* [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues
2024-03-13 13:24 [PATCH rdma-next 0/4] Define and use mana queues for CQs and WQs Konstantin Taranov
@ 2024-03-13 13:24 ` Konstantin Taranov
2024-03-15 16:45 ` Long Li
2024-03-17 6:42 ` Zhu Yanjun
2024-03-13 13:24 ` [PATCH rdma-next 2/4] RDMA:mana_ib: Use struct mana_ib_queue for CQs Konstantin Taranov
` (2 subsequent siblings)
3 siblings, 2 replies; 13+ messages in thread
From: Konstantin Taranov @ 2024-03-13 13:24 UTC (permalink / raw)
To: kotaranov, sharmaajay, longli, jgg, leon; +Cc: linux-rdma, linux-kernel
From: Konstantin Taranov <kotaranov@microsoft.com>
Intoduce helpers to work with mana ib queues (struct mana_ib_queue).
A queue always consists of umem, gdma_region, and id.
A queue can be used for a WQ or a CQ.
Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
---
drivers/infiniband/hw/mana/main.c | 40 ++++++++++++++++++++++++++++
drivers/infiniband/hw/mana/mana_ib.h | 10 +++++++
2 files changed, 50 insertions(+)
diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
index 71e33feee..0ec940b97 100644
--- a/drivers/infiniband/hw/mana/main.c
+++ b/drivers/infiniband/hw/mana/main.c
@@ -237,6 +237,46 @@ void mana_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
ibdev_dbg(ibdev, "Failed to destroy doorbell page %d\n", ret);
}
+int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
+ struct mana_ib_queue *queue)
+{
+ struct ib_umem *umem;
+ int err;
+
+ queue->umem = NULL;
+ queue->id = INVALID_QUEUE_ID;
+ queue->gdma_region = GDMA_INVALID_DMA_REGION;
+
+ umem = ib_umem_get(&mdev->ib_dev, addr, size, IB_ACCESS_LOCAL_WRITE);
+ if (IS_ERR(umem)) {
+ err = PTR_ERR(umem);
+ ibdev_dbg(&mdev->ib_dev, "Failed to get umem, %d\n", err);
+ return err;
+ }
+
+ err = mana_ib_create_zero_offset_dma_region(mdev, umem, &queue->gdma_region);
+ if (err) {
+ ibdev_dbg(&mdev->ib_dev, "Failed to create dma region, %d\n", err);
+ goto free_umem;
+ }
+ queue->umem = umem;
+
+ ibdev_dbg(&mdev->ib_dev,
+ "create_dma_region ret %d gdma_region 0x%llx\n",
+ err, queue->gdma_region);
+
+ return 0;
+free_umem:
+ ib_umem_release(umem);
+ return err;
+}
+
+void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct mana_ib_queue *queue)
+{
+ mana_ib_gd_destroy_dma_region(mdev, queue->gdma_region);
+ ib_umem_release(queue->umem);
+}
+
static int
mana_ib_gd_first_dma_region(struct mana_ib_dev *dev,
struct gdma_context *gc,
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index f83390eeb..859fd3bfc 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -45,6 +45,12 @@ struct mana_ib_adapter_caps {
u32 max_inline_data_size;
};
+struct mana_ib_queue {
+ struct ib_umem *umem;
+ u64 gdma_region;
+ u64 id;
+};
+
struct mana_ib_dev {
struct ib_device ib_dev;
struct gdma_dev *gdma_dev;
@@ -169,6 +175,10 @@ int mana_ib_create_dma_region(struct mana_ib_dev *dev, struct ib_umem *umem,
int mana_ib_gd_destroy_dma_region(struct mana_ib_dev *dev,
mana_handle_t gdma_region);
+int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
+ struct mana_ib_queue *queue);
+void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct mana_ib_queue *queue);
+
struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
struct ib_wq_init_attr *init_attr,
struct ib_udata *udata);
--
2.43.0
^ permalink raw reply related [flat|nested] 13+ messages in thread* RE: [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues
2024-03-13 13:24 ` [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues Konstantin Taranov
@ 2024-03-15 16:45 ` Long Li
2024-03-17 6:42 ` Zhu Yanjun
1 sibling, 0 replies; 13+ messages in thread
From: Long Li @ 2024-03-15 16:45 UTC (permalink / raw)
To: Konstantin Taranov, Konstantin Taranov, sharmaajay, jgg, leon
Cc: linux-rdma, linux-kernel
> -----Original Message-----
> From: Konstantin Taranov <kotaranov@linux.microsoft.com>
> Sent: Wednesday, March 13, 2024 6:25 AM
> To: Konstantin Taranov <kotaranov@microsoft.com>;
> sharmaajay@microsoft.com; Long Li <longli@microsoft.com>; jgg@ziepe.ca;
> leon@kernel.org
> Cc: linux-rdma@vger.kernel.org; linux-kernel@vger.kernel.org
> Subject: [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and
> destroy mana queues
>
> From: Konstantin Taranov <kotaranov@microsoft.com>
>
> Intoduce helpers to work with mana ib queues (struct mana_ib_queue).
> A queue always consists of umem, gdma_region, and id.
> A queue can be used for a WQ or a CQ.
>
> Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Reviewed-by: Long Li <longli@microsoft.com>
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues
2024-03-13 13:24 ` [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues Konstantin Taranov
2024-03-15 16:45 ` Long Li
@ 2024-03-17 6:42 ` Zhu Yanjun
2024-03-18 9:31 ` [EXTERNAL] " Konstantin Taranov
1 sibling, 1 reply; 13+ messages in thread
From: Zhu Yanjun @ 2024-03-17 6:42 UTC (permalink / raw)
To: Konstantin Taranov, kotaranov, sharmaajay, longli, jgg, leon
Cc: linux-rdma, linux-kernel
在 2024/3/13 14:24, Konstantin Taranov 写道:
> From: Konstantin Taranov <kotaranov@microsoft.com>
>
> Intoduce helpers to work with mana ib queues (struct mana_ib_queue).
> A queue always consists of umem, gdma_region, and id.
> A queue can be used for a WQ or a CQ.
