct padata_priv *padata;
+
+ padata = list_entry(local_list.next, struct padata_priv,
+ csd.list);
+
+ list_del_init(&padata->csd.list);
+
+ padata->parallel(padata);
+ }
+}
+
+static int __devinit pcrypt_cpu_callback(struct notifier_block *nfb,
+ unsigned long action, void *hcpu)
+{
+ int cpu = (unsigned long)hcpu;
+
+ switch (action) {
+ case CPU_ONLINE:
+ case CPU_ONLINE_FROZEN:
+ padata_add_cpu(AEAD_ENC_PADATA, cpu);
+ padata_add_cpu(AEAD_DEC_PADATA, cpu);
+ break;
+
+ case CPU_DEAD:
+ case CPU_DEAD_FROZEN:
+ padata_remove_cpu(AEAD_ENC_PADATA, cpu);
+ padata_remove_cpu(AEAD_DEC_PADATA, cpu);
+
+ break;
+ }
+
+ return NOTIFY_OK;
+}
+
+static int __init pcrypt_init_padata(void)
+{
+ open_softirq(AEAD_ENC_SOFTIRQ, aead_enc_action);
+ open_softirq(AEAD_DEC_SOFTIRQ, aead_dec_action);
+
+ padata_init(AEAD_ENC_PADATA, cpu_online_map);
+ padata_init(AEAD_DEC_PADATA, cpu_online_map);
+
+ hotcpu_notifier(pcrypt_cpu_callback, 0);
+
+ return 0;
+}
+subsys_initcall(pcrypt_init_padata);
diff --git a/include/crypto/pcrypt.h b/include/crypto/pcrypt.h
new file mode 100644
index 0000000..65c1f94
--- /dev/null
+++ b/include/crypto/pcrypt.h
@@ -0,0 +1,56 @@
+/*
+ * pcrypt - Parallel crypto engine.
+ *
+ * Copyright (C) 2009 secunet Security Networks AG
+ * Copyright (C) 2009 Steffen Klassert <steffen.klassert@secunet.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#ifndef _CRYPTO_PCRYPT_H
+#define _CRYPTO_PCRYPT_H
+
+#include <linux/crypto.h>
+#include <linux/kernel.h>
+#include <linux/padata.h>
+
+struct pcrypt_request {
+ struct padata_priv padata;
+ void *data;
+ void *__ctx[] CRYPTO_MINALIGN_ATTR;
+};
+
+static inline void *pcrypt_request_ctx(struct pcrypt_request *req)
+{
+ return req->__ctx;
+}
+
+static inline
+struct padata_priv *pcrypt_request_padata(struct pcrypt_request *req)
+{
+ return &req->padata;
+}
+
+static inline
+struct pcrypt_request *pcrypt_padata_request(struct padata_priv *padata)
+{
+ return container_of(padata, struct pcrypt_request, padata);
+}
+
+struct crypto_aead *pcrypt_alloc_aead_tfm(const char *alg_name, u32 type,
+ u32 mask);
+
+struct crypto_ablkcipher *pcrypt_alloc_ablkcipher_tfm(const char *alg_name,
+ u32 type, u32 mask);
+#endif
diff --git a/include/linux/crypto.h b/include/linux/crypto.h
index ec29fa2..69c2655 100644
--- a/include/linux/crypto.h
+++ b/include/linux/crypto.h
@@ -71,6 +71,8 @@
#define CRYPTO_ALG_TESTED 0x00000400
+#define CRYPTO_ALG_PCRYPT 0x00000800
+
/*
* Transform masks and values (for crt_flags).
