lkml.org 
[lkml]   [2018]   [May]   [23]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH rdma-next 2/5] RDMA/hns: Modify uar allocation algorithm to avoid bitmap exhaust
From
Date


On 2018/5/23 15:00, Leon Romanovsky wrote:
> On Wed, May 23, 2018 at 02:49:35PM +0800, Wei Hu (Xavier) wrote:
>>
>> On 2018/5/23 14:05, Leon Romanovsky wrote:
>>> On Thu, May 17, 2018 at 04:02:50PM +0800, Wei Hu (Xavier) wrote:
>>>> This patch modified uar allocation algorithm in hns_roce_uar_alloc
>>>> function to avoid bitmap exhaust.
>>>>
>>>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>>>> ---
>>>> drivers/infiniband/hw/hns/hns_roce_device.h | 1 +
>>>> drivers/infiniband/hw/hns/hns_roce_pd.c | 10 ++++++----
>>>> 2 files changed, 7 insertions(+), 4 deletions(-)
>>>>
>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
>>>> index 53c2f1b..412297d4 100644
>>>> --- a/drivers/infiniband/hw/hns/hns_roce_device.h
>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_device.h
>>>> @@ -214,6 +214,7 @@ enum {
>>>> struct hns_roce_uar {
>>>> u64 pfn;
>>>> unsigned long index;
>>>> + unsigned long logic_idx;
>>>> };
>>>>
>>>> struct hns_roce_ucontext {
>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_pd.c b/drivers/infiniband/hw/hns/hns_roce_pd.c
>>>> index 4b41e04..b9f2c87 100644
>>>> --- a/drivers/infiniband/hw/hns/hns_roce_pd.c
>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_pd.c
>>>> @@ -107,13 +107,15 @@ int hns_roce_uar_alloc(struct hns_roce_dev *hr_dev, struct hns_roce_uar *uar)
>>>> int ret = 0;
>>>>
>>>> /* Using bitmap to manager UAR index */
>>>> - ret = hns_roce_bitmap_alloc(&hr_dev->uar_table.bitmap, &uar->index);
>>>> + ret = hns_roce_bitmap_alloc(&hr_dev->uar_table.bitmap, &uar->logic_idx);
>>>> if (ret == -1)
>>>> return -ENOMEM;
>>>>
>>>> - if (uar->index > 0)
>>>> - uar->index = (uar->index - 1) %
>>>> + if (uar->logic_idx > 0 && hr_dev->caps.phy_num_uars > 1)
>>>> + uar->index = (uar->logic_idx - 1) %
>>>> (hr_dev->caps.phy_num_uars - 1) + 1;
>>>> + else
>>>> + uar->index = 0;
>>>>
>>> Sorry, but maybe I didn't understand this change fully, but logic_idx is
>>> not initialized at all and one of two (needs to check your uar
>>> allocation): the logic_idx is always zero -> index will be zero too,
>>> or logic_idx is random variable -> index will be random too.
>>>
>>> What did you want to do?
>>>
>> Hi, Leon
>>
>> The prototype of hns_roce_bitmap_alloc as belows:
>> int hns_roce_bitmap_alloc(struct hns_roce_bitmap *bitmap,
>> unsigned long *obj);
>> In this statement, we evaluate uar->logic_idx.
>> ret = hns_roce_bitmap_alloc(&hr_dev->uar_table.bitmap,
>> &uar->logic_idx);
>>
>> In hip06, hr_dev->caps.phy_num_uars equals 8,
>> if (uar->logic_idx > 0)
>> uar-> index = 0;
>> else
>> uar-> index =(uar->logic_idx - 1) %
>> (hr_dev->caps.phy_num_uars - 1) + 1;
>> In hip08, hr_dev->caps.phy_num_uars equals 1, uar-> index = 0;
>>
>> Regards
> Where did you change/set logic_idx?
In hns_roce_uar_alloc,
ret = hns_roce_bitmap_alloc(&hr_dev->uar_table.bitmap, &uar->logic_idx);
In hns_roce_uar_free,
hns_roce_bitmap_free(&hr_dev->uar_table.bitmap, uar->logic_idx,
BITMAP_NO_RR);

Thanks
> Thanks
>
>
>> Wei Hu
>>>> if (!dev_is_pci(hr_dev->dev)) {
>>>> res = platform_get_resource(hr_dev->pdev, IORESOURCE_MEM, 0);
>>>> @@ -132,7 +134,7 @@ int hns_roce_uar_alloc(struct hns_roce_dev *hr_dev, struct hns_roce_uar *uar)
>>>>
>>>> void hns_roce_uar_free(struct hns_roce_dev *hr_dev, struct hns_roce_uar *uar)
>>>> {
>>>> - hns_roce_bitmap_free(&hr_dev->uar_table.bitmap, uar->index,
>>>> + hns_roce_bitmap_free(&hr_dev->uar_table.bitmap, uar->logic_idx,
>>>> BITMAP_NO_RR);
>>>> }
>>>>
>>>> --
>>>> 1.9.1
>>>>
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at http://vger.kernel.org/majordomo-info.html


\
 
 \ /
  Last update: 2018-05-23 09:13    [W:0.127 / U:0.464 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site