From: Logan Gunthorpe <logang@deltatee.com>
To: Marc Smith <msmith626@gmail.com>
Cc: ntb@lists.linux.dev, Kelvin Cao <kelvin.cao@microchip.com>,
kelvincao@outlook.com
Subject: Re: ntb_netdev Communication Failure Issue
Date: Wed, 9 Mar 2022 11:31:32 -0700 [thread overview]
Message-ID: <bec56b4f-1eb5-506b-2cca-7bd8bc8cbe37@deltatee.com> (raw)
In-Reply-To: <CAH6h+hfNWgXCguK7xtKCQOm9F_Xc-XCSwRyfrS5nQZGoh7PKFQ@mail.gmail.com>
On 2022-03-09 11:26, Marc Smith wrote:
> On Wed, Mar 9, 2022 at 11:52 AM Logan Gunthorpe <logang@deltatee.com> wrote:
>>
>>
>>
>> On 2022-03-09 07:35, Marc Smith wrote:
>>> No difference between the two above, and actually, comparing the
>>> entire 4 MiB of GAS dump taken while working/functional versus broken
>>> state shows no differences:
>>> # cmp switchtec0_gas_dump_after_ifup_working.1 switchtec0_gas_dump_broken.1
>>
>> Well I'm stumped. It looks like the receive side is getting the
>> doorbell, but when it reads the memory window the done flag is not set,
>> so it doesn't process any packets. The transmit side then fills up the
>> memory window and breaks.
>>
>> So it sounds like the read side is not reading the correct data from the
>> memory window. Like it's cache or something (which it shouldn't be). But
>> I have no idea what could cause that.
>
> Anything possibly to blame in the PCIe switch firmware that could
> cause this? Working with the vendor to ensure we're at the least, but
> still TBD at the moment.
I can't say anything for certain. It certainly worth getting the latest
firmware. But as far as I understand the memory windows (if that is
indeed the problem) are mostly hardware related so if they are setup
correctly by the firmware they shouldn't stop working if the firmware is
right.
So I'm skeptical a firmware fix would solve the issue, but I could
definitely be wrong about that.
Logan
prev parent reply other threads:[~2022-03-09 18:31 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-02-16 16:16 ntb_netdev Communication Failure Issue Marc Smith
2022-02-16 17:14 ` Logan Gunthorpe
2022-02-17 15:22 ` Marc Smith
2022-02-17 16:31 ` Eric Pilmore
2022-02-17 16:49 ` Logan Gunthorpe
2022-03-09 14:35 ` Marc Smith
2022-03-09 16:52 ` Logan Gunthorpe
2022-03-09 18:26 ` Marc Smith
2022-03-09 18:31 ` Logan Gunthorpe [this message]
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=bec56b4f-1eb5-506b-2cca-7bd8bc8cbe37@deltatee.com \
--to=logang@deltatee.com \
--cc=kelvin.cao@microchip.com \
--cc=kelvincao@outlook.com \
--cc=msmith626@gmail.com \
--cc=ntb@lists.linux.dev \
/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 read-only IMAP folder(s) and NNTP newsgroup(s).