| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <201204152047.51019.rjw@sisk.pl>
Date: Sun, 15 Apr 2012 20:47:50 +0200
From: "Rafael J. Wysocki" <rjw@...k.pl>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: Bjorn Helgaas <bhelgaas@...gle.com>,
Mikko Vinni <mmvinni@...oo.com>,
"linux-input@...r.kernel.org" <linux-input@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
Dmitry Torokhov <dmitry.torokhov@...il.com>,
Allen Kay <allen.m.kay@...el.com>,
Jesse Barnes <jbarnes@...tuousgeek.org>,
"linux-pci@...r.kernel.org" <linux-pci@...r.kernel.org>
Subject: Re: [PATCH] PCI: Fix regression in pci_restore_state()
On Sunday, April 15, 2012, Linus Torvalds wrote:
> On Sun, Apr 15, 2012 at 11:32 AM, Rafael J. Wysocki <rjw@...k.pl> wrote:
> >
> > To fix the regression, make pci_restore_state() only retry the
> > writes for BAR registers and only wait if the first read from
> > the register doesn't return the written value. Additionally, make
> > it use msleep() instead of mdelay() for waiting.
>
> WHAT?
>
> msleep() is the one that can sleep. So that part of the change is just
> crazy. There's no way you can sleep during PCI register restore.
OK, point taken.
mdelay(10) doesn't really look good either to me in this case, though.
---
From: Rafael J. Wysocki <rjw@...k.pl>
Subject: PCI: Fix regression in pci_restore_state(), v2
Commit 26f41062f28de65e11d3cf353e52d0be73442be1
PCI: check for pci bar restore completion and retry
attempted to address problems with PCI BAR restoration on systems
where FLR had not been completed before pci_restore_state() was
called, but it did that in an utterly wrong way.
First off, instead of retrying the writes for the BAR registers
only, it did that for all of the PCI config space of the device,
including the status register (whose value after the write quite
obviously need not be the same as the written one). Second, it
added arbitrary delay to pci_restore_state() even for systems
where the PCI config space restoration was successful at first
attempt.
All of this actually caused resume failures for some devices on
the Mikko's system.
To fix the regression, make pci_restore_state() only retry the
writes for BAR registers and only wait if the first read from
the register doesn't return the written value.
Reported-by: Mikko Vinni <mmvinni@...oo.com>
Signed-off-by: Rafael J. Wysocki <rjw@...k.pl>
---
drivers/pci/pci.c | 50 ++++++++++++++++++++++++++++++++------------------
1 file changed, 32 insertions(+), 18 deletions(-)
Index: linux/drivers/pci/pci.c
===================================================================
--- linux.orig/drivers/pci/pci.c
+++ linux/drivers/pci/pci.c
@@ -967,16 +967,40 @@ pci_save_state(struct pci_dev *dev)
return 0;
}
+static void pci_restore_config_space(struct pci_dev *pdev, int start, int end,
+ int retry)
+{
+ int index;
+
+ for (index = end; index >= start; index--) {
+ u32 val, saved;
+ int offset = 4 * index;
+ int count = retry;
+
+ saved = pdev->saved_config_space[index];
+ pci_read_config_dword(pdev, offset, &val);
+ while (saved != val) {
+ dev_dbg(&pdev->dev, "restoring config space at offset "
+ "%#x (was %#x, writing %#x)\n",
+ offset, val, saved);
+ pci_write_config_dword(pdev, offset, saved);
+ if (count-- > 0) {
+ pci_read_config_dword(pdev, offset, &val);
+ if (saved != val)
+ mdelay(10);
+ } else {
+ break;
+ }
+ }
+ }
+}
+
/**
* pci_restore_state - Restore the saved state of a PCI device
* @dev: - PCI device that we're dealing with
*/
void pci_restore_state(struct pci_dev *dev)
{
- int i;
- u32 val;
- int tries;
-
if (!dev->state_saved)
return;
@@ -984,24 +1008,14 @@ void pci_restore_state(struct pci_dev *d
pci_restore_pcie_state(dev);
pci_restore_ats_state(dev);
+ pci_restore_config_space(dev, 10, 15, 0);
/*
* The Base Address register should be programmed before the command
* register(s)
*/
- for (i = 15; i >= 0; i--) {
- pci_read_config_dword(dev, i * 4, &val);
- tries = 10;
- while (tries && val != dev->saved_config_space[i]) {
- dev_dbg(&dev->dev, "restoring config "
- "space at offset %#x (was %#x, writing %#x)\n",
- i, val, (int)dev->saved_config_space[i]);
- pci_write_config_dword(dev,i * 4,
- dev->saved_config_space[i]);
- pci_read_config_dword(dev, i * 4, &val);
- mdelay(10);
- tries--;
- }
- }
+ pci_restore_config_space(dev, 4, 9, 10);
+ pci_restore_config_space(dev, 0, 3, 0);
+
pci_restore_pcix_state(dev);
pci_restore_msi_state(dev);
pci_restore_iov_state(dev);
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists