All of lore.kernel.org
 help / color / mirror / Atom feed
From: Sergei Miroshnichenko <s.miroshnichenko@yadro.com>
To: <linux-pci@vger.kernel.org>
Cc: "Bjorn Helgaas" <helgaas@kernel.org>,
	"Lukas Wunner" <lukas@wunner.de>, "Stefan Roese" <sr@denx.de>,
	"Andy Lavr" <andy.lavr@gmail.com>,
	"Christian König" <christian.koenig@amd.com>,
	"David Laight" <David.Laight@ACULAB.COM>,
	"Rajat Jain" <rajatja@google.com>,
	linux@yadro.com,
	"Sergei Miroshnichenko" <s.miroshnichenko@yadro.com>,
	linux-nvme@lists.infradead.org, "Christoph Hellwig" <hch@lst.de>
Subject: [PATCH v9 24/26] nvme-pci: Handle movable BARs
Date: Fri, 18 Dec 2020 20:40:09 +0300	[thread overview]
Message-ID: <20201218174011.340514-25-s.miroshnichenko@yadro.com> (raw)
In-Reply-To: <20201218174011.340514-1-s.miroshnichenko@yadro.com>

Hot-added devices can affect the existing ones by moving their BARs. The
PCI subsystem will inform the NVME driver about this by invoking the
.rescan_prepare() and .rescan_done() hooks, so the BARs can by re-mapped.

Tested under the "randrw" mode of the fio tool, and when using an NVME
drive as a root filesystem storage. Before the hot-adding:

  % sudo cat /proc/iomem
  ...
                3fe800000000-3fe8007fffff : PCI Bus 0020:0b
                  3fe800000000-3fe8007fffff : PCI Bus 0020:18
                    3fe800000000-3fe8000fffff : 0020:18:00.0
                      3fe800000000-3fe8000fffff : nvme
                      ^^^^^^^^^^^^^^^^^^^^^^^^^
                    3fe800100000-3fe80017ffff : 0020:18:00.0
  ...

Then another NVME drive was hot-added, so BARs of the 0020:18:00.0 are
moved:

  % sudo cat /proc/iomem
    ...
                3fe800000000-3fe800ffffff : PCI Bus 0020:0b
                  3fe800000000-3fe8007fffff : PCI Bus 0020:10
                    3fe800000000-3fe800003fff : 0020:10:00.0
                      3fe800000000-3fe800003fff : nvme
                    3fe800010000-3fe80001ffff : 0020:10:00.0
                  3fe800800000-3fe800ffffff : PCI Bus 0020:18
                    3fe800800000-3fe8008fffff : 0020:18:00.0
                      3fe800800000-3fe8008fffff : nvme
                      ^^^^^^^^^^^^^^^^^^^^^^^^^
                    3fe800900000-3fe80097ffff : 0020:18:00.0
    ...

During the rescanning, both READ and WRITE speeds drop to zero for a while
due to driver's pause, then restore.

Cc: linux-nvme@lists.infradead.org
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sergei Miroshnichenko <s.miroshnichenko@yadro.com>
---
 drivers/nvme/host/pci.c | 27 ++++++++++++++++++++++++++-
 1 file changed, 26 insertions(+), 1 deletion(-)

diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index b4385cb0ff60..7993062b11b6 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -1656,7 +1656,7 @@ static int nvme_remap_bar(struct nvme_dev *dev, unsigned long size)
 {
 	struct pci_dev *pdev = to_pci_dev(dev->dev);
 
-	if (size <= dev->bar_mapped_size)
+	if (dev->bar && size <= dev->bar_mapped_size)
 		return 0;
 	if (size > pci_resource_len(pdev, 0))
 		return -ENOMEM;
@@ -3152,6 +3152,28 @@ static void nvme_error_resume(struct pci_dev *pdev)
 	flush_work(&dev->ctrl.reset_work);
 }
 
+static bool nvme_bar_fixed(struct pci_dev *pdev, int resno)
+{
+	return false;
+}
+
+static void nvme_rescan_prepare(struct pci_dev *pdev)
+{
+	struct nvme_dev *dev = pci_get_drvdata(pdev);
+
+	nvme_dev_disable(dev, true);
+	nvme_dev_unmap(dev);
+	dev->bar = NULL;
+}
+
+static void nvme_rescan_done(struct pci_dev *pdev)
+{
+	struct nvme_dev *dev = pci_get_drvdata(pdev);
+
+	nvme_dev_map(dev);
+	nvme_reset_ctrl(&dev->ctrl);
+}
+
 static const struct pci_error_handlers nvme_err_handler = {
 	.error_detected	= nvme_error_detected,
 	.slot_reset	= nvme_slot_reset,
@@ -3238,6 +3260,9 @@ static struct pci_driver nvme_driver = {
 #endif
 	.sriov_configure = pci_sriov_configure_simple,
 	.err_handler	= &nvme_err_handler,
+	.rescan_prepare	= nvme_rescan_prepare,
+	.rescan_done	= nvme_rescan_done,
+	.bar_fixed	= nvme_bar_fixed,
 };
 
