[libata] ahci: SB600 workaround is suspect... play it safe for now

At least one report claims that a878539ef994787c447a98c2e3ba0fe3dad984ec
failed to solve lockups, whereas the old limit-to-32-bit trick worked.

Restore the 32-bit limit, but also leave the 255-sector limit in place,
because we know that's needed as well.

Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
index 17ee6ed985d9198ad356a9b68efebf406a4b11f6..b1eb4e24c86a90a4382be0fd21cbc2765e14ef96 100644 (file)
--- a/drivers/ata/ahci.c
+++ b/drivers/ata/ahci.c
@@ -433,6 +433,7 @@ static const struct ata_port_info ahci_port_info[] = {
        /* board_ahci_sb600 */
        {
                AHCI_HFLAGS     (AHCI_HFLAG_IGN_SERR_INTERNAL |
+                                AHCI_HFLAG_32BIT_ONLY |
                                 AHCI_HFLAG_SECT255 | AHCI_HFLAG_NO_PMP),
                .flags          = AHCI_FLAG_COMMON,
                .link_flags     = AHCI_LFLAG_COMMON,
@@ -1217,8 +1218,11 @@ static void ahci_dev_config(struct ata_device *dev)
 {
        struct ahci_host_priv *hpriv = dev->link->ap->host->private_data;
 
-       if (hpriv->flags & AHCI_HFLAG_SECT255)
+       if (hpriv->flags & AHCI_HFLAG_SECT255) {
                dev->max_sectors = 255;
+               ata_dev_printk(dev, KERN_INFO,
+                              "SB600 AHCI: limiting to 255 sectors per cmd\n");
+       }
 }
 
 static unsigned int ahci_dev_classify(struct ata_port *ap)