>
> Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
> ---
> drivers/infiniband/hw/mana/main.c | 40 ++++++++++++++++++++++++++++
> drivers/infiniband/hw/mana/mana_ib.h | 10 +++++++
> 2 files changed, 50 insertions(+)
>
> diff --git a/drivers/infiniband/hw/mana/main.c b/drivers/infiniband/hw/mana/main.c
> index 71e33feee..0ec940b97 100644
> --- a/drivers/infiniband/hw/mana/main.c
> +++ b/drivers/infiniband/hw/mana/main.c
> @@ -237,6 +237,46 @@ void mana_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
> ibdev_dbg(ibdev, "Failed to destroy doorbell page %d\n", ret);
> }
>
> +int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
> + struct mana_ib_queue *queue)
> +{
> + struct ib_umem *umem;
> + int err;
> +
> + queue->umem = NULL;
> + queue->id = INVALID_QUEUE_ID;
> + queue->gdma_region = GDMA_INVALID_DMA_REGION;
> +
> + umem = ib_umem_get(&mdev->ib_dev, addr, size, IB_ACCESS_LOCAL_WRITE);
> + if (IS_ERR(umem)) {
> + err = PTR_ERR(umem);
> + ibdev_dbg(&mdev->ib_dev, "Failed to get umem, %d\n", err);
> + return err;
> + }
> +
> + err = mana_ib_create_zero_offset_dma_region(mdev, umem, &queue->gdma_region);
> + if (err) {
> + ibdev_dbg(&mdev->ib_dev, "Failed to create dma region, %d\n", err);
> + goto free_umem;
> + }
> + queue->umem = umem;
> +
> + ibdev_dbg(&mdev->ib_dev,
> + "create_dma_region ret %d gdma_region 0x%llx\n",
> + err, queue->gdma_region);
> +
> + return 0;
> +free_umem:
> + ib_umem_release(umem);
> + return err;
> +}
> +
> +void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct mana_ib_queue *queue)
> +{
> + mana_ib_gd_destroy_dma_region(mdev, queue->gdma_region);
The function mana_ib_gd_destroy_dma_region will call
mana_gd_destroy_dma_region. In the function mana_gd_destroy_dma_region,
the function mana_gd_send_request will return the error -EPROTO.
The procedure is as below. So the function mana_ib_destroy_queue should
also handle this error?
mana_ib_gd_destroy_dma_region --- > mana_gd_destroy_dma_region
693 int mana_gd_destroy_dma_region(struct gdma_context *gc, u64
dma_region_handle)
694 {
...
706 err = mana_gd_send_request(gc, sizeof(req), &req,
sizeof(resp), &resp);
707 if (err || resp.hdr.status) {
708 dev_err(gc->dev, "Failed to destroy DMA region:
%d, 0x%x\n",
709 err, resp.hdr.status);
710 return -EPROTO;
711 }
...
714 }
Zhu Yanjun
> + ib_umem_release(queue->umem);
> +}
> +
> static int
> mana_ib_gd_first_dma_region(struct mana_ib_dev *dev,
> struct gdma_context *gc,
> diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
> index f83390eeb..859fd3bfc 100644
> --- a/drivers/infiniband/hw/mana/mana_ib.h
> +++ b/drivers/infiniband/hw/mana/mana_ib.h
> @@ -45,6 +45,12 @@ struct mana_ib_adapter_caps {
> u32 max_inline_data_size;
> };
>
> +struct mana_ib_queue {
> + struct ib_umem *umem;
> + u64 gdma_region;
> + u64 id;
> +};
> +
> struct mana_ib_dev {
> struct ib_device ib_dev;
> struct gdma_dev *gdma_dev;
> @@ -169,6 +175,10 @@ int mana_ib_create_dma_region(struct mana_ib_dev *dev, struct ib_umem *umem,
> int mana_ib_gd_destroy_dma_region(struct mana_ib_dev *dev,
> mana_handle_t gdma_region);
>
> +int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
> + struct mana_ib_queue *queue);
> +void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct mana_ib_queue *queue);
> +
> struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
> struct ib_wq_init_attr *init_attr,
> struct ib_udata *udata);
^ permalink raw reply [flat|nested] 13+ messages in thread* RE: [EXTERNAL] Re: [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues
2024-03-17 6:42 ` Zhu Yanjun
@ 2024-03-18 9:31 ` Konstantin Taranov
2024-03-18 13:56 ` Zhu Yanjun
0 siblings, 1 reply; 13+ messages in thread
From: Konstantin Taranov @ 2024-03-18 9:31 UTC (permalink / raw)
To: Zhu Yanjun, Konstantin Taranov, sharmaajay, Long Li, jgg, leon
Cc: linux-rdma, linux-kernel
> > From: Konstantin Taranov <kotaranov@microsoft.com>
> >
> > Intoduce helpers to work with mana ib queues (struct mana_ib_queue).
> > A queue always consists of umem, gdma_region, and id.
> > A queue can be used for a WQ or a CQ.
> >
> > Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
> > ---
> > drivers/infiniband/hw/mana/main.c | 40
> ++++++++++++++++++++++++++++
> > drivers/infiniband/hw/mana/mana_ib.h | 10 +++++++
> > 2 files changed, 50 insertions(+)
> >
> > diff --git a/drivers/infiniband/hw/mana/main.c
> > b/drivers/infiniband/hw/mana/main.c
> > index 71e33feee..0ec940b97 100644
> > --- a/drivers/infiniband/hw/mana/main.c
> > +++ b/drivers/infiniband/hw/mana/main.c
> > @@ -237,6 +237,46 @@ void mana_ib_dealloc_ucontext(struct
> ib_ucontext *ibcontext)
> > ibdev_dbg(ibdev, "Failed to destroy doorbell page %d\n", ret);
> > }
> >
> > +int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
> > + struct mana_ib_queue *queue) {
> > + struct ib_umem *umem;
> > + int err;
> > +
> > + queue->umem = NULL;
> > + queue->id = INVALID_QUEUE_ID;
> > + queue->gdma_region = GDMA_INVALID_DMA_REGION;
> > +
> > + umem = ib_umem_get(&mdev->ib_dev, addr, size,
> IB_ACCESS_LOCAL_WRITE);
> > + if (IS_ERR(umem)) {
> > + err = PTR_ERR(umem);
> > + ibdev_dbg(&mdev->ib_dev, "Failed to get umem, %d\n", err);
> > + return err;
> > + }
> > +
> > + err = mana_ib_create_zero_offset_dma_region(mdev, umem, &queue-
> >gdma_region);
> > + if (err) {
> > + ibdev_dbg(&mdev->ib_dev, "Failed to create dma region, %d\n",
> err);
> > + goto free_umem;
> > + }
> > + queue->umem = umem;
> > +
> > + ibdev_dbg(&mdev->ib_dev,
> > + "create_dma_region ret %d gdma_region 0x%llx\n",
> > + err, queue->gdma_region);
> > +
> > + return 0;
> > +free_umem:
> > + ib_umem_release(umem);
> > + return err;
> > +}
> > +
> > +void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct
> > +mana_ib_queue *queue) {
> > + mana_ib_gd_destroy_dma_region(mdev, queue->gdma_region);
>
> The function mana_ib_gd_destroy_dma_region will call
> mana_gd_destroy_dma_region. In the function
> mana_gd_destroy_dma_region, the function mana_gd_send_request will
> return the error -EPROTO.
