NTB function for PCIe RC to EP connection
Subsystems (6)
v2 cover NTB function for PCIe RC to EP connection
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐
│ │ │ │
├────────────┤ │ ┌──────────────┤
│ NTB │ │ │ NTB │
│ NetDev │ │ │ NetDev │
├────────────┤ │ ├──────────────┤
│ NTB │ │ │ NTB │
│ Transfer │ │ │ Transfer │
├────────────┤ │ ├──────────────┤
│ │ │ │ │
│ PCI NTB │ │ │ │
│ EPF │ │ │ │
│ Driver │ │ │ PCI Virtual │
│ │ ├───────────────┐ │ NTB Driver │
│ │ │ PCI EP NTB │◄────►│ │
│ │ │ FN Driver │ │ │
├────────────┤ ├───────────────┤ ├──────────────┤
│ │ │ │ │ │
│ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │
│ │ PCI │ │ │ BUS │
└────────────┘ └───────────────┴──────┴──────────────┘
PCI RC PCI EPFrank Li (4):
PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function device
Documentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c
Signature · 1 line
2.24.0.rc1
PATCH v2 1/4 PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
ntb_mw_set_trans() will set memory map window after endpoint function driver bind. The inbound map address need be updated dynamically when using NTB by PCIe Root Port and PCIe Endpoint connection.
Checking if iatu already assigned to the BAR, if yes, using assigned iatu number to update inbound address map and skip set BAR's register.
---
Change from V1:
- improve commit message
drivers/pci/controller/dwc/pcie-designware-ep.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/pci/controller/dwc/pcie-designware-ep.c +9 -1
@@ -161,7 +161,11 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no,u32 free_win;struct dw_pcie *pci =to_dw_pcie_from_ep (ep); - free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows); +if (!ep->bar_to_atu[bar]) + free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows); +else + free_win = ep->bar_to_atu[bar]; +if (free_win >= pci->num_ib_windows) {dev_err (pci->dev,"No free inbound window\n" );return -EINVAL ;
@@ -215,6 +219,7 @@ static void dw_pcie_ep_clear_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,dw_pcie_disable_atu (pci, atu_index,DW_PCIE_REGION_INBOUND );clear_bit (atu_index, ep->ib_window_map); ep->epf_bar[bar] =NULL ; + ep->bar_to_atu[bar] =0 ; } static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,
@@ -244,6 +249,9 @@ static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,if (ret)return ret; +if (ep->epf_bar[bar]) +return 0 ; +dw_pcie_dbi_ro_wr_en (pci);dw_pcie_writel_dbi2 (pci, reg,lower_32_bits (size -1 ));
Signature · 1 line
2.24.0.rc1
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Hello Frank,
Quoted 21 lines · Frank Li
ntb_mw_set_trans()will set memory map window after endpoint function driver bind. The inbound map address need be updated dynamically when using NTB by PCIe Root Port and PCIe Endpoint connection.Checking if iatu already assigned to the BAR, if yes, using assigned iatu number to update inbound address map and skip set BAR's register.
Signed-off-by Frank Li <[email protected]>---
Change from V1:
- improve commit messagedrivers/pci/controller/dwc/pcie-designware-ep.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-1@@ -161,7 +161,11 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no,u32 free_win;struct dw_pcie *pci =to_dw_pcie_from_ep (ep); - free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);
find_first_zero_bit() can return 0, representing bit 0, which is a perfectly valid return value.
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 1 line
drivers/pci/controller/dwc/pcie-designware-ep.c +1
-0+if (!ep->bar_to_atu[bar])
so this check is not correct.
For platforms that has a core_init_notifier, e.g. qcom and tegra, what will happen is that, on first boot:
BAR0: iATU0
BAR1: iATU1
BAR2: iATU2
BAR3: iATU3
BAR4: iATU4
BAR5: iATU5
Rebooting the RC, will cause a perst assertion + deassertion, after which:
BAR0: iATU6
BAR1: iATU1
BAR2: iATU2
BAR3: iATU3
BAR4: iATU4
BAR5: iATU5
This is obviously a bug, because you cannot use:
if (!ep->bar_to_atu[bar])
when 0 is a valid return value.
My proposed fix is:
@@ -172,16 +176,26 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type,
{
int ret;
u32 free_win;
+ u32 saved_atu;
struct dw_pcie *pci = to_dw_pcie_from_ep(ep);- if (!ep->bar_to_atu[bar])
+ saved_atu = ep->bar_to_atu[bar];
+ if (!saved_atu) {
free_win = find_first_zero_bit(ep->ib_window_map, pci->num_ib_windows);- else
- free_win = ep->bar_to_atu[bar];
-
- if (free_win >= pci->num_ib_windows) {
- dev_err(pci->dev, "No free inbound window\n");
- return -EINVAL;
+ //pr_err("%s BAR: %d, found no ATU, using first free, index: %d\n", __func__, bar, free_win);
+ if (free_win >= pci->num_ib_windows) {+ dev_err(pci->dev, "No free inbound window\n");
+ return -EINVAL;
+ }
+
+ /*
+ * In order for bar_to_atu[bar] == 0 to equal no iATU, offset + * the saved value with 1. + */
+ saved_atu = free_win + 1;
+ } else {
+ free_win = saved_atu - 1;
+ //pr_err("%s BAR: %d, already has ATU, index: %d\n", __func__, bar, free_win);
}
ret = dw_pcie_prog_ep_inbound_atu(pci, func_no, free_win, type,
@@ -191,7 +205,7 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type,
return ret;
}- ep->bar_to_atu[bar] = free_win;
+ ep->bar_to_atu[bar] = saved_atu;
set_bit(free_win, ep->ib_window_map);
return 0;drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no, · line 254
Quoted 8 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +3
-0static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,@@ -244,6 +249,9 @@ static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,if (ret)return ret; +if (ep->epf_bar[bar]) +return 0 ; +
However, here you ignore writing the settings if (ep->epf_bar[bar]), again, this will not work for a platform with a core_init_notifier, as they perform a full core reset using reset_control_assert(), when perst is asserted + deasserted, so AFAICT, these settings will get cleared for those platforms, so they will need to be re-written.
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no, · line 257
Quoted 3 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0dw_pcie_dbi_ro_wr_en (pci);dw_pcie_writel_dbi2 (pci, reg,lower_32_bits (size -1 ));
Kind regards,
Niklas
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 27 lines · Niklas Cassel
Hello Frank,
Quoted 24 lines · Frank Li
ntb_mw_set_trans()will set memory map window after endpoint function driver bind. The inbound map address need be updated dynamically when using NTB by PCIe Root Port and PCIe Endpoint connection.Checking if iatu already assigned to the BAR, if yes, using assigned iatu number to update inbound address map and skip set BAR's register.
Signed-off-by Frank Li <[email protected]>---
Change from V1:
- improve commit messagedrivers/pci/controller/dwc/pcie-designware-ep.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0@@ -161,7 +161,11 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no,u32 free_win;struct dw_pcie *pci =to_dw_pcie_from_ep (ep);
Quoted 9 lines
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 1 line
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-1- free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);
find_first_zero_bit()can return 0, representing bit 0, which is a perfectly valid return value.drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 1 line
drivers/pci/controller/dwc/pcie-designware-ep.c +1
-0+if (!ep->bar_to_atu[bar])so this check is not correct.
Please sent out your fixed patch. If want to me fix it, please tell me reproduce steps.
Frank
Quoted 98 lines
For platforms that has a
core_init_notifier, e.g. qcom and tegra, what will happen is that, on first boot:BAR0: iATU0
BAR1: iATU1
BAR2: iATU2
BAR3: iATU3
BAR4: iATU4
BAR5: iATU5Rebooting the RC, will cause a perst assertion + deassertion, after which:
BAR0: iATU6
BAR1: iATU1
BAR2: iATU2
BAR3: iATU3
BAR4: iATU4
BAR5: iATU5This is obviously a bug, because you cannot use:
if (!ep->bar_to_atu[bar])
when 0 is a valid return value.
My proposed fix is:
(quoted diff) +18
-8@@ -172,16 +176,26 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type, { int ret; u32 free_win; + u32 saved_atu; struct dw_pcie *pci = to_dw_pcie_from_ep(ep); - if (!ep->bar_to_atu[bar]) + saved_atu = ep->bar_to_atu[bar]; + if (!saved_atu) { free_win = find_first_zero_bit(ep->ib_window_map, pci->num_ib_windows); - else - free_win = ep->bar_to_atu[bar]; - - if (free_win >= pci->num_ib_windows) { - dev_err(pci->dev, "No free inbound window\n"); - return -EINVAL; + //pr_err("%s BAR: %d, found no ATU, using first free, index: %d\n", __func__, bar, free_win); + if (free_win >= pci->num_ib_windows) { + dev_err(pci->dev, "No free inbound window\n"); + return -EINVAL; + } + + /* + * In order for bar_to_atu[bar] == 0 to equal no iATU, offset + * the saved value with 1. + */ + saved_atu = free_win + 1; + } else { + free_win = saved_atu - 1; + //pr_err("%s BAR: %d, already has ATU, index: %d\n", __func__, bar, free_win); } ret = dw_pcie_prog_ep_inbound_atu(pci, func_no, free_win, type,@@ -191,7 +205,7 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type, return ret; } - ep->bar_to_atu[bar] = free_win; + ep->bar_to_atu[bar] = saved_atu; set_bit(free_win, ep->ib_window_map); return 0;drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no, · line 254
Quoted 10 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +3
-0static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,@@ -244,6 +249,9 @@ static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,if (ret)return ret; +if (ep->epf_bar[bar]) +return 0 ; +However, here you ignore writing the settings if (ep->epf_bar[bar]), again, this will not work for a platform with a
core_init_notifier, as they perform a full core reset usingreset_control_assert(), when perst is asserted + deasserted, so AFAICT, these settings will get cleared for those platforms, so they will need to be re-written.drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no, · line 257
Quoted 3 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0dw_pcie_dbi_ro_wr_en (pci);dw_pcie_writel_dbi2 (pci, reg,lower_32_bits (size -1 ));Kind regards,
Niklas
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 33 lines · Frank Li
Quoted 32 lines · Niklas Cassel
Hello Frank,
Quoted 21 lines · Frank Li
ntb_mw_set_trans()will set memory map window after endpoint function driver bind. The inbound map address need be updated dynamically when using NTB by PCIe Root Port and PCIe Endpoint connection.Checking if iatu already assigned to the BAR, if yes, using assigned iatu number to update inbound address map and skip set BAR's register.
Signed-off-by Frank Li <[email protected]>---
Change from V1:
- improve commit messagedrivers/pci/controller/dwc/pcie-designware-ep.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-1@@ -161,7 +161,11 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no,u32 free_win;struct dw_pcie *pci =to_dw_pcie_from_ep (ep); - free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);
find_first_zero_bit()can return 0, representing bit 0, which is a perfectly valid return value.
Quoted 7 lines
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 1 line
drivers/pci/controller/dwc/pcie-designware-ep.c +1
-0+if (!ep->bar_to_atu[bar])so this check is not correct.
Please sent out your fixed patch. If want to me fix it, please tell me reproduce steps.
Reproduce steps are simple:
1) Add debug messages to dw_pcie_ep_inbound_atu() to see the iATU index for each BAR.
2) Boot an EP platform with a core_init_notifier.
3) Boot the RC.
4) Reboot the RC, which will assert + deassert PERST, and will call pci_epc_init_notify(), which will call .core_init (pci_epf_test_core_init()) which will set the BARs.
In step 3) BAR0 will use iATU0.
In step 4) BAR0 will use iATU6 instead of iATU0. There is no reason for this, as it should really reuse the same iATU index as before, just like all the other BARs do. (This is because of find_first_zero_bit() misusage.)
I could send out my patch, but from inspecting the code, it looks like:
Quoted 2 lines
Quoted 2 lines
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_set_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no, · line 253
Quoted 2 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +2
-0+if (ep->epf_bar[bar]) +return 0 ;
from dw_pcie_ep_set_bar(), also needs to be dropped, so that the iATU settings will be re-written for platforms with core_init_notifiers.
Right now, for a platform with a core_init_notifier, if you run pci_endpoint_test + reboot the RC (so that PERST is asserted + deasserted), and then run pci_endpoint_test again, then I'm quite sure that pci_endpoint_test will not pass the second time (because iATU settings were not rewritten after reset).
It would be nice if Mani or Vidya could confirm this.
I guess that you added this statement for some reason, so I assume that we can't just drop this line without breaking something else.
I guess one option would be modify dw_pcie_ep_init_notify() to call dw_pcie_ep_clear_bar() on all non-NULL BARs stored in ep->epf_bar[], before calling pci_epc_init_notify(). That way the second .core_init (pci_epf_test_core_init()) call will use write the settings, because ep->epf_bar[] will be empty, so the "write the settings only the first time" approach will then also work for core_init_notifier platforms.
Kind regards,
Niklas
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 54 lines · Niklas Cassel
Quoted 33 lines · Frank Li
Quoted 32 lines · Niklas Cassel
Hello Frank,
Quoted 21 lines · Frank Li
ntb_mw_set_trans()will set memory map window after endpoint function driver bind. The inbound map address need be updated dynamically when using NTB by PCIe Root Port and PCIe Endpoint connection.Checking if iatu already assigned to the BAR, if yes, using assigned iatu number to update inbound address map and skip set BAR's register.
Signed-off-by Frank Li <[email protected]>---
Change from V1:
- improve commit messagedrivers/pci/controller/dwc/pcie-designware-ep.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-1@@ -161,7 +161,11 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no,u32 free_win;struct dw_pcie *pci =to_dw_pcie_from_ep (ep); - free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);
find_first_zero_bit()can return 0, representing bit 0, which is a perfectly valid return value.Quoted 7 lines
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 1 line
drivers/pci/controller/dwc/pcie-designware-ep.c +1
-0+if (!ep->bar_to_atu[bar])so this check is not correct.
Please sent out your fixed patch. If want to me fix it, please tell me reproduce steps.
Reproduce steps are simple:
1) Add debug messages todw_pcie_ep_inbound_atu()to see the iATU index for each BAR.
2) Boot an EP platform with acore_init_notifier.
3) Boot the RC.
4) Reboot the RC, which will assert + deassert PERST, and will callpci_epc_init_notify(), which will call .core_init (pci_epf_test_core_init()) which will set the BARs.In step 3) BAR0 will use iATU0.
Quoted 10 lines
In step 4) BAR0 will use iATU6 instead of iATU0. There is no reason for this, as it should really reuse the same iATU index as before, just like all the other BARs do. (This is because of
find_first_zero_bit()misusage.)I could send out my patch, but from inspecting the code, it looks like:
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +2
-0+if (ep->epf_bar[bar]) +return 0 ;
I checked current code.
dw_pcie_ep_inbound_atu()
{
if (!ep->bar_to_atu[bar])
free_win = find_first_zero_bit(ep->ib_window_map, pci->num_ib_windows);
else
free_win = ep->bar_to_atu[bar];
}I missed conside '0' is validate windows number. I think we should init bar_to_atu to -1.
if (ep->bar_to_atu[bar] < 0)
Origial change want dw_pcie_ep_inbound_atu() can be call twice to update bar map address. vntb use "bar3" as memory map windows, so have not trigger this problem.
Frank
Quoted 26 lines
from
dw_pcie_ep_set_bar(), also needs to be dropped, so that the iATU settings will be re-written for platforms withcore_init_notifiers.Right now, for a platform with a
core_init_notifier, if you runpci_endpoint_test+ reboot the RC (so that PERST is asserted + deasserted), and then runpci_endpoint_testagain, then I'm quite sure thatpci_endpoint_testwill not pass the second time (because iATU settings were not rewritten after reset).It would be nice if Mani or Vidya could confirm this.
I guess that you added this statement for some reason, so I assume that we can't just drop this line without breaking something else.
I guess one option would be modify
dw_pcie_ep_init_notify()to calldw_pcie_ep_clear_bar()on all non-NULL BARs stored in ep->epf_bar[], before callingpci_epc_init_notify(). That way the second .core_init (pci_epf_test_core_init()) call will use write the settings, because ep->epf_bar[] will be empty, so the "write the settings only the first time" approach will then also work forcore_init_notifierplatforms.Kind regards,
Niklas
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 79 lines · Frank Li
Quoted 54 lines · Niklas Cassel
Quoted 33 lines · Frank Li
Quoted 32 lines · Niklas Cassel
Hello Frank,
Quoted 21 lines · Frank Li
ntb_mw_set_trans()will set memory map window after endpoint function driver bind. The inbound map address need be updated dynamically when using NTB by PCIe Root Port and PCIe Endpoint connection.Checking if iatu already assigned to the BAR, if yes, using assigned iatu number to update inbound address map and skip set BAR's register.
Signed-off-by Frank Li <[email protected]>---
Change from V1:
- improve commit messagedrivers/pci/controller/dwc/pcie-designware-ep.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-1@@ -161,7 +161,11 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no,u32 free_win;struct dw_pcie *pci =to_dw_pcie_from_ep (ep); - free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);
find_first_zero_bit()can return 0, representing bit 0, which is a perfectly valid return value.Quoted 7 lines
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 1 line
drivers/pci/controller/dwc/pcie-designware-ep.c +1
-0+if (!ep->bar_to_atu[bar])so this check is not correct.
Please sent out your fixed patch. If want to me fix it, please tell me reproduce steps.
Reproduce steps are simple:
1) Add debug messages todw_pcie_ep_inbound_atu()to see the iATU index for each BAR.
2) Boot an EP platform with acore_init_notifier.
3) Boot the RC.
4) Reboot the RC, which will assert + deassert PERST, and will callpci_epc_init_notify(), which will call .core_init (pci_epf_test_core_init()) which will set the BARs.In step 3) BAR0 will use iATU0.
Quoted 10 lines
In step 4) BAR0 will use iATU6 instead of iATU0. There is no reason for this, as it should really reuse the same iATU index as before, just like all the other BARs do. (This is because of
find_first_zero_bit()misusage.)I could send out my patch, but from inspecting the code, it looks like:
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +2
-0+if (ep->epf_bar[bar]) +return 0 ;I checked current code.
dw_pcie_ep_inbound_atu()
{
if (!ep->bar_to_atu[bar])free_win = find_first_zero_bit(ep->ib_window_map, pci->num_ib_windows); else free_win = ep->bar_to_atu[bar]; }I missed conside '0' is validate windows number. I think we should init
bar_to_atuto -1.
Quoted 6 lines
if (ep->bar_to_atu[bar] < 0)Origial change want
dw_pcie_ep_inbound_atu()can be call twice to update bar map address. vntb use "bar3" as memory map windows, so have not trigger this problem.
Does below change fix your problem?
drivers/pci/controller/dwc/pcie-designware-ep.c +3 -3
@@ -177,7 +177,7 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type,if (!ep->bar_to_atu[bar]) free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);else - free_win = ep->bar_to_atu[bar]; + free_win = ep->bar_to_atu[bar] -1 ;if (free_win >= pci->num_ib_windows) {dev_err (pci->dev,"No free inbound window\n" );
@@ -191,7 +191,7 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type,return ret; } - ep->bar_to_atu[bar] = free_win; + ep->bar_to_atu[bar] = free_win +1 ;set_bit (free_win, ep->ib_window_map);return 0 ;
@@ -228,7 +228,7 @@ static void dw_pcie_ep_clear_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,struct dw_pcie_ep *ep =epc_get_drvdata (epc);struct dw_pcie *pci =to_dw_pcie_from_ep (ep);enum pci_barno bar = epf_bar->barno; -u32 atu_index = ep->bar_to_atu[bar]; +u32 atu_index = ep->bar_to_atu[bar] -1 ;__dw_pcie_ep_reset_bar (pci, func_no, bar, epf_bar->flags);
Quoted 29 lines
Frank
Quoted 26 lines
from
dw_pcie_ep_set_bar(), also needs to be dropped, so that the iATU settings will be re-written for platforms withcore_init_notifiers.Right now, for a platform with a
core_init_notifier, if you runpci_endpoint_test+ reboot the RC (so that PERST is asserted + deasserted), and then runpci_endpoint_testagain, then I'm quite sure thatpci_endpoint_testwill not pass the second time (because iATU settings were not rewritten after reset).It would be nice if Mani or Vidya could confirm this.
I guess that you added this statement for some reason, so I assume that we can't just drop this line without breaking something else.
I guess one option would be modify
dw_pcie_ep_init_notify()to calldw_pcie_ep_clear_bar()on all non-NULL BARs stored in ep->epf_bar[], before callingpci_epc_init_notify(). That way the second .core_init (pci_epf_test_core_init()) call will use write the settings, because ep->epf_bar[] will be empty, so the "write the settings only the first time" approach will then also work forcore_init_notifierplatforms.Kind regards,
Niklas
Niklas Cassel
Tested-byRe: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 1 line · Frank Li
(snip)
Quoted 1 line
Does below change fix your problem?
It is basically the same fix as I sent out earlier in this thread, but yes, it does solve 1 out of 2 problems introduced by the patch in $subject, so:
Quoted 23 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +2
-2@@ -177,7 +177,7 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type,if (!ep->bar_to_atu[bar]) free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);else - free_win = ep->bar_to_atu[bar]; + free_win = ep->bar_to_atu[bar] -1 ;if (free_win >= pci->num_ib_windows) {dev_err (pci->dev,"No free inbound window\n" );@@ -191,7 +191,7 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type,return ret; } - ep->bar_to_atu[bar] = free_win; + ep->bar_to_atu[bar] = free_win +1 ;set_bit (free_win, ep->ib_window_map);return 0 ;
drivers/pci/controller/dwc/pcie-designware-ep.c static void dw_pcie_ep_clear_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no, · line 231
Quoted 5 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +1
-1@@ -228,7 +228,7 @@ static void dw_pcie_ep_clear_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,struct dw_pcie_ep *ep =epc_get_drvdata (epc);struct dw_pcie *pci =to_dw_pcie_from_ep (ep);enum pci_barno bar = epf_bar->barno; -u32 atu_index = ep->bar_to_atu[bar]; +u32 atu_index = ep->bar_to_atu[bar] -1 ;
You probably want to add a:
if (!ep->bar_to_atu[bar])
return;
here, so that dw_pcie_ep_clear_bar() will never try to use -1 as index. (E.g. if clear_bar() is called twice on the same BAR.)
Quoted 3 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0__dw_pcie_ep_reset_bar (pci, func_no, bar, epf_bar->flags);
(snip)
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
from
dw_pcie_ep_set_bar(), also needs to be dropped, so that the iATU settings will be re-written for platforms withcore_init_notifiers.Right now, for a platform with a
core_init_notifier, if you runpci_endpoint_test+ reboot the RC (so that PERST is asserted + deasserted), and then runpci_endpoint_testagain, then I'm quite sure thatpci_endpoint_testwill not pass the second time (because iATU settings were not rewritten after reset).It would be nice if Mani or Vidya could confirm this.
So problem 2 out of 2 introduced by the patch in $subject is that DWC drivers with a .core_init_notifier, will perform a reset_control_assert() to reset the core (which will reset both sticky and non-sticky registers), which means that the early return in dw_pcie_ep_set_bar():
https://github.com/torvalds/linux/blob/v6.7-rc5/drivers/pci/controller/dwc/pcie-designware-ep.c#L268-L269
while returning after the iATU settings have been written, it will return before:
dw_pcie_writel_dbi2(pci, reg_dbi2, lower_32_bits(size - 1)); dw_pcie_writel_dbi(pci, reg, flags);
Which means that, for drivers with a .core_init_notifier, BARx_REG and BARx_MASK registers will not be written. This means that they will have reset values for these registers, which means that e.g. the BAR_SIZE (which is defined by BARx_MASK) might be incorrect for these platforms because this function returns early.
I will not send a fix for this problem, I will leave that to you, or Mani, or Vidya, and hope that people are happy that I simply reported this issue.
Here is my suggested solution in case anyone wants to take a stab at it:
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
I guess one option would be modify
dw_pcie_ep_init_notify()to calldw_pcie_ep_clear_bar()on all non-NULL BARs stored in ep->epf_bar[], before callingpci_epc_init_notify(). That way the second .core_init (pci_epf_test_core_init()) call will use write the settings, because ep->epf_bar[] will be empty, so the "write the settings only the first time" approach will then also work forcore_init_notifierplatforms.
Kind regards,
Niklas
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 79 lines · Niklas Cassel
Quoted 1 line · Frank Li
(snip)
Quoted 1 line
Does below change fix your problem?
It is basically the same fix as I sent out earlier in this thread, but yes, it does solve 1 out of 2 problems introduced by the patch in $subject, so:
Tested-by Niklas Cassel <[email protected]>Quoted 23 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +2
-2@@ -177,7 +177,7 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type,if (!ep->bar_to_atu[bar]) free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);else - free_win = ep->bar_to_atu[bar]; + free_win = ep->bar_to_atu[bar] -1 ;if (free_win >= pci->num_ib_windows) {dev_err (pci->dev,"No free inbound window\n" );@@ -191,7 +191,7 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, int type,return ret; } - ep->bar_to_atu[bar] = free_win; + ep->bar_to_atu[bar] = free_win +1 ;set_bit (free_win, ep->ib_window_map);return 0 ;drivers/pci/controller/dwc/pcie-designware-ep.c static void dw_pcie_ep_clear_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no, · line 231
Quoted 5 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +1
-1@@ -228,7 +228,7 @@ static void dw_pcie_ep_clear_bar(struct pci_epc *epc, u8 func_no, u8 vfunc_no,struct dw_pcie_ep *ep =epc_get_drvdata (epc);struct dw_pcie *pci =to_dw_pcie_from_ep (ep);enum pci_barno bar = epf_bar->barno; -u32 atu_index = ep->bar_to_atu[bar]; +u32 atu_index = ep->bar_to_atu[bar] -1 ;You probably want to add a:
if (!ep->bar_to_atu[bar])
return;here, so that
dw_pcie_ep_clear_bar()will never try to use -1 as index. (E.g. ifclear_bar()is called twice on the same BAR.)Quoted 3 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0__dw_pcie_ep_reset_bar (pci, func_no, bar, epf_bar->flags);(snip)
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
from
dw_pcie_ep_set_bar(), also needs to be dropped, so that the iATU settings will be re-written for platforms withcore_init_notifiers.Right now, for a platform with a
core_init_notifier, if you runpci_endpoint_test+ reboot the RC (so that PERST is asserted + deasserted), and then runpci_endpoint_testagain, then I'm quite sure thatpci_endpoint_testwill not pass the second time (because iATU settings were not rewritten after reset).It would be nice if Mani or Vidya could confirm this.
So problem 2 out of 2 introduced by the patch in $subject is that DWC drivers with a .core_init_notifier, will perform a
reset_control_assert()to reset the core (which will reset both sticky and non-sticky registers), which means that the early return indw_pcie_ep_set_bar():
https://github.com/torvalds/linux/blob/v6.7-rc5/drivers/pci/controller/dwc/pcie-designware-ep.c#L268-L269while returning after the iATU settings have been written, it will return before:
dw_pcie_writel_dbi2(pci, reg_dbi2, lower_32_bits(size - 1)); dw_pcie_writel_dbi(pci, reg, flags);
Quoted 9 lines
Which means that, for drivers with a .core_init_notifier,
BARx_REGandBARx_MASKregisters will not be written. This means that they will have reset values for these registers, which means that e.g. theBAR_SIZE(which is defined byBARx_MASK) might be incorrect for these platforms because this function returns early.I will not send a fix for this problem, I will leave that to you, or Mani, or Vidya, and hope that people are happy that I simply reported this issue.
The fix for this issue shouldn't be in the DWC driver but rather in the EPF drivers. Because, EPF drivers are the ones calling pci_epc_set_bar() during bind(). So they have to properly cleanup the resources once the perst assertion happens. This issue also applies to other resources such as DMA channels.
The problem here is, there is a disconnect between DWC (EPC) and EPF drivers. When the perst assertion happens, the event is not passed to EPF drivers resulting in the EPF drivers having no knowledge that they need to give up the resources.
We need to fix this in a sane way and I'm looking into it.
But I really appreciate your finding here and in other thread where we discussed similar issue.
- Mani
Quoted 13 lines
Here is my suggested solution in case anyone wants to take a stab at it:
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
I guess one option would be modify
dw_pcie_ep_init_notify()to calldw_pcie_ep_clear_bar()on all non-NULL BARs stored in ep->epf_bar[], before callingpci_epc_init_notify(). That way the second .core_init (pci_epf_test_core_init()) call will use write the settings, because ep->epf_bar[] will be empty, so the "write the settings only the first time" approach will then also work forcore_init_notifierplatforms.Kind regards,
Niklas
Signature · 1 line
மணிவண்ணன் சதாசிவம்
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 43 lines · Manivannan Sadhasivam
Quoted 23 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
from
dw_pcie_ep_set_bar(), also needs to be dropped, so that the iATU settings will be re-written for platforms withcore_init_notifiers.Right now, for a platform with a
core_init_notifier, if you runpci_endpoint_test+ reboot the RC (so that PERST is asserted + deasserted), and then runpci_endpoint_testagain, then I'm quite sure thatpci_endpoint_testwill not pass the second time (because iATU settings were not rewritten after reset).It would be nice if Mani or Vidya could confirm this.
So problem 2 out of 2 introduced by the patch in $subject is that DWC drivers with a .core_init_notifier, will perform a
reset_control_assert()to reset the core (which will reset both sticky and non-sticky registers), which means that the early return indw_pcie_ep_set_bar():
https://github.com/torvalds/linux/blob/v6.7-rc5/drivers/pci/controller/dwc/pcie-designware-ep.c#L268-L269while returning after the iATU settings have been written, it will return before:
dw_pcie_writel_dbi2(pci, reg_dbi2, lower_32_bits(size - 1)); dw_pcie_writel_dbi(pci, reg, flags);Quoted 9 lines
Which means that, for drivers with a .core_init_notifier,
BARx_REGandBARx_MASKregisters will not be written. This means that they will have reset values for these registers, which means that e.g. theBAR_SIZE(which is defined byBARx_MASK) might be incorrect for these platforms because this function returns early.I will not send a fix for this problem, I will leave that to you, or Mani, or Vidya, and hope that people are happy that I simply reported this issue.
The fix for this issue shouldn't be in the DWC driver but rather in the EPF drivers. Because, EPF drivers are the ones calling
pci_epc_set_bar()duringbind(). So they have to properly cleanup the resources once the perst assertion happens. This issue also applies to other resources such as DMA channels.The problem here is, there is a disconnect between DWC (EPC) and EPF drivers. When the perst assertion happens, the event is not passed to EPF drivers resulting in the EPF drivers having no knowledge that they need to give up the resources.
Quoted 4 lines
We need to fix this in a sane way and I'm looking into it.
But I really appreciate your finding here and in other thread where we discussed similar issue.
We have core_init and link_up notifiers for EPF drivers. Adding link_down and core_exit notifiers would make a lot of sense :) A core_reset notifier could also be a solution.
Adding that would also help EPF drivers to handle links going down temporarily (which can happen). Right now, an EPF driver can only deal with such event by tracking if it gets multiple link_up events.
Signature · 2 lines
Damien Le Moal Western Digital Research
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 44 lines · Damien Le Moal
Quoted 43 lines · Manivannan Sadhasivam
Quoted 23 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
from
dw_pcie_ep_set_bar(), also needs to be dropped, so that the iATU settings will be re-written for platforms withcore_init_notifiers.Right now, for a platform with a
core_init_notifier, if you runpci_endpoint_test+ reboot the RC (so that PERST is asserted + deasserted), and then runpci_endpoint_testagain, then I'm quite sure thatpci_endpoint_testwill not pass the second time (because iATU settings were not rewritten after reset).It would be nice if Mani or Vidya could confirm this.
So problem 2 out of 2 introduced by the patch in $subject is that DWC drivers with a .core_init_notifier, will perform a
reset_control_assert()to reset the core (which will reset both sticky and non-sticky registers), which means that the early return indw_pcie_ep_set_bar():
https://github.com/torvalds/linux/blob/v6.7-rc5/drivers/pci/controller/dwc/pcie-designware-ep.c#L268-L269while returning after the iATU settings have been written, it will return before:
dw_pcie_writel_dbi2(pci, reg_dbi2, lower_32_bits(size - 1)); dw_pcie_writel_dbi(pci, reg, flags);Quoted 9 lines
Which means that, for drivers with a .core_init_notifier,
BARx_REGandBARx_MASKregisters will not be written. This means that they will have reset values for these registers, which means that e.g. theBAR_SIZE(which is defined byBARx_MASK) might be incorrect for these platforms because this function returns early.I will not send a fix for this problem, I will leave that to you, or Mani, or Vidya, and hope that people are happy that I simply reported this issue.
The fix for this issue shouldn't be in the DWC driver but rather in the EPF drivers. Because, EPF drivers are the ones calling
pci_epc_set_bar()duringbind(). So they have to properly cleanup the resources once the perst assertion happens. This issue also applies to other resources such as DMA channels.The problem here is, there is a disconnect between DWC (EPC) and EPF drivers. When the perst assertion happens, the event is not passed to EPF drivers resulting in the EPF drivers having no knowledge that they need to give up the resources.
Quoted 9 lines
Quoted 4 lines
We need to fix this in a sane way and I'm looking into it.
But I really appreciate your finding here and in other thread where we discussed similar issue.
We have
core_initandlink_upnotifiers for EPF drivers. Addinglink_downandcore_exitnotifiers would make a lot of sense :) Acore_resetnotifier could also be a solution.
Yeah, I'm thinking along those lines.
Quoted 4 lines
Adding that would also help EPF drivers to handle links going down temporarily (which can happen). Right now, an EPF driver can only deal with such event by tracking if it gets multiple
link_upevents.
There is already link_down notifier that I introduced a while back. But it is only used by MHI_EPF driver.
- Mani
Quoted 6 lines
Signature · 3 lines
-- Damien Le Moal Western Digital Research
Signature · 1 line
மணிவண்ணன் சதாசிவம்
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
Quoted 7 lines · Manivannan Sadhasivam
Quoted 4 lines
We have
core_initandlink_upnotifiers for EPF drivers. Addinglink_downandcore_exitnotifiers would make a lot of sense :) Acore_resetnotifier could also be a solution.Yeah, I'm thinking along those lines.
Quoted 7 lines
Quoted 4 lines
Adding that would also help EPF drivers to handle links going down temporarily (which can happen). Right now, an EPF driver can only deal with such event by tracking if it gets multiple
link_upevents.There is already
link_downnotifier that I introduced a while back. But it is only used byMHI_EPFdriver.
I missed that. I need that notifier :)
Signature · 2 lines
Damien Le Moal Western Digital Research
Re: [PATCH v2 1/4] PCI: designware-ep: Allow pci_epc_set_bar() update inbound map address
In reply to Frank LiQuoted 79 lines · Frank Li
Quoted 54 lines · Niklas Cassel
Quoted 33 lines · Frank Li
Quoted 32 lines · Niklas Cassel
Hello Frank,
Quoted 21 lines · Frank Li
ntb_mw_set_trans()will set memory map window after endpoint function driver bind. The inbound map address need be updated dynamically when using NTB by PCIe Root Port and PCIe Endpoint connection.Checking if iatu already assigned to the BAR, if yes, using assigned iatu number to update inbound address map and skip set BAR's register.
Signed-off-by Frank Li <[email protected]>---
Change from V1:
- improve commit messagedrivers/pci/controller/dwc/pcie-designware-ep.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)drivers/pci/controller/dwc/pcie-designware-ep.c +0
-0drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +0
-1@@ -161,7 +161,11 @@ static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no,u32 free_win;struct dw_pcie *pci =to_dw_pcie_from_ep (ep); - free_win =find_first_zero_bit (ep->ib_window_map, pci->num_ib_windows);
find_first_zero_bit()can return 0, representing bit 0, which is a perfectly valid return value.Quoted 7 lines
Quoted 4 lines
drivers/pci/controller/dwc/pcie-designware-ep.c static int dw_pcie_ep_inbound_atu(struct dw_pcie_ep *ep, u8 func_no, · line 164
Quoted 1 line
drivers/pci/controller/dwc/pcie-designware-ep.c +1
-0+if (!ep->bar_to_atu[bar])so this check is not correct.
Please sent out your fixed patch. If want to me fix it, please tell me reproduce steps.
Reproduce steps are simple:
1) Add debug messages todw_pcie_ep_inbound_atu()to see the iATU index for each BAR.
2) Boot an EP platform with acore_init_notifier.
3) Boot the RC.
4) Reboot the RC, which will assert + deassert PERST, and will callpci_epc_init_notify(), which will call .core_init (pci_epf_test_core_init()) which will set the BARs.In step 3) BAR0 will use iATU0.
Quoted 10 lines
In step 4) BAR0 will use iATU6 instead of iATU0. There is no reason for this, as it should really reuse the same iATU index as before, just like all the other BARs do. (This is because of
find_first_zero_bit()misusage.)I could send out my patch, but from inspecting the code, it looks like:
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
drivers/pci/controller/dwc/pcie-designware-ep.c +2
-0+if (ep->epf_bar[bar]) +return 0 ;I checked current code.
dw_pcie_ep_inbound_atu()
{
if (!ep->bar_to_atu[bar])free_win = find_first_zero_bit(ep->ib_window_map, pci->num_ib_windows); else free_win = ep->bar_to_atu[bar]; }I missed conside '0' is validate windows number. I think we should init
bar_to_atuto -1.
Quoted 6 lines
if (ep->bar_to_atu[bar] < 0)Origial change want
dw_pcie_ep_inbound_atu()can be call twice to update bar map address. vntb use "bar3" as memory map windows, so have not trigger this problem.
Feel free to send out a patch, I just wanted to inform people about the problem.
I can understand that the point of the patch in $subject was to re-use the exact same iATU for a specific BAR, however, does the change:
if (ep->epf_bar[bar])
return 0;
to dw_pcie_ep_set_bar(), really provide any value?
I guess it will avoid clearing the BAR_REG, if the RC has assigned an PCI address to that BAR, which I guess was important for vntb?
If so, I don't see any alternative but to let DWC drivers with a core_init_notifier to call dw_pcie_ep_clear_bar() when asserting perst, or to modify dw_pcie_ep_init_notify() to call dw_pcie_ep_clear_bar().
Kind regards,
Niklas
PATCH v2 2/4 NTB: epf: Allow more flexibility in the memory BAR map method
In reply to Frank LiSupport the below BAR configuration methods for epf NTB.
BAR 0: config and scratchpad
BAR 2: doorbell
BAR 4: memory map windows
Set difference BAR number information into struct ntb_epf_data. So difference VID/PID can choose different BAR configurations. There are difference BAR map method between epf NTB and epf vNTB Endpoint function.
