| Message ID | 20230927130919.25683-3-rohan.g.thomas@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | net: stmmac: TBS support for platform driver | expand |
Qualcomm had similar discussions with respect to enabling of TBS for a particular queue. We had similar discussion on these terms yesterday with Redhat. Adding Andrew from Redhat here What we discovered as part of the discussions is listed below. 1. Today upstream stmmac code is designed in such a way that TBS flag is put as part of queue configurations(see below snippet) and as well know that stmmac queue configuration comes from the dtsi file. //ndo_open => stmmac_open int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en;(comes from tx_queues_cfg) /* Setup per-TXQ tbs flag before TX descriptor alloc */ tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; 2. There is a no way to do this dynamically from user space because we don't have any API exposed which can do it from user space and also TBS rely on special descriptors aka enhanced desc this cannot be done run time and stmmac has to be aware of it before we do DMA/MAC/MTL start. To do this dynamically would only mean stopping DMA/MAC/MTL realloc resources for enhanced desc and the starting MAC/DMA/MTL. This means we are disrupting other traffic(By stopping MAC block). 3. I dont think there is a way we can enable this dynamically today. I would like upstream community to share your thoughts as well. 4. I agree with Rohan's patch here and want upstream community to accept it. This will allow use to configure the queues where TBS needs to be enabled as hardcoding in the code unless upstream has better way to this using userspace. Please let us know if you think otherwise. On 9/27/2023 6:09 AM, Rohan G Thomas wrote: > Enable Time Based Scheduling(TBS) support for Tx queues through the > stmmac platform driver. For this a new per-queue tx-config property, > tbs-enabled is added to the devicetree. > > Commit 7eadf57290ec ("net: stmmac: pci: Enable TBS on GMAC5 IPK PCI > entry") enables similar support for the stmmac pci driver. > > Also add check whether TBS support is available for a Tx DMA channel > before enabling TBS support for that Tx DMA channel. > > Signed-off-by: Rohan G Thomas <rohan.g.thomas@intel.com> > --- > .../net/ethernet/stmicro/stmmac/stmmac_main.c | 25 +++++++++++++++---- > .../ethernet/stmicro/stmmac/stmmac_platform.c | 4 +++ > 2 files changed, 24 insertions(+), 5 deletions(-) > > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c > index 81b6f3ecdf92..7333f0640b3d 100644 > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c > @@ -3773,12 +3773,18 @@ stmmac_setup_dma_desc(struct stmmac_priv *priv, unsigned int mtu) > dma_conf->dma_rx_size = DMA_DEFAULT_RX_SIZE; > > /* Earlier check for TBS */ > - for (chan = 0; chan < priv->plat->tx_queues_to_use; chan++) { > - struct stmmac_tx_queue *tx_q = &dma_conf->tx_queue[chan]; > - int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; > + if (priv->dma_cap.tbssel) { > + /* TBS is available only for tbs_ch_num of Tx DMA channels, > + * starting from the highest Tx DMA channel. > + */ > + chan = priv->dma_cap.number_tx_channel - priv->dma_cap.tbs_ch_num; > + for (; chan < priv->plat->tx_queues_to_use; chan++) { > + struct stmmac_tx_queue *tx_q = &dma_conf->tx_queue[chan]; > + int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; > > - /* Setup per-TXQ tbs flag before TX descriptor alloc */ > - tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; > + /* Setup per-TXQ tbs flag before TX descriptor alloc */ > + tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; > + } > } > > ret = alloc_dma_desc_resources(priv, dma_conf); > @@ -7505,6 +7511,15 @@ int stmmac_dvr_probe(struct device *device, > } > } > > + /* If TBS feature is supported(i.e. tbssel is true), then at least 1 Tx > + * DMA channel supports TBS. So if tbs_ch_num is 0 and tbssel is true, > + * assume all Tx DMA channels support TBS. TBS_CH field, which gives > + * number of Tx DMA channels with TBS support is only available only for > + * DW xGMAC IP. For other DWMAC IPs all Tx DMA channels can support TBS. > + */ > + if (priv->dma_cap.tbssel && !priv->dma_cap.tbs_ch_num) > + priv->dma_cap.tbs_ch_num = priv->dma_cap.number_tx_channel; > + > ndev->features |= ndev->hw_features | NETIF_F_HIGHDMA; > ndev->watchdog_timeo = msecs_to_jiffies(watchdog); > #ifdef STMMAC_VLAN_TAG_USED > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > index 843bd8804bfa..6c0191c84071 100644 > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > @@ -279,6 +279,10 @@ static int stmmac_mtl_setup(struct platform_device *pdev, > plat->tx_queues_cfg[queue].coe_unsupported = > of_property_read_bool(q_node, "snps,coe-unsupported"); > > + /* Select TBS for supported queues */ > + plat->tx_queues_cfg[queue].tbs_en = > + of_property_read_bool(q_node, "snps,tbs-enabled"); > + > queue++; > } > if (queue != plat->tx_queues_to_use) {
