From: Bjorn Helgaas <bhelgaas@google.com>
To: Sebastian Ott <sebott@linux.vnet.ibm.com>
Cc: linux-pci@vger.kernel.org,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Subject: Re: [Resend] pcibios_add_platform_entries usage
Date: Mon, 14 Apr 2014 11:35:41 -0600	[thread overview]
Message-ID: <20140414173541.GA4417@google.com> (raw)
In-Reply-To: <alpine.LFD.2.11.1404141101500.1529@denkbrett>
[+cc Greg]
On Mon, Apr 14, 2014 at 11:03:42AM +0200, Sebastian Ott wrote:
> Hello Bjorn,
> 
> for pci on s390 we currently use pcibios_add_platform_entries to add
> some arch specific attributes to pdevs. This has 2 downsides - it will
> race with userspace which is triggered by udev events and expecting
> these attributes (but that's a theoretical issue). More important to
> me is that one cannot use attribute_groups with this. Both issues could
> be addressed by using pdev->dev.groups and let the driver core handle
> attribute creation.
> 
> So would it be ok if we set pdev->dev.groups in pcibios_add_device?
> (It should be since it's not used by pci common code which uses bus_type,
> dev_type, and class groups).
Hi Sebastian,
Sorry, I meant to respond to this earlier, but forgot.  This sounds
reasonable to me, but Greg can give you a much better answer than I can.
Documentation/driver-model/device.txt says the dev->groups pointer
should be set before calling device_register().  PCI calls
device_initialize() and device_add() instead of using device_register(),
and pcibios_add_device() looks like it happens at the right time:
    pci_scan_root_bus
      pci_scan_child_bus
        pci_scan_slot
          pci_scan_single_device
            pci_device_add
              device_initialize
              pcibios_add_device                # <---
              device_add
      pci_bus_add_devices
        pci_bus_add_device
          pci_create_sysfs_dev_files
            pcibios_add_platform_entries        # 8d4cd0833107 (benh)
          device_attach
I'm not sure why pci_create_sysfs_dev_files() is done later.  It seems
like that should be done before device_add() as well.  Maybe it's
because BARs might not be valid yet (that doesn't seem like a very good
excuse, but it's all I can think of).
I assume that if you change s390, you'll also change microblaze and
powerpc?  They look structurally similar to s390.
Bjorn
next prev parent reply	other threads:[~2014-04-14 17:35 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-03-28 18:41 pcibios_add_platform_entries usage Sebastian Ott
2014-04-14  9:03 ` [Resend] " Sebastian Ott
2014-04-14 17:35   ` Bjorn Helgaas [this message]
2014-04-14 18:07     ` Sebastian Ott
2014-04-17 17:46       ` Sebastian Ott
2014-04-17 21:02         ` Benjamin Herrenschmidt
2014-04-18 10:10           ` Sebastian Ott
2014-04-18 21:31             ` Benjamin Herrenschmidt
2014-04-22  8:26               ` Sebastian Ott
2014-04-23  6:48                 ` Michal Simek
2014-04-23  9:00                   ` Sebastian Ott
2014-04-29 23:30         ` Bjorn Helgaas
2014-04-30 14:40           ` Sebastian Ott
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=20140414173541.GA4417@google.com \
    --to=bhelgaas@google.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=sebott@linux.vnet.ibm.com \
    /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).