All of lore.kernel.org
 help / color / mirror / Atom feed
From: Bjorn Helgaas <bhelgaas@google.com>
To: linux-pci@vger.kernel.org
Cc: Rob Herring <robh@kernel.org>, Heiko Stuebner <heiko@sntech.de>,
	Gabriele Paoloni <gabriele.paoloni@huawei.com>,
	Jon Mason <jonmason@broadcom.com>, Ray Jui <rjui@broadcom.com>,
	Shawn Lin <shawn.lin@rock-chips.com>,
	Wenrui Li <wenrui.li@rock-chips.com>,
	Scott Branden <sbranden@broadcom.com>,
	Russell King <linux@armlinux.org.uk>,
	Zhou Wang <wangzhou1@hisilicon.com>,
	Tanmay Inamdar <tinamdar@apm.com>,
	Greg Ungerer <gerg@uclinux.org>,
	Thierry Reding <treding@nvidia.com>
Subject: [PATCH] PCI: Warn on possible RW1C corruption for sub-32 bit config writes
Date: Mon, 31 Oct 2016 16:39:02 -0500	[thread overview]
Message-ID: <20161031213902.6340.96123.stgit@bhelgaas-glaptop.roam.corp.google.com> (raw)

Hardware that supports only 32-bit config writes is not spec-compliant.
For example, if software performs a 16-bit write, we must do a 32-bit read,
merge in the 16 bits we intend to write, followed by a 32-bit write.  If
the 16 bits we *don't* intend to write happen to have any RW1C (write-one-
to-clear) bits set, we just inadvertently cleared something we shouldn't
have.

Add a rate-limited warning when we do sub-32 bit config writes.  Remove
similar probe-time warnings from some of the affected host bridge drivers.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
---
 drivers/pci/access.c             |   17 +++++++++++++++--
 drivers/pci/host/pcie-hisi.c     |    2 --
 drivers/pci/host/pcie-rockchip.c |    3 ---
 3 files changed, 15 insertions(+), 7 deletions(-)

diff --git a/drivers/pci/access.c b/drivers/pci/access.c
index d11cdbb..e0cd124 100644
--- a/drivers/pci/access.c
+++ b/drivers/pci/access.c
@@ -142,10 +142,23 @@ int pci_generic_config_write32(struct pci_bus *bus, unsigned int devfn,
 	if (size == 4) {
 		writel(val, addr);
 		return PCIBIOS_SUCCESSFUL;
-	} else {
-		mask = ~(((1 << (size * 8)) - 1) << ((where & 0x3) * 8));
 	}
 
+	/*
+	 * N.B.  In general, hardware that supports only 32-bit writes on
+	 * PCI is not spec-compliant.  For example, software may perform a
+	 * 16-bit write.  If the hardware only supports 32-bit accesses, we
+	 * must do a 32-bit read, merge in the 16 bits we intend to write,
+	 * followed by a 32-bit write.  If the 16 bits we *don't* intend to
+	 * write happen to have any RW1C (write-one-to-clear) bits set, we
+	 * just inadvertently cleared something we shouldn't have.
+	 */
+
+	dev_warn_ratelimited(&bus->dev, "%d-byte config write to %04x:%02x:%02x.%d offset %#x may corrupt adjacent RW1C bits\n",
+			     size, pci_domain_nr(bus), bus->number,
+			     PCI_SLOT(devfn), PCI_FUNC(devfn), where);
+
+	mask = ~(((1 << (size * 8)) - 1) << ((where & 0x3) * 8));
 	tmp = readl(addr) & mask;
 	tmp |= val << ((where & 0x3) * 8);
 	writel(tmp, addr);
diff --git a/drivers/pci/host/pcie-hisi.c b/drivers/pci/host/pcie-hisi.c
index 56154c2..5b5901d 100644
--- a/drivers/pci/host/pcie-hisi.c
+++ b/drivers/pci/host/pcie-hisi.c
@@ -194,8 +194,6 @@ static int hisi_pcie_probe(struct platform_device *pdev)
 	if (ret)
 		return ret;
 
-	dev_warn(dev, "only 32-bit config accesses supported; smaller writes may corrupt adjacent RW1C fields\n");
-
 	return 0;
 }
 
diff --git a/drivers/pci/host/pcie-rockchip.c b/drivers/pci/host/pcie-rockchip.c
index e0b22da..6419d8c 100644
--- a/drivers/pci/host/pcie-rockchip.c
+++ b/drivers/pci/host/pcie-rockchip.c
@@ -1187,9 +1187,6 @@ static int rockchip_pcie_probe(struct platform_device *pdev)
 		pcie_bus_configure_settings(child);
 
 	pci_bus_add_devices(bus);
-
-	dev_warn(dev, "only 32-bit config accesses supported; smaller writes may corrupt adjacent RW1C fields\n");
-
 	return err;
 
 err_vpcie:


             reply	other threads:[~2016-10-31 21:39 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-10-31 21:39 Bjorn Helgaas [this message]
2016-11-01  9:34 ` [PATCH] PCI: Warn on possible RW1C corruption for sub-32 bit config writes Russell King - ARM Linux
2016-11-01  9:59 ` Shawn Lin
2016-11-02 17:45 ` Thierry Reding
2016-11-10 17:27 ` Bjorn Helgaas

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=20161031213902.6340.96123.stgit@bhelgaas-glaptop.roam.corp.google.com \
    --to=bhelgaas@google.com \
    --cc=gabriele.paoloni@huawei.com \
    --cc=gerg@uclinux.org \
    --cc=heiko@sntech.de \
    --cc=jonmason@broadcom.com \
    --cc=linux-pci@vger.kernel.org \
    --cc=linux@armlinux.org.uk \
    --cc=rjui@broadcom.com \
    --cc=robh@kernel.org \
    --cc=sbranden@broadcom.com \
    --cc=shawn.lin@rock-chips.com \
    --cc=tinamdar@apm.com \
    --cc=treding@nvidia.com \
    --cc=wangzhou1@hisilicon.com \
    --cc=wenrui.li@rock-chips.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 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.