On Wed, 10 Jan 2024 20:20:39 +0000 : Abhishek Chauhan (ABC) wrote: > Qualcomm had similar discussions with respect to enabling of TBS for a > particular queue. > We had similar discussion on these terms yesterday with Redhat. Adding > Andrew from Redhat here > > What we discovered as part of the discussions is listed below. > > 1. Today upstream stmmac code is designed in such a way that TBS flag is put > as part of queue configurations(see below snippet) and as well know that > stmmac queue configuration comes from the dtsi file. > > //ndo_open => stmmac_open > int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en;(comes from > tx_queues_cfg) > > /* Setup per-TXQ tbs flag before TX descriptor alloc */ tx_q->tbs |= tbs_en ? > STMMAC_TBS_AVAIL : 0; > > 2. There is a no way to do this dynamically from user space because we don't > have any API exposed which can do it from user space and also TBS rely on > special descriptors aka enhanced desc this cannot be done run time and > stmmac has to be aware of it before we do DMA/MAC/MTL start. To do this > dynamically would only mean stopping DMA/MAC/MTL realloc resources for > enhanced desc and the starting MAC/DMA/MTL. This means we are disrupting > other traffic(By stopping MAC block). > > 3. I dont think there is a way we can enable this dynamically today. I would like > upstream community to share your thoughts as well. > > 4. I agree with Rohan's patch here and want upstream community to accept it. > This will allow use to configure the queues where TBS needs to be enabled as > hardcoding in the code unless upstream has better way to this using > userspace. > > Please let us know if you think otherwise. > > Hi Abhishek, Thanks for bringing this to discussion and a better explanation from your side. I agree with you. Best Regards, Rohan
Rohan G Thomas <rohan.g.thomas@intel.com> writes: > Enable Time Based Scheduling(TBS) support for Tx queues through the > stmmac platform driver. For this a new per-queue tx-config property, > tbs-enabled is added to the devicetree. > > Commit 7eadf57290ec ("net: stmmac: pci: Enable TBS on GMAC5 IPK PCI > entry") enables similar support for the stmmac pci driver. > > Also add check whether TBS support is available for a Tx DMA channel > before enabling TBS support for that Tx DMA channel. > > Signed-off-by: Rohan G Thomas <rohan.g.thomas@intel.com> > --- > .../net/ethernet/stmicro/stmmac/stmmac_main.c | 25 +++++++++++++++---- > .../ethernet/stmicro/stmmac/stmmac_platform.c | 4 +++ > 2 files changed, 24 insertions(+), 5 deletions(-) > > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c > index 81b6f3ecdf92..7333f0640b3d 100644 > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c > @@ -3773,12 +3773,18 @@ stmmac_setup_dma_desc(struct stmmac_priv *priv, unsigned int mtu) > dma_conf->dma_rx_size = DMA_DEFAULT_RX_SIZE; > > /* Earlier check for TBS */ > - for (chan = 0; chan < priv->plat->tx_queues_to_use; chan++) { > - struct stmmac_tx_queue *tx_q = &dma_conf->tx_queue[chan]; > - int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; > + if (priv->dma_cap.tbssel) { > + /* TBS is available only for tbs_ch_num of Tx DMA channels, > + * starting from the highest Tx DMA channel. > + */ > + chan = priv->dma_cap.number_tx_channel - priv->dma_cap.tbs_ch_num; > + for (; chan < priv->plat->tx_queues_to_use; chan++) { > + struct stmmac_tx_queue *tx_q = &dma_conf->tx_queue[chan]; > + int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; > > - /* Setup per-TXQ tbs flag before TX descriptor alloc */ > - tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; > + /* Setup per-TXQ tbs flag before TX descriptor alloc */ > + tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; > + } > } > > ret = alloc_dma_desc_resources(priv, dma_conf); > @@ -7505,6 +7511,15 @@ int stmmac_dvr_probe(struct device *device, > } > } > > + /* If TBS feature is supported(i.e. tbssel is true), then at least 1 Tx > + * DMA channel supports TBS. So if tbs_ch_num is 0 and tbssel is true, > + * assume all Tx DMA channels support TBS. TBS_CH field, which gives > + * number of Tx DMA channels with TBS support is only available only for > + * DW xGMAC IP. For other DWMAC IPs all Tx DMA channels can support TBS. The Ethernet QOS controllers found in various i.MX socs does not support TBS on TX queue 0. I believe this patch would break the dwmac driver for these platforms. > + */ > + if (priv->dma_cap.tbssel && !priv->dma_cap.tbs_ch_num) > + priv->dma_cap.tbs_ch_num = priv->dma_cap.number_tx_channel; > + > ndev->features |= ndev->hw_features | NETIF_F_HIGHDMA; > ndev->watchdog_timeo = msecs_to_jiffies(watchdog); > #ifdef STMMAC_VLAN_TAG_USED > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > index 843bd8804bfa..6c0191c84071 100644 > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c > @@ -279,6 +279,10 @@ static int stmmac_mtl_setup(struct platform_device *pdev, > plat->tx_queues_cfg[queue].coe_unsupported = > of_property_read_bool(q_node, "snps,coe-unsupported"); > > + /* Select TBS for supported queues */ > + plat->tx_queues_cfg[queue].tbs_en = > + of_property_read_bool(q_node, "snps,tbs-enabled"); > + > queue++; > } > if (queue != plat->tx_queues_to_use) { /Esben
"Abhishek Chauhan (ABC)" <quic_abchauha@quicinc.com> writes: > Qualcomm had similar discussions with respect to enabling of TBS for a > particular queue. We had similar discussion on these terms yesterday > with Redhat. Adding Andrew from Redhat here > > What we discovered as part of the discussions is listed below. > > 1. Today upstream stmmac code is designed in such a way that TBS flag > is put as part of queue configurations(see below snippet) and as well > know that stmmac queue configuration comes from the dtsi file. > > //ndo_open => stmmac_open > int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en;(comes from tx_queues_cfg) > > /* Setup per-TXQ tbs flag before TX descriptor alloc */ > tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; > > 2. There is a no way to do this dynamically from user space because we don't have any > API exposed which can do it from user space Not now. But why not extend ethtool API to allow enabling TBS for supported controllers? > and also TBS rely on special descriptors aka enhanced desc this cannot > be done run time and stmmac has to be aware of it before we do > DMA/MAC/MTL start. Isn't this somewhat similar to changing the RX/TX ring parameters, which I believe also is quite difficult to do at run time, and ethtool therefore requires the interface to be down in oroer to change them? > To do this dynamically would only mean stopping DMA/MAC/MTL realloc > resources for enhanced desc and the starting MAC/DMA/MTL. This means > we are disrupting other traffic(By stopping MAC block). Yes. But you would be disrupting traffic less than by requiring a complete reboot of the target which is needed if the devicetree must be changed. > 3. I dont think there is a way we can enable this dynamically today. I > would like upstream community to share your thoughts as well. Hereby done. Could we investigate the possibility of using ethtool to change TBS enable/disable "run-time"? > 4. I agree with Rohan's patch here and want upstream community to > accept it. This will allow use to configure the queues where TBS needs > to be enabled as hardcoding in the code unless upstream has better way > to this using userspace. > > Please let us know if you think otherwise. /Esben
From: Rohan G Thomas <rohan.g.thomas@intel.com> On Fri, 26 Jan 2024 09:35:01 +0100, Esben Haabendal wrote: Hi Esben, Thanks for your comments. Please find my reply inline. > > + /* If TBS feature is supported(i.e. tbssel is true), then at least 1 Tx > > + * DMA channel supports TBS. So if tbs_ch_num is 0 and tbssel is true, > > + * assume all Tx DMA channels support TBS. TBS_CH field, which gives > > + * number of Tx DMA channels with TBS support is only available only > for > > + * DW xGMAC IP. For other DWMAC IPs all Tx DMA channels can > support TBS. > > The Ethernet QOS controllers found in various i.MX socs does not support TBS > on TX queue 0. I believe this