From: "Wei Hu (Xavier)" <xavier.huwei-WVlzvzqoTvw@public.gmane.org>
To: Leon Romanovsky <leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>,
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
lijun_nudt-9Onoh4P/yGk@public.gmane.org,
oulijun-hv44wF8Li93QT0dZR+AlfA@public.gmane.org,
charles.chenxin-hv44wF8Li93QT0dZR+AlfA@public.gmane.org,
liuyixian-hv44wF8Li93QT0dZR+AlfA@public.gmane.org,
xushaobo2-hv44wF8Li93QT0dZR+AlfA@public.gmane.org,
zhangxiping3-hv44wF8Li93QT0dZR+AlfA@public.gmane.org,
xavier.huwei-hv44wF8Li93QT0dZR+AlfA@public.gmane.org,
linuxarm-hv44wF8Li93QT0dZR+AlfA@public.gmane.org
Subject: Re: [PATCH for-next 05/20] RDMA/hns: Add command queue support for hip08 RoCE driver
Date: Wed, 27 Sep 2017 05:12:20 +0800 [thread overview]
Message-ID: <e3fe5e55-0d16-deee-e8a6-01399593963d@tom.com> (raw)
In-Reply-To: <5514bf6d-3a98-a6fe-ea90-476f5ae1f623-WVlzvzqoTvw@public.gmane.org>
On 2017/9/27 0:13, Wei Hu (Xavier) wrote:
>
>
> On 2017/9/26 23:51, Leon Romanovsky wrote:
>> On Tue, Sep 26, 2017 at 11:24:41PM +0800, Wei Hu (Xavier) wrote:
>>>
>>> On 2017/9/26 21:13, Wei Hu (Xavier) wrote:
>>>>
>>>> On 2017/9/26 1:36, Doug Ledford wrote:
>>>>> On Mon, 2017-09-25 at 20:18 +0300, Leon Romanovsky wrote:
>>>>>> On Mon, Sep 25, 2017 at 01:06:53PM -0400, Doug Ledford wrote:
>>>>>>> On Wed, 2017-08-30 at 17:23 +0800, Wei Hu (Xavier) wrote:
>>>>>>>
>>>>>>>> + /*
>>>>>>>> + * If the command is sync, wait for the firmware to
>>>>>>>> write
>>>>>>>> back,
>>>>>>>> + * if multi descriptors to be sent, use the first one to
>>>>>>>> check
>>>>>>>> + */
>>>>>>>> + if ((desc->flag) & HNS_ROCE_CMD_FLAG_NO_INTR) {
>>>>>>>> + do {
>>>>>>>> + if (hns_roce_cmq_csq_done(hr_dev))
>>>>>>>> + break;
>>>>>>>> + usleep_range(1000, 2000);
>>>>>>>> + timeout++;
>>>>>>>> + } while (timeout < priv->cmq.tx_timeout);
>>>>>>>> + }
>>>>>>> then we spin here for a maximum amount of time between 200 and
>>>>>>> 400ms,
>>>>>>> so 1/4 to 1/2 a second. All the time we are holding the bh lock on
>>>>>>> this CPU. That seems excessive to me. If we are going to spin
>>>>>>> that
>>>>>>> long, can you find a way to allocate/reserve your resources, send
>>>>>>> the
>>>>>>> command, then drop the bh lock while you spin, and retake it before
>>>>>>> you
>>>>>>> complete once the spinning is done?
>>>>>> They don't allocate anything in this loop, but checking the pointers
>>>>>> are
>>>>>> the same, see hns_roce_cmq_csq_done.
>>>>> I'm not sure I understand your intended implication of your
>>>>> comment. I
>>>>> wasn't concerned about them allocating anything, only that if the
>>>>> hardware is hung, then this loop will hang out for 1/4 to 1/2 a
>>>>> second
>>>>> and hold up all bottom half processing on this CPU in the meantime.
>>>>> That's the sort of things that provides poor overall system behavior.
