Comment 5 for bug 1973167

Alas! I spoke to seen. The seeming "workaround" only lasted for one boot, and after that, the ~180 seconds delay is back to stay, although the Intel IOMMU is still disabled.

So this issue is "sporadic". Maybe even a race condition... :(

And maybe not even related to the intel-lpss driver: The line:

"intel-lpss 0000:00:15.0: enabling device (0000 -> 0002)"

seems to come from drivers/pci/setup-res.c:

 if (cmd != old_cmd) {
  dev_info(&dev->dev, "enabling device (%04x -> %04x)\n",
    old_cmd, cmd);
  pci_write_config_word(dev, PCI_COMMAND, cmd);
 }

So it just enables the PCI device, which at some point leads to drivers/mfd/intel-lpss.c#intel_lpss_probe() being called which requests the DMA module leading to drivers/dma/idma64.c#idma64_probe() being called which finally outputs:

dev_info(chip->dev, "Found Intel integrated DMA 64-bit\n");

So a lot of code between these two log lines:

[ 6.439056] intel-lpss 0000:00:15.0: enabling device (0000 -> 0002)
[ 187.141427] idma64 idma64.0: Found Intel integrated DMA 64-bit