patch would break the dwmac driver for these > platforms. AFAIU from Synopsys DWMAC5 Databook, all queues support TBS. But TBS cannot coexist with TSO. So all glue drivers enabling TBS feature are avoiding queue 0 to support TSO. Also packets requesting TSO are always directed to queue 0 by stmmac driver. > > > + */ > > + if (priv->dma_cap.tbssel && !priv->dma_cap.tbs_ch_num) > > + priv->dma_cap.tbs_ch_num = priv- > >dma_cap.number_tx_channel; > > + I don't think updating tbs_ch_num with number_tx_channel would break the dwmac driver for i.MX socs. tbs_ch_num is used by 'Earlier check for TBS' loop to set tbs flag. > > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c > > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c > > @@ -3773,12 +3773,18 @@ stmmac_setup_dma_desc(struct stmmac_priv > *priv, unsigned int mtu) > > dma_conf->dma_rx_size = DMA_DEFAULT_RX_SIZE; > > > > /* Earlier check for TBS */ > > - for (chan = 0; chan < priv->plat->tx_queues_to_use; chan++) { > > - struct stmmac_tx_queue *tx_q = &dma_conf- > >tx_queue[chan]; > > - int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; > > + if (priv->dma_cap.tbssel) { > > + /* TBS is available only for tbs_ch_num of Tx DMA channels, > > + * starting from the highest Tx DMA channel. > > + */ > > + chan = priv->dma_cap.number_tx_channel - priv- > >dma_cap.tbs_ch_num; For IPs which don't have tbs_ch_num, this loop goes from 0 to number_tx_channel to check if tbs_enable is set by glue driver. Existing logic is also the same. Unless you set tbs_en flag of queue 0 from the glue driver or dts configuration this patch doesn't set tbs flag for queue 0. This is a sanity check to avoid wrong configuration for IPs which support tbs only in a few number of queues. > > + for (; chan < priv->plat->tx_queues_to_use; chan++) { > > + struct stmmac_tx_queue *tx_q = &dma_conf- > >tx_queue[chan]; > > + int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; > > > > - /* Setup per-TXQ tbs flag before TX descriptor alloc */ > > - tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; > > + /* Setup per-TXQ tbs flag before TX descriptor alloc > */ > > + tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; > > + } > > } Please correct me if I've misstated anything. BR, Rohan
rohan.g.thomas@intel.com writes: > From: Rohan G Thomas <rohan.g.thomas@intel.com> > > On Fri, 26 Jan 2024 09:35:01 +0100, Esben Haabendal wrote: > >> > + /* If TBS feature is supported(i.e. tbssel is true), then at least 1 Tx >> > + * DMA channel supports TBS. So if tbs_ch_num is 0 and tbssel is true, >> > + * assume all Tx DMA channels support TBS. TBS_CH field, which gives >> > + * number of Tx DMA channels with TBS support is only available only >> for >> > + * DW xGMAC IP. For other DWMAC IPs all Tx DMA channels can >> support TBS. >> >> The Ethernet QOS controllers found in various i.MX socs does not support TBS >> on TX queue 0. I believe this patch would break the dwmac driver for these >> platforms. > > AFAIU from Synopsys DWMAC5 Databook, all queues support TBS. But TBS > cannot coexist with TSO. So all glue drivers enabling TBS feature are > avoiding queue 0 to support TSO. Also packets requesting TSO are > always directed to queue 0 by stmmac driver. After re-reading the i.MX8MP documentation, and making a few experiments, I have to agree with you. Enabling TBS (enhanced descriptors) for Q0 should be ok on i.MX. >> > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c >> > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c >> > @@ -3773,12 +3773,18 @@ stmmac_setup_dma_desc(struct stmmac_priv >> *priv, unsigned int mtu) >> > dma_conf->dma_rx_size = DMA_DEFAULT_RX_SIZE; >> > >> > /* Earlier check for TBS */ >> > - for (chan = 0; chan < priv->plat->tx_queues_to_use; chan++) { >> > - struct stmmac_tx_queue *tx_q = &dma_conf- >> >tx_queue[chan]; >> > - int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; >> > + if (priv->dma_cap.tbssel) { >> > + /* TBS is available only for tbs_ch_num of Tx DMA channels, >> > + * starting from the highest Tx DMA channel. >> > + */ >> > + chan = priv->dma_cap.number_tx_channel - priv- >> >dma_cap.tbs_ch_num; > > For IPs which don't have tbs_ch_num, this loop goes from 0 to > number_tx_channel to check if tbs_enable is set by glue driver. > Existing logic is also the same. Unless you set tbs_en flag of > queue 