linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Yinghai Lu <yinghai@kernel.org>
To: Bjorn Helgaas <bhelgaas@google.com>
Cc: Len Brown <lenb@kernel.org>,
	Taku Izumi <izumi.taku@jp.fujitsu.com>,
	Jiang Liu <jiang.liu@huawei.com>, x86 <x86@kernel.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-acpi@vger.kernel.org, Dave Airlie <airlied@redhat.com>,
	Julia Lawall <julia@diku.dk>, Matthew Garrett <mjg@redhat.com>
Subject: Re: [PATCH 01/40] PCI: fix default vga ref_count
Date: Fri, 21 Sep 2012 14:18:00 -0700	[thread overview]
Message-ID: <CAE9FiQVP7Q2p2zN8pXubcgx=7CFAt9pw0tWcWpuPddO9U7pR5A@mail.gmail.com> (raw)
In-Reply-To: <CAErSpo7S3DCMq1Rsj0Ya856ebyqFT_1pF2yirzaHrGPSSUJinQ@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 900 bytes --]

On Fri, Sep 21, 2012 at 1:52 PM, Bjorn Helgaas <bhelgaas@google.com> wrote:
> On Wed, Sep 19, 2012 at 12:54 PM, Yinghai Lu <yinghai@kernel.org> wrote:
>> --- a/drivers/gpu/vga/vgaarb.c
>> +++ b/drivers/gpu/vga/vgaarb.c
>> @@ -141,6 +141,12 @@ EXPORT_SYMBOL_GPL(vga_default_device);
>>
>>  void vga_set_default_device(struct pci_dev *pdev)
>>  {
>> +       if (vga_default)
>> +               pci_dev_put(vga_default);
>> +
>> +       if (pdev)
>> +               pdev = pci_dev_get(pdev);
>> +
>>         vga_default = pdev;
>
> I think this is equivalent to:
>
>     pci_dev_put(vga_default);
>     vga_default = pci_dev_get(pdev);

ah, i missed that, pci_dev_put and pci_dev_get check that inside.

>
> I haven't seen an answer to Matthew's question about whether we want
> to check for "vga_default == pdev".

yes, that could avoid the extra put/get pair.

Please check updated version.

-Yinghai

[-- Attachment #2: fix_vgarab_add_x.patch --]
[-- Type: application/octet-stream, Size: 2560 bytes --]

Subject: [PATCH] PCI: fix default vga ref_count

when __ARCH_HAS_VGA_DEFAULT_DEVICE is not defined, aka EFIFB is not used,
for static path, vga_default setting is through vga_arbiter_add_pci_device.
and later x86 pci_fixup_video, will skip setting again.
- subsys_initcall(vga_arb_device_init) come first to call vga_arbiter_add_pci_device. It will call pci_get_dev to hold one reference.

for hotplug add path, even vga_arbiter_add_pci_device is called via notifier,
but it will check VGA_RSRC_LEGACY_MASK that is not set for hotplug path.
So x86 pci_fixup_video will take over to call vga_set_default_device().
It will not hold one refrence.

Later for hotplug remove path, vga_arbiter_del_pci_device that does not check
VGA_RSRC_LEGACY_MASK will call put_device and it will cause ref_count to
decrease extra. that will have that pci device get deleted early wrongly.

Need to make get/put balance for both cases.

-v2: According to Bjorn, update vga_set_default_device instead...
-v3: Remove extra checking about null that pointed out by Bjorn.
     Add extra check for same value, to avoid extra put/get pair, suggested by
	Matthew.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Cc: x86@kernel.org
Cc: Dave Airlie <airlied@redhat.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Julia Lawall <julia@diku.dk>
Cc: Matthew Garrett <mjg@redhat.com>

---
 drivers/gpu/vga/vgaarb.c |   14 ++++++++------
 1 file changed, 8 insertions(+), 6 deletions(-)

Index: linux-2.6/drivers/gpu/vga/vgaarb.c
===================================================================
--- linux-2.6.orig/drivers/gpu/vga/vgaarb.c
+++ linux-2.6/drivers/gpu/vga/vgaarb.c
@@ -141,7 +141,11 @@ EXPORT_SYMBOL_GPL(vga_default_device);
 
 void vga_set_default_device(struct pci_dev *pdev)
 {
-	vga_default = pdev;
+	if (vga_default == pdev)
+		return;
+
+	pci_dev_put(vga_default);
+	vga_default = pci_dev_get(pdev);
 }
 #endif
 
