* [PATCH 1/4] io-wq: make the io_wq ref counted
2020-01-24 21:31 [PATCHSET v2 0/4] Add support for shared io-wq backends Jens Axboe
@ 2020-01-24 21:31 ` Jens Axboe
2020-01-24 21:31 ` [PATCH 2/4] io-wq: add 'id' to io_wq Jens Axboe
` (2 subsequent siblings)
3 siblings, 0 replies; 10+ messages in thread
From: Jens Axboe @ 2020-01-24 21:31 UTC (permalink / raw)
To: io-uring; +Cc: Jens Axboe
In preparation for sharing an io-wq across different users, add a
reference count that manages destruction of it.
Signed-off-by: Jens Axboe <[email protected]>
---
fs/io-wq.c | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/fs/io-wq.c b/fs/io-wq.c
index 4d902c19ee5f..54e270ae12ab 100644
--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -113,6 +113,8 @@ struct io_wq {
struct mm_struct *mm;
refcount_t refs;
struct completion done;
+
+ refcount_t use_refs;
};
static bool io_worker_get(struct io_worker *worker)
@@ -1073,6 +1075,7 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
ret = -ENOMEM;
goto err;
}
+ refcount_set(&wq->use_refs, 1);
reinit_completion(&wq->done);
return wq;
}
@@ -1093,7 +1096,7 @@ static bool io_wq_worker_wake(struct io_worker *worker, void *data)
return false;
}
-void io_wq_destroy(struct io_wq *wq)
+static void __io_wq_destroy(struct io_wq *wq)
{
int node;
@@ -1113,3 +1116,9 @@ void io_wq_destroy(struct io_wq *wq)
kfree(wq->wqes);
kfree(wq);
}
+
+void io_wq_destroy(struct io_wq *wq)
+{
+ if (refcount_dec_and_test(&wq->use_refs))
+ __io_wq_destroy(wq);
+}
--
2.25.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 2/4] io-wq: add 'id' to io_wq
2020-01-24 21:31 [PATCHSET v2 0/4] Add support for shared io-wq backends Jens Axboe
2020-01-24 21:31 ` [PATCH 1/4] io-wq: make the io_wq ref counted Jens Axboe
@ 2020-01-24 21:31 ` Jens Axboe
2020-01-24 21:31 ` [PATCH 3/4] io-wq: allow lookup of existing io_wq with given id Jens Axboe
2020-01-24 21:31 ` [PATCH 4/4] io_uring: add support for sharing kernel io-wq workqueue Jens Axboe
3 siblings, 0 replies; 10+ messages in thread
From: Jens Axboe @ 2020-01-24 21:31 UTC (permalink / raw)
To: io-uring; +Cc: Jens Axboe
Add each io_wq to a global list, and assign an id for each of them. This
is in preparation for attaching to an existing io_wq, rather than
creating a new one.
