sata_sil corruption / lockup fix

Silicon Image contributed a patch which should help some of the situations that users were seeing. If you are having problems with sata_sil, please do try out this patch.

I'm concerned that the sata_sil blacklist has been growing beyond the older Seagate drives which definitely had buggy firmware; concerned that the Mod15Write fix was simply "fixing" the problem addressed by this patch, simply by hiding the problem behind slow performance. [note: the only way to really know for sure is with ATA bus traces]

On platforms where the SiI BIOS isn't executed (non-x86), this patch is probably more critical. On x86, it is purported to only be needed on a single motherboard.

Test results (to [email protected]) would be appreciated, particularly from users with newer Seagate drives.

Finally, there are also a few reports of problems of "screaming interrupts" on configurations with SiI 311x + Seagate NCQ drives. This is a separate problem, and I haven't looked into it yet.

Jeff

#
# ChangeSet
# 2005/03/24 23:32:42-05:00 [email protected]
# [PATCH] sata_sil: Fix FIFO PCI Bus Arbitration
#

#   This patch set default values for the FIFO PCI Bus Arbitration to avoid
#   data corruption. The root cause is due to our PCI bus master handling
#   mismatch with the chipset PCI bridge during DMA xfer (write data to the
#   device). The patch is to setup the DMA fifo threshold so that there is
#   no chance for the DMA engine to change protocol. We have seen this
#   problem only on one motherboard.

#
# Signed-off-by: Silicon Image Corporation <[email protected]>
# Signed-off-by: Jeff Garzik <[email protected]>
#

diff -Nru a/drivers/scsi/sata_sil.c b/drivers/scsi/sata_sil.c
@@ -38,12 +38,21 @@
 #include <linux/libata.h>
 
 #define DRV_NAME	"sata_sil"
-#define DRV_VERSION	"0.8"
+#define DRV_VERSION	"0.9"
 
 enum {
 	sil_3112		= 0,
 	sil_3114		= 1,
 
+	SIL_FIFO_R0		= 0x40,
+	SIL_FIFO_W0		= 0x41,
+	SIL_FIFO_R1		= 0x44,
+	SIL_FIFO_W1		= 0x45,
+	SIL_FIFO_R2		= 0x240,
+	SIL_FIFO_W2		= 0x241,
+	SIL_FIFO_R3		= 0x244,
+	SIL_FIFO_W3		= 0x245,
+
 	SIL_SYSCFG		= 0x48,
 	SIL_MASK_IDE0_INT	= (1 << 22),
 	SIL_MASK_IDE1_INT	= (1 << 23),

@@ -199,6 +208,13 @@
 MODULE_DEVICE_TABLE(pci, sil_pci_tbl);
 MODULE_VERSION(DRV_VERSION);
 
+static unsigned char sil_get_device_cache_line(struct pci_dev *pdev)
+{
+	u8 cache_line = 0;
+	pci_read_config_byte(pdev, PCI_CACHE_LINE_SIZE, &cache_line);
+	return cache_line;
+}
+
 static void sil_post_set_mode (struct ata_port *ap)
 {
 	struct ata_host_set *host_set = ap->host_set;

@@ -341,6 +357,7 @@
 	unsigned int i;
 	int pci_dev_busy = 0;
 	u32 tmp, irq_mask;
+	u8 cls;
 
 	if (!printed_version++)
 		printk(KERN_DEBUG DRV_NAME " version " DRV_VERSION "\n");

@@ -404,6 +421,15 @@
 		probe_ent->port[i].scr_addr = base + sil_port[i].scr;
 		ata_std_ports(&probe_ent->port[i]);
 	}
+
+	/* Initialize FIFO PCI bus arbitration */
+	cls = sil_get_device_cache_line(pdev);
+	cls >>= 3;
+	cls++;  /* cls = (line_size/8)+1 */
+	writeb(cls, mmio_base + SIL_FIFO_R0);
+	writeb(cls, mmio_base + SIL_FIFO_W0);
+	writeb(cls, mmio_base + SIL_FIFO_R1);
+	writeb(cls, mmio_base + SIL_FIFO_W2);
 
 	if (ent->driver_data == sil_3114) {
 		irq_mask = SIL_MASK_4PORT;

Hello, Jeff!

Yes, you are right that the list grew too much.
I, for example, have a sil 3112 controller that made problems (mod15) in the past (crash on big tranfers). But I upgraded the BIOS (and it shows me that even sil firmware was updated) and the problem disappeard.

So, I'll be glad to test this patch, but I can't because now it works very good (I manualy remove the disks from blaklist because they have a match in the list).

Thanks goes to Silicon Image people that, finally, they release a patch to fix this problem.

My motherboard is Intel.

Thanks Jeff.

---
Catalin(ux aka Dino) BOIE
catab at deuroconsult.ro
http://kernel.umbrella.ro/