@@ -577,7 +581,7 @@ static bool vga_arbiter_add_pci_device(s
 #ifndef __ARCH_HAS_VGA_DEFAULT_DEVICE
 	if (vga_default == NULL &&
 	    ((vgadev->owns & VGA_RSRC_LEGACY_MASK) == VGA_RSRC_LEGACY_MASK))
-		vga_default = pci_dev_get(pdev);
+		vga_set_default_device(pdev);
 #endif
 
 	vga_arbiter_check_bridge_sharing(vgadev);
@@ -613,10 +617,8 @@ static bool vga_arbiter_del_pci_device(s
 	}
 
 #ifndef __ARCH_HAS_VGA_DEFAULT_DEVICE
-	if (vga_default == pdev) {
-		pci_dev_put(vga_default);
-		vga_default = NULL;
-	}
+	if (vga_default == pdev)
+		vga_set_default_device(NULL);
 #endif
 
 	if (vgadev->decodes & (VGA_RSRC_LEGACY_IO | VGA_RSRC_LEGACY_MEM))

  reply	other threads:[~2012-09-21 21:18 UTC|newest]

Thread overview: 60+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-09-02 21:50 [PATCH part2 0/6] PCI, x86: pci root bus hotplug support - part2 Yinghai Lu
2012-09-02 21:50 ` [PATCH part2 1/6] x86, PCI: Separate pcibios_allocate_bridge_resources() Yinghai Lu
2012-09-02 21:50 ` [PATCH part2 2/6] x86, PCI: Separate pcibios_allocate_dev_resources() Yinghai Lu
2012-09-02 21:50 ` [PATCH part2 3/6] x86, PCI: Let pcibios_allocate_bus_resources() take bus instead Yinghai Lu
2012-09-02 21:50 ` [PATCH part2 4/6] x86, PCI: Separate rom resource claim out Yinghai Lu
2012-09-17 23:38   ` Bjorn Helgaas
2012-09-18  3:18     ` Yinghai Lu
2012-09-02 21:50 ` [PATCH part2 5/6] PCI, x86: Add pcibios_fw_addr_done Yinghai Lu
2012-09-02 21:50 ` [PATCH part2 6/6] PCI: Claim hw/fw allocated resources in hot add path Yinghai Lu
2012-09-18  0:12   ` Bjorn Helgaas
2012-09-18  4:44     ` Yinghai Lu
2012-09-18  5:36       ` Yinghai Lu
2012-09-19 18:54         ` [PATCH 00/40] PCI, ACPI, x86: pci root bus hotplug support Yinghai Lu
2012-09-19 18:54           ` [PATCH 01/40] PCI: fix default vga ref_count Yinghai Lu
2012-09-21 20:52             ` Bjorn Helgaas
2012-09-21 21:18               ` Yinghai Lu [this message]
2012-09-23 15:25             ` Jiang Liu
2012-09-19 18:54           ` [PATCH 02/40] PCI, x86: clear initial value for root info resources Yinghai Lu
2012-09-19 18:54           ` [PATCH 03/40] PCI, ia64: " Yinghai Lu
2012-09-19 18:54           ` [PATCH 04/40] PCI, acpiphp: Add is_hotplug_bridge detection Yinghai Lu
2012-09-19 18:54           ` [PATCH 05/40] PCI: Add root bus children dev's res to fail list Yinghai Lu
2012-09-19 18:54           ` [PATCH 06/40] PCI: Split out stop_bus_device and remove_bus_dev again Yinghai Lu
2012-09-19 18:54           ` [PATCH 07/40] x86, PCI: Separate pcibios_allocate_bridge_resources() Yinghai Lu
2012-09-19 18:54           ` [PATCH 08/40] x86, PCI: Separate pcibios_allocate_dev_resources() Yinghai Lu
2012-09-19 18:54           ` [PATCH 09/40] x86, PCI: Let pcibios_allocate_bus_resources() take bus instead Yinghai Lu
2012-09-19 18:54           ` [PATCH 10/40] x86, PCI: Separate rom resource claim out Yinghai Lu
2012-09-19 18:54           ` [PATCH 11/40] PCI, x86: Add pcibios_fw_addr_done Yinghai Lu
2012-09-19 18:54           ` [PATCH 12/40] PCI, x86: Remove __init for hw/fw allocated functions Yinghai Lu
2012-09-19 18:54           ` [PATCH 13/40] PCI: Claim hw/fw allocated resources in hot add path Yinghai Lu
2012-09-19 18:54           ` [PATCH 14/40] PCI: Separate out pci_assign_unassigned_bus_resources() Yinghai Lu
2012-09-19 18:54           ` [PATCH 15/40] PCI: Move back pci_rescan_bus() to probe.c Yinghai Lu
