| Message ID | 20241114102002.481081-1-wojackbb@gmail.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [net-next,v2,net-next,v2] net: wwan: t7xx: Change PM_AUTOSUSPEND_MS to 5000 | expand |
On Thu, Nov 14, 2024 at 06:20:02PM +0800, wojackbb@gmail.com wrote: > From: Jack Wu <wojackbb@gmail.com> > > Because optimizing the power consumption of t7XX, > change auto suspend time to 5000. > > The Tests uses a script to loop through the power_state > of t7XX. > (for example: /sys/bus/pci/devices/0000\:72\:00.0/power_state) > > * If Auto suspend is 20 seconds, > test script show power_state have 0~5% of the time was in D3 state > when host don't have data packet transmission. > > * Changed auto suspend time to 5 seconds, > test script show power_state have 50%~80% of the time was in D3 state > when host don't have data packet transmission. > > We tested Fibocom FM350 and our products using the t7xx and they all > benefited from this. > > Signed-off-by: Jack Wu <wojackbb@gmail.com> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Andrew
Hello Jack, On 14.11.2024 12:20, wojackbb@gmail.com wrote: > From: Jack Wu <wojackbb@gmail.com> > > Because optimizing the power consumption of t7XX, > change auto suspend time to 5000. > > The Tests uses a script to loop through the power_state > of t7XX. > (for example: /sys/bus/pci/devices/0000\:72\:00.0/power_state) > > * If Auto suspend is 20 seconds, > test script show power_state have 0~5% of the time was in D3 state > when host don't have data packet transmission. > > * Changed auto suspend time to 5 seconds, > test script show power_state have 50%~80% of the time was in D3 state > when host don't have data packet transmission. > > We tested Fibocom FM350 and our products using the t7xx and they all > benefited from this. Possible negative outcomes for data transmission still need clarification. Let me repeat it here. On 06.11.2024 13:10, 吳逼逼 wrote: > Receiving or sending data will cause PCIE to change D3 Cold to D0 state. Am I understand it correctly that receiving IP packets on downlink will cause PCIe link re-activation? I am concerned about a TCP connection that can be idle for a long period of time. For example, an established SSH connection can stay idle for minutes. If I connected to a server and execute something like this: user@host$ sleep 20 && echo "Done" Will I eventually see the "Done" message or will the autosuspended modem effectively block any incoming traffic? And how long does it take for the modem to wake up and deliver a downlink packet to the host? Have you measured StDev change? > Signed-off-by: Jack Wu <wojackbb@gmail.com> > --- > V2: > * supplementary commit information > --- > --- > drivers/net/wwan/t7xx/t7xx_pci.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/net/wwan/t7xx/t7xx_pci.c b/drivers/net/wwan/t7xx/t7xx_pci.c > index e556e5bd49ab..dcadd615a025 100644 > --- a/drivers/net/wwan/t7xx/t7xx_pci.c > +++ b/drivers/net/wwan/t7xx/t7xx_pci.c > @@ -48,7 +48,7 @@ > #define T7XX_INIT_TIMEOUT 20 > #define PM_SLEEP_DIS_TIMEOUT_MS 20 > #define PM_ACK_TIMEOUT_MS 1500 > -#define PM_AUTOSUSPEND_MS 20000 > +#define PM_AUTOSUSPEND_MS 5000 > #define PM_RESOURCE_POLL_TIMEOUT_US 10000 > #define PM_RESOURCE_POLL_STEP_US 100 >
On Thu, 14 Nov 2024 20:54:20 +0200 Sergey Ryazanov wrote: > > We tested Fibocom FM350 and our products using the t7xx and they all > > benefited from this. > > Possible negative outcomes for data transmission still need > clarification. Let me repeat it here. > > On 06.11.2024 13:10, 吳逼逼 wrote: > > Receiving or sending data will cause PCIE to change D3 Cold to D0 state. > > Am I understand it correctly that receiving IP packets on downlink will > cause PCIe link re-activation? > > > I am concerned about a TCP connection that can be idle for a long period > of time. For example, an established SSH connection can stay idle for > minutes. If I connected to a server and execute something like this: > > user@host$ sleep 20 && echo "Done" > > Will I eventually see the "Done" message or will the autosuspended modem > effectively block any incoming traffic? And how long does it take for > the modem to wake up and deliver a downlink packet to the host? Have you > measured StDev change? He's decreasing the sleep timer from 20 to 5 sec, both of which are very high for networking, anyway. You appear to be questioning autosuspend itself but it seems to have been added 2 years ago already. What am I missing?
