On Thu, Jul 06, 2006 at 09:21:39AM +0800, Zhang, Yanmin wrote:On Thu, 2006-07-06 at 03:44, Linas Vepstas wrote:On Wed, Jul 05, 2006 at 08:49:27AM -0700, Auke Kok wrote:In the future, we might move the pci error recovery codes to generic toZhang, Yanmin wrote:On the pSeries, its harmless to try to do i/o; the i/o will e blocked.On Fri, 2006-06-30 at 00:26, Linas Vepstas wrote:Moreover, it was Linas who wrote this documentation in the first place :)Adds PCI Error recovery callbacks to the Intel 10-gigabit ethernetBoth pci_disable_device and ixgb_down would access the device. It doesn't
ixgb device driver. Lightly tested, works.
follow Documentation/pci-error-recovery.txt that error_detected shouldn't do
any access to the device.
support other platforms which might not block I/O. So it's better to follow
Documentation/pci-error-recovery.txt when adding error recovery codes into driver.
Or we could change the documentation. The point was that doing
unexpected i/o after the aapter reset is likely to wedge the adapter
again, leading to an inf loop of resets. As a practical matter, I found that, while developing this patch, and the other related
patches, that this was indeed the usual failure mode: incorrect bringup
just lead to more errors.
What I really want to do is to perform as clean a shut-down as possible, reset the adapter, and then bring it back up. I'm concerned that changing the order to "reset"-"shutdown-"bringup" would be inappropriate.
Perhaps the right fix is to figure out what parts of the driver do i/o
during shutdown, and then add a line "if(wedged) skip i/o;" to those
places?