*/
diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h
index a17679c..93e0c9f 100644
--- a/include/linux/interrupt.h
+++ b/include/linux/interrupt.h
@@ -338,6 +338,8 @@ enum
TASKLET_SOFTIRQ,
SCHED_SOFTIRQ,
HRTIMER_SOFTIRQ,
+ AEAD_ENC_SOFTIRQ,
+ AEAD_DEC_SOFTIRQ,
PADATA_SOFTIRQ,
RCU_SOFTIRQ, /* Preferable RCU should always be the last softirq */
diff --git a/kernel/softirq.c b/kernel/softirq.c
index 12c9b64..c76ace0 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -56,7 +56,8 @@ static DEFINE_PER_CPU(struct task_struct *, ksoftirqd);
char *softirq_to_name[NR_SOFTIRQS] = {
"HI", "TIMER", "NET_TX", "NET_RX", "BLOCK",
- "TASKLET", "SCHED", "HRTIMER", "PADATA", "RCU"
+ "TASKLET", "SCHED", "HRTIMER", "AEAD_ENC",
+ "AEAD_DEC", "PADATA", "RCU"
};
/*
--
1.5.4.2
^ permalink raw reply related [flat|nested] 16+ messages in thread* Re: [RFC] [PATCH v2 2/4] pcrypt: Add pcrypt crypto parallelization engine
2009-04-24 10:27 ` [RFC] [PATCH v2 2/4] pcrypt: Add pcrypt crypto parallelization engine Steffen Klassert
@ 2009-04-27 8:56 ` Herbert Xu
2009-04-28 6:23 ` Steffen Klassert
0 siblings, 1 reply; 16+ messages in thread
From: Herbert Xu @ 2009-04-27 8:56 UTC (permalink / raw)
To: Steffen Klassert; +Cc: David Miller, linux-crypto
On Fri, Apr 24, 2009 at 12:27:20PM +0200, Steffen Klassert wrote:
>
> + alg = crypto_get_attr_alg(tb, algt->type,
> + (algt->mask & CRYPTO_ALG_TYPE_MASK)
> + | CRYPTO_ALG_PCRYPT);
Do you really need to exclude pcrypt algorithms here? If the user
wants to added another layer of pcrypt then we should obey.
Otherwise this looks pretty good to me.
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply [flat|nested] 16+ messages in thread* Re: [RFC] [PATCH v2 2/4] pcrypt: Add pcrypt crypto parallelization engine
2009-04-27 8:56 ` Herbert Xu
@ 2009-04-28 6:23 ` Steffen Klassert
2009-04-28 6:31 ` Herbert Xu
0 siblings, 1 reply; 16+ messages in thread
From: Steffen Klassert @ 2009-04-28 6:23 UTC (permalink / raw)
To: Herbert Xu; +Cc: David Miller, linux-crypto
On Mon, Apr 27, 2009 at 04:56:20PM +0800, Herbert Xu wrote:
> On Fri, Apr 24, 2009 at 12:27:20PM +0200, Steffen Klassert wrote:
> >
> > + alg = crypto_get_attr_alg(tb, algt->type,
> > + (algt->mask & CRYPTO_ALG_TYPE_MASK)
> > + | CRYPTO_ALG_PCRYPT);
>
> Do you really need to exclude pcrypt algorithms here? If the user
> wants to added another layer of pcrypt then we should obey.
>
I don't need to exclude it. I just wanted to avoid another layer
of pcrypt here because it does not make too much sense to parallelize
a second time. If the aead algorithm is already parallel it would just
add overhead if we parallelize an underlying algorithm too.
Anyway, I would not mind to allow more than one pcrypt layer.
