On 08/02/2018 11:07 AM, Lukas Wunner wrote:
[cc += Thomas Tai]
Hi Lukas,
Thank you very much for cc me.
On Thu, Aug 02, 2018 at 10:46:57AM +0200, Lukas Wunner wrote:
On Thu, Aug 02, 2018 at 12:59:18PM +0530, gokul cg wrote:
I am suspecting a possible race condition in the kernel between PCI driver
and AER handling.
The solution is to acquire a ref on each device in add_error_device().
Then release the ref aer_process_err_devices() by calling pci_dev_put().
So in case it wasn't clear, the below is what I had in mind.
Completely untested though. Does this work for you?
For v3.10 compatibility, cherry-pick 89ee9f768003 (or alternatively
cherry-pick 8496e85c20e7 and replace pci_dev_is_disconnected(dev)
with !pci_device_is_present(dev)).
-- >8 --
Subject: [PATCH] PCI/AER: Fix use-after-free on surprise removal
The work item to consume errors, aer_isr(), walks the hierarchy using
pci_walk_bus() and stores a pointer to PCI devices which reported an
error in an array. As long as pci_walk_bus() runs, those pointers are
valid because pci_bus_sem is held. But once pci_walk_bus() finishes,
nothing prevents the pointers from becoming invalid, e.g. through
unplugging of the PCI devices. The unprotected pointers are then
dereferenced in aer_process_err_devices(), which may oops:
I like your idea to increment the refcount during pci_walk_bus(), that
should fix the use-after-free issue. We just need Gokul to confirm if it
fixes his issue or not.
Thanks,
Thomas
#5 general_protection at ffffffff8176cdf2
[exception RIP: pci_bus_read_config_dword+100]
#6 pci_find_next_ext_capability at ffffffff81345d7b
#7 pci_find_ext_capability at ffffffff81347225
#8 get_device_error_info at ffffffff81356c4d
#9 aer_isr at ffffffff81357a38
Fix by holding a ref on the devices until they have been processed.
Skip processing of unplugged devices.
Reported-by: gokul cg <gokuljnpr@xxxxxxxxx>
Signed-off-by: Lukas Wunner <lukas@xxxxxxxxx>
---
drivers/pci/pcie/aer.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/pci/pcie/aer.c b/drivers/pci/pcie/aer.c
index a2e8838..937592e 100644
--- a/drivers/pci/pcie/aer.c
+++ b/drivers/pci/pcie/aer.c
@@ -657,7 +657,7 @@ void cper_print_aer(struct pci_dev *dev, int aer_severity,
static int add_error_device(struct aer_err_info *e_info, struct pci_dev *dev)
{
if (e_info->error_dev_num < AER_MAX_MULTI_ERR_DEVICES) {
- e_info->dev[e_info->error_dev_num] = dev;
+ e_info->dev[e_info->error_dev_num] = pci_dev_get(dev);
e_info->error_dev_num++;
return 0;
}
@@ -898,6 +898,9 @@ static int get_device_error_info(struct pci_dev *dev, struct aer_err_info *info)
if (!pos)
return 0;
+ if (pci_dev_is_disconnected(dev))
+ return 0;
+
if (info->severity == AER_CORRECTABLE) {
pci_read_config_dword(dev, pos + PCI_ERR_COR_STATUS,
&info->status);
@@ -948,6 +951,7 @@ static inline void aer_process_err_devices(struct aer_err_info *e_info)
for (i = 0; i < e_info->error_dev_num && e_info->dev[i]; i++) {
if (get_device_error_info(e_info->dev[i], e_info))
handle_error_source(e_info->dev[i], e_info);
+ pci_dev_put(e_info->dev[i]);
}
}
