From: Alex Williamson <alex.williamson@redhat.com>
To: Sinan Kaya <okaya@codeaurora.org>
Cc: Linux PCI <linux-pci@vger.kernel.org>
Subject: Re: IO Space Allocation Question on ARM64
Date: Mon, 22 May 2017 14:14:16 -0600 [thread overview]
Message-ID: <20170522141416.4fdc09a3@w520.home> (raw)
In-Reply-To: <31f48cf6-fd97-98fe-ac7c-ad80870df68b@codeaurora.org>
On Mon, 22 May 2017 13:50:31 -0400
Sinan Kaya <okaya@codeaurora.org> wrote:
> We are seeing IO space allocation failure on QDF2400 with multiple switch hierarchy
> below. This card has 20 PCI bridges with endpoints attached to each one to increase
> the number of endpoints supported.
>
> QDF2400 supports a maximum of 64k IO space aka 16 bit IO on the root port.
>
> -+-[0003:00]---00.0-[01-16]----00.0-[02-16]--+-00.0-[03]----00.0
> | +-01.0-[04]----00.0
> | +-02.0-[05]----00.0
> | +-03.0-[06]----00.0
> | +-04.0-[07]----00.0
> | +-05.0-[08]----00.0
> | +-06.0-[09]----00.0
> | +-07.0-[0a]----00.0
> | +-08.0-[0b]--
> | +-09.0-[0c]--
> | +-0a.0-[0d]--
> | +-0b.0-[0e]--
> | +-0c.0-[0f]--
> | +-0d.0-[10]--
> | +-0e.0-[11]--
> | +-0f.0-[12]--
> | +-10.0-[13]--
> | +-11.0-[14]--
> | +-12.0-[15]--
> | \-13.0-[16]--
>
> What we so far is that Linux is assigning an IO resource for every single bridge
> that it finds even though there is nothing connected behind on some of the bridges.
> Do we know the reason for this behavior?
>
> [ 4.288269] pci 0003:00:00.0: BAR 13: failed to assign [io size 0x15000]
> [ 4.317175] pci 0003:01:00.0: BAR 13: failed to assign [io size 0x14000]
> [ 4.644518] pci 0003:02:00.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.657539] pci 0003:02:01.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.670560] pci 0003:02:02.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.683581] pci 0003:02:03.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.696602] pci 0003:02:04.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.709623] pci 0003:02:05.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.722644] pci 0003:02:06.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.735664] pci 0003:02:07.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.748685] pci 0003:02:08.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.761706] pci 0003:02:09.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.774727] pci 0003:02:0a.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.787748] pci 0003:02:0b.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.800768] pci 0003:02:0c.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.813789] pci 0003:02:0d.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.826810] pci 0003:02:0e.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.839831] pci 0003:02:0f.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.852852] pci 0003:02:10.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.865873] pci 0003:02:11.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.878893] pci 0003:02:12.0: BAR 13: failed to assign [io size 0x1000]
> [ 4.891914] pci 0003:02:13.0: BAR 13: failed to assign [io size 0x1000]
>
> UEFI seems to be assigning an IO resource to bridge only if finds an endpoint
> behind it with an actual IO BAR space.
If it's anything like we saw with similar issues on QEMU, it might be
because the slots report themselves as hotplug capable and Linux tries
to allocate minimum apertures for any hotplug slot. See hpiosize
and hpmemsize options for pci=. Thanks,
Alex
next prev parent reply other threads:[~2017-05-22 20:14 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-05-22 17:50 IO Space Allocation Question on ARM64 Sinan Kaya
2017-05-22 20:14 ` Alex Williamson [this message]
2017-05-22 21:20 ` Sinan Kaya
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=20170522141416.4fdc09a3@w520.home \
--to=alex.williamson@redhat.com \
--cc=linux-pci@vger.kernel.org \
--cc=okaya@codeaurora.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).