[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <d37309bd-c7e0-8e15-bae9-9341f4f9192d@nvidia.com>
Date: Sun, 15 May 2022 10:36:07 +0300
From: Mark Bloch <mbloch@...dia.com>
To: Leon Romanovsky <leon@...nel.org>,
Rohit Sajan Kumar <rohit.sajan.kumar@...cle.com>
Cc: "matanb@...lanox.com" <matanb@...lanox.com>,
"dledford@...hat.com" <dledford@...hat.com>,
"sean.hefty@...el.com" <sean.hefty@...el.com>,
"hal.rosenstock@...il.com" <hal.rosenstock@...il.com>,
Manjunath Patil <manjunath.b.patil@...cle.com>,
Rama Nichanamatlu <rama.nichanamatlu@...cle.com>,
"linux-rdma@...r.kernel.org" <linux-rdma@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] net/mlx5: Add sysfs entry for vhca to
/sys/class/infiniband/mlx5
On 5/14/2022 08:09, Leon Romanovsky wrote:
> On Fri, May 13, 2022 at 05:46:16PM +0000, Rohit Sajan Kumar wrote:
>> Hey,
>>
>> Sending this as a gentle reminder to review the patch sent earlier this week which can be found in this email chain.
>
> Patches that sent in HTML format, to wrong addresses and not visible
> in patchworks/ML, without target net-next/rdma-next/e.t.c., with wrong
> title are generally ignored.
>
> Why vhca_id that returned from MLX5_IB_METHOD_QUERY_PORT is not enough?
The driver returns that only when in switchdev mode.
I don't see why that can't be changed but that's the state today.
>
> Anyway, sysfs file in IB driver for the property of mlx5_core is no-go.
For vhca_id a nicer route is probably to use DEVX and and run a mlx5dv_devx_general_cmd()
where you query HCA_CAP and take vhca_id from there. The added bonus is that
there is no change required in rdma-core/kernel.
Mark
>
> Thanks
>
>>
>> Thank you.
>>
>> Best,
>> Rohit.
>> ________________________________
>> From: Rohit Nair <rohit.sajan.kumar@...cle.com>
>> Sent: Monday, May 9, 2022 4:00 PM
>> To: matanb@...lanox.com <matanb@...lanox.com>; leonro@...lanox.com <leonro@...lanox.com>; dledford@...hat.com <dledford@...hat.com>; sean.hefty@...el.com <sean.hefty@...el.com>; hal.rosenstock@...il.com <hal.rosenstock@...il.com>; Rohit Sajan Kumar <rohit.sajan.kumar@...cle.com>; Manjunath Patil <manjunath.b.patil@...cle.com>; Rama Nichanamatlu <rama.nichanamatlu@...cle.com>
>> Cc: linux-rdma@...r.kernel.org <linux-rdma@...r.kernel.org>; linux-kernel@...r.kernel.org <linux-kernel@...r.kernel.org>
>> Subject: [PATCH] net/mlx5: Add sysfs entry for vhca to /sys/class/infiniband/mlx5
>>
>> While collecting diagnostic information (Ex:wqdump) in virtual
>> environment, we need vhca id to collect data belonging a particular VF.
>> This patch adds a sysfs entry to show the vhca id inside guest.
>>
>> Signed-off-by: Rohit Nair <rohit.sajan.kumar@...cle.com>
>> ---
>> drivers/infiniband/hw/mlx5/main.c | 12 ++++++++++++
>> 1 file changed, 12 insertions(+)
>>
>> diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c
>> index 32a0ea8..dd935bc 100644
>> --- a/drivers/infiniband/hw/mlx5/main.c
>> +++ b/drivers/infiniband/hw/mlx5/main.c
>> @@ -2499,12 +2499,24 @@ static ssize_t board_id_show(struct device *device,
>> }
>> static DEVICE_ATTR_RO(board_id);
>>
>> +static ssize_t vhca_id_show(struct device *device,
>> + struct device_attribute *attr, char *buf)
>> +{
>> + struct mlx5_ib_dev *dev =
>> + container_of(device, struct mlx5_ib_dev, ib_dev.dev);
>> + return sysfs_emit(buf, "%d [0x%x]\n",
>> + MLX5_CAP_GEN(dev->mdev, vhca_id),
>> + MLX5_CAP_GEN(dev->mdev, vhca_id));
>> +}
>> +static DEVICE_ATTR_RO(vhca_id);
>> +
>> static struct attribute *mlx5_class_attributes[] = {
>> &dev_attr_hw_rev.attr,
>> &dev_attr_hca_type.attr,
>> &dev_attr_board_id.attr,
>> &dev_attr_fw_pages.attr,
>> &dev_attr_reg_pages.attr,
>> + &dev_attr_vhca_id.attr,
>> NULL,
>> };
>>
>> --
>> 1.8.3.1
>>
Powered by blists - more mailing lists