> The procedure is as below. So the function mana_ib_destroy_queue should
> also handle this error?
Thanks for the comment!
This error can be ignored and it was ignored before this commit.
I checked the corresponding Windows driver code, and it is also intentionally ignored there.
I can add a comment that the error is ignored intentionally if you want.
>
> mana_ib_gd_destroy_dma_region --- > mana_gd_destroy_dma_region
>
> 693 int mana_gd_destroy_dma_region(struct gdma_context *gc, u64
> dma_region_handle)
> 694 {
>
> ...
>
> 706 err = mana_gd_send_request(gc, sizeof(req), &req,
> sizeof(resp), &resp);
> 707 if (err || resp.hdr.status) {
> 708 dev_err(gc->dev, "Failed to destroy DMA region:
> %d, 0x%x\n",
> 709 err, resp.hdr.status);
> 710 return -EPROTO;
> 711 }
>
> ...
>
> 714 }
>
> Zhu Yanjun
>
> > + ib_umem_release(queue->umem);
> > +}
> > +
> > static int
> > mana_ib_gd_first_dma_region(struct mana_ib_dev *dev,
> > struct gdma_context *gc, diff --git
> > a/drivers/infiniband/hw/mana/mana_ib.h
> > b/drivers/infiniband/hw/mana/mana_ib.h
> > index f83390eeb..859fd3bfc 100644
> > --- a/drivers/infiniband/hw/mana/mana_ib.h
> > +++ b/drivers/infiniband/hw/mana/mana_ib.h
> > @@ -45,6 +45,12 @@ struct mana_ib_adapter_caps {
> > u32 max_inline_data_size;
> > };
> >
> > +struct mana_ib_queue {
> > + struct ib_umem *umem;
> > + u64 gdma_region;
> > + u64 id;
> > +};
> > +
> > struct mana_ib_dev {
> > struct ib_device ib_dev;
> > struct gdma_dev *gdma_dev;
> > @@ -169,6 +175,10 @@ int mana_ib_create_dma_region(struct
> mana_ib_dev *dev, struct ib_umem *umem,
> > int mana_ib_gd_destroy_dma_region(struct mana_ib_dev *dev,
> > mana_handle_t gdma_region);
> >
> > +int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
> > + struct mana_ib_queue *queue); void
> > +mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct
> mana_ib_queue
> > +*queue);
> > +
> > struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
> > struct ib_wq_init_attr *init_attr,
> > struct ib_udata *udata);
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [EXTERNAL] Re: [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues
2024-03-18 9:31 ` [EXTERNAL] " Konstantin Taranov
@ 2024-03-18 13:56 ` Zhu Yanjun
0 siblings, 0 replies; 13+ messages in thread
From: Zhu Yanjun @ 2024-03-18 13:56 UTC (permalink / raw)
To: Konstantin Taranov, Konstantin Taranov, sharmaajay, Long Li, jgg, leon
Cc: linux-rdma, linux-kernel
On 18.03.24 10:31, Konstantin Taranov wrote:
>>> From: Konstantin Taranov <kotaranov@microsoft.com>
>>>
>>> Intoduce helpers to work with mana ib queues (struct mana_ib_queue).
>>> A queue always consists of umem, gdma_region, and id.
>>> A queue can be used for a WQ or a CQ.
>>>
>>> Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
>>> ---
>>> drivers/infiniband/hw/mana/main.c | 40
>> ++++++++++++++++++++++++++++
>>> drivers/infiniband/hw/mana/mana_ib.h | 10 +++++++
>>> 2 files changed, 50 insertions(+)
>>>
>>> diff --git a/drivers/infiniband/hw/mana/main.c
>>> b/drivers/infiniband/hw/mana/main.c
>>> index 71e33feee..0ec940b97 100644
>>> --- a/drivers/infiniband/hw/mana/main.c
>>> +++ b/drivers/infiniband/hw/mana/main.c
>>> @@ -237,6 +237,46 @@ void mana_ib_dealloc_ucontext(struct
>> ib_ucontext *ibcontext)
>>> ibdev_dbg(ibdev, "Failed to destroy doorbell page %d\n", ret);
>>> }
>>>
>>> +int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
>>> + struct mana_ib_queue *queue) {
>>> + struct ib_umem *umem;
>>> + int err;
>>> +
>>> + queue->umem = NULL;
>>> + queue->id = INVALID_QUEUE_ID;
>>> + queue->gdma_region = GDMA_INVALID_DMA_REGION;
>>> +
>>> + umem = ib_umem_get(&mdev->ib_dev, addr, size,
>> IB_ACCESS_LOCAL_WRITE);
>>> + if (IS_ERR(umem)) {
>>> + err = PTR_ERR(umem);
>>> + ibdev_dbg(&mdev->ib_dev, "Failed to get umem, %d\n", err);
>>> + return err;
>>> + }
>>> +
>>> + err = mana_ib_create_zero_offset_dma_region(mdev, umem, &queue-
>>> gdma_region);
>>> + if (err) {
>>> + ibdev_dbg(&mdev->ib_dev, "Failed to create dma region, %d\n",
>> err);
>>> + goto free_umem;
>>> + }
>>> + queue->umem = umem;
>>> +
>>> + ibdev_dbg(&mdev->ib_dev,
>>> + "create_dma_region ret %d gdma_region 0x%llx\n",
>>> + err, queue->gdma_region);
>>> +
>>> + return 0;
>>> +free_umem:
>>> + ib_umem_release(umem);
>>> + return err;
>>> +}
>>> +
>>> +void mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct
>>> +mana_ib_queue *queue) {
>>> + mana_ib_gd_destroy_dma_region(mdev, queue->gdma_region);
>> The function mana_ib_gd_destroy_dma_region will call
>> mana_gd_destroy_dma_region. In the function
>> mana_gd_destroy_dma_region, the function mana_gd_send_request will
>> return the error -EPROTO.
