In 2009, While running "cache read" performance test of drives behind
SII PMP we encountered a "all 5 drives" timeout on more than 30% of the
machines under test. This patch reduces the rate by a factor of about 70.
Low enough that we didn't care to further investigate the issue.
Performance impact with any sort of "normal" use was ~2%+ CPU and less
than 1% throughput degradation. Worst case impact (cached read) was
6% IOPS reduction. This is with NCQ off (q=1) but I believe FIS based
switching enabled in the SATA driver.
The patch disables "Early ACK" in the 3726/4726 port multiplier.
"Early ACK" is issued when device sends a FIS to the host (via PMP)
and the PMP sends an ACK immediately back to the device - well before
the host gets the response. Under worst case IOPs load (cached read
test) and more than 2 PMPs connected to a 4-port SATA controller,
I suspect the time to service all of the PMPs is exceeding the PMPs
ability to keep track of outstanding FIS it owes the Host. Reducing
the number of PMPs to 2 (or 1) reduces the frequency by several orders
of magnitude. Kudos to Gwendal for initial debugging of this issue.
[Any errors in the description are mine, not his.]
Patch is currently in production on Google servers.
Signed-off-by: Grant Grundler <[email protected]>
Signed-off-by: Gwendal Grignou <[email protected]>
---
And my apologies in advance for using gmail. Patch appended below
is also attached since I expect gmail will mangle the inlined content.
diff --git a/drivers/ata/libata-pmp.c b/drivers/ata/libata-pmp.c
index 00305f4..69bb203 100644
--- a/drivers/ata/libata-pmp.c
+++ b/drivers/ata/libata-pmp.c
@@ -235,6 +235,8 @@ static int sata_pmp_configure(struct ata_device
*dev, int print_info)
{
struct ata_port *ap = dev->link->ap;
u32 *gscr = dev->gscr;
+ u16 vendor = sata_pmp_gscr_vendor(gscr);
+ u16 devid = sata_pmp_gscr_devid(gscr);
unsigned int err_mask = 0;
const char *reason;
int nr_ports, rc;
@@ -260,12 +262,28 @@ static int sata_pmp_configure(struct ata_device
*dev, int print_info)
goto fail;
}
+ /* For Sil3726, disable sending Early R_OK */
+ if (vendor == 0x1095 && devid == 0x3726) {
+ u32 reg;
+ err_mask = sata_pmp_read(&ap->link,
SATA_PMP_GSCR_SII_POL, ®);
+ if (err_mask) {
+ rc = -EIO;
+ reason = "failed to read Sil3726 Private Register";
+ goto fail;
+ }
+ reg &= ~0x1;
+ err_mask = sata_pmp_write(&ap->link,
SATA_PMP_GSCR_SII_POL, reg);
+ if (err_mask) {
+ rc = -EIO;
+ reason = "failed to write Sil3726 Private Register";
+ goto fail;
+ }
+ }
+
if (print_info) {
ata_dev_printk(dev, KERN_INFO, "Port Multiplier %s, "
"0x%04x:0x%04x r%d, %d ports, feat 0x%x/0x%x\n",
- sata_pmp_spec_rev_str(gscr),
- sata_pmp_gscr_vendor(gscr),
- sata_pmp_gscr_devid(gscr),
+ sata_pmp_spec_rev_str(gscr), vendor, devid,
sata_pmp_gscr_rev(gscr),
nr_ports, gscr[SATA_PMP_GSCR_FEAT_EN],
gscr[SATA_PMP_GSCR_FEAT]);
diff --git a/include/linux/ata.h b/include/linux/ata.h
index 700c5b9..18fc208 100644
--- a/include/linux/ata.h
+++ b/include/linux/ata.h
@@ -386,6 +386,7 @@ enum {
SATA_PMP_GSCR_ERROR_EN = 33,
SATA_PMP_GSCR_FEAT = 64,
SATA_PMP_GSCR_FEAT_EN = 96,
+ SATA_PMP_GSCR_SII_POL = 129, /* VENDOR : Silicon Image 3726/4726 */
SATA_PMP_PSCR_STATUS = 0,
SATA_PMP_PSCR_ERROR = 1,
Hello,
On 04/14/2010 09:46 AM, Grant Grundler wrote:
> In 2009, While running "cache read" performance test of drives behind
> SII PMP we encountered a "all 5 drives" timeout on more than 30% of the
> machines under test. This patch reduces the rate by a factor of about 70.