Signed-off-by: Jens Axboe <[email protected]>
---
fs/io-wq.c | 25 ++++++++++++++++++++++++-
1 file changed, 24 insertions(+), 1 deletion(-)
diff --git a/fs/io-wq.c b/fs/io-wq.c
index 54e270ae12ab..51b3677d6fbd 100644
--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -16,9 +16,14 @@
#include <linux/slab.h>
#include <linux/kthread.h>
#include <linux/rculist_nulls.h>
+#include <linux/idr.h>
#include "io-wq.h"
+static LIST_HEAD(wq_list);
+static DEFINE_MUTEX(wq_lock);
+static DEFINE_IDR(wq_idr);
+
#define WORKER_IDLE_TIMEOUT (5 * HZ)
enum {
@@ -115,6 +120,8 @@ struct io_wq {
struct completion done;
refcount_t use_refs;
+ struct list_head wq_list;
+ unsigned int id;
};
static bool io_worker_get(struct io_worker *worker)
@@ -1076,6 +1083,15 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
goto err;
}
refcount_set(&wq->use_refs, 1);
+
+ /* if we're out of IDs or fail to get one, use 0 */
+ mutex_lock(&wq_lock);
+ wq->id = idr_alloc(&wq_idr, wq, 1, INT_MAX, GFP_KERNEL);
+ if (wq->id < 0)
+ wq->id = 0;
+
+ list_add(&wq->wq_list, &wq_list);
+ mutex_unlock(&wq_lock);
reinit_completion(&wq->done);
return wq;
}
@@ -1119,6 +1135,13 @@ static void __io_wq_destroy(struct io_wq *wq)
void io_wq_destroy(struct io_wq *wq)
{
- if (refcount_dec_and_test(&wq->use_refs))
+ if (refcount_dec_and_test(&wq->use_refs)) {
+ mutex_lock(&wq_lock);
+ if (wq->id)
+ idr_remove(&wq_idr, wq->id);
+ list_del(&wq->wq_list);
+ mutex_unlock(&wq_lock);
+
__io_wq_destroy(wq);
+ }
}
--
2.25.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 3/4] io-wq: allow lookup of existing io_wq with given id
2020-01-24 21:31 [PATCHSET v2 0/4] Add support for shared io-wq backends Jens Axboe
2020-01-24 21:31 ` [PATCH 1/4] io-wq: make the io_wq ref counted Jens Axboe
2020-01-24 21:31 ` [PATCH 2/4] io-wq: add 'id' to io_wq Jens Axboe
@ 2020-01-24 21:31 ` Jens Axboe
2020-01-24 21:31 ` [PATCH 4/4] io_uring: add support for sharing kernel io-wq workqueue Jens Axboe
3 siblings, 0 replies; 10+ messages in thread
From: Jens Axboe @ 2020-01-24 21:31 UTC (permalink / raw)
To: io-uring; +Cc: Jens Axboe
If the id and user/creds match, return an existing io_wq if we can safely
grab a reference to it.
Signed-off-by: Jens Axboe <[email protected]>
---
fs/io-wq.c | 37 ++++++++++++++++++++++++++++++++++++-
fs/io-wq.h | 11 ++++++++++-
2 files changed, 46 insertions(+), 2 deletions(-)
diff --git a/fs/io-wq.c b/fs/io-wq.c
index 51b3677d6fbd..08ef69b498f6 100644
--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -1024,7 +1024,7 @@ void io_wq_flush(struct io_wq *wq)
}
}
-struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
+static struct io_wq *__io_wq_create(unsigned bounded, struct io_wq_data *data)
{
int ret = -ENOMEM, node;
struct io_wq *wq;
@@ -1106,6 +1106,36 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
return ERR_PTR(ret);
}
+static bool io_wq_match(struct io_wq *wq, struct io_wq_data *data)
+{
+ if (data->creds != wq->creds || data->user != wq->user)
+ return false;
+ if (data->get_work != wq->get_work || data->put_work != wq->put_work)
+ return false;
+ return refcount_inc_not_zero(&wq->use_refs);
+}
+
+/*
+ * Find and return io_wq with given id and grab a reference to it.
