On 02/15/2017 06:56 PM, Bjorn Helgaas wrote:
> On Wed, Feb 15, 2017 at 11:53:08AM -0500, Prarit Bhargava wrote:
>> No response ... trying again.
>>
>> P.
>>
>> ---8<---
>>
>> During device setup, msix_setup_entries() and msi_setup_entry() allocate
>> msi_desc by calling alloc_msi_entry(). alloc_msi_entry() can also allocate a
>> affinity cpumask. During device teardown free_msi_irqs() is called and the
>> msi_desc is freed, but the affinity cpumask is leaked.
>>
>> Fix it by calling free_msi_entry() which frees both the msi_desc and the
>> affinity cpumask.
>
> Sorry for the delay, Prarit. I started looking at this when you first
> posted it. The more I looked at it, the less I liked the setup/free
> paths -- they're not very parallel and I don't remember what else.
>
Yes, I noticed this as well. As you say ...
> But I guess it's probably a longer term effort to clean that up.
>
... there's a larger cleanup that's required here.
> Do you have a bug report or anything for this, or did you just find
> this by code inspection? If there's a bugzilla or even a symptom, we
> could mention it in the changelog.
>
This was noticed during code inspection while Myron & I were debugging
an issue in RHEL7.
P.
>> Signed-off-by: Prarit Bhargava <prarit@xxxxxxxxxx>
>> Cc: mstowe@xxxxxxxxxx
>> Cc: Bjorn Helgaas <bhelgaas@xxxxxxxxxx>
>> ---
>> drivers/pci/msi.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c
>> index 50c5003295ca..3d709311052d 100644
>> --- a/drivers/pci/msi.c
>> +++ b/drivers/pci/msi.c
>> @@ -379,7 +379,7 @@ static void free_msi_irqs(struct pci_dev *dev)
>> }
>>
>> list_del(&entry->list);
>> - kfree(entry);
>> + free_msi_entry(entry);
>> }
>>
>> if (dev->msi_irq_groups) {
>> --
>> 1.7.9.3
>>
