PCI/MSI: Prevent recursive locking in pci_msix_write_tph_tag()

pci_msix_write_tph_tag() takes the per device MSI descriptor mutex and then
invokes msi_domain_get_virq(), which takes the same mutex again. That
obviously results in a system hang which is exposed by a softlockup or
lockdep warning.

Move the lock guard after the invocation of msi_domain_get_virq() to fix
this.

[ tglx: Massage changelog by adding a proper explanation and removing the
  	not really useful stacktrace ]

Fixes: d5124a9957b2 ("PCI/MSI: Provide a sane mechanism for TPH")
Reported-by: Jorge Lopez <jorge.jo.lopez@oracle.com>
Suggested-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Jorge Lopez <jorge.jo.lopez@oracle.com>
Link: https://lore.kernel.org/all/20250708222530.1041477-1-himanshu.madhani@oracle.com

diff --git a/drivers/pci/msi/msi.c b/drivers/pci/msi/msi.c
index 6ede55a7c5e652c80b51b10e58f0290eb6556430..d686488f4111dc6f98e38fd82ce716caa2b3edfc 100644 (file)
--- a/drivers/pci/msi/msi.c
+++ b/drivers/pci/msi/msi.c
@@ -934,10 +934,12 @@ int pci_msix_write_tph_tag(struct pci_dev *pdev, unsigned int index, u16 tag)
        if (!pdev->msix_enabled)
                return -ENXIO;
 
-       guard(msi_descs_lock)(&pdev->dev);
        virq = msi_get_virq(&pdev->dev, index);
        if (!virq)
                return -ENXIO;
+
+       guard(msi_descs_lock)(&pdev->dev);
+
        /*
         * This is a horrible hack, but short of implementing a PCI
         * specific interrupt chip callback and a huge pile of