+ */
+struct io_wq *io_wq_create_id(unsigned bounded, struct io_wq_data *data,
+ unsigned int id)
+{
+ if (id) {
+ struct io_wq *wq;
+
+ mutex_lock(&wq_lock);
+ wq = idr_find(&wq_idr, id);
+ if (wq && io_wq_match(wq, data)) {
+ mutex_unlock(&wq_lock);
+ return wq;
+ }
+ mutex_unlock(&wq_lock);
+ }
+
+ return __io_wq_create(bounded, data);
+}
+
static bool io_wq_worker_wake(struct io_worker *worker, void *data)
{
wake_up_process(worker->task);
@@ -1145,3 +1175,8 @@ void io_wq_destroy(struct io_wq *wq)
__io_wq_destroy(wq);
}
}
+
+unsigned int io_wq_id(struct io_wq *wq)
+{
+ return wq->id;
+}
diff --git a/fs/io-wq.h b/fs/io-wq.h
index 1cd039af8813..dee7bb3b7cd4 100644
--- a/fs/io-wq.h
+++ b/fs/io-wq.h
@@ -97,8 +97,17 @@ struct io_wq_data {
put_work_fn *put_work;
};
-struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data);
+struct io_wq *io_wq_create_id(unsigned bounded, struct io_wq_data *data,
+ unsigned int id);
+
+static inline struct io_wq *io_wq_create(unsigned bounded,
+ struct io_wq_data *data)
+{
+ return io_wq_create_id(bounded, data, 0);
+}
+
void io_wq_destroy(struct io_wq *wq);
+unsigned int io_wq_id(struct io_wq *wq);
void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work);
void io_wq_enqueue_hashed(struct io_wq *wq, struct io_wq_work *work, void *val);
--
2.25.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* [PATCH 4/4] io_uring: add support for sharing kernel io-wq workqueue
2020-01-24 21:31 [PATCHSET v2 0/4] Add support for shared io-wq backends Jens Axboe
` (2 preceding siblings ...)
2020-01-24 21:31 ` [PATCH 3/4] io-wq: allow lookup of existing io_wq with given id Jens Axboe
@ 2020-01-24 21:31 ` Jens Axboe
2020-01-25 7:45 ` Stefan Metzmacher
3 siblings, 1 reply; 10+ messages in thread
From: Jens Axboe @ 2020-01-24 21:31 UTC (permalink / raw)
To: io-uring; +Cc: Jens Axboe
An id field is added to io_uring_params, which always returns the ID of
the io-wq backend that is associated with an io_uring context. If an 'id'
is provided and IORING_SETUP_SHARED is set in the creation flags, then
we attempt to attach to an existing io-wq instead of setting up a new one.
This allows creation of "sibling" io_urings, where we prefer to keep the
SQ/CQ private, but want to share the async backend to minimize the amount
of overhead associated with having multiple rings that belong to the same
backend.
Signed-off-by: Jens Axboe <[email protected]>
---
fs/io_uring.c | 20 +++++++++++++++++---
include/uapi/linux/io_uring.h | 4 +++-
2 files changed, 20 insertions(+), 4 deletions(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 9f73586dcfb8..3dad12906db3 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -5673,7 +5673,7 @@ static int io_sq_offload_start(struct io_ring_ctx *ctx,
{
struct io_wq_data data;
unsigned concurrency;
- int ret;
+ int ret, id;
init_waitqueue_head(&ctx->sqo_wait);
mmgrab(current->mm);
@@ -5724,13 +5724,23 @@ static int io_sq_offload_start(struct io_ring_ctx *ctx,
/* Do QD, or 4 * CPUS, whatever is smallest */
concurrency = min(ctx->sq_entries, 4 * num_online_cpus());
- ctx->io_wq = io_wq_create(concurrency, &data);
+
+ id = 0;
+ if (ctx->flags & IORING_SETUP_ATTACH_WQ) {
+ id = p->id;
+ if (!id) {
+ ret = -EINVAL;
+ goto err;
+ }
+ }
+ ctx->io_wq = io_wq_create_id(concurrency, &data, id);
if (IS_ERR(ctx->io_wq)) {
ret = PTR_ERR(ctx->io_wq);
ctx->io_wq = NULL;
goto err;
}
+ p->id = io_wq_id(ctx->io_wq);
return 0;
err:
io_finish_async(ctx);
