ALSA: hda - Force polling mode on CFL for fixing codec communication

We've observed too long probe time with Coffee Lake (CFL) machines,
and the likely cause is some communication problem between the
HD-audio controller and the codec chips.  While the controller expects
an IRQ wakeup for each codec response, it seems sometimes missing, and
it takes one second for the controller driver to time out and read the
response in the polling mode.

Although we aren't sure about the real culprit yet, in this patch, we
put a workaround by forcing the polling mode as default for CFL
machines; the polling mode itself isn't too heavy, and much better
than other workarounds initially suggested (e.g. disabling
power-save), at least.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=199007
Fixes: e79b0006c45c ("ALSA: hda - Add Coffelake PCI ID")
Reported-and-tested-by: Hui Wang <hui.wang@canonical.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>

diff --git a/sound/pci/hda/hda_intel.c b/sound/pci/hda/hda_intel.c
index d5017adf9febc4f591e54ef26447603975da8282..c507c69029e31f9fe8715c384dab8f98114db7bb 100644 (file)
--- a/sound/pci/hda/hda_intel.c
+++ b/sound/pci/hda/hda_intel.c
@@ -375,6 +375,7 @@ enum {
                                        ((pci)->device == 0x160c))
 
 #define IS_BXT(pci) ((pci)->vendor == 0x8086 && (pci)->device == 0x5a98)
+#define IS_CFL(pci) ((pci)->vendor == 0x8086 && (pci)->device == 0xa348)
 
 static char *driver_short_names[] = {
        [AZX_DRIVER_ICH] = "HDA Intel",
@@ -1744,6 +1745,10 @@ static int azx_create(struct snd_card *card, struct pci_dev *pci,
        else
                chip->bdl_pos_adj = bdl_pos_adj[dev];
 
+       /* Workaround for a communication error on CFL (bko#199007) */
+       if (IS_CFL(pci))
+               chip->polling_mode = 1;
+
        err = azx_bus_init(chip, model[dev], &pci_hda_io_ops);
        if (err < 0) {
                kfree(hda);