Steffen
^ permalink raw reply [flat|nested] 16+ messages in thread
* [RFC] [PATCH v2 3/4] xfrm: Add a netlink attribute for software crypto accelerators
2009-04-24 10:24 [RFC] [PATCH v2 0/4] Parallel IPsec Steffen Klassert
2009-04-24 10:26 ` [RFC] [PATCH v2 1/4] padata: generic interface for parallel processing Steffen Klassert
2009-04-24 10:27 ` [RFC] [PATCH v2 2/4] pcrypt: Add pcrypt crypto parallelization engine Steffen Klassert
@ 2009-04-24 10:28 ` Steffen Klassert
2009-04-27 8:53 ` Herbert Xu
2009-04-24 10:29 ` [RFC] [PATCH v2 4/4] esp: Use pcrypt if it is selected as software crypto accelerator Steffen Klassert
2009-04-25 8:38 ` [RFC] [PATCH v2 0/4] Parallel IPsec Evgeniy Polyakov
4 siblings, 1 reply; 16+ messages in thread
From: Steffen Klassert @ 2009-04-24 10:28 UTC (permalink / raw)
To: Herbert Xu; +Cc: David Miller, linux-crypto
This patch adds a netlink attribute for software crypto
accelerators like pcrypt. This makes it possible to choose a
crypto accelerator template by it's name from the userspace.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
---
include/linux/xfrm.h | 10 ++++++++++
include/net/xfrm.h | 3 +++
net/xfrm/xfrm_user.c | 8 ++++++++
3 files changed, 21 insertions(+), 0 deletions(-)
diff --git a/include/linux/xfrm.h b/include/linux/xfrm.h
index 2d4ec15..e20b74e 100644
--- a/include/linux/xfrm.h
+++ b/include/linux/xfrm.h
@@ -240,6 +240,15 @@ struct xfrm_encap_tmpl {
xfrm_address_t encap_oa;
};
+struct xfrm_accl {
+ char name[64];
+ __u8 type;
+#define XFRM_ACCL_ENC 1
+#define XFRM_ACCL_AUTH 2
+#define XFRM_ACCL_COMP 4
+#define XFRM_ACCL_AEAD 8
+};
+
/* AEVENT flags */
enum xfrm_ae_ftype_t {
XFRM_AE_UNSPEC,
@@ -283,6 +292,7 @@ enum xfrm_attr_type_t {
XFRMA_MIGRATE,
XFRMA_ALG_AEAD, /* struct xfrm_algo_aead */
XFRMA_KMADDRESS, /* struct xfrm_user_kmaddress */
+ XFRMA_CRYPTO_ACCL, /* struct xfrm_accl */
__XFRMA_MAX
#define XFRMA_MAX (__XFRMA_MAX - 1)
diff --git a/include/net/xfrm.h b/include/net/xfrm.h
index 2e9f5c0..b574147 100644
--- a/include/net/xfrm.h
+++ b/include/net/xfrm.h
@@ -171,6 +171,9 @@ struct xfrm_state
/* Data for care-of address */
xfrm_address_t *coaddr;
+ /* Data for crypto accelerator */
+ struct xfrm_accl *accl;
+
/* IPComp needs an IPIP tunnel for handling uncompressed packets */
struct xfrm_state *tunnel;
diff --git a/net/xfrm/xfrm_user.c b/net/xfrm/xfrm_user.c
index b95a2d6..5272cbc 100644
--- a/net/xfrm/xfrm_user.c
+++ b/net/xfrm/xfrm_user.c
@@ -359,6 +359,13 @@ static struct xfrm_state *xfrm_state_construct(struct net *net,
goto error;
}
+ if (attrs[XFRMA_CRYPTO_ACCL]) {
+ x->accl = kmemdup(nla_data(attrs[XFRMA_CRYPTO_ACCL]),
+ sizeof(*x->accl), GFP_KERNEL);
+ if (x->accl == NULL)
+ goto error;
+ }
+
err = xfrm_init_state(x);
if (err)
goto error;
@@ -1958,6 +1965,7 @@ static const struct nla_policy xfrma_policy[XFRMA_MAX+1] = {
[XFRMA_POLICY_TYPE] = { .len = sizeof(struct xfrm_userpolicy_type)},
[XFRMA_MIGRATE] = { .len = sizeof(struct xfrm_user_migrate) },
[XFRMA_KMADDRESS] = { .len = sizeof(struct xfrm_user_kmaddress) },
+ [XFRMA_CRYPTO_ACCL] = { .len = sizeof(struct xfrm_accl) },
};
static struct xfrm_link {
--
1.5.4.2
^ permalink raw reply related [flat|nested] 16+ messages in thread* Re: [RFC] [PATCH v2 3/4] xfrm: Add a netlink attribute for software crypto accelerators
2009-04-24 10:28 ` [RFC] [PATCH v2 3/4] xfrm: Add a netlink attribute for software crypto accelerators Steffen Klassert
@ 2009-04-27 8:53 ` Herbert Xu
2009-04-28 10:11 ` Steffen Klassert
0 siblings, 1 reply; 16+ messages in thread
From: Herbert Xu @ 2009-04-27 8:53 UTC (permalink / raw)
To: Steffen Klassert; +Cc: David Miller, linux-crypto
On Fri, Apr 24, 2009 at 12:28:28PM +0200, Steffen Klassert wrote:
>
> +struct xfrm_accl {
> + char name[64];
> + __u8 type;
> +#define XFRM_ACCL_ENC 1
> +#define XFRM_ACCL_AUTH 2
> +#define XFRM_ACCL_COMP 4
> +#define XFRM_ACCL_AEAD 8
> +};
While this should work for pcrypt, I'd like this to be solved
more generally. The crux of the issue is that we can't specify
an arbitrary implementation of a given algorithm. So the obvious
solution is to specify the driver name along with the algorithm
name.