@@ -6554,7 +6564,11 @@ static long io_uring_setup(u32 entries, struct io_uring_params __user *params)
if (p.flags & ~(IORING_SETUP_IOPOLL | IORING_SETUP_SQPOLL |
IORING_SETUP_SQ_AFF | IORING_SETUP_CQSIZE |
- IORING_SETUP_CLAMP))
+ IORING_SETUP_CLAMP | IORING_SETUP_ATTACH_WQ))
+ return -EINVAL;
+
+ /* id isn't valid without ATTACH_WQ being set */
+ if (!(p.flags & IORING_SETUP_ATTACH_WQ) && p.id)
return -EINVAL;
ret = io_uring_create(entries, &p);
diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
index 57d05cc5e271..f66e53c74a3d 100644
--- a/include/uapi/linux/io_uring.h
+++ b/include/uapi/linux/io_uring.h
@@ -75,6 +75,7 @@ enum {
#define IORING_SETUP_SQ_AFF (1U << 2) /* sq_thread_cpu is valid */
#define IORING_SETUP_CQSIZE (1U << 3) /* app defines CQ size */
#define IORING_SETUP_CLAMP (1U << 4) /* clamp SQ/CQ ring sizes */
+#define IORING_SETUP_ATTACH_WQ (1U << 5) /* attach to existing wq */
enum {
IORING_OP_NOP,
@@ -183,7 +184,8 @@ struct io_uring_params {
__u32 sq_thread_cpu;
__u32 sq_thread_idle;
__u32 features;
- __u32 resv[4];
+ __u32 id;
+ __u32 resv[3];
struct io_sqring_offsets sq_off;
struct io_cqring_offsets cq_off;
};
--
2.25.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH 4/4] io_uring: add support for sharing kernel io-wq workqueue
2020-01-24 21:31 ` [PATCH 4/4] io_uring: add support for sharing kernel io-wq workqueue Jens Axboe
@ 2020-01-25 7:45 ` Stefan Metzmacher
2020-01-25 16:44 ` Jens Axboe
0 siblings, 1 reply; 10+ messages in thread
From: Stefan Metzmacher @ 2020-01-25 7:45 UTC (permalink / raw)
To: Jens Axboe, io-uring
[-- Attachment #1.1: Type: text/plain, Size: 1167 bytes --]
Am 24.01.20 um 22:31 schrieb Jens Axboe:
> An id field is added to io_uring_params, which always returns the ID of
> the io-wq backend that is associated with an io_uring context. If an 'id'
> is provided and IORING_SETUP_SHARED is set in the creation flags, then
> we attempt to attach to an existing io-wq instead of setting up a new one.
Use the new name here too.
> ret = io_uring_create(entries, &p);
> diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
> index 57d05cc5e271..f66e53c74a3d 100644
> --- a/include/uapi/linux/io_uring.h
> +++ b/include/uapi/linux/io_uring.h
> @@ -75,6 +75,7 @@ enum {
> #define IORING_SETUP_SQ_AFF (1U << 2) /* sq_thread_cpu is valid */
> #define IORING_SETUP_CQSIZE (1U << 3) /* app defines CQ size */
> #define IORING_SETUP_CLAMP (1U << 4) /* clamp SQ/CQ ring sizes */
> +#define IORING_SETUP_ATTACH_WQ (1U << 5) /* attach to existing wq */
>
> enum {
> IORING_OP_NOP,
> @@ -183,7 +184,8 @@ struct io_uring_params {
> __u32 sq_thread_cpu;
> __u32 sq_thread_idle;
> __u32 features;
> - __u32 resv[4];
> + __u32 id;
I think this should be wq_id;
metze
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 4/4] io_uring: add support for sharing kernel io-wq workqueue
2020-01-25 7:45 ` Stefan Metzmacher
@ 2020-01-25 16:44 ` Jens Axboe
0 siblings, 0 replies; 10+ messages in thread
From: Jens Axboe @ 2020-01-25 16:44 UTC (permalink / raw)
To: Stefan Metzmacher, io-uring
On 1/25/20 12:45 AM, Stefan Metzmacher wrote:
> Am 24.01.20 um 22:31 schrieb Jens Axboe:
>> An id field is added to io_uring_params, which always returns the ID of
>> the io-wq backend that is associated with an io_uring context. If an 'id'
>> is provided and IORING_SETUP_SHARED is set in the creation flags, then
>> we attempt to attach to an existing io-wq instead of setting up a new one.