0 from the glue driver or dts configuration this patch doesn't > set tbs flag for queue 0. This is a sanity check to avoid wrong > configuration for IPs which support tbs only in a few number of > queues. Sounds good. >> > + for (; chan < priv->plat->tx_queues_to_use; chan++) { >> > + struct stmmac_tx_queue *tx_q = &dma_conf- >> >tx_queue[chan]; >> > + int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; >> > >> > - /* Setup per-TXQ tbs flag before TX descriptor alloc */ >> > - tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; >> > + /* Setup per-TXQ tbs flag before TX descriptor alloc >> */ >> > + tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; >> > + } >> > } > > Please correct me if I've misstated anything. No corrections for now :) /Esben
Hi Esben, My responses are inline to the queries. I hope i have justified my problem statement. On 1/26/2024 12:43 AM, Esben Haabendal wrote: > "Abhishek Chauhan (ABC)" <quic_abchauha@quicinc.com> writes: > >> Qualcomm had similar discussions with respect to enabling of TBS for a >> particular queue. We had similar discussion on these terms yesterday >> with Redhat. Adding Andrew from Redhat here >> >> What we discovered as part of the discussions is listed below. >> >> 1. Today upstream stmmac code is designed in such a way that TBS flag >> is put as part of queue configurations(see below snippet) and as well >> know that stmmac queue configuration comes from the dtsi file. >> >> //ndo_open => stmmac_open >> int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en;(comes from tx_queues_cfg) >> >> /* Setup per-TXQ tbs flag before TX descriptor alloc */ >> tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; >> >> 2. There is a no way to do this dynamically from user space because we don't have any >> API exposed which can do it from user space > > Not now. But why not extend ethtool API to allow enabling TBS for > supported controllers? > ethtool API can be implemented but that still doesn't solve the problem of stopping the entire MAC block because of enhanced desc allocation. 1. We can either allocate enhanced desc for all channels at bootup and then choose to switch to enable TBS mode at runtime (Additional memory usage) 2. Live with the disruption of traffic for a brief duration of time. Which is not a good solution for priority and critical traffic. >> and also TBS rely on special descriptors aka enhanced desc this cannot >> be done run time and stmmac has to be aware of it before we do >> DMA/MAC/MTL start. > > Isn't this somewhat similar to changing the RX/TX ring parameters, which > I believe also is quite difficult to do at run time, and ethtool > therefore requires the interface to be down in oroer to change them? > >> To do this dynamically would only mean stopping DMA/MAC/MTL realloc >> resources for enhanced desc and the starting MAC/DMA/MTL. This means >> we are disrupting other traffic(By stopping MAC block). > > Yes. But you would be disrupting traffic less than by requiring a > complete reboot of the target which is needed if the devicetree must be > changed. > any DTS solution today anyway requires completely loading the boot image and rebooting the device, but once the device is functional, End user can activate TBS, as he knows the exact usecase and requirements. I understand the solution is not scalable, but at this point we don't have a solution to activate TBS at runtime. >> 3. I dont think there is a way we can enable this dynamically today. I >> would like upstream community to share your thoughts as well. > > Hereby done. Could we investigate the possibility of using ethtool to > change TBS enable/disable "run-time"? > We can either allocate enhanced desc for all channels at bootup and then choose to switch to enable TBS mode at runtime >> 4. I agree with Rohan's patch here and want upstream community to >> accept it. This will allow use to configure the queues where TBS needs >> to be enabled as hardcoding in the code unless upstream has better way >> to this using userspace. >> >> Please let us know if you think otherwise. > > /Esben Best Regards ABC
"Abhishek Chauhan (ABC)" <quic_abchauha@quicinc.com> writes: > On 1/26/2024 12:43 AM, Esben Haabendal wrote: >> "Abhishek Chauhan (ABC)" <quic_abchauha@quicinc.com> writes: >> >>> Qualcomm had similar discussions with respect to enabling of TBS for a >>> particular queue. We had similar discussion on these terms yesterday >>> with Redhat. Adding Andrew from Redhat here >>> >>> What we discovered as part of the discussions is listed below. >>> >>> 1. Today upstream stmmac code is designed in such a way that TBS flag >>> is put as part of queue configurations(see below snippet) and as well >>> know that stmmac queue configuration comes from the dtsi file. >>> >>> //ndo_open => stmmac_open >>> int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en;(comes from tx_queues_cfg) >>> >>> /* Setup per-TXQ tbs flag before TX descriptor alloc */ >>> tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; >>> >>> 2. There is a no way to do this dynamically from user space because >>> we don't have any API exposed which can do it from user space >> >> Not now. But why not extend ethtool API to allow enabling TBS for >> supported controllers? > > ethtool API can be implemented but that still doesn't solve the > problem of stopping the entire MAC block because of enhanced desc > allocation. I am not sure what you exact point is here. If you look at the implementation of ethtool API for changing ring parameters, you have stmmac_set_ringparam() which calls stmmac_reinit_ringparam(), which again calls stmmac_release() if the interface is up (stopping the entire MAC), and then stmmac_open() which reinitializes everything. The same pattern could be applied to changes to enable enhanced descriptor allocation. I don't see why that will not be acceptable. Why would anyone have to do that while critical traffic is flowing? In your system you should be able to know which queues needs enhanced descriptors before starting communication. > 1. We can either allocate enhanced desc for all channels at bootup and > then choose to switch to enable TBS mode at runtime (Additional memory > usage) A good default would IMHO be to enable enhanced descriptors for all but TX queue 0. This will allow use of TBS without needing to change anything. If the rather minimal extra memory usage is disturbing anyone, then they can tune it at boot time before bringing the interface up. > 2. Live with the disruption of traffic for a brief duration of time. > Which is not a good solution for priority and critical traffic. As mentioned above, I don't see why anyone would need to modify the descriptor allocation while critical traffic is flowing. If you are able put this information in your device tree, you definitely will be able to put it in an boot script in some form. >>> and also TBS rely on special descriptors aka enhanced desc this >>> cannot be done run time and stmmac has to be aware of it before we >>> do DMA/MAC/MTL start. >> >> Isn't this somewhat similar to changing the RX/TX ring parameters, >> which I believe also is quite difficult to do at run time, and >> ethtool therefore requires the interface to be down in oroer to >> change them? >> >>> To do this dynamically would only mean stopping DMA/MAC/MTL realloc >>> resources for enhanced desc and the starting MAC/DMA/MTL. This means >>> we are disrupting other traffic(By stopping MAC block). >> >> Yes. But you would be disrupting traffic less than by requiring a >> complete reboot of the target which is needed if the devicetree must be >> changed. >> > any DTS solution today anyway requires completely loading the boot > image and rebooting the device, but once the device is functional, End > user can activate TBS, as he knows the exact usecase and requirements. > I understand the solution is not scalable, but at this point we don't > have a solution to activate TBS at runtime. Exactly. We are discussing a solution to activate enhanced descriptors at "runtime". But I propose to do it in a similar way as changing ring parameters, so it is in runtime seen from a CPU perspective, but the interface will be shortly brought down when changing it. >>> 3. I dont think there is a way we can enable this dynamically today. I >>> would like upstream community to share your thoughts as well. >> >> Hereby done. Could we investigate the possibility of using ethtool to >> change TBS enable/disable "run-time"? >> > We can either allocate enhanced desc for all channels at bootup > and then choose to switch to enable TBS mode at runtime. I think we should do something like this: https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=3b12ec8f618e for all glue drivers, so a sane default is established that allows using TBS from boot up. But in addition to that, I think it would make sense to create an ethtool API to change it from that default. And it will bring down the interface while applying the change. /Esben
