[PATCH 10/13] tg3: Increase the PCI MRRS

[PATCH 10/13] tg3: Increase the PCI MRRS

[Matt Carlson]

Previous devices hardcoded the PCI Maximum Read Request Size to 4K.  To
better comply with the PCI spec, the hardware now defaults the MRRS to
512 bytes.  This will yield poor driver performance if left untouched.
This patch increases the MRRS to 4K on driver initialization.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>

diff --git a/drivers/net/tg3.c b/drivers/net/tg3.c
index ecd64a2..72db78b 100644
--- a/drivers/net/tg3.c
+++ b/drivers/net/tg3.c
@@ -5098,12 +5098,15 @@ static void tg3_restore_pci_state(struct tg3 *tp)
 
 	pci_write_config_word(tp->pdev, PCI_COMMAND, tp->pci_cmd);
 
-	if (!(tp->tg3_flags2 & TG3_FLG2_PCI_EXPRESS)) {
+	if (tp->tg3_flags2 & TG3_FLG2_PCI_EXPRESS)
+		pcie_set_readrq(tp->pdev, 4096);
+	else {
 		pci_write_config_byte(tp->pdev, PCI_CACHE_LINE_SIZE,
 				      tp->pci_cacheline_sz);
 		pci_write_config_byte(tp->pdev, PCI_LATENCY_TIMER,
 				      tp->pci_lat_timer);
 	}
+
 	/* Make sure PCI-X relaxed ordering bit is clear. */
 	if (tp->pcix_cap) {
 		u16 pcix_cmd;
@@ -11215,6 +11218,9 @@ static int __devinit tg3_get_invariants(struct tg3 *tp)
 	pcie_cap = pci_find_capability(tp->pdev, PCI_CAP_ID_EXP);
 	if (pcie_cap != 0) {
 		tp->tg3_flags2 |= TG3_FLG2_PCI_EXPRESS;
+
+		pcie_set_readrq(tp->pdev, 4096);
+
 		if (GET_ASIC_REV(tp->pci_chip_rev_id) == ASIC_REV_5906) {
 			u16 lnkctl;
 

Re: [PATCH 10/13] tg3: Increase the PCI MRRS

[David Miller]

From: "Matt Carlson" <mcarlson@broadcom.com>
Date: Fri, 09 Nov 2007 16:39:01 -0800

> Previous devices hardcoded the PCI Maximum Read Request Size to 4K.  To
> better comply with the PCI spec, the hardware now defaults the MRRS to
> 512 bytes.  This will yield poor driver performance if left untouched.
> This patch increases the MRRS to 4K on driver initialization.
> 
> Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
> Signed-off-by: Michael Chan <mchan@broadcom.com>

I've applied this patch, but...

I sense that the PCI spec wants devices to use an MRRS value of 512 in
order to get better fairness on a PCI-E segment amongst multiple
devices.

>From that perspective, jacking up the MRRS to 4096 unilaterally seems
like a very bad idea.  If this was necessary for good performance, I'm
sure the PCI spec folks would have choosen a higher value.

Or is this some tg3 specific performance issue?

Re: [PATCH 10/13] tg3: Increase the PCI MRRS

[Matt Carlson]

On Mon, 2007-11-12 at 21:21 -0800, David Miller wrote:
> From: "Matt Carlson" <mcarlson@broadcom.com>
> Date: Fri, 09 Nov 2007 16:39:01 -0800
> 
> > Previous devices hardcoded the PCI Maximum Read Request Size to 4K.  To
> > better comply with the PCI spec, the hardware now defaults the MRRS to
> > 512 bytes.  This will yield poor driver performance if left untouched.
> > This patch increases the MRRS to 4K on driver initialization.
> > 
> > Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
> > Signed-off-by: Michael Chan <mchan@broadcom.com>
> 
> I've applied this patch, but...
> 
> I sense that the PCI spec wants devices to use an MRRS value of 512 in
> order to get better fairness on a PCI-E segment amongst multiple
> devices.
> 
> From that perspective, jacking up the MRRS to 4096 unilaterally seems
> like a very bad idea.  If this was necessary for good performance, I'm
> sure the PCI spec folks would have choosen a higher value.
> 
> Or is this some tg3 specific performance issue?