 static int __init nvme_init(void)
-- 
2.24.1


WARNING: multiple messages have this Message-ID (diff)
From: Sergei Miroshnichenko <s.miroshnichenko@yadro.com>
To: <linux-pci@vger.kernel.org>
Cc: "Rajat Jain" <rajatja@google.com>,
	"Andy Lavr" <andy.lavr@gmail.com>,
	"David Laight" <David.Laight@ACULAB.COM>,
	"Sergei Miroshnichenko" <s.miroshnichenko@yadro.com>,
	linux-nvme@lists.infradead.org, linux@yadro.com,
	"Lukas Wunner" <lukas@wunner.de>,
	"Bjorn Helgaas" <helgaas@kernel.org>, "Stefan Roese" <sr@denx.de>,
	"Christoph Hellwig" <hch@lst.de>,
	"Christian König" <christian.koenig@amd.com>
Subject: [PATCH v9 24/26] nvme-pci: Handle movable BARs
Date: Fri, 18 Dec 2020 20:40:09 +0300	[thread overview]
Message-ID: <20201218174011.340514-25-s.miroshnichenko@yadro.com> (raw)
In-Reply-To: <20201218174011.340514-1-s.miroshnichenko@yadro.com>

Hot-added devices can affect the existing ones by moving their BARs. The
PCI subsystem will inform the NVME driver about this by invoking the
.rescan_prepare() and .rescan_done() hooks, so the BARs can by re-mapped.

Tested under the "randrw" mode of the fio tool, and when using an NVME
drive as a root filesystem storage. Before the hot-adding:

  % sudo cat /proc/iomem
  ...
                3fe800000000-3fe8007fffff : PCI Bus 0020:0b
                  3fe800000000-3fe8007fffff : PCI Bus 0020:18
                    3fe800000000-3fe8000fffff : 0020:18:00.0
                      3fe800000000-3fe8000fffff : nvme
                      ^^^^^^^^^^^^^^^^^^^^^^^^^
                    3fe800100000-3fe80017ffff : 0020:18:00.0
  ...

Then another NVME drive was hot-added, so BARs of the 0020:18:00.0 are
moved:

  % sudo cat /proc/iomem
    ...
                3fe800000000-3fe800ffffff : PCI Bus 0020:0b
                  3fe800000000-3fe8007fffff : PCI Bus 0020:10
                    3fe800000000-3fe800003fff : 0020:10:00.0
                      3fe800000000-3fe800003fff : nvme
                    3fe800010000-3fe80001ffff : 0020:10:00.0
                  3fe800800000-3fe800ffffff : PCI Bus 0020:18
                    3fe800800000-3fe8008fffff : 0020:18:00.0
                      3fe800800000-3fe8008fffff : nvme
                      ^^^^^^^^^^^^^^^^^^^^^^^^^
                    3fe800900000-3fe80097ffff : 0020:18:00.0
    ...

During the rescanning, both READ and WRITE speeds drop to zero for a while
due to driver's pause, then restore.

Cc: linux-nvme@lists.infradead.org
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sergei Miroshnichenko <s.miroshnichenko@yadro.com>
---
 drivers/nvme/host/pci.c | 27 ++++++++++++++++++++++++++-
 1 file changed, 26 insertions(+), 1 deletion(-)

diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index b4385cb0ff60..7993062b11b6 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -1656,7 +1656,7 @@ static int nvme_remap_bar(struct nvme_dev *dev, unsigned long size)
 {
 	struct pci_dev *pdev = to_pci_dev(dev->dev);
 
-	if (size <= dev->bar_mapped_size)
+	if (dev->bar && size <= dev->bar_mapped_size)
 		return 0;
 	if (size > pci_resource_len(pdev, 0))
 		return -ENOMEM;
@@ -3152,6 +3152,28 @@ static void nvme_error_resume(struct pci_dev *pdev)
 	flush_work(&dev->ctrl.reset_work);
 }
 