>> The procedure is as below. So the function mana_ib_destroy_queue should
>> also handle this error?
> Thanks for the comment!
> This error can be ignored and it was ignored before this commit.
> I checked the corresponding Windows driver code, and it is also intentionally ignored there.
> I can add a comment that the error is ignored intentionally if you want.
Sure. Thanks a lot.
Zhu Yanjun
>
>> mana_ib_gd_destroy_dma_region --- > mana_gd_destroy_dma_region
>>
>> 693 int mana_gd_destroy_dma_region(struct gdma_context *gc, u64
>> dma_region_handle)
>> 694 {
>>
>> ...
>>
>> 706 err = mana_gd_send_request(gc, sizeof(req), &req,
>> sizeof(resp), &resp);
>> 707 if (err || resp.hdr.status) {
>> 708 dev_err(gc->dev, "Failed to destroy DMA region:
>> %d, 0x%x\n",
>> 709 err, resp.hdr.status);
>> 710 return -EPROTO;
>> 711 }
>>
>> ...
>>
>> 714 }
>>
>> Zhu Yanjun
>>
>>> + ib_umem_release(queue->umem);
>>> +}
>>> +
>>> static int
>>> mana_ib_gd_first_dma_region(struct mana_ib_dev *dev,
>>> struct gdma_context *gc, diff --git
>>> a/drivers/infiniband/hw/mana/mana_ib.h
>>> b/drivers/infiniband/hw/mana/mana_ib.h
>>> index f83390eeb..859fd3bfc 100644
>>> --- a/drivers/infiniband/hw/mana/mana_ib.h
>>> +++ b/drivers/infiniband/hw/mana/mana_ib.h
>>> @@ -45,6 +45,12 @@ struct mana_ib_adapter_caps {
>>> u32 max_inline_data_size;
>>> };
>>>
>>> +struct mana_ib_queue {
>>> + struct ib_umem *umem;
>>> + u64 gdma_region;
>>> + u64 id;
>>> +};
>>> +
>>> struct mana_ib_dev {
>>> struct ib_device ib_dev;
>>> struct gdma_dev *gdma_dev;
>>> @@ -169,6 +175,10 @@ int mana_ib_create_dma_region(struct
>> mana_ib_dev *dev, struct ib_umem *umem,
>>> int mana_ib_gd_destroy_dma_region(struct mana_ib_dev *dev,
>>> mana_handle_t gdma_region);
>>>
>>> +int mana_ib_create_queue(struct mana_ib_dev *mdev, u64 addr, u32 size,
>>> + struct mana_ib_queue *queue); void
>>> +mana_ib_destroy_queue(struct mana_ib_dev *mdev, struct
>> mana_ib_queue
>>> +*queue);
>>> +
>>> struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
>>> struct ib_wq_init_attr *init_attr,
>>> struct ib_udata *udata);
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH rdma-next 2/4] RDMA:mana_ib: Use struct mana_ib_queue for CQs
2024-03-13 13:24 [PATCH rdma-next 0/4] Define and use mana queues for CQs and WQs Konstantin Taranov
2024-03-13 13:24 ` [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues Konstantin Taranov
@ 2024-03-13 13:24 ` Konstantin Taranov
2024-03-15 16:54 ` Long Li
2024-03-13 13:24 ` [PATCH rdma-next 3/4] RDMA/mana_ib: Use struct mana_ib_queue for WQs Konstantin Taranov
2024-03-13 13:24 ` [PATCH rdma-next 4/4] RDMA/mana_ib: Use struct mana_ib_queue for RAW QPs Konstantin Taranov
3 siblings, 1 reply; 13+ messages in thread
From: Konstantin Taranov @ 2024-03-13 13:24 UTC (permalink / raw)
To: kotaranov, sharmaajay, longli, jgg, leon; +Cc: linux-rdma, linux-kernel
From: Konstantin Taranov <kotaranov@microsoft.com>
Use struct mana_ib_queue and its helpers for CQs
Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
---
drivers/infiniband/hw/mana/cq.c | 52 ++++++----------------------
drivers/infiniband/hw/mana/mana_ib.h | 4 +--
drivers/infiniband/hw/mana/qp.c | 26 +++++++-------
3 files changed, 24 insertions(+), 58 deletions(-)
diff --git a/drivers/infiniband/hw/mana/cq.c b/drivers/infiniband/hw/mana/cq.c
index 4a71e678d..c9129218f 100644
--- a/drivers/infiniband/hw/mana/cq.c
+++ b/drivers/infiniband/hw/mana/cq.c
@@ -39,37 +39,13 @@ int mana_ib_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr,
}
cq->cqe = attr->cqe;
- cq->umem = ib_umem_get(ibdev, ucmd.buf_addr, cq->cqe * COMP_ENTRY_SIZE,
- IB_ACCESS_LOCAL_WRITE);
- if (IS_ERR(cq->umem)) {
- err = PTR_ERR(cq->umem);
- ibdev_dbg(ibdev, "Failed to get umem for create cq, err %d\n",
- err);
- return err;
- }
-
- err = mana_ib_create_zero_offset_dma_region(mdev, cq->umem, &cq->gdma_region);
+ err = mana_ib_create_queue(mdev, ucmd.buf_addr, cq->cqe * COMP_ENTRY_SIZE, &cq->queue);
if (err) {
- ibdev_dbg(ibdev,
- "Failed to create dma region for create cq, %d\n",
- err);
- goto err_release_umem;
+ ibdev_dbg(ibdev, "Failed to create queue for create cq, %d\n", err);
+ return err;
}
- ibdev_dbg(ibdev,
- "create_dma_region ret %d gdma_region 0x%llx\n",
- err, cq->gdma_region);
-
- /*
- * The CQ ID is not known at this time. The ID is generated at create_qp
- */
- cq->id = INVALID_QUEUE_ID;
-
return 0;
-
-err_release_umem:
- ib_umem_release(cq->umem);
- return err;
}
int mana_ib_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
@@ -78,24 +54,16 @@ int mana_ib_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata)
struct ib_device *ibdev = ibcq->device;
struct mana_ib_dev *mdev;
struct gdma_context *gc;
- int err;
mdev = container_of(ibdev, struct mana_ib_dev, ib_dev);