>>>>>
>>>>> Now, since they are really only checking to see if the hardware has
>>>>> gotten around to their particular command, and their command is
>>>>> part of
>>>>> a ring structure, it's possible to record the original head command,
>>>>> and our new head command, and then release the spin_lock_bh around
>>>>> the
>>>>> entire do{ }while construct, and in hns_roce_cmd_csq_done() you could
>>>>> check that head is not in the range old_head:new_head. That would
>>>>> protect you in case something in the bottom half processing queued up
>>>>> some more commands and from one sleep to the next the head jumped
>>>>> from
>>>>> something other than the new_head to something past new_head, so that
>>>>> head == priv->cmq.csq.next_to_use ends up being perpetually false.
>>>>> But, that's just from a quick read of the code, I could easily be
>>>>> missing something here...
>>>> Hi, Doug
>>>> Driver issues the cmds in cmq, and firmware gets and processes
>>>> them.
>>>> The firmware process only one cmd at the same time, and it
>>>> will take
>>>> about serveral to 200 us in one cmd currently, so the driver need
>>>> not to use stream mode to issue cmd.
>>>>
>>>> Regards
>>>> Wei Hu
>>> Hi, Doug
>>> We can replace usleep_range(1000, 2000); with the following
>>> statement:
>>> udelay(1);
>>> to avoid spinning too long time and using usleep_range function in
>>> spin_lock_bh locked region.
>>>
>>> Can you give som=e suggestions?
>> He already suggested number of options, while the simple one was to
>> reduce
>> HNS_ROCE_CMQ_TX_TIMEOUT from 200 to be something sane (for example 5).
>>
>> Thanks
> Thanks, Leon
> Actually We want to use another option:
> replace usleep_range(1000, 2000); with the following statement:
> udelay(1);
> And if so, we can avoid using usleep_range function in
> spin_lock_bh spin region,
> because it probally cause calltrace by rcu.
>
Can we send a following up patch to fix this problem? or send patch v2?
Thanks
> Best regards
> Wei Hu
>>> Regards
>>> Wei Hu
>>>>>>>> +#define HNS_ROCE_CMQ_TX_TIMEOUT 200
>>>>>>> or you could reduce the size of this define...
>>>>>>>
>>>>>>> --
>>>>>>> Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
>>>>>>> GPG KeyID: B826A3330E572FDD
>>>>>>> Key fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333
>>>>>>> 0E57
>>>>>>> 2FDD
>>>>>>>
>>>>>>> --
>>>>>>> To unsubscribe from this list: send the line "unsubscribe linux-
>>>>>>> rdma" in
>>>>>>> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
>>>>>>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>>>>
>>>> --
>>>> To unsubscribe from this list: send the line "unsubscribe
>>>> linux-rdma" in
>>>> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
>>>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2017-09-26 21:12 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-08-30 9:22 [PATCH for-next 00/20] RDMA/hns: Add hip08 RoCE driver support Wei Hu (Xavier)
[not found] ` <1504084998-64397-1-git-send-email-xavier.huwei-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
2017-08-30 9:22 ` [PATCH for-next 01/20] RDMA/hns: Split hw v1 driver from hns roce driver Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 02/20] RDMA/hns: Move priv in order to add multiple hns_roce support Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 03/20] RDMA/hns: Initialize the PCI device for hip08 RoCE Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 04/20] RDMA/hns: Modify assignment device variable to support both PCI device and platform device Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 05/20] RDMA/hns: Add command queue support for hip08 RoCE driver Wei Hu (Xavier)
[not found] ` <1504084998-64397-6-git-send-email-xavier.huwei-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
2017-09-25 17:06 ` Doug Ledford
[not found] ` <1506359213.120853.75.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-09-25 17:18 ` Leon Romanovsky
[not found] ` <20170925171821.GQ25094-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2017-09-25 17:36 ` Doug Ledford
[not found] ` <1506361015.120853.81.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-09-26 5:15 ` Leon Romanovsky
2017-09-26 13:13 ` Wei Hu (Xavier)
[not found] ` <59CA5261.80209-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
2017-09-26 15:24 ` Wei Hu (Xavier)
[not found] ` <e99f8917-1906-697b-3dcd-5f024b444750-WVlzvzqoTvw@public.gmane.org>
2017-09-26 15:51 ` Leon Romanovsky
