2006-10-11 21:48:12

by Jeff Garzik

Subject: [PATCH] drivers/{mmc,misc}: handle PCI errors on resume

Since pci_enable_device() is one of the first things called in the
resume step, take the minimalist approach and return immediately, if
pci_enable_device() fails during resume.

Also, in sdhci: don't check for impossible condition (chip==NULL)

Signed-off-by: Jeff Garzik <[email protected]>

---

drivers/misc/tifm_7xx1.c | 7 ++++++-
drivers/mmc/sdhci.c | 11 +++++------

diff --git a/drivers/misc/tifm_7xx1.c b/drivers/misc/tifm_7xx1.c
index 1ba8754..03988b9 100644
--- a/drivers/misc/tifm_7xx1.c
+++ b/drivers/misc/tifm_7xx1.c
@@ -273,9 +273,14 @@ static int tifm_7xx1_resume(struct pci_d
{
struct tifm_adapter *fm = pci_get_drvdata(dev);
unsigned long flags;
+ int rc;

pci_restore_state(dev);
- pci_enable_device(dev);
+
+ rc = pci_enable_device(dev);
+ if (rc)
+ return rc;
+
pci_set_power_state(dev, PCI_D0);
pci_set_master(dev);

diff --git a/drivers/mmc/sdhci.c b/drivers/mmc/sdhci.c
index 9a7d39b..b27fff9 100644
--- a/drivers/mmc/sdhci.c
+++ b/drivers/mmc/sdhci.c
@@ -1090,18 +1090,17 @@ static int sdhci_suspend (struct pci_dev

static int sdhci_resume (struct pci_dev *pdev)
{
- struct sdhci_chip *chip;
+ struct sdhci_chip *chip = pci_get_drvdata(pdev);
int i, ret;

- chip = pci_get_drvdata(pdev);
- if (!chip)
- return 0;
-
DBG("Resuming...\n");

pci_set_power_state(pdev, PCI_D0);
pci_restore_state(pdev);
- pci_enable_device(pdev);
+
+ ret = pci_enable_device(pdev);
+ if (ret)
+ return ret;

for (i = 0;i < chip->num_slots;i++) {
if (!chip->hosts[i])

2006-10-13 08:00:45

by Russell King

Subject: Re: [PATCH] drivers/{mmc,misc}: handle PCI errors on resume

On Wed, Oct 11, 2006 at 05:48:09PM -0400, Jeff Garzik wrote:
> Since pci_enable_device() is one of the first things called in the
> resume step, take the minimalist approach and return immediately, if
> pci_enable_device() fails during resume.

NAK, for the same reason as for drivers/serial.

--
Russell King
Linux kernel 2.6 ARM Linux - http://www.arm.linux.org.uk/
maintainer of: 2.6 Serial core

2006-10-21 10:24:45

by Pierre Ossman

Subject: Re: [PATCH] drivers/{mmc,misc}: handle PCI errors on resume

Russell King wrote:
> On Wed, Oct 11, 2006 at 05:48:09PM -0400, Jeff Garzik wrote:
>> Since pci_enable_device() is one of the first things called in the
>> resume step, take the minimalist approach and return immediately, if
>> pci_enable_device() fails during resume.
>
> NAK, for the same reason as for drivers/serial.
>

So what would be the prudent course of action in this case?

Rgds
--
-- Pierre Ossman

Linux kernel, MMC maintainer http://www.kernel.org
PulseAudio, core developer http://pulseaudio.org
rdesktop, core developer http://www.rdesktop.org