This is in fact pretty much what you've done, but I'd just like
it to be generalised. In particular, instead of having just a
single name per SA, we should allow one to be set for each algorithm
type.
On another note, I don't expect this to be the primary mechanism
for activating parallel processing. Doing it manually on each
SA is just painful. This should be used for testing or when you
want to specify it for a subset of SAs only.
When the admin wants to turn the entire system over to pcrypt,
it should be done at the crypto layer, by simply registering
the pcrypt version of the algorithm in question, and having it
as the default implementation of that algorithm.
In fact, this mechanism should then be able to allow specific
SAs to not use parallel processing, which means that it should
definitely not be called accl :)
Thanks,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply [flat|nested] 16+ messages in thread* Re: [RFC] [PATCH v2 3/4] xfrm: Add a netlink attribute for software crypto accelerators
2009-04-27 8:53 ` Herbert Xu
@ 2009-04-28 10:11 ` Steffen Klassert
0 siblings, 0 replies; 16+ messages in thread
From: Steffen Klassert @ 2009-04-28 10:11 UTC (permalink / raw)
To: Herbert Xu; +Cc: David Miller, linux-crypto
On Mon, Apr 27, 2009 at 04:53:46PM +0800, Herbert Xu wrote:
>
> While this should work for pcrypt, I'd like this to be solved
> more generally. The crux of the issue is that we can't specify
> an arbitrary implementation of a given algorithm. So the obvious
> solution is to specify the driver name along with the algorithm
> name.
So how general should it be? For the moment I would see pcrypt and maybe
cryptd as possible candidates to use this mechanism. I'm just wondering
if it is worth to set up a list of crypto templates that can be choosen
from userspace, similar to the xfrm_algo_list.
>
> This is in fact pretty much what you've done, but I'd just like
> it to be generalised. In particular, instead of having just a
> single name per SA, we should allow one to be set for each algorithm
> type.
Just to get you right, do you think about adding a netlink attribute for
each algorithm type?
>
> On another note, I don't expect this to be the primary mechanism
> for activating parallel processing. Doing it manually on each
> SA is just painful. This should be used for testing or when you
> want to specify it for a subset of SAs only.
>
> When the admin wants to turn the entire system over to pcrypt,
> it should be done at the crypto layer, by simply registering
> the pcrypt version of the algorithm in question, and having it
> as the default implementation of that algorithm.
That's not really clear to me how to let the user register the pcrypt
version of the algorithm, so what's the desired way do this.
>
> In fact, this mechanism should then be able to allow specific
> SAs to not use parallel processing, which means that it should
> definitely not be called accl :)
>
Yes, I think I'll find a better name :)
Steffen
^ permalink raw reply [flat|nested] 16+ messages in thread
* [RFC] [PATCH v2 4/4] esp: Use pcrypt if it is selected as software crypto accelerator
2009-04-24 10:24 [RFC] [PATCH v2 0/4] Parallel IPsec Steffen Klassert
` (2 preceding siblings ...)
2009-04-24 10:28 ` [RFC] [PATCH v2 3/4] xfrm: Add a netlink attribute for software crypto accelerators Steffen Klassert
@ 2009-04-24 10:29 ` Steffen Klassert
2009-04-25 8:38 ` [RFC] [PATCH v2 0/4] Parallel IPsec Evgeniy Polyakov
4 siblings, 0 replies; 16+ messages in thread
From: Steffen Klassert @ 2009-04-24 10:29 UTC (permalink / raw)
To: Herbert Xu; +Cc: David Miller, linux-crypto
This patch adds a function that sets up the format string for
authenc algorithms. If pcrypt is coosen as accelerator,
the format string is set up to use pcrypt. If no accelerator
is choosen, the default format string is set up.
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
---
net/ipv4/esp4.c | 40 ++++++++++++++++++++++++++++++++++++----
net/ipv6/esp6.c | 40 ++++++++++++++++++++++++++++++++++++----
2 files changed, 72 insertions(+), 8 deletions(-)
diff --git a/net/ipv4/esp4.c b/net/ipv4/esp4.c
index 18bb383..bff5609 100644
--- a/net/ipv4/esp4.c
+++ b/net/ipv4/esp4.c
@@ -467,6 +467,40 @@ error:
return err;
}
+static int esp_authenc_name(struct xfrm_state *x, char *authenc_name)
+{
+ struct xfrm_accl *accl = x->accl;
+ int err = 0;
+
+ if (!accl) {
+ if (snprintf(authenc_name, CRYPTO_MAX_ALG_NAME,
+ "authenc(%s,%s)",
+ x->aalg ? x->aalg->alg_name : "digest_null",
+ x->ealg->alg_name) >= CRYPTO_MAX_ALG_NAME)
+ err = -ENAMETOOLONG;
+
+ goto out;
+ }
+
+ /* Set aead as the default accl type if type is unspecified */
+ if (!accl->type)
+ accl->type |= XFRM_ACCL_AEAD;
+
+ if (!strcmp(accl->name, "pcrypt") && (accl->type & XFRM_ACCL_AEAD)) {
+ if (snprintf(authenc_name, CRYPTO_MAX_ALG_NAME,
+ "pcrypt(authenc(%s,eseqiv(%s)))",
+ x->aalg ? x->aalg->alg_name : "digest_null",
+ x->ealg->alg_name) >= CRYPTO_MAX_ALG_NAME)
+ err = -ENAMETOOLONG;
+
+ goto out;
+ }
+
+ err = -EINVAL;
+out:
+ return err;
+}
+
static int esp_init_authenc(struct xfrm_state *x)
{
struct esp_data *esp = x->data;
@@ -483,10 +517,8 @@ static int esp_init_authenc(struct xfrm_state *x)
if (x->ealg == NULL)
goto error;
- err = -ENAMETOOLONG;
- if (snprintf(authenc_name, CRYPTO_MAX_ALG_NAME, "authenc(%s,%s)",
- x->aalg ? x->aalg->alg_name : "digest_null",
- x->ealg->alg_name) >= CRYPTO_MAX_ALG_NAME)
+ err = esp_authenc_name(x, authenc_name);
+ if (err)
goto error;
aead = crypto_alloc_aead(authenc_name, 0, 0);
diff --git a/net/ipv6/esp6.c b/net/ipv6/esp6.c
index c2f2501..ac5ab90 100644
--- a/net/ipv6/esp6.c
+++ b/net/ipv6/esp6.c
@@ -410,6 +410,40 @@ error:
return err;
}
+static int esp_authenc_name(struct xfrm_state *x, char *authenc_name)
+{
+ struct xfrm_accl *accl = x->accl;
+ int err = 0;
+
+ if (!accl) {
+ if (snprintf(authenc_name, CRYPTO_MAX_ALG_NAME,
+ "authenc(%s,%s)",
+ x->aalg ? x->aalg->alg_name : "digest_null",
+ x->ealg->alg_name) >= CRYPTO_MAX_ALG_NAME)
+ err = -ENAMETOOLONG;
+
+ goto out;
+ }
+
+ /* Set aead as the default accl type if type is unspecified */
+ if (!accl->type)
+ accl->type |= XFRM_ACCL_AEAD;
+
+ if (!strcmp(accl->name, "pcrypt") && (accl->type & XFRM_ACCL_AEAD)) {
+ if (snprintf(authenc_name, CRYPTO_MAX_ALG_NAME,
+ "pcrypt(authenc(%s,eseqiv(%s)))",
+ x->aalg ? x->aalg->alg_name : "digest_null",
+ x->ealg->alg_name) >= CRYPTO_MAX_ALG_NAME)
+ err = -ENAMETOOLONG;
+
+ goto out;
+ }
+
+ err = -EINVAL;
+out:
+ return err;
+}
+
static int esp_init_authenc(struct xfrm_state *x)
{
struct esp_data *esp = x->data;
@@ -426,10 +460,8 @@ static int esp_init_authenc(struct xfrm_state *x)
if (x->ealg == NULL)
goto error;
- err = -ENAMETOOLONG;
- if (snprintf(authenc_name, CRYPTO_MAX_ALG_NAME, "authenc(%s,%s)",
- x->aalg ? x->aalg->alg_name : "digest_null",
- x->ealg->alg_name) >= CRYPTO_MAX_ALG_NAME)
+ err = esp_authenc_name(x, authenc_name);
+ if (err)
goto error;
aead = crypto_alloc_aead(authenc_name, 0, 0);
--
1.5.4.2
^ permalink raw reply related [flat|nested] 16+ messages in thread* Re: [RFC] [PATCH v2 0/4] Parallel IPsec
2009-04-24 10:24 [RFC] [PATCH v2 0/4] Parallel IPsec Steffen Klassert
` (3 preceding siblings ...)