gc = mdev_to_gc(mdev);
- err = mana_ib_gd_destroy_dma_region(mdev, cq->gdma_region);
- if (err) {
- ibdev_dbg(ibdev,
- "Failed to destroy dma region, %d\n", err);
- return err;
- }
-
- if (cq->id != INVALID_QUEUE_ID) {
- kfree(gc->cq_table[cq->id]);
- gc->cq_table[cq->id] = NULL;
+ if (cq->queue.id != INVALID_QUEUE_ID) {
+ kfree(gc->cq_table[cq->queue.id]);
+ gc->cq_table[cq->queue.id] = NULL;
}
- ib_umem_release(cq->umem);
+ mana_ib_destroy_queue(mdev, &cq->queue);
return 0;
}
@@ -114,7 +82,7 @@ int mana_ib_install_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
struct gdma_queue *gdma_cq;
/* Create CQ table entry */
- WARN_ON(gc->cq_table[cq->id]);
+ WARN_ON(gc->cq_table[cq->queue.id]);
gdma_cq = kzalloc(sizeof(*gdma_cq), GFP_KERNEL);
if (!gdma_cq)
return -ENOMEM;
@@ -122,7 +90,7 @@ int mana_ib_install_cq_cb(struct mana_ib_dev *mdev, struct mana_ib_cq *cq)
gdma_cq->cq.context = cq;
gdma_cq->type = GDMA_CQ;
gdma_cq->cq.callback = mana_ib_cq_handler;
- gdma_cq->id = cq->id;
- gc->cq_table[cq->id] = gdma_cq;
+ gdma_cq->id = cq->queue.id;
+ gc->cq_table[cq->queue.id] = gdma_cq;
return 0;
}
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index 859fd3bfc..6acb5c281 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -88,10 +88,8 @@ struct mana_ib_mr {
struct mana_ib_cq {
struct ib_cq ibcq;
- struct ib_umem *umem;
+ struct mana_ib_queue queue;
int cqe;
- u64 gdma_region;
- u64 id;
u32 comp_vector;
};
diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c
index 6e7627745..d7485ee6a 100644
--- a/drivers/infiniband/hw/mana/qp.c
+++ b/drivers/infiniband/hw/mana/qp.c
@@ -197,7 +197,7 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
wq_spec.gdma_region = wq->gdma_region;
wq_spec.queue_size = wq->wq_buf_size;
- cq_spec.gdma_region = cq->gdma_region;
+ cq_spec.gdma_region = cq->queue.gdma_region;
cq_spec.queue_size = cq->cqe * COMP_ENTRY_SIZE;
cq_spec.modr_ctx_id = 0;
eq = &mpc->ac->eqs[cq->comp_vector % gc->max_num_queues];
@@ -213,16 +213,16 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
/* The GDMA regions are now owned by the WQ object */
wq->gdma_region = GDMA_INVALID_DMA_REGION;
- cq->gdma_region = GDMA_INVALID_DMA_REGION;
+ cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
wq->id = wq_spec.queue_index;
- cq->id = cq_spec.queue_index;
+ cq->queue.id = cq_spec.queue_index;
ibdev_dbg(&mdev->ib_dev,
"ret %d rx_object 0x%llx wq id %llu cq id %llu\n",
- ret, wq->rx_object, wq->id, cq->id);
+ ret, wq->rx_object, wq->id, cq->queue.id);
- resp.entries[i].cqid = cq->id;
+ resp.entries[i].cqid = cq->queue.id;
resp.entries[i].wqid = wq->id;
mana_ind_table[i] = wq->rx_object;
@@ -232,7 +232,7 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
if (ret)
goto fail;
- gdma_cq_allocated[i] = gc->cq_table[cq->id];
+ gdma_cq_allocated[i] = gc->cq_table[cq->queue.id];
}
resp.num_entries = i;
@@ -264,7 +264,7 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
wq = container_of(ibwq, struct mana_ib_wq, ibwq);
cq = container_of(ibcq, struct mana_ib_cq, ibcq);
- gc->cq_table[cq->id] = NULL;
+ gc->cq_table[cq->queue.id] = NULL;
kfree(gdma_cq_allocated[i]);
mana_destroy_wq_obj(mpc, GDMA_RQ, wq->rx_object);
@@ -374,7 +374,7 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
wq_spec.gdma_region = qp->sq_gdma_region;
wq_spec.queue_size = ucmd.sq_buf_size;
- cq_spec.gdma_region = send_cq->gdma_region;
+ cq_spec.gdma_region = send_cq->queue.gdma_region;
cq_spec.queue_size = send_cq->cqe * COMP_ENTRY_SIZE;
cq_spec.modr_ctx_id = 0;
eq_vec = send_cq->comp_vector % gc->max_num_queues;
@@ -392,10 +392,10 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
/* The GDMA regions are now owned by the WQ object */
qp->sq_gdma_region = GDMA_INVALID_DMA_REGION;
- send_cq->gdma_region = GDMA_INVALID_DMA_REGION;
+ send_cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
qp->sq_id = wq_spec.queue_index;
- send_cq->id = cq_spec.queue_index;
+ send_cq->queue.id = cq_spec.queue_index;
/* Create CQ table entry */
err = mana_ib_install_cq_cb(mdev, send_cq);
@@ -404,10 +404,10 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
ibdev_dbg(&mdev->ib_dev,
"ret %d qp->tx_object 0x%llx sq id %llu cq id %llu\n", err,
- qp->tx_object, qp->sq_id, send_cq->id);
+ qp->tx_object, qp->sq_id, send_cq->queue.id);
resp.sqid = qp->sq_id;
- resp.cqid = send_cq->id;
+ resp.cqid = send_cq->queue.id;
resp.tx_vp_offset = pd->tx_vp_offset;
err = ib_copy_to_udata(udata, &resp, sizeof(resp));
@@ -422,7 +422,7 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
err_release_gdma_cq:
kfree(gdma_cq);
- gc->cq_table[send_cq->id] = NULL;
+ gc->cq_table[send_cq->queue.id] = NULL;
err_destroy_wq_obj:
mana_destroy_wq_obj(mpc, GDMA_SQ, qp->tx_object);
--
2.43.0
^ permalink raw reply related [flat|nested] 13+ messages in thread* RE: [PATCH rdma-next 2/4] RDMA:mana_ib: Use struct mana_ib_queue for CQs