[not found] ` <20170926155149.GE6816-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2017-09-26 16:13 ` Wei Hu (Xavier)
[not found] ` <5514bf6d-3a98-a6fe-ea90-476f5ae1f623-WVlzvzqoTvw@public.gmane.org>
2017-09-26 21:12 ` Wei Hu (Xavier) [this message]
2017-09-26 16:18 ` Doug Ledford
[not found] ` <81dd332d-e060-d7e3-bec9-1791511c5470-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-09-27 2:46 ` Wei Hu (Xavier)
[not found] ` <9172f8c5-3dd6-a573-8e28-1b3ae4b1726b-WVlzvzqoTvw@public.gmane.org>
2017-09-27 12:21 ` Doug Ledford
[not found] ` <1b8bda3b-c514-7e46-08bf-3ea50ea68096-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-09-27 12:41 ` Doug Ledford
2017-09-28 4:34 ` Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 06/20] RDMA/hns: Add profile support for hip08 driver Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 07/20] RDMA/hns: Add mailbox's implementation for hip08 RoCE driver Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 08/20] RDMA/hns: Add the interfaces to support multi hop addressing for the contexts in hip08 Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 09/20] RDMA/hns: Configure BT BA and BT attribute " Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 10/20] RDMA/hns: Update the interfaces for MTT/CQE multi hop addressing " Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 11/20] RDMA/hns: Split CQE from MTT " Wei Hu (Xavier)
[not found] ` <1504084998-64397-12-git-send-email-xavier.huwei-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
2017-09-13 17:52 ` Leon Romanovsky
[not found] ` <20170913175259.GW3405-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2017-09-15 1:09 ` Wei Hu (Xavier)
[not found] ` <59BB2848.6080802-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
2017-09-20 2:48 ` Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 12/20] RDMA/hns: Support multi hop addressing for PBL " Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 13/20] RDMA/hns: Configure mac&gid and user access region for hip08 RoCE driver Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 14/20] RDMA/hns: Add CQ operations support " Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 15/20] RDMA/hns: Add QP operations support for hip08 SoC Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 16/20] RDMA/hns: Add support for processing send wr and receive wr Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 17/20] RDMA/hns: Configure the MTPT in hip08 Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 18/20] RDMA/hns: Add releasing resource operation in error branch Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 19/20] RDMA/hns: Replace condition statement using hardware version information Wei Hu (Xavier)
2017-08-30 9:23 ` [PATCH for-next 20/20] RDMA/hns: Fix inconsistent warning Wei Hu (Xavier)
2017-09-13 17:55 ` [PATCH for-next 00/20] RDMA/hns: Add hip08 RoCE driver support Leon Romanovsky
[not found] ` <20170913175554.GX3405-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2017-09-14 7:45 ` Wei Hu (Xavier)
[not found] ` <59BA33B1.8030300-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
2017-09-14 12:43 ` Leon Romanovsky
[not found] ` <20170914124341.GY3405-U/DQcQFIOTAAJjI8aNfphQ@public.gmane.org>
2017-09-15 1:12 ` Wei Hu (Xavier)
[not found] ` <59BB28F1.9040007-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
2017-09-20 2:50 ` Wei Hu (Xavier)
2017-09-25 6:18 ` Wei Hu (Xavier)
[not found] ` <59C89FD0.9050606-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
2017-09-25 15:57 ` Doug Ledford
[not found] ` <1506355051.120853.70.camel-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2017-09-25 17:37 ` Doug Ledford
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e3fe5e55-0d16-deee-e8a6-01399593963d@tom.com \
--to=xavier.huwei-wvlzvzqotvw@public.gmane.org \
--cc=charles.chenxin-hv44wF8Li93QT0dZR+AlfA@public.gmane.org \
--cc=dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
--cc=leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org \
--cc=lijun_nudt-9Onoh4P/yGk@public.gmane.org \
--cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=linuxarm-hv44wF8Li93QT0dZR+AlfA@public.gmane.org \
--cc=liuyixian-hv44wF8Li93QT0dZR+AlfA@public.gmane.org \
--cc=oulijun-hv44wF8Li93QT0dZR+AlfA@public.gmane.org \
--cc=xavier.huwei-hv44wF8Li93QT0dZR+AlfA@public.gmane.org \
--cc=xushaobo2-hv44wF8Li93QT0dZR+AlfA@public.gmane.org \
--cc=zhangxiping3-hv44wF8Li93QT0dZR+AlfA@public.gmane.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).