2009-04-24 10:29 ` [RFC] [PATCH v2 4/4] esp: Use pcrypt if it is selected as software crypto accelerator Steffen Klassert
@ 2009-04-25 8:38 ` Evgeniy Polyakov
2009-04-25 9:21 ` Herbert Xu
4 siblings, 1 reply; 16+ messages in thread
From: Evgeniy Polyakov @ 2009-04-25 8:38 UTC (permalink / raw)
To: Steffen Klassert; +Cc: Herbert Xu, David Miller, linux-crypto
Hi.
On Fri, Apr 24, 2009 at 12:24:51PM +0200, Steffen Klassert (steffen.klassert@secunet.com) wrote:
> This patchset adds the 'pcrypt' parallel crypto template. With this template it
> is possible to process the crypto requests of a transform in parallel without
> getting request reorder. This is in particular interesting for IPsec.
Why can't it be used by default for all crypto operations instead of
synchronous one?
--
Evgeniy Polyakov
^ permalink raw reply [flat|nested] 16+ messages in thread* Re: [RFC] [PATCH v2 0/4] Parallel IPsec
2009-04-25 8:38 ` [RFC] [PATCH v2 0/4] Parallel IPsec Evgeniy Polyakov
@ 2009-04-25 9:21 ` Herbert Xu
2009-04-25 10:10 ` Evgeniy Polyakov
0 siblings, 1 reply; 16+ messages in thread
From: Herbert Xu @ 2009-04-25 9:21 UTC (permalink / raw)
To: Evgeniy Polyakov; +Cc: Steffen Klassert, David Miller, linux-crypto
On Sat, Apr 25, 2009 at 12:38:49PM +0400, Evgeniy Polyakov wrote:
> Hi.
>
> On Fri, Apr 24, 2009 at 12:24:51PM +0200, Steffen Klassert (steffen.klassert@secunet.com) wrote:
> > This patchset adds the 'pcrypt' parallel crypto template. With this template it
> > is possible to process the crypto requests of a transform in parallel without
> > getting request reorder. This is in particular interesting for IPsec.
>
> Why can't it be used by default for all crypto operations instead of
> synchronous one?
PCI-based drivers will not benefit from spreading the requests
across CPUs. If anything they will suffer from the synchronisation.
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
^ permalink raw reply [flat|nested] 16+ messages in thread* Re: [RFC] [PATCH v2 0/4] Parallel IPsec
2009-04-25 9:21 ` Herbert Xu
@ 2009-04-25 10:10 ` Evgeniy Polyakov
2009-04-25 10:56 ` Herbert Xu
0 siblings, 1 reply; 16+ messages in thread
From: Evgeniy Polyakov @ 2009-04-25 10:10 UTC (permalink / raw)
To: Herbert Xu; +Cc: Steffen Klassert, David Miller, linux-crypto
On Sat, Apr 25, 2009 at 05:21:42PM +0800, Herbert Xu (herbert@gondor.apana.org.au) wrote:
> > Why can't it be used by default for all crypto operations instead of
> > synchronous one?
>
> PCI-based drivers will not benefit from spreading the requests
> across CPUs. If anything they will suffer from the synchronisation.
What's the deal PCI drivers have with CPUs data comes from/to?
They do not touch cache, just run DMA transfer and complete the request.
--
Evgeniy Polyakov
^ permalink raw reply [flat|nested] 16+ messages in thread