Linux-ACPI Archive on lore.kernel.org
 help / color / Atom feed
* [PATCH v4 12/12] PCI: Remove '*val = 0' from pcie_capability_read_*()
       [not found] <20200731114329.100848-1-refactormyself@gmail.com>
@ 2020-07-31 11:43 ` Saheed O. Bolarinwa
  0 siblings, 0 replies; only message in thread
From: Saheed O. Bolarinwa @ 2020-07-31 11:43 UTC (permalink / raw)
  To: helgaas
  Cc: Saheed O. Bolarinwa, bjorn, skhan, linux-kernel-mentees,
	linux-pci, linux-kernel, linux-rdma, QCA ath9k Development,
	linux-wireless, netdev, linux-acpi, linuxppc-dev,
	Mike Marciniszyn, Dennis Dalessandro, Doug Ledford,
	Jason Gunthorpe, Arnd Bergmann, Greg Kroah-Hartman,
	David S. Miller, Kalle Valo, Jakub Kicinski, Rafael J. Wysocki,
	Len Brown, Russell Currey, Sam Bobroff, Oliver O'Halloran

There are several reasons why a PCI capability read may fail whether the
device is present or not. If this happens, pcie_capability_read_*() will
return -EINVAL/PCIBIOS_BAD_REGISTER_NUMBER or PCIBIOS_DEVICE_NOT_FOUND
and *val is set to 0.

This behaviour if further ensured by this code inside
pcie_capability_read_*()

 ret = pci_read_config_dword(dev, pci_pcie_cap(dev) + pos, val);
 /*
  * Reset *val to 0 if pci_read_config_dword() fails, it may
  * have been written as 0xFFFFFFFF if hardware error happens
  * during pci_read_config_dword().
  */
 if (ret)
	 *val = 0;
 return ret;

a) Since all pci_generic_config_read() does is read a register value,
it may return success after reading a ~0 which *may* have been fabricated
by the PCI host bridge due to a read timeout. Hence pci_read_config_*()
will return success with a fabricated ~0 in *val, indicating a problem.
In this case, the assumed behaviour of  pcie_capability_read_*() will be
wrong. To avoid error slipping through, more checks are necessary.

b) pci_read_config_*() will return PCIBIOS_DEVICE_NOT_FOUND only if
dev->error_state = pci_channel_io_perm_failure (i.e.
pci_dev_is_disconnected()) or if pci_generic_config_read() can't find the
device. In both cases *val is initially set to ~0 but as shown in the code
above pcie_capability_read_*() resets it back to 0. Even with this effort,
drivers still have to perform validation checks more so if 0 is a valid
value.

Most drivers only consider the case (b) and in some cases, there is the
expectation that on timeout *val has a fabricated value of ~0, which *may*
not always be true as explained in (a).

In any case, checks need to be done to validate the value read and maybe
confirm which error has occurred. It is better left to the drivers to do.

Remove the reset of *val to 0 when pci_read_config_*() fails.

Suggested-by: Bjorn Helgaas <bjorn@helgaas.com>
Signed-off-by: Saheed O. Bolarinwa <refactormyself@gmail.com>
---
 drivers/pci/access.c | 14 --------------
 1 file changed, 14 deletions(-)

diff --git a/drivers/pci/access.c b/drivers/pci/access.c
index 79c4a2ef269a..ec95edbb1ac8 100644
--- a/drivers/pci/access.c
+++ b/drivers/pci/access.c
@@ -413,13 +413,6 @@ int pcie_capability_read_word(struct pci_dev *dev, int pos, u16 *val)
 
 	if (pcie_capability_reg_implemented(dev, pos)) {
 		ret = pci_read_config_word(dev, pci_pcie_cap(dev) + pos, val);
-		/*
-		 * Reset *val to 0 if pci_read_config_word() fails, it may
-		 * have been written as 0xFFFF if hardware error happens
-		 * during pci_read_config_word().
-		 */
-		if (ret)
-			*val = 0;
 		return ret;
 	}
 
@@ -448,13 +441,6 @@ int pcie_capability_read_dword(struct pci_dev *dev, int pos, u32 *val)
 
 	if (pcie_capability_reg_implemented(dev, pos)) {
 		ret = pci_read_config_dword(dev, pci_pcie_cap(dev) + pos, val);
-		/*
-		 * Reset *val to 0 if pci_read_config_dword() fails, it may
-		 * have been written as 0xFFFFFFFF if hardware error happens
-		 * during pci_read_config_dword().
-		 */
-		if (ret)
-			*val = 0;
 		return ret;
 	}
 
-- 
2.18.4


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, back to index

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <20200731114329.100848-1-refactormyself@gmail.com>
2020-07-31 11:43 ` [PATCH v4 12/12] PCI: Remove '*val = 0' from pcie_capability_read_*() Saheed O. Bolarinwa

Linux-ACPI Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/linux-acpi/0 linux-acpi/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 linux-acpi linux-acpi/ https://lore.kernel.org/linux-acpi \
		linux-acpi@vger.kernel.org
	public-inbox-index linux-acpi

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.kernel.vger.linux-acpi


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git