2024-03-13 13:24 ` [PATCH rdma-next 2/4] RDMA:mana_ib: Use struct mana_ib_queue for CQs Konstantin Taranov
@ 2024-03-15 16:54 ` Long Li
0 siblings, 0 replies; 13+ messages in thread
From: Long Li @ 2024-03-15 16:54 UTC (permalink / raw)
To: Konstantin Taranov, Konstantin Taranov, sharmaajay, jgg, leon
Cc: linux-rdma, linux-kernel
> -----Original Message-----
> From: Konstantin Taranov <kotaranov@linux.microsoft.com>
> Sent: Wednesday, March 13, 2024 6:25 AM
> To: Konstantin Taranov <kotaranov@microsoft.com>;
> sharmaajay@microsoft.com; Long Li <longli@microsoft.com>; jgg@ziepe.ca;
> leon@kernel.org
> Cc: linux-rdma@vger.kernel.org; linux-kernel@vger.kernel.org
> Subject: [PATCH rdma-next 2/4] RDMA:mana_ib: Use struct mana_ib_queue for
> CQs
>
> From: Konstantin Taranov <kotaranov@microsoft.com>
>
> Use struct mana_ib_queue and its helpers for CQs
>
> Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Reviewed-by: Long Li <longli@microsoft.com>
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH rdma-next 3/4] RDMA/mana_ib: Use struct mana_ib_queue for WQs
2024-03-13 13:24 [PATCH rdma-next 0/4] Define and use mana queues for CQs and WQs Konstantin Taranov
2024-03-13 13:24 ` [PATCH rdma-next 1/4] RDMA/mana_ib: Introduce helpers to create and destroy mana queues Konstantin Taranov
2024-03-13 13:24 ` [PATCH rdma-next 2/4] RDMA:mana_ib: Use struct mana_ib_queue for CQs Konstantin Taranov
@ 2024-03-13 13:24 ` Konstantin Taranov
2024-03-15 16:56 ` Long Li
2024-03-13 13:24 ` [PATCH rdma-next 4/4] RDMA/mana_ib: Use struct mana_ib_queue for RAW QPs Konstantin Taranov
3 siblings, 1 reply; 13+ messages in thread
From: Konstantin Taranov @ 2024-03-13 13:24 UTC (permalink / raw)
To: kotaranov, sharmaajay, longli, jgg, leon; +Cc: linux-rdma, linux-kernel
From: Konstantin Taranov <kotaranov@microsoft.com>
Use struct mana_ib_queue and its helpers for WQs
Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
---
drivers/infiniband/hw/mana/mana_ib.h | 4 +---
drivers/infiniband/hw/mana/qp.c | 10 ++++-----
drivers/infiniband/hw/mana/wq.c | 31 ++++------------------------
3 files changed, 10 insertions(+), 35 deletions(-)
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index 6acb5c281..a8953ee80 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -59,11 +59,9 @@ struct mana_ib_dev {
struct mana_ib_wq {
struct ib_wq ibwq;
- struct ib_umem *umem;
+ struct mana_ib_queue queue;
int wqe;
u32 wq_buf_size;
- u64 gdma_region;
- u64 id;
mana_handle_t rx_object;
};
diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c
index d7485ee6a..f606caa75 100644
--- a/drivers/infiniband/hw/mana/qp.c
+++ b/drivers/infiniband/hw/mana/qp.c
@@ -194,7 +194,7 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
ibcq = ibwq->cq;
cq = container_of(ibcq, struct mana_ib_cq, ibcq);
- wq_spec.gdma_region = wq->gdma_region;
+ wq_spec.gdma_region = wq->queue.gdma_region;
wq_spec.queue_size = wq->wq_buf_size;
cq_spec.gdma_region = cq->queue.gdma_region;
@@ -212,18 +212,18 @@ static int mana_ib_create_qp_rss(struct ib_qp *ibqp, struct ib_pd *pd,
}
/* The GDMA regions are now owned by the WQ object */
- wq->gdma_region = GDMA_INVALID_DMA_REGION;
+ wq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
- wq->id = wq_spec.queue_index;
+ wq->queue.id = wq_spec.queue_index;
cq->queue.id = cq_spec.queue_index;
ibdev_dbg(&mdev->ib_dev,
"ret %d rx_object 0x%llx wq id %llu cq id %llu\n",
- ret, wq->rx_object, wq->id, cq->queue.id);
+ ret, wq->rx_object, wq->queue.id, cq->queue.id);
resp.entries[i].cqid = cq->queue.id;
- resp.entries[i].wqid = wq->id;
+ resp.entries[i].wqid = wq->queue.id;
mana_ind_table[i] = wq->rx_object;
diff --git a/drivers/infiniband/hw/mana/wq.c b/drivers/infiniband/hw/mana/wq.c
index 7c9c69962..f959f4b92 100644
--- a/drivers/infiniband/hw/mana/wq.c
+++ b/drivers/infiniband/hw/mana/wq.c
@@ -13,7 +13,6 @@ struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
container_of(pd->device, struct mana_ib_dev, ib_dev);
struct mana_ib_create_wq ucmd = {};
struct mana_ib_wq *wq;
- struct ib_umem *umem;
int err;
if (udata->inlen < sizeof(ucmd))
@@ -32,39 +31,18 @@ struct ib_wq *mana_ib_create_wq(struct ib_pd *pd,
ibdev_dbg(&mdev->ib_dev, "ucmd wq_buf_addr 0x%llx\n", ucmd.wq_buf_addr);
- umem = ib_umem_get(pd->device, ucmd.wq_buf_addr, ucmd.wq_buf_size,
- IB_ACCESS_LOCAL_WRITE);
- if (IS_ERR(umem)) {
- err = PTR_ERR(umem);
+ err = mana_ib_create_queue(mdev, ucmd.wq_buf_addr, ucmd.wq_buf_size, &wq->queue);
+ if (err) {
ibdev_dbg(&mdev->ib_dev,
- "Failed to get umem for create wq, err %d\n", err);
+ "Failed to create queue for create wq, %d\n", err);
goto err_free_wq;
}
- wq->umem = umem;
wq->wqe = init_attr->max_wr;
wq->wq_buf_size = ucmd.wq_buf_size;
wq->rx_object = INVALID_MANA_HANDLE;
-
- err = mana_ib_create_zero_offset_dma_region(mdev, wq->umem, &wq->gdma_region);
- if (err) {
- ibdev_dbg(&mdev->ib_dev,
- "Failed to create dma region for create wq, %d\n",
- err);
- goto err_release_umem;
- }
-
- ibdev_dbg(&mdev->ib_dev,
- "create_dma_region ret %d gdma_region 0x%llx\n",
- err, wq->gdma_region);
-
- /* WQ ID is returned at wq_create time, doesn't know the value yet */
-
return &wq->ibwq;
-err_release_umem:
- ib_umem_release(umem);
-
err_free_wq:
kfree(wq);
@@ -86,8 +64,7 @@ int mana_ib_destroy_wq(struct ib_wq *ibwq, struct ib_udata *udata)
mdev = container_of(ib_dev, struct mana_ib_dev, ib_dev);
- mana_ib_gd_destroy_dma_region(mdev, wq->gdma_region);
- ib_umem_release(wq->umem);
+ mana_ib_destroy_queue(mdev, &wq->queue);
kfree(wq);
--
2.43.0
^ permalink raw reply related [flat|nested] 13+ messages in thread* RE: [PATCH rdma-next 3/4] RDMA/mana_ib: Use struct mana_ib_queue for WQs
2024-03-13 13:24 ` [PATCH rdma-next 3/4] RDMA/mana_ib: Use struct mana_ib_queue for WQs Konstantin Taranov
@ 2024-03-15 16:56 ` Long Li
0 siblings, 0 replies; 13+ messages in thread
From: Long Li @ 2024-03-15 16:56 UTC (permalink / raw)
To: Konstantin Taranov, Konstantin Taranov, sharmaajay, jgg, leon
Cc: linux-rdma, linux-kernel
> -----Original Message-----
> From: Konstantin Taranov <kotaranov@linux.microsoft.com>
> Sent: Wednesday, March 13, 2024 6:25 AM
> To: Konstantin Taranov <kotaranov@microsoft.com>;
> sharmaajay@microsoft.com; Long Li <longli@microsoft.com>; jgg@ziepe.ca;
> leon@kernel.org
> Cc: linux-rdma@vger.kernel.org; linux-kernel@vger.kernel.org
> Subject: [PATCH rdma-next 3/4] RDMA/mana_ib: Use struct mana_ib_queue for
> WQs
>
> From: Konstantin Taranov <kotaranov@microsoft.com>
>
> Use struct mana_ib_queue and its helpers for WQs
>
> Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
Reviewed-by: Long Li <longli@microsoft.com>
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH rdma-next 4/4] RDMA/mana_ib: Use struct mana_ib_queue for RAW QPs
2024-03-13 13:24 [PATCH rdma-next 0/4] Define and use mana queues for CQs and WQs Konstantin Taranov
` (2 preceding siblings ...)
2024-03-13 13:24 ` [PATCH rdma-next 3/4] RDMA/mana_ib: Use struct mana_ib_queue for WQs Konstantin Taranov
@ 2024-03-13 13:24 ` Konstantin Taranov
2024-03-15 17:03 ` Long Li
3 siblings, 1 reply; 13+ messages in thread
From: Konstantin Taranov @ 2024-03-13 13:24 UTC (permalink / raw)
To: kotaranov, sharmaajay, longli, jgg, leon; +Cc: linux-rdma, linux-kernel
From: Konstantin Taranov <kotaranov@microsoft.com>
Use struct mana_ib_queue and its helpers for RAW QPs
Signed-off-by: Konstantin Taranov <kotaranov@microsoft.com>
---
drivers/infiniband/hw/mana/mana_ib.h | 13 +++----
drivers/infiniband/hw/mana/qp.c | 54 ++++++++--------------------
2 files changed, 22 insertions(+), 45 deletions(-)
diff --git a/drivers/infiniband/hw/mana/mana_ib.h b/drivers/infiniband/hw/mana/mana_ib.h
index a8953ee80..ddb6e73e3 100644
--- a/drivers/infiniband/hw/mana/mana_ib.h
+++ b/drivers/infiniband/hw/mana/mana_ib.h
@@ -91,15 +91,16 @@ struct mana_ib_cq {
u32 comp_vector;
};
+struct mana_ib_raw_qp {
+ /* Work queue info */
+ struct mana_ib_queue queue;
+ mana_handle_t tx_object;
+};
+
struct mana_ib_qp {
struct ib_qp ibqp;
- /* Work queue info */
- struct ib_umem *sq_umem;
- int sqe;
- u64 sq_gdma_region;
- u64 sq_id;
- mana_handle_t tx_object;
+ struct mana_ib_raw_qp sq;
/* The port on the IB device, starting with 1 */
u32 port;
diff --git a/drivers/infiniband/hw/mana/qp.c b/drivers/infiniband/hw/mana/qp.c
index f606caa75..5818b665e 100644
--- a/drivers/infiniband/hw/mana/qp.c
+++ b/drivers/infiniband/hw/mana/qp.c
@@ -297,7 +297,6 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
struct mana_obj_spec cq_spec = {};
struct mana_port_context *mpc;
struct net_device *ndev;
- struct ib_umem *umem;
struct mana_eq *eq;
int eq_vec;
u32 port;
@@ -346,32 +345,15 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
ibdev_dbg(&mdev->ib_dev, "ucmd sq_buf_addr 0x%llx port %u\n",
ucmd.sq_buf_addr, ucmd.port);
- umem = ib_umem_get(ibpd->device, ucmd.sq_buf_addr, ucmd.sq_buf_size,
- IB_ACCESS_LOCAL_WRITE);
- if (IS_ERR(umem)) {
- err = PTR_ERR(umem);
- ibdev_dbg(&mdev->ib_dev,
- "Failed to get umem for create qp-raw, err %d\n",
- err);
- goto err_free_vport;
- }
- qp->sq_umem = umem;
-
- err = mana_ib_create_zero_offset_dma_region(mdev, qp->sq_umem,
- &qp->sq_gdma_region);
+ err = mana_ib_create_queue(mdev, ucmd.sq_buf_addr, ucmd.sq_buf_size, &qp->sq.queue);
if (err) {
ibdev_dbg(&mdev->ib_dev,
- "Failed to create dma region for create qp-raw, %d\n",
- err);
- goto err_release_umem;
+ "Failed to create queue for create qp-raw, err %d\n", err);
+ goto err_free_vport;
}
- ibdev_dbg(&mdev->ib_dev,