On 2/1/2024 12:26 AM, Esben Haabendal wrote: > "Abhishek Chauhan (ABC)" <quic_abchauha@quicinc.com> writes: >> On 1/26/2024 12:43 AM, Esben Haabendal wrote: >>> "Abhishek Chauhan (ABC)" <quic_abchauha@quicinc.com> writes: >>> >>>> Qualcomm had similar discussions with respect to enabling of TBS for a >>>> particular queue. We had similar discussion on these terms yesterday >>>> with Redhat. Adding Andrew from Redhat here >>>> >>>> What we discovered as part of the discussions is listed below. >>>> >>>> 1. Today upstream stmmac code is designed in such a way that TBS flag >>>> is put as part of queue configurations(see below snippet) and as well >>>> know that stmmac queue configuration comes from the dtsi file. >>>> >>>> //ndo_open => stmmac_open >>>> int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en;(comes from tx_queues_cfg) >>>> >>>> /* Setup per-TXQ tbs flag before TX descriptor alloc */ >>>> tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; >>>> >>>> 2. There is a no way to do this dynamically from user space because >>>> we don't have any API exposed which can do it from user space >>> >>> Not now. But why not extend ethtool API to allow enabling TBS for >>> supported controllers? >> >> ethtool API can be implemented but that still doesn't solve the >> problem of stopping the entire MAC block because of enhanced desc >> allocation. > > I am not sure what you exact point is here. > > If you look at the implementation of ethtool API for changing ring > parameters, you have stmmac_set_ringparam() which calls > stmmac_reinit_ringparam(), which again calls stmmac_release() if the > interface is up (stopping the entire MAC), and then stmmac_open() which > reinitializes everything. > > The same pattern could be applied to changes to enable enhanced > descriptor allocation. > > I don't see why that will not be acceptable. Why would anyone have to do > that while critical traffic is flowing? In your system you should be > able to know which queues needs enhanced descriptors before starting > communication. > >> 1. We can either allocate enhanced desc for all channels at bootup and >> then choose to switch to enable TBS mode at runtime (Additional memory >> usage) > > A good default would IMHO be to enable enhanced descriptors for all but > TX queue 0. This will allow use of TBS without needing to change > anything. If the rather minimal extra memory usage is disturbing anyone, > then they can tune it at boot time before bringing the interface up. > >> 2. Live with the disruption of traffic for a brief duration of time. >> Which is not a good solution for priority and critical traffic. > > As mentioned above, I don't see why anyone would need to modify the > descriptor allocation while critical traffic is flowing. > > If you are able put this information in your device tree, you definitely > will be able to put it in an boot script in some form. > >>>> and also TBS rely on special descriptors aka enhanced desc this >>>> cannot be done run time and stmmac has to be aware of it before we >>>> do DMA/MAC/MTL start. >>> >>> Isn't this somewhat similar to changing the RX/TX ring parameters, >>> which I believe also is quite difficult to do at run time, and >>> ethtool therefore requires the interface to be down in oroer to >>> change them? >>> >>>> To do this dynamically would only mean stopping DMA/MAC/MTL realloc >>>> resources for enhanced desc and the starting MAC/DMA/MTL. This means >>>> we are disrupting other traffic(By stopping MAC block). >>> >>> Yes. But you would be disrupting traffic less than by requiring a >>> complete reboot of the target which is needed if the devicetree must be >>> changed. >>> >> any DTS solution today anyway requires completely loading the boot >> image and rebooting the device, but once the device is functional, End >> user can activate TBS, as he knows the exact usecase and requirements. >> I understand the solution is not scalable, but at this point we don't >> have a solution to activate TBS at runtime. > > Exactly. We are discussing a solution