Keeping the MRRS at 512 introduces DMA latencies that effectively
prevent us from achieving linerate.  With a packet size of ~1.5K and the
MRRS at 512 bytes, the DMA will be broken into at least 3 DMA reads.
Each DMA read takes ~1usec to initiate.  It is this overhead that starts
to cut into total throughput.

Re: [PATCH 10/13] tg3: Increase the PCI MRRS

[David Miller]

From: "Matt Carlson" <mcarlson@broadcom.com>
Date: Thu, 15 Nov 2007 14:20:10 -0800

> On Mon, 2007-11-12 at 21:21 -0800, David Miller wrote:
> > From: "Matt Carlson" <mcarlson@broadcom.com>
> > Date: Fri, 09 Nov 2007 16:39:01 -0800
> > 
> > > Previous devices hardcoded the PCI Maximum Read Request Size to 4K.  To
> > > better comply with the PCI spec, the hardware now defaults the MRRS to
> > > 512 bytes.  This will yield poor driver performance if left untouched.
> > > This patch increases the MRRS to 4K on driver initialization.
> > > 
> > > Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
> > > Signed-off-by: Michael Chan <mchan@broadcom.com>
> > 
> > I've applied this patch, but...
> > 
> > I sense that the PCI spec wants devices to use an MRRS value of 512 in
> > order to get better fairness on a PCI-E segment amongst multiple
> > devices.
> > 
> > From that perspective, jacking up the MRRS to 4096 unilaterally seems
> > like a very bad idea.  If this was necessary for good performance, I'm
> > sure the PCI spec folks would have choosen a higher value.
> > 
> > Or is this some tg3 specific performance issue?
> 
> Keeping the MRRS at 512 introduces DMA latencies that effectively
> prevent us from achieving linerate.  With a packet size of ~1.5K and the
> MRRS at 512 bytes, the DMA will be broken into at least 3 DMA reads.
> Each DMA read takes ~1usec to initiate.  It is this overhead that starts
> to cut into total throughput.

Ok, but wouldn't every networking device on PCI need to do this then?

I want to hear what you think about this wrt. what I mentioned about
fairness above.  What's the point of PCI specifying a limit to comply
with if nobody complies with the limit for localized performance
reasons?

I think this is an important issue.  Someone down the road is going to
see bad disk throughput when doing lots of network transfers and
wonder why that is.  It will be hard to debug, but it won't be
difficult for us to do something proactive about this right now to
prevent that problem from happening in the first place.

Thanks.

Re: [PATCH 10/13] tg3: Increase the PCI MRRS

[Michael Chan]

On Thu, 2007-11-15 at 14:41 -0800, David Miller wrote:
> From: "Matt Carlson" <mcarlson@broadcom.com>
> Date: Thu, 15 Nov 2007 14:20:10 -0800
> >
> > Keeping the MRRS at 512 introduces DMA latencies that effectively
> > prevent us from achieving linerate.  With a packet size of ~1.5K and the
> > MRRS at 512 bytes, the DMA will be broken into at least 3 DMA reads.
> > Each DMA read takes ~1usec to initiate.  It is this overhead that starts
> > to cut into total throughput.
> 
> Ok, but wouldn't every networking device on PCI need to do this then?

No, it depends on the design.  For example, a bigger maximum payload
size will alleviate the problem (tg3 hardware is using 128).  Multiple
DMA read channels to pipeline the multiple read requests will also help.

We don't need to increase the MRRS on bnx2 hardware to get line-rate,
for example.

> 
> I want to hear what you think about this wrt. what I mentioned about
> fairness above.  What's the point of PCI specifying a limit to comply
> with if nobody complies with the limit for localized performance
> reasons?

Fairness is harder to know because it depends on chipset behavior.  PCIE
is not shared so there's no fairness issue at the local PCIE.  Any
fairness issue will be at the bridge.

I don't know what's the exact rationale for defaulting to 512, but I
will try to find out.

> 
> I think this is an important issue.  Someone down the road is going to
> see bad disk throughput when doing lots of network transfers and
> wonder why that is.  It will be hard to debug, but it won't be
> difficult for us to do something proactive about this right now to
> prevent that problem from happening in the first place.
> 
> Thanks.
> 

Re: [PATCH 10/13] tg3: Increase the PCI MRRS

[David Miller]

From: "Michael Chan" <mchan@broadcom.com>
Date: Thu, 15 Nov 2007 15:51:31 -0800

> I don't know what's the exact rationale for defaulting to 512, but I
> will try to find out.