2012-09-19 18:54           ` [PATCH 16/40] PCI: pci_bus_size_bridges() should not size own bridge Yinghai Lu
2012-09-19 18:54           ` [PATCH 17/40] PCI: Use __pci_bus_size_bridges() directly in pci_assign_unassigned_bus_resources() Yinghai Lu
2012-09-19 18:54           ` [PATCH 18/40] PCI: Rescan bus using callback method too Yinghai Lu
2012-09-19 18:54           ` [PATCH 19/40] PCI, sysfs: Clean up rescan/remove with schedule_callback Yinghai Lu
2012-09-19 18:54           ` [PATCH 20/40] PCI: Fix a device reference count leakage issue in pci_dev_present() Yinghai Lu
2012-09-19 18:54           ` [PATCH 21/40] PCI: Add pci_stop_and_remove_root_bus() Yinghai Lu
2012-09-19 18:54           ` [PATCH 22/40] PCI, acpiphp: Separate out hot-add support of pci host bridge Yinghai Lu
2012-10-12 10:36             ` Tang Chen
2012-09-19 18:54           ` [PATCH 23/40] PCI, ACPI: Pass device instead of handle when config root bridge Yinghai Lu
2012-09-19 18:54           ` [PATCH 24/40] PCI, acpi: Update acpi_pci_driver add/remove interface Yinghai Lu
2012-09-19 18:54           ` [PATCH 25/40] PCI, ACPI: Make acpi_pci_root_remove remove pci root bus too Yinghai Lu
2012-09-19 18:54           ` [PATCH 26/40] PCI, ACPI: del root bus prt Yinghai Lu
2012-09-19 18:54           ` [PATCH 27/40] ACPI: acpi_bus_trim to support two steps Yinghai Lu
2012-09-19 18:54           ` [PATCH 28/40] PCI, ACPI: Add pci_root_hp hot removal notification support Yinghai Lu
2012-09-19 18:54           ` [PATCH 29/40] PCI, ACPI: Add alloc_acpi_hp_work() Yinghai Lu
2012-09-19 18:54           ` [PATCH 30/40] PCI, acpiphp: Use acpi_hp_work Yinghai Lu
2012-09-19 18:54           ` [PATCH 31/40] PCI, pci_root_hp: " Yinghai Lu
2012-09-19 18:54           ` [PATCH 32/40] PCI, ACPI: Make kacpi_hotplug_wq static Yinghai Lu
2012-09-19 18:54           ` [PATCH 33/40] ACPI, PCI: Use normal list for struct acpi_pci_driver Yinghai Lu
2012-09-19 18:54           ` [PATCH 34/40] ACPI, PCI: Notify acpi_pci_drivers when hot-plugging PCI root bridges Yinghai Lu
2012-09-19 18:54           ` [PATCH 35/40] ACPI, PCI: Protect global lists in drivers/acpi/pci_root.c Yinghai Lu
2012-09-19 18:54           ` [PATCH 36/40] PCI: Set dev_node early for pci_dev Yinghai Lu
2012-09-19 18:54           ` [PATCH 37/40] PCI, x86: Move pci_enable_bridges() down Yinghai Lu
2012-09-19 18:54           ` [PATCH 38/40] ACPI, PCI: Skip extra pci_enable_bridges for non hot-add root Yinghai Lu
2012-09-19 18:54           ` [PATCH 39/40] PCI, acpiphp: Don't ailout even no slots found yet Yinghai Lu
2012-09-19 18:54           ` [PATCH 40/40] ACPI: Enable SCI_EMULATE to manually simulate physical hotplug testing Yinghai Lu
2012-10-17  7:50           ` [PATCH 00/40] PCI, ACPI, x86: pci root bus hotplug support Yijing Wang
2012-10-17 16:19             ` Yinghai Lu
2012-10-18  0:51               ` Yijing Wang

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='CAE9FiQVP7Q2p2zN8pXubcgx=7CFAt9pw0tWcWpuPddO9U7pR5A@mail.gmail.com' \
    --to=yinghai@kernel.org \
    --cc=airlied@redhat.com \
    --cc=akpm@linux-foundation.org \
    --cc=bhelgaas@google.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=izumi.taku@jp.fujitsu.com \
    --cc=jiang.liu@huawei.com \
    --cc=julia@diku.dk \
    --cc=lenb@kernel.org \
    --cc=linux-acpi@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=mjg@redhat.com \
    --cc=torvalds@linux-foundation.org \
    --cc=x86@kernel.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).