2012-11-15 22:03:46

by Christian Gmeiner

Subject: [PATCH] pata_cs5536: add quirk for broken udma

I am working on a device which uses the cs5536 pata driver. There
are some broken hardware revisions out in the field, which can be
detected via DMI. On older versions with an embedded BIOS I
used libata.dma=0 to disable dma completely.
Now we are switching to a coreboot/seabios based BIOS where we
have DMI support and so I think its a good idea to get rid of
all those hacky kernel parameters as the same image
is used other devices where libata.dma=0 is not a good idea.

Signed-off-by: Christian Gmeiner <[email protected]>
---
drivers/ata/pata_cs5536.c | 32 ++++++++++++++++++++++++++++++
+-
1 file changed, 31 insertions(+), 1 deletion(-)

diff --git a/drivers/ata/pata_cs5536.c b/drivers/ata/pata_cs5536.c
index dec1b6c..0448860 100644
--- a/drivers/ata/pata_cs5536.c
+++ b/drivers/ata/pata_cs5536.c
@@ -38,6 +38,7 @@
#include <linux/delay.h>
#include <linux/libata.h>
#include <scsi/scsi_host.h>
+#include <linux/dmi.h>

#ifdef CONFIG_X86_32
#include <asm/msr.h>
@@ -80,6 +81,21 @@ enum {
IDE_ETC_UDMA_MASK = 0xc0,
};

+/* Some Bachmann OT200 devices have a non working UDMA support due a
+ * missing resistor.
+ */
+static const struct dmi_system_id udma_quirk_dmi_table[] = {
+ {
+ .ident = "Bachmann electronic OT200",
+ .matches = {
+ DMI_MATCH(DMI_SYS_VENDOR, "Bachmann electronic"),
+ DMI_MATCH(DMI_PRODUCT_NAME, "OT200"),
+ DMI_MATCH(DMI_PRODUCT_VERSION, "1")
+ },
+ },
+ { }
+};
+
static int cs5536_read(struct pci_dev *pdev, int reg, u32 *val)
{
if (unlikely(use_msr)) {
@@ -242,9 +258,23 @@ static int cs5536_init_one(struct pci_dev *dev,
const struct pci_device_id *id)
.port_ops = &cs5536_port_ops,
};

- const struct ata_port_info *ppi[] = { &info, &ata_dummy_port_info };
+ static const struct ata_port_info no_udma_info = {
+ .flags = ATA_FLAG_SLAVE_POSS,
+ .pio_mask = ATA_PIO4,
+ .port_ops = &cs5536_port_ops,
+ };
+
+
+ const struct ata_port_info *ppi[2];
u32 cfg;

+ if (dmi_check_system(udma_quirk_dmi_table))
+ ppi[0] = &no_udma_info;
+ else
+ ppi[0] = &info;
+
+ ppi[1] = &ata_dummy_port_info;
+
if (use_msr)
printk(KERN_ERR DRV_NAME ": Using MSR regs instead of PCI\n");


--
1.7.12.2.421.g261b511

2012-11-16 04:23:00

by Jeff Garzik

Subject: Re: [PATCH] pata_cs5536: add quirk for broken udma

On 11/15/2012 05:03 PM, Christian Gmeiner wrote:
> I am working on a device which uses the cs5536 pata driver. There
> are some broken hardware revisions out in the field, which can be
> detected via DMI. On older versions with an embedded BIOS I
> used libata.dma=0 to disable dma completely.
> Now we are switching to a coreboot/seabios based BIOS where we
> have DMI support and so I think its a good idea to get rid of
> all those hacky kernel parameters as the same image
> is used other devices where libata.dma=0 is not a good idea.
>
> Signed-off-by: Christian Gmeiner <[email protected]>
> ---
> drivers/ata/pata_cs5536.c | 32 ++++++++++++++++++++++++++++++
> +-
> 1 file changed, 31 insertions(+), 1 deletion(-)

ACK, but patch is word-wrapped-mangled