- "create_dma_region ret %d gdma_region 0x%llx\n",
- err, qp->sq_gdma_region);
-
/* Create a WQ on the same port handle used by the Ethernet */
- wq_spec.gdma_region = qp->sq_gdma_region;
+ wq_spec.gdma_region = qp->sq.queue.gdma_region;
wq_spec.queue_size = ucmd.sq_buf_size;
cq_spec.gdma_region = send_cq->queue.gdma_region;
@@ -382,19 +364,19 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
cq_spec.attached_eq = eq->eq->id;
err = mana_create_wq_obj(mpc, mpc->port_handle, GDMA_SQ, &wq_spec,
- &cq_spec, &qp->tx_object);
+ &cq_spec, &qp->sq.tx_object);
if (err) {
ibdev_dbg(&mdev->ib_dev,
"Failed to create wq for create raw-qp, err %d\n",
err);
- goto err_destroy_dma_region;
+ goto err_destroy_queue;
}
/* The GDMA regions are now owned by the WQ object */
- qp->sq_gdma_region = GDMA_INVALID_DMA_REGION;
+ qp->sq.queue.gdma_region = GDMA_INVALID_DMA_REGION;
send_cq->queue.gdma_region = GDMA_INVALID_DMA_REGION;
- qp->sq_id = wq_spec.queue_index;
+ qp->sq.queue.id = wq_spec.queue_index;
send_cq->queue.id = cq_spec.queue_index;
/* Create CQ table entry */
@@ -404,9 +386,9 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
ibdev_dbg(&mdev->ib_dev,
"ret %d qp->tx_object 0x%llx sq id %llu cq id %llu\n", err,
- qp->tx_object, qp->sq_id, send_cq->queue.id);
+ qp->sq.tx_object, qp->sq.queue.id, send_cq->queue.id);
- resp.sqid = qp->sq_id;
+ resp.sqid = qp->sq.queue.id;
resp.cqid = send_cq->queue.id;
resp.tx_vp_offset = pd->tx_vp_offset;
@@ -425,13 +407,10 @@ static int mana_ib_create_qp_raw(struct ib_qp *ibqp, struct ib_pd *ibpd,
gc->cq_table[send_cq->queue.id] = NULL;
err_destroy_wq_obj:
- mana_destroy_wq_obj(mpc, GDMA_SQ, qp->tx_object);
+ mana_destroy_wq_obj(mpc, GDMA_SQ, qp->sq.tx_object);
-err_destroy_dma_region:
- mana_ib_gd_destroy_dma_region(mdev, qp->sq_gdma_region);
-
-err_release_umem:
- ib_umem_release(umem);
+err_destroy_queue:
+ mana_ib_destroy_queue(mdev, &qp->sq.queue);
err_free_vport:
mana_ib_uncfg_vport(mdev, pd, port);
@@ -505,12 +484,9 @@ static int mana_ib_destroy_qp_raw(struct mana_ib_qp *qp, struct ib_udata *udata)
mpc = netdev_priv(ndev);
pd = container_of(ibpd, struct mana_ib_pd, ibpd);
- mana_destroy_wq_obj(mpc, GDMA_SQ, qp->tx_object);
+ mana_destroy_wq_obj(mpc, GDMA_SQ, qp->sq.tx_object);
- if (qp->sq_umem) {
- mana_ib_gd_destroy_dma_region(mdev, qp->sq_gdma_region);
- ib_umem_release(qp->sq_umem);
- }
+ mana_ib_destroy_queue(mdev, &qp->sq.queue);
mana_ib_uncfg_vport(mdev, pd, qp->port);
--
2.43.0
^ permalink raw reply related [flat|nested] 13+ messages in thread* RE: [PATCH rdma-next 4/4] RDMA/mana_ib: Use struct mana_ib_queue for RAW QPs
2024-03-13 13:24 ` [PATCH rdma-next 4/4] RDMA/mana_ib: Use struct mana_ib_queue for RAW QPs Konstantin Taranov
@ 2024-03-15 17:03 ` Long Li
2024-03-18 9:21 ` Konstantin Taranov
0 siblings, 1 reply; 13+ messages in thread
From: Long Li @ 2024-03-15 17:03 UTC (permalink / raw)
To: Konstantin Taranov, Konstantin Taranov, sharmaajay, jgg, leon
Cc: linux-rdma, linux-kernel
> +struct mana_ib_raw_qp {
> + /* Work queue info */
> + struct mana_ib_queue queue;
> + mana_handle_t tx_object;
> +};
> +
> struct mana_ib_qp {
> struct ib_qp ibqp;
>
> - /* Work queue info */
> - struct ib_umem *sq_umem;
> - int sqe;
> - u64 sq_gdma_region;
> - u64 sq_id;
> - mana_handle_t tx_object;
> + struct mana_ib_raw_qp sq;
What's the naming scheme for RC? Maybe use raw_sq here?
^ permalink raw reply [flat|nested] 13+ messages in thread* RE: [PATCH rdma-next 4/4] RDMA/mana_ib: Use struct mana_ib_queue for RAW QPs
2024-03-15 17:03 ` Long Li
@ 2024-03-18 9:21 ` Konstantin Taranov
0 siblings, 0 replies; 13+ messages in thread
From: Konstantin Taranov @ 2024-03-18 9:21 UTC (permalink / raw)
To: Long Li, Konstantin Taranov, sharmaajay, jgg, leon
Cc: linux-rdma, linux-kernel
> From: Long Li <longli@microsoft.com>
> Sent: Friday, 15 March 2024 18:04
> To: Konstantin Taranov <kotaranov@linux.microsoft.com>; Konstantin
> Taranov <kotaranov@microsoft.com>; sharmaajay@microsoft.com;
> jgg@ziepe.ca; leon@kernel.org
> Cc: linux-rdma@vger.kernel.org; linux-kernel@vger.kernel.org
> Subject: RE: [PATCH rdma-next 4/4] RDMA/mana_ib: Use struct
> mana_ib_queue for RAW QPs
>
> > +struct mana_ib_raw_qp {
> > + /* Work queue info */
> > + struct mana_ib_queue queue;
> > + mana_handle_t tx_object;
> > +};
> > +
> > struct mana_ib_qp {
> > struct ib_qp ibqp;
> >
> > - /* Work queue info */
> > - struct ib_umem *sq_umem;
> > - int sqe;
> > - u64 sq_gdma_region;
> > - u64 sq_id;
> > - mana_handle_t tx_object;
> > + struct mana_ib_raw_qp sq;
>
> What's the naming scheme for RC? Maybe use raw_sq here?
The plan is to use struct mana_ib_rc_qp for RC QP.
But I think mana_ib_raw_sq is a good proposal for RAW packets. I will make it in v2.
Thanks!
^ permalink raw reply [flat|nested] 13+ messages in thread