+static bool nvme_bar_fixed(struct pci_dev *pdev, int resno)
+{
+	return false;
+}
+
+static void nvme_rescan_prepare(struct pci_dev *pdev)
+{
+	struct nvme_dev *dev = pci_get_drvdata(pdev);
+
+	nvme_dev_disable(dev, true);
+	nvme_dev_unmap(dev);
+	dev->bar = NULL;
+}
+
+static void nvme_rescan_done(struct pci_dev *pdev)
+{
+	struct nvme_dev *dev = pci_get_drvdata(pdev);
+
+	nvme_dev_map(dev);
+	nvme_reset_ctrl(&dev->ctrl);
+}
+
 static const struct pci_error_handlers nvme_err_handler = {
 	.error_detected	= nvme_error_detected,
 	.slot_reset	= nvme_slot_reset,
@@ -3238,6 +3260,9 @@ static struct pci_driver nvme_driver = {
 #endif
 	.sriov_configure = pci_sriov_configure_simple,
 	.err_handler	= &nvme_err_handler,
+	.rescan_prepare	= nvme_rescan_prepare,
+	.rescan_done	= nvme_rescan_done,
+	.bar_fixed	= nvme_bar_fixed,
 };
 
 static int __init nvme_init(void)
-- 
2.24.1


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

  parent reply	other threads:[~2020-12-18 17:43 UTC|newest]

Thread overview: 46+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-12-18 17:39 [PATCH v9 00/26] PCI: Allow BAR movement during boot and hotplug Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 01/26] PCI: Fix race condition in pci_enable/disable_device() Sergei Miroshnichenko
2020-12-19 19:01   ` kernel test robot
2020-12-28 15:37   ` Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 02/26] PCI: Ensure a bridge has I/O and MEM access for hot-added devices Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 03/26] PCI: hotplug: Initial support of the movable BARs feature Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 04/26] PCI: Add version of release_child_resources() aware of fixed BARs Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 05/26] PCI: hotplug: Fix reassigning the released BARs Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 06/26] PCI: hotplug: Recalculate every bridge window during rescan Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 07/26] PCI: hotplug: Don't allow hot-added devices to steal resources Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 08/26] PCI: Reassign BARs if BIOS/bootloader had assigned not all of them Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 09/26] PCI: Movable BARs: Make just a single attempt to assign bridge resources Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 10/26] PCI: hotplug: Calculate fixed parts of bridge windows Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 11/26] PCI: Include fixed BARs into the bus size calculating Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 12/26] PCI: Make sure bridge windows include their fixed BARs Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 13/26] PCI: hotplug: Add support of fixed BARs to pci_assign_resource() Sergei Miroshnichenko
2020-12-18 17:39 ` [PATCH v9 14/26] PCI: hotplug: Sort fixed BARs before assignment Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 15/26] x86/PCI/ACPI: Fix up PCIBIOS_MIN_MEM if value computed from e820 is invalid Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 16/26] PCI: hotplug: Configure MPS after manual bus rescan Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 17/26] PCI: hotplug: Don't disable the released bridge windows immediately Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 18/26] PCI: pciehp: Trigger a domain rescan on hp events when enabled movable BARs Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 19/26] PCI: Don't claim fixed BARs Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 20/26] PCI: hotplug: Retry bus assignment without reserved space Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 21/26] PCI: Rescan the bus to resize a BAR Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 22/26] PCI: hotplug: Enable the movable BARs feature by default Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 23/26] PCI/portdrv: Declare support of movable BARs Sergei Miroshnichenko
2020-12-18 17:40 ` Sergei Miroshnichenko [this message]
2020-12-18 17:40   ` [PATCH v9 24/26] nvme-pci: Handle " Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 25/26] PCI: Add a message for updating BARs Sergei Miroshnichenko
2020-12-18 17:40 ` [PATCH v9 26/26] resource: increase max nesting level for /proc/iomem Sergei Miroshnichenko
2021-01-28 14:53 ` [PATCH v9 00/26] PCI: Allow BAR movement during boot and hotplug Bjorn Helgaas
2021-01-28 20:39   ` Lukas Wunner
2021-02-01 12:55     ` Mika Westerberg
2021-02-03 20:17       ` Sergei Miroshnichenko
2021-02-04 10:49         ` mika.westerberg
2021-02-10 19:40           ` Sergei Miroshnichenko
2021-02-10 21:46             ` Lukas Wunner
2021-02-11 11:39             ` mika.westerberg
2021-02-11 17:45               ` Sergei Miroshnichenko
2021-02-12 12:52                 ` mika.westerberg
2021-02-12 20:54                   ` Sergei Miroshnichenko
2021-02-15 14:56                     ` mika.westerberg
2021-02-03 20:01     ` Sergei Miroshnichenko
2021-02-04  9:34       ` David Laight
2021-02-03 19:59   ` Sergei Miroshnichenko
2021-02-04  8:26     ` Hinko Kocevar

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=20201218174011.340514-25-s.miroshnichenko@yadro.com \
    --to=s.miroshnichenko@yadro.com \
    --cc=David.Laight@ACULAB.COM \
    --cc=andy.lavr@gmail.com \
    --cc=christian.koenig@amd.com \
    --cc=hch@lst.de \
    --cc=helgaas@kernel.org \
    --cc=linux-nvme@lists.infradead.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=linux@yadro.com \
    --cc=lukas@wunner.de \
    --cc=rajatja@google.com \
    --cc=sr@denx.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.