> Low enough that we didn't care to further investigate the issue.
Interesting.
> Performance impact with any sort of "normal" use was ~2%+ CPU and less
> than 1% throughput degradation. Worst case impact (cached read) was
> 6% IOPS reduction. This is with NCQ off (q=1) but I believe FIS based
> switching enabled in the SATA driver.
Given the general flakiness of the device, I'm more than willing to
pay some overhead for stability and the performance hit seems pretty
small.
> @@ -260,12 +262,28 @@ static int sata_pmp_configure(struct ata_device
> *dev, int print_info)
> goto fail;
> }
>
> + /* For Sil3726, disable sending Early R_OK */
> + if (vendor == 0x1095 && devid == 0x3726) {
Isn't this also applicable to 4723 and 4726?
> diff --git a/include/linux/ata.h b/include/linux/ata.h
> index 700c5b9..18fc208 100644
> --- a/include/linux/ata.h
> +++ b/include/linux/ata.h
> @@ -386,6 +386,7 @@ enum {
> SATA_PMP_GSCR_ERROR_EN = 33,
> SATA_PMP_GSCR_FEAT = 64,
> SATA_PMP_GSCR_FEAT_EN = 96,
> + SATA_PMP_GSCR_SII_POL = 129, /* VENDOR : Silicon Image 3726/4726 */
Can you plesae define it inside libata-pmp.c? ata.h only contains
stuff defined by the ATA standard.
Other than the above nitpicks,
Acked-by: Tejun Heo <[email protected]>
Thanks.
--
tejun
On Wed, Apr 14, 2010 at 5:31 PM, Tejun Heo <[email protected]> wrote:
> Hello,
>
> On 04/14/2010 09:46 AM, Grant Grundler wrote:
>> In 2009, While running "cache read" performance test of drives behind
>> SII PMP we encountered a "all 5 drives" timeout on more than 30% of the
>> machines under test. This patch reduces the rate by a factor of about 70.
>> Low enough that we didn't care to further investigate the issue.
>
> Interesting.
>
>> Performance impact with any sort of "normal" use was ~2%+ CPU and less
>> than 1% throughput degradation. Worst case impact (cached read) was
>> 6% IOPS reduction. This is with NCQ off (q=1) but I believe FIS based
>> switching enabled in the SATA driver.
>
> Given the general flakiness of the device, I'm more than willing to
> pay some overhead for stability and the performance hit seems pretty
> small.
>
>> @@ -260,12 +262,28 @@ static int sata_pmp_configure(struct ata_device
>> *dev, int print_info)
>> goto fail;
>> }
>>
>> + /* For Sil3726, disable sending Early R_OK */
>> + if (vendor == 0x1095 && devid == 0x3726) {
>
> Isn't this also applicable to 4723 and 4726?
Sorry, I don't know why I included 4726 in the Subject line since I
only tested with 3726.
I expect it would apply to 4726 but someone who knows the internals of
both would have to confirm that.
I don't know anything about 4723.
>
>> diff --git a/include/linux/ata.h b/include/linux/ata.h
>> index 700c5b9..18fc208 100644
>> --- a/include/linux/ata.h
>> +++ b/include/linux/ata.h
>> @@ -386,6 +386,7 @@ enum {
>> SATA_PMP_GSCR_ERROR_EN = 33,
>> SATA_PMP_GSCR_FEAT = 64,
>> SATA_PMP_GSCR_FEAT_EN = 96,
>> + SATA_PMP_GSCR_SII_POL = 129, /* VENDOR : Silicon Image 3726/4726 */
>
> Can you plesae define it inside libata-pmp.c?
Yes. I'll resubmit with "SII_POL" defined "locally".
> ata.h only contains stuff defined by the ATA standard.
Ah ok. I didn't realize that.
>
> Other than the above nitpicks,
>
> Acked-by: Tejun Heo <[email protected]>
Thanks for the Ack/Review!
grant
>
> Thanks.
>
> --
> tejun
>