to activate enhanced descriptors > at "runtime". But I propose to do it in a similar way as changing ring > parameters, so it is in runtime seen from a CPU perspective, but the > interface will be shortly brought down when changing it. > >>>> 3. I dont think there is a way we can enable this dynamically today. I >>>> would like upstream community to share your thoughts as well. >>> >>> Hereby done. Could we investigate the possibility of using ethtool to >>> change TBS enable/disable "run-time"? >>> >> We can either allocate enhanced desc for all channels at bootup >> and then choose to switch to enable TBS mode at runtime. > > I think we should do something like this: > https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=3b12ec8f618e > > for all glue drivers, so a sane default is established that allows using > TBS from boot up. > > But in addition to that, I think it would make sense to create an > ethtool API to change it from that default. And it will bring down the > interface while applying the change. > Agreed. Okay, We can go with this approach. I will evaluate the changes from ethtool perspective. ethtool approach will be favorable to everyone as most of the users rely on ethtool to configure the NIC parameters. For now glue drivers can exclude queue0 for tbs and have all other queues to have tbs enabled. > /Esben
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index 81b6f3ecdf92..7333f0640b3d 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -3773,12 +3773,18 @@ stmmac_setup_dma_desc(struct stmmac_priv *priv, unsigned int mtu) dma_conf->dma_rx_size = DMA_DEFAULT_RX_SIZE; /* Earlier check for TBS */ - for (chan = 0; chan < priv->plat->tx_queues_to_use; chan++) { - struct stmmac_tx_queue *tx_q = &dma_conf->tx_queue[chan]; - int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; + if (priv->dma_cap.tbssel) { + /* TBS is available only for tbs_ch_num of Tx DMA channels, + * starting from the highest Tx DMA channel. + */ + chan = priv->dma_cap.number_tx_channel - priv->dma_cap.tbs_ch_num; + for (; chan < priv->plat->tx_queues_to_use; chan++) { + struct stmmac_tx_queue *tx_q = &dma_conf->tx_queue[chan]; + int tbs_en = priv->plat->tx_queues_cfg[chan].tbs_en; - /* Setup per-TXQ tbs flag before TX descriptor alloc */ - tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; + /* Setup per-TXQ tbs flag before TX descriptor alloc */ + tx_q->tbs |= tbs_en ? STMMAC_TBS_AVAIL : 0; + } } ret = alloc_dma_desc_resources(priv, dma_conf); @@ -7505,6 +7511,15 @@ int stmmac_dvr_probe(struct device *device, } } + /* If TBS feature is supported(i.e. tbssel is true), then at least 1 Tx + * DMA channel supports TBS. So if tbs_ch_num is 0 and tbssel is true, + * assume all Tx DMA channels support TBS. TBS_CH field, which gives + * number of Tx DMA channels with TBS support is only available only for + * DW xGMAC IP. For other DWMAC IPs all Tx DMA channels can support TBS. + */ + if (priv->dma_cap.tbssel && !priv->dma_cap.tbs_ch_num) + priv->dma_cap.tbs_ch_num = priv->dma_cap.number_tx_channel; + ndev->features |= ndev->hw_features | NETIF_F_HIGHDMA; ndev->watchdog_timeo = msecs_to_jiffies(watchdog); #ifdef STMMAC_VLAN_TAG_USED diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c index 843bd8804bfa..6c0191c84071 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c @@ -279,6 +279,10 @@ static int stmmac_mtl_setup(struct platform_device *pdev, plat->tx_queues_cfg[queue].coe_unsupported = of_property_read_bool(q_node, "snps,coe-unsupported"); + /* Select TBS for supported queues */ + plat->tx_queues_cfg[queue].tbs_en = + of_property_read_bool(q_node, "snps,tbs-enabled"); + queue++; } if (queue != plat->tx_queues_to_use) {
Enable Time Based Scheduling(TBS) support for Tx queues through the stmmac platform driver. For this a new per-queue tx-config property, tbs-enabled is added to the devicetree. Commit 7eadf57290ec ("net: stmmac: pci: Enable TBS on GMAC5 IPK PCI entry") enables similar support for the stmmac pci driver. Also add check whether TBS support is available for a Tx DMA channel before enabling TBS support for that Tx DMA channel. Signed-off-by: Rohan G Thomas <rohan.g.thomas@intel.com> --- .../net/ethernet/stmicro/stmmac/stmmac_main.c | 25 +++++++++++++++---- .../ethernet/stmicro/stmmac/stmmac_platform.c | 4 +++ 2 files changed, 24 insertions(+), 5 deletions(-)