>
> Use the new name here too.
Already fixed that up yesterday.
>> ret = io_uring_create(entries, &p);
>> diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h
>> index 57d05cc5e271..f66e53c74a3d 100644
>> --- a/include/uapi/linux/io_uring.h
>> +++ b/include/uapi/linux/io_uring.h
>> @@ -75,6 +75,7 @@ enum {
>> #define IORING_SETUP_SQ_AFF (1U << 2) /* sq_thread_cpu is valid */
>> #define IORING_SETUP_CQSIZE (1U << 3) /* app defines CQ size */
>> #define IORING_SETUP_CLAMP (1U << 4) /* clamp SQ/CQ ring sizes */
>> +#define IORING_SETUP_ATTACH_WQ (1U << 5) /* attach to existing wq */
>>
>> enum {
>> IORING_OP_NOP,
>> @@ -183,7 +184,8 @@ struct io_uring_params {
>> __u32 sq_thread_cpu;
>> __u32 sq_thread_idle;
>> __u32 features;
>> - __u32 resv[4];
>> + __u32 id;
>
> I think this should be wq_id;
Yeah, probably clearer, I'll make that change, thanks.
--
Jens Axboe
^ permalink raw reply [flat|nested] 10+ messages in thread
* [PATCH 3/4] io-wq: allow lookup of existing io_wq with given id
2020-01-23 23:16 [PATCHSET 0/4] Add support for shared io-wq backends Jens Axboe
@ 2020-01-23 23:16 ` Jens Axboe
2020-01-24 9:54 ` Stefan Metzmacher
0 siblings, 1 reply; 10+ messages in thread
From: Jens Axboe @ 2020-01-23 23:16 UTC (permalink / raw)
To: io-uring; +Cc: Jens Axboe
If the id and user/creds match, return an existing io_wq if we can safely
grab a reference to it.
Signed-off-by: Jens Axboe <[email protected]>
---
fs/io-wq.c | 42 +++++++++++++++++++++++++++++++++++++++++-
fs/io-wq.h | 3 +++
2 files changed, 44 insertions(+), 1 deletion(-)
diff --git a/fs/io-wq.c b/fs/io-wq.c
index 8cb0cff5f15c..a9985856033d 100644
--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -1024,7 +1024,7 @@ void io_wq_flush(struct io_wq *wq)
}
}
-struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
+static struct io_wq *__io_wq_create(unsigned bounded, struct io_wq_data *data)
{
int ret = -ENOMEM, node, id;
struct io_wq *wq;
@@ -1107,6 +1107,41 @@ struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
return ERR_PTR(ret);
}
+struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data)
+{
+ return __io_wq_create(bounded, data);
+}
+
+/*
+ * Find and return io_wq with given id and grab a reference to it.