On 16.11.2024 01:21, Jakub Kicinski wrote: > On Thu, 14 Nov 2024 20:54:20 +0200 Sergey Ryazanov wrote: >>> We tested Fibocom FM350 and our products using the t7xx and they all >>> benefited from this. >> >> Possible negative outcomes for data transmission still need >> clarification. Let me repeat it here. >> >> On 06.11.2024 13:10, 吳逼逼 wrote: >>> Receiving or sending data will cause PCIE to change D3 Cold to D0 state. >> >> Am I understand it correctly that receiving IP packets on downlink will >> cause PCIe link re-activation? >> >> >> I am concerned about a TCP connection that can be idle for a long period >> of time. For example, an established SSH connection can stay idle for >> minutes. If I connected to a server and execute something like this: >> >> user@host$ sleep 20 && echo "Done" >> >> Will I eventually see the "Done" message or will the autosuspended modem >> effectively block any incoming traffic? And how long does it take for >> the modem to wake up and deliver a downlink packet to the host? Have you >> measured StDev change? > > He's decreasing the sleep timer from 20 to 5 sec, both of which > are very high for networking, anyway. You appear to be questioning > autosuspend itself but it seems to have been added 2 years ago already. > > What am I missing? Some possible funny side-effect of sleeping with this chipset. Like loosing network connection and dropping TCP sessions. I hope that 20 seconds was putted on purpose. Suddenly, I don't have this modem at hand and want to be sure that we are not going to receive a stream of bug reports. -- Sergey
On Tue, 19 Nov 2024 03:01:47 +0200 Sergey Ryazanov wrote: > > He's decreasing the sleep timer from 20 to 5 sec, both of which > > are very high for networking, anyway. You appear to be questioning > > autosuspend itself but it seems to have been added 2 years ago already. > > > > What am I missing? > > Some possible funny side-effect of sleeping with this chipset. Like > loosing network connection and dropping TCP sessions. I hope that 20 > seconds was putted on purpose. > > Suddenly, I don't have this modem at hand and want to be sure that we > are not going to receive a stream of bug reports. Power saving is always tricky, but they say they tested. I think we should give it a go, worst case - it will be an easy revert.
On 19.11.2024 03:44, Jakub Kicinski wrote: > On Tue, 19 Nov 2024 03:01:47 +0200 Sergey Ryazanov wrote: >>> He's decreasing the sleep timer from 20 to 5 sec, both of which >>> are very high for networking, anyway. You appear to be questioning >>> autosuspend itself but it seems to have been added 2 years ago already. >>> >>> What am I missing? >> >> Some possible funny side-effect of sleeping with this chipset. Like >> loosing network connection and dropping TCP sessions. I hope that 20 >> seconds was putted on purpose. >> >> Suddenly, I don't have this modem at hand and want to be sure that we >> are not going to receive a stream of bug reports. > > Power saving is always tricky, but they say they tested. I think we > should give it a go, worst case - it will be an easy revert. Make sense. Then let's merge it. Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Hello: This patch was applied to netdev/net-next.git (main) by Jakub Kicinski <kuba@kernel.org>: On Thu, 14 Nov 2024 18:20:02 +0800 you wrote: > From: Jack Wu <wojackbb@gmail.com> > > Because optimizing the power consumption of t7XX, > change auto suspend time to 5000. > > The Tests uses a script to loop through the power_state > of t7XX. > (for example: /sys/bus/pci/devices/0000\:72\:00.0/power_state) > > [...] Here is the summary with links: - [net-next,v2,net-next,v2] net: wwan: t7xx: Change PM_AUTOSUSPEND_MS to 5000 https://git.kernel.org/netdev/net-next/c/a0c80d5108ab You are awesome, thank you!
diff --git a/drivers/net/wwan/t7xx/t7xx_pci.c b/drivers/net/wwan/t7xx/t7xx_pci.c index e556e5bd49ab..dcadd615a025 100644 --- a/drivers/net/wwan/t7xx/t7xx_pci.c +++ b/drivers/net/wwan/t7xx/t7xx_pci.c @@ -48,7 +48,7 @@ #define T7XX_INIT_TIMEOUT 20 #define PM_SLEEP_DIS_TIMEOUT_MS 20 #define PM_ACK_TIMEOUT_MS 1500 -#define PM_AUTOSUSPEND_MS 20000 +#define PM_AUTOSUSPEND_MS 5000 #define PM_RESOURCE_POLL_TIMEOUT_US 10000 #define PM_RESOURCE_POLL_STEP_US 100