From: Bjorn Helgaas <helgaas@kernel.org>
To: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>,
Maxime Ripard <mripard@kernel.org>,
Thomas Zimmermann <tzimmermann@suse.de>
Cc: David Airlie <airlied@linux.ie>, Daniel Vetter <daniel@ffwll.ch>,
Xuefeng Li <lixuefeng@loongson.cn>,
Huacai Chen <chenhuacai@loongson.cn>,
linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org,
dri-devel@lists.freedesktop.org
Subject: [PATCH v9 04/11] PCI/VGA: Factor out default VGA device selection
Date: Thu, 24 Feb 2022 16:47:46 -0600 [thread overview]
Message-ID: <20220224224753.297579-5-helgaas@kernel.org> (raw)
In-Reply-To: <20220224224753.297579-1-helgaas@kernel.org>
From: Huacai Chen <chenhuacai@loongson.cn>
Default VGA device selection fails when PCI devices are enumerated after
the vga_arb_device_init() subsys_initcall.
vga_arbiter_add_pci_device() selects the first fully enabled device to
which legacy VGA resources are routed as the default VGA device. This is
an ADD_DEVICE notifier, so it runs after every PCI device is enumerated.
vga_arb_select_default_device() may select framebuffer devices, partially
enabled GPUs, or non-legacy devices that don't have legacy VGA resources
routed to them as the default VGA device. But this only happens once, from
the vga_arb_device_init() subsys_initcall, so it doesn't consider devices
enumerated after that:
acpi_init
acpi_scan_init
acpi_pci_root_init # PCI device enumeration (ACPI systems)
vga_arb_device_init
for_each_pci_device
vga_arbiter_add_pci_device # ADD_DEVICE notifier
if (VGA-owner)
vga_set_default_device <-- set default VGA
vga_arb_select_default_device # only called ONCE
for_each_vga_device
if (framebuffer)
vga_set_default_device <-- set default VGA to framebuffer
if (!vga_default_device())
if (non-legacy, integrated GPU, etc)
vga_set_default_device <-- set default VGA
if (!vga_default_device())
vga_set_default_device <-- set default VGA
pcibios_init
pcibios_scanbus # PCI device enumeration (non-ACPI systems)
...
vga_arbiter_add_pci_device # ADD_DEVICE notification
if (VGA-owner)
vga_set_default_device <-- set default VGA
Note that on non-ACPI systems, vga_arb_select_default_device() runs before
pcibios_init(), so it sees no VGA devices and can never set a framebuffer
device, a non-legacy integrated GPU, etc., as the default device.
Factor out the default VGA device selection to vga_is_boot_device(), called
from vga_arbiter_add_pci_device().
Then we can migrate the default device selection from
vga_arb_select_default_device() to the vga_arbiter_add_pci_device() path.
[bhelgaas: commit log, split to separate patch]
Link: https://lore.kernel.org/r/20211015061512.2941859-4-chenhuacai@loongson.cn
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
---
drivers/pci/vgaarb.c | 45 ++++++++++++++++++++++++++++++++++++++------
1 file changed, 39 insertions(+), 6 deletions(-)
diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
index 3f8fead49197..58e0a12e623b 100644
--- a/drivers/pci/vgaarb.c
+++ b/drivers/pci/vgaarb.c
@@ -628,6 +628,41 @@ static bool vga_arb_integrated_gpu(struct device *dev)
#endif
}
+/*
+ * Return true if vgadev is a better default VGA device than the best one
+ * we've seen so far.
+ */
+static bool vga_is_boot_device(struct vga_device *vgadev)
+{
+ struct vga_device *boot_vga = vgadev_find(vga_default_device());
+
+ /*
+ * We select the default VGA device in this order:
+ * Firmware framebuffer (see vga_arb_select_default_device())
+ * Legacy VGA device (owns VGA_RSRC_LEGACY_MASK)
+ * Non-legacy integrated device (see vga_arb_select_default_device())
+ * Non-legacy discrete device (see vga_arb_select_default_device())
+ * Other device (see vga_arb_select_default_device())
+ */
+
+ /*
+ * A legacy VGA device has MEM and IO enabled and any bridges
+ * leading to it have PCI_BRIDGE_CTL_VGA enabled so the legacy
+ * resources ([mem 0xa0000-0xbffff], [io 0x3b0-0x3bb], etc) are
+ * routed to it.
+ *
+ * We use the first one we find, so if we've already found one,
+ * vgadev is no better.
+ */
+ if (boot_vga)
+ return false;
+
+ if ((vgadev->owns & VGA_RSRC_LEGACY_MASK) == VGA_RSRC_LEGACY_MASK)
+ return true;
+
+ return false;
+}
+
/*
* Rules for using a bridge to control a VGA descendant decoding: if a bridge
* has only one VGA descendant then it can be used to control the VGA routing
@@ -755,12 +790,10 @@ static bool vga_arbiter_add_pci_device(struct pci_dev *pdev)
bus = bus->parent;
}
- /* Deal with VGA default device. Use first enabled one
- * by default if arch doesn't have it's own hook
- */
- if (vga_default == NULL &&
- ((vgadev->owns & VGA_RSRC_LEGACY_MASK) == VGA_RSRC_LEGACY_MASK)) {
- vgaarb_info(&pdev->dev, "setting as boot VGA device\n");
+ if (vga_is_boot_device(vgadev)) {
+ vgaarb_info(&pdev->dev, "setting as boot VGA device%s\n",
+ vga_default_device() ?
+ " (overriding previous)" : "");
vga_set_default_device(pdev);
}
--
2.25.1
next prev parent reply other threads:[~2022-02-24 22:48 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-02-24 22:47 [PATCH v9 00/11] vgaarb: Rework default VGA device selection Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 01/11] PCI/VGA: Move vgaarb to drivers/pci Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 02/11] PCI/VGA: Move vga_arb_integrated_gpu() earlier in file Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 03/11] PCI/VGA: Factor out vga_select_framebuffer_device() Bjorn Helgaas
2022-02-24 22:47 ` Bjorn Helgaas [this message]
2022-02-24 22:47 ` [PATCH v9 05/11] PCI/VGA: Move firmware default device detection to ADD_DEVICE path Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 06/11] PCI/VGA: Move non-legacy VGA " Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 07/11] PCI/VGA: Move disabled VGA device " Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 08/11] PCI/VGA: Remove empty vga_arb_device_card_gone() Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 09/11] PCI/VGA: Log bridge control messages when adding devices Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 10/11] PCI/VGA: Use unsigned format string to print lock counts Bjorn Helgaas
2022-02-24 22:47 ` [PATCH v9 11/11] PCI/VGA: Replace full MIT license text with SPDX identifier Bjorn Helgaas
2022-02-25 22:15 ` [PATCH v9 00/11] vgaarb: Rework default VGA device selection Bjorn Helgaas
2022-03-09 16:29 ` Bjorn Helgaas
2022-03-10 3:56 ` Huacai Chen
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=20220224224753.297579-5-helgaas@kernel.org \
--to=helgaas@kernel.org \
--cc=airlied@linux.ie \
--cc=chenhuacai@loongson.cn \
--cc=daniel@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=lixuefeng@loongson.cn \
--cc=maarten.lankhorst@linux.intel.com \
--cc=mripard@kernel.org \
--cc=tzimmermann@suse.de \
/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).