+ */
+struct io_wq *io_wq_create_id(unsigned bounded, struct io_wq_data *data,
+ unsigned int id)
+{
+ struct io_wq *wq, *ret = NULL;
+
+ mutex_lock(&wq_lock);
+ list_for_each_entry(wq, &wq_list, wq_list) {
+ if (id != wq->id)
+ continue;
+ if (data->creds != wq->creds || data->user != wq->user)
+ continue;
+ if (data->get_work != wq->get_work ||
+ data->put_work != wq->put_work)
+ continue;
+ if (!refcount_inc_not_zero(&wq->use_refs))
+ continue;
+ ret = wq;
+ break;
+ }
+ mutex_unlock(&wq_lock);
+
+ if (!ret)
+ ret = io_wq_create(bounded, data);
+
+ return ret;
+}
+
static bool io_wq_worker_wake(struct io_worker *worker, void *data)
{
wake_up_process(worker->task);
@@ -1145,3 +1180,8 @@ void io_wq_destroy(struct io_wq *wq)
__io_wq_destroy(wq);
}
}
+
+unsigned int io_wq_id(struct io_wq *wq)
+{
+ return wq->id;
+}
diff --git a/fs/io-wq.h b/fs/io-wq.h
index 1cd039af8813..7abe2c56b535 100644
--- a/fs/io-wq.h
+++ b/fs/io-wq.h
@@ -98,7 +98,10 @@ struct io_wq_data {
};
struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data);
+struct io_wq *io_wq_create_id(unsigned bounded, struct io_wq_data *data,
+ unsigned int id);
void io_wq_destroy(struct io_wq *wq);
+unsigned int io_wq_id(struct io_wq *wq);
void io_wq_enqueue(struct io_wq *wq, struct io_wq_work *work);
void io_wq_enqueue_hashed(struct io_wq *wq, struct io_wq_work *work, void *val);
--
2.25.0
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH 3/4] io-wq: allow lookup of existing io_wq with given id
2020-01-23 23:16 ` [PATCH 3/4] io-wq: allow lookup of existing io_wq with given id Jens Axboe
@ 2020-01-24 9:54 ` Stefan Metzmacher
2020-01-24 16:41 ` Jens Axboe
0 siblings, 1 reply; 10+ messages in thread
From: Stefan Metzmacher @ 2020-01-24 9:54 UTC (permalink / raw)
To: Jens Axboe, io-uring
[-- Attachment #1.1: Type: text/plain, Size: 803 bytes --]
> +/*
> + * Find and return io_wq with given id and grab a reference to it.
> + */
> +struct io_wq *io_wq_create_id(unsigned bounded, struct io_wq_data *data,
> + unsigned int id)
> +{
> + struct io_wq *wq, *ret = NULL;
> +
> + mutex_lock(&wq_lock);
> + list_for_each_entry(wq, &wq_list, wq_list) {
> + if (id != wq->id)
> + continue;
> + if (data->creds != wq->creds || data->user != wq->user)
> + continue;
> + if (data->get_work != wq->get_work ||
> + data->put_work != wq->put_work)
> + continue;
> + if (!refcount_inc_not_zero(&wq->use_refs))
> + continue;
> + ret = wq;
> + break;
> + }
> + mutex_unlock(&wq_lock);
Isn't there a more efficient ida_find function in order to avoid
the loop, which won't really scale in the long run.
metze
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH 3/4] io-wq: allow lookup of existing io_wq with given id
2020-01-24 9:54 ` Stefan Metzmacher
@ 2020-01-24 16:41 ` Jens Axboe
0 siblings, 0 replies; 10+ messages in thread
From: Jens Axboe @ 2020-01-24 16:41 UTC (permalink / raw)
To: Stefan Metzmacher, io-uring
On 1/24/20 2:54 AM, Stefan Metzmacher wrote:
>> +/*
>> + * Find and return io_wq with given id and grab a reference to it.
>> + */
>> +struct io_wq *io_wq_create_id(unsigned bounded, struct io_wq_data *data,
>> + unsigned int id)
>> +{
>> + struct io_wq *wq, *ret = NULL;
>> +
>> + mutex_lock(&wq_lock);
>> + list_for_each_entry(wq, &wq_list, wq_list) {
>> + if (id != wq->id)
>> + continue;
>> + if (data->creds != wq->creds || data->user != wq->user)
>> + continue;
>> + if (data->get_work != wq->get_work ||
>> + data->put_work != wq->put_work)
>> + continue;
>> + if (!refcount_inc_not_zero(&wq->use_refs))
>> + continue;
>> + ret = wq;
>> + break;
>> + }
>> + mutex_unlock(&wq_lock);
>
> Isn't there a more efficient ida_find function in order to avoid
> the loop, which won't really scale in the long run.
Yeah that would be a lot better - I initially just used a sequence
for this, but since I'm now using ida, might as well use it for
lookup as well. I'll make the change.
--
Jens Axboe
^ permalink raw reply [flat|nested] 10+ messages in thread