Re: [PATCH v20 4/4] PCI/MSI: Enable memory decoding before restoring MSI-X messages

[Farhan Ali <alifm@linux.ibm.com>]

On 6/22/2026 1:29 PM, Thomas Gleixner wrote:
> On Mon, Jun 22 2026 at 10:18, Farhan Ali wrote:
>> The current MSI-X restoration path assumes the Command register Memory bit
>> is enabled when writing MSI-X messages. But its possible the last saved and
>> restored state of device may not have the Memory bit enabled, even if a
>> device driver later enables Memory bit and MSI-X. Attempting to access
>> Memory space without Memory bit enabled can lead to Unsupported Request
>> (UR) from the device. Fix this by enabling Memory bit and restore
>> it afterwards.
>>
>> Signed-off-by: Farhan Ali <alifm@linux.ibm.com>
>> ---
>>   drivers/pci/msi/msi.c | 4 ++++
>>   1 file changed, 4 insertions(+)
>>
>> diff --git a/drivers/pci/msi/msi.c b/drivers/pci/msi/msi.c
>> index 81d24a270a79..46a0d9f68a57 100644
>> --- a/drivers/pci/msi/msi.c
>> +++ b/drivers/pci/msi/msi.c
>> @@ -874,6 +874,7 @@ void __pci_restore_msix_state(struct pci_dev *dev)
>>   {
>>   	struct msi_desc *entry;
>>   	bool write_msg;
>> +	u16 cmd;
>>   
>>   	if (!dev->msix_enabled)
>>   		return;
>> @@ -882,6 +883,8 @@ void __pci_restore_msix_state(struct pci_dev *dev)
>>   	pci_intx_for_msi(dev, 0);
>>   	pci_msix_clear_and_set_ctrl(dev, 0,
>>   				PCI_MSIX_FLAGS_ENABLE | PCI_MSIX_FLAGS_MASKALL);
>> +	pci_read_config_word(dev, PCI_COMMAND, &cmd);
>> +	pci_write_config_word(dev, PCI_COMMAND, cmd | PCI_COMMAND_MEMORY);
> Can we please have a comment there which explains this? Three month down
> the road this will results in head scratching otherwise.
>
> I agree with Niklas that this wants a Fixes and a Cc:stable tag.
>
> Other than that:
>
> Reviewed-by: Thomas Gleixner <tglx@kernel.org>
>
> Thanks,
>
>          tglx
>
I can add a comment, how about something like below?

"The restored device state may not have Memory decoding enabled in 
Command register. Since the MSI-X was enabled for the device, enable 
Memory decoding before restoring MSI-X"

For the Fixes tag, do you have a suggestion for a commit? This behavior 
has been present since 41017f0cac92 ("[PATCH] PCI: MSI(X) save/restore 
for suspend/resume" which introduced these restore functions. So should 
be Fixes against 41017f0cac92?

Thanks

Farhan