lkml.org 
[lkml]   [2013]   [Mar]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Subject[PATCH v5 0/5] Fix aer_inject tool bug
Date
Hi Tony,
Can you pick up these bug fix patches into your ras tree? I sent them to PCI subsystem maillist, but Bjorn does not seem
very interested in these aer_inject tool bug fix patches. Huang Ying <ying.huang@intel.com> is the author of aer_inject tool.
And I had a long discussion with Huang Ying about these patches, and got his reviewed-by. I do not know what people are really
interested in these patches, So I try to send them to you. Thank you very much!

This series of patch mainly to fix the aer_inject bug described as below:

-+-[0000:40]-+-00.0-[0000:41]--
| +-01.0-[0000:42]--+-00.0 Intel Corporation 82576 Gigabit Network Connection
| | \-00.1 Intel Corporation 82576 Gigabit Network Connection
| +-03.0-[0000:43]----00.0 LSI Logic / Symbios Logic SAS1064ET PCI-Express Fusion-MPT SAS
| +-04.0-[0000:44]--
| +-05.0-[0000:45]--
| +-07.0-[0000:46-49]----00.0-[0000:47-49]--+-02.0-[0000:48]--+-00.0 Intel Corporation 82576 Gigabit Network Connection
| | | \-00.1 Intel Corporation 82576 Gigabit Network Connection
| | \-04.0-[0000:49]--+-00.0 Intel Corporation 82576 Gigabit Network Connection
| | \-00.1 Intel Corporation 82576 Gigabit Network Connection

my steps:
1)modprobe aer_inject
2)inject aer errors to pcie device 0000:48:00.0
3)modprobe pciehp
4)hot remove Network Card in slot(port 0000:40:07.0)
5)hot add Network Card in slot(port 0000:40:07.0)
6)system panic

in step 2) the pci_ops of bus 0000:48 and bus 0000:40 will be assigned to pci_ops_aer
in step 5) the pci_ops of the newly created bus 0000:46 will be assigned to pci_ops_aer(inherited by parent pci_ops),
but this pci_ops(0000:46) is not tracked in pci_bus_ops_list in aer_inject module. So every access to pci_config space
by pci_ops of 0000:46 will cause system panic, Since pci_ops_aer cannot find its original pci_ops, thus , a NULL pci_ops return;

The first patch fix this bug by finding parent pci_ops(tracked in pci_ops_list) instead of returning NULL in step 5);
The second patch fix a small race condition window in aer_inject_exit;
The Third patch to find and clean all untracked pci_ops_aer in system when aer_inject module exit
The rest two patch mainly about to clean bus_ops;

Yijing Wang (5):
PCI/AER: Fix pci_ops return NULL in pci_read/write_aer
PCI/AER: use list_for_each_entry to avoid a small race condition
window
PCI/AER: clean all untracked pci_ops_aer when rmmod aer_inject
PCI/AER: clean pci_bus_ops when related pci bus was removed
PCI/AER: free pci_bus_ops_list and remove pci_bus_ops_pop

drivers/pci/pcie/aer/aer_inject.c | 123 +++++++++++++++++++++++++++++++------
1 files changed, 103 insertions(+), 20 deletions(-)




\
 
 \ /
  Last update: 2013-03-26 04:21    [W:0.042 / U:0.744 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site