Thank you.

Re: [PATCH 10/13] tg3: Increase the PCI MRRS

[Rick Jones]

>>>I sense that the PCI spec wants devices to use an MRRS value of 512 in
>>>order to get better fairness on a PCI-E segment amongst multiple
>>>devices.

Unless there are PCIe switches, there is only ever one device on the 
PCIe segment yes?  Even on the biggest PCI-X or PCIe systems with which 
I am familiar, each PCIe slot is a separate PCIe bus, and I think that 
holds for the most recent PCI-X one as well.

Now, "core" I/O may be another matter - for example, on the rx3600 and 
rx6600 there are some shared PCI-X bus slots (two per) and on the 
"combo" versions in addition to the two independent PCIe x8's there is 
one pair of PCIe's on a switched bus.

However, none of the shared bus slots are recommended for "high 
performance cards" (that wonderfully moving target of a devinition :).


>>>From that perspective, jacking up the MRRS to 4096 unilaterally seems
>>>like a very bad idea.  If this was necessary for good performance, I'm
>>>sure the PCI spec folks would have choosen a higher value.
>>>
>>>Or is this some tg3 specific performance issue?
>>
>>Keeping the MRRS at 512 introduces DMA latencies that effectively
>>prevent us from achieving linerate.  With a packet size of ~1.5K and the
>>MRRS at 512 bytes, the DMA will be broken into at least 3 DMA reads.
>>Each DMA read takes ~1usec to initiate.  It is this overhead that starts
>>to cut into total throughput.

Reminds me of Tigon2 on original PCI.

> Ok, but wouldn't every networking device on PCI need to do this then?

I'm going to get very rapidly out of my PCI depth, but on one or the 
other (e vs X) isn't is possible from the standpoint of PCI for a device 
to have multiple transactions outstanding at a time?

If a given device can only have one outstanding at a time and happens to 
take a while to setup the DMA (perhaps there is firmware down there or 
something) I could see where it would want to make the transactions as 
large as it could, but another device, either able to have multiple in 
flight, or perhaps quicker on the DMA setup draw might not need it so large.

And even if the device itself is reasonably quick on the DMA setup draw, 
there may be systems, particularly large ones, where the rest of the 
setup of the DMA isn't instantaneous.

> I want to hear what you think about this wrt. what I mentioned about
> fairness above.  What's the point of PCI specifying a limit to comply
> with if nobody complies with the limit for localized performance
> reasons?

I'm _guessing_ that much of those limits were set with shared busses in 
mind.  The "trend" seems to be towards single-slot busses.

> I think this is an important issue.  Someone down the road is going to
> see bad disk throughput when doing lots of network transfers and
> wonder why that is.  It will be hard to debug, but it won't be
> difficult for us to do something proactive about this right now to
> prevent that problem from happening in the first place.

Does the current value of the MRRS get displayed in lspci output?  It 
wouldn't be a slam dunk, but if someone were looking at that and saw the 
value large they might make an educated guess.

rick jones

Re: [PATCH 10/13] tg3: Increase the PCI MRRS

[Michael Chan]

On Thu, 2007-11-15 at 16:32 -0800, Rick Jones wrote:

> I'm going to get very rapidly out of my PCI depth, but on one or the 
> other (e vs X) isn't is possible from the standpoint of PCI for a device 
> to have multiple transactions outstanding at a time?

Yes, see my other response.  Multiple outstanding transactions and a
bigger maximum payload size will increase the throughput without the
need to increase the MRRS.

> 
> Does the current value of the MRRS get displayed in lspci output?  It 
> wouldn't be a slam dunk, but if someone were looking at that and saw the 
> value large they might make an educated guess.
> 

Yes:
        Capabilities: [ac] Express Endpoint IRQ 0
                Device: Supported: MaxPayload 512 bytes, PhantFunc 0, ExtTag-
                Device: Latency L0s <1us, L1 <2us
                Device: AtnBtn- AtnInd- PwrInd-
                Device: Errors: Correctable- Non-Fatal- Fatal- Unsupported-
                Device: RlxdOrd+ ExtTag- PhantFunc- AuxPwr+ NoSnoop+
                Device: MaxPayload 256 bytes, MaxReadReq 512 bytes