---
Change from v1:
- Improve commit message
drivers/ntb/hw/epf/ntb_hw_epf.c | 48 ++++++++++++++++++++++++--------- 1 file changed, 35 insertions(+), 13 deletions(-)
drivers/ntb/hw/epf/ntb_hw_epf.c +35 -13
@@ -45,7 +45,6 @@#define NTB_EPF_MIN_DB_COUNT 3#define NTB_EPF_MAX_DB_COUNT 31 -#define NTB_EPF_MW_OFFSET 2#define NTB_EPF_COMMAND_TIMEOUT 1000/* 1 Sec */
@@ -67,6 +66,7 @@ struct ntb_epf_dev {enum pci_barno ctrl_reg_bar;enum pci_barno peer_spad_reg_bar;enum pci_barno db_reg_bar; +enum pci_barno mw_bar;unsigned int mw_count;unsigned int spad_count;
@@ -92,6 +92,8 @@ struct ntb_epf_data {enum pci_barno peer_spad_reg_bar;/* BAR that contains Doorbell region and Memory window '1' */ enum pci_barno db_reg_bar; +/* BAR that contains memory windows*/ +enum pci_barno mw_bar; }; static int ntb_epf_send_command(struct ntb_epf_dev *ndev, u32 command,
@@ -411,7 +413,7 @@ static int ntb_epf_mw_set_trans(struct ntb_dev *ntb, int pidx, int idx,return -EINVAL ; } - bar = idx +NTB_EPF_MW_OFFSET ; + bar = idx + ndev->mw_bar; mw_size =pci_resource_len (ntb->pdev, bar);
@@ -453,7 +455,7 @@ static int ntb_epf_peer_mw_get_addr(struct ntb_dev *ntb, int idx,if (idx ==0 ) offset =readl (ndev->ctrl_reg +NTB_EPF_MW1_OFFSET ); - bar = idx +NTB_EPF_MW_OFFSET ; + bar = idx + ndev->mw_bar;if (base) *base =pci_resource_start (ndev->ntb.pdev, bar) + offset;
@@ -565,6 +567,7 @@ static int ntb_epf_init_pci(struct ntb_epf_dev *ndev, struct pci_dev *pdev) {struct device *dev = ndev->dev; +size_t spad_sz, spad_off;int ret;pci_set_drvdata (pdev, ndev);
@@ -599,10 +602,16 @@ static int ntb_epf_init_pci(struct ntb_epf_dev *ndev, goto err_dma_mask; } - ndev->peer_spad_reg =pci_iomap (pdev, ndev->peer_spad_reg_bar,0 ); -if (!ndev->peer_spad_reg) { - ret = -EIO ; - goto err_dma_mask; +if (ndev->peer_spad_reg_bar) { + ndev->peer_spad_reg =pci_iomap (pdev, ndev->peer_spad_reg_bar,0 ); +if (!ndev->peer_spad_reg) { + ret = -EIO ; + goto err_dma_mask; + } + }else { + spad_sz =4 *readl (ndev->ctrl_reg +NTB_EPF_SPAD_COUNT ); + spad_off =readl (ndev->ctrl_reg +NTB_EPF_SPAD_OFFSET ); + ndev->peer_spad_reg = ndev->ctrl_reg + spad_off + spad_sz; } ndev->db_reg =pci_iomap (pdev, ndev->db_reg_bar,0 );
@@ -657,6 +666,7 @@ static int ntb_epf_pci_probe(struct pci_dev *pdev,enum pci_barno peer_spad_reg_bar =BAR_1 ;enum pci_barno ctrl_reg_bar =BAR_0 ;enum pci_barno db_reg_bar =BAR_2 ; +enum pci_barno mw_bar =BAR_2 ;struct device *dev = &pdev->dev;struct ntb_epf_data *data;struct ntb_epf_dev *ndev;
@@ -671,17 +681,16 @@ static int ntb_epf_pci_probe(struct pci_dev *pdev, data = (struct ntb_epf_data *)id->driver_data;if (data) { -if (data->peer_spad_reg_bar) - peer_spad_reg_bar = data->peer_spad_reg_bar; -if (data->ctrl_reg_bar) - ctrl_reg_bar = data->ctrl_reg_bar; -if (data->db_reg_bar) - db_reg_bar = data->db_reg_bar; + peer_spad_reg_bar = data->peer_spad_reg_bar; + ctrl_reg_bar = data->ctrl_reg_bar; + db_reg_bar = data->db_reg_bar; + mw_bar = data->mw_bar; } ndev->peer_spad_reg_bar = peer_spad_reg_bar; ndev->ctrl_reg_bar = ctrl_reg_bar; ndev->db_reg_bar = db_reg_bar; + ndev->mw_bar = mw_bar; ndev->dev = dev;ntb_epf_init_struct (ndev, pdev);
@@ -729,6 +738,14 @@ static const struct ntb_epf_data j721e_data = { .ctrl_reg_bar = BAR_0, .peer_spad_reg_bar = BAR_1, .db_reg_bar = BAR_2, + .mw_bar = BAR_2, +}; + +static const struct ntb_epf_data mx8_data = { + .ctrl_reg_bar =BAR_0 , + .peer_spad_reg_bar =BAR_0 , + .db_reg_bar =BAR_2 , + .mw_bar =BAR_4 , };static const struct pci_device_id ntb_epf_pci_tbl[] = {
@@ -737,6 +754,11 @@ static const struct pci_device_id ntb_epf_pci_tbl[] = { .class = PCI_CLASS_MEMORY_RAM << 8, .class_mask = 0xffff00, .driver_data = (kernel_ulong_t)&j721e_data, }, + { + PCI_DEVICE(PCI_VENDOR_ID_FREESCALE, 0x0809), + .class = PCI_CLASS_MEMORY_RAM << 8, .class_mask = 0xffff00, + .driver_data = (kernel_ulong_t)&mx8_data, + }, { }, };
Signature · 1 line
2.24.0.rc1
Re: [PATCH v2 2/4] NTB: epf: Allow more flexibility in the memory BAR map method
Quoted 7 lines · Frank Li
Support the below BAR configuration methods for epf NTB.
BAR 0: config and scratchpad
BAR 2: doorbell
BAR 4: memory map windows
Quoted 6 lines
Set difference BAR number information into struct
ntb_epf_data. So difference VID/PID can choose different BAR configurations. There are difference BAR map method between epf NTB and epf vNTB Endpoint function.Signed-off-by Frank Li <[email protected]>---
Update ntb mail list.
Quoted 144 lines
Change from v1:
- Improve commit messagedrivers/ntb/hw/epf/ntb_hw_epf.c | 48 ++++++++++++++++++++++++--------- 1 file changed, 35 insertions(+), 13 deletions(-)drivers/ntb/hw/epf/ntb_hw_epf.c +35
-14@@ -45,7 +45,6 @@#define NTB_EPF_MIN_DB_COUNT 3#define NTB_EPF_MAX_DB_COUNT 31 -#define NTB_EPF_MW_OFFSET 2#define NTB_EPF_COMMAND_TIMEOUT 1000/* 1 Sec */ @@ -67,6 +66,7 @@ struct ntb_epf_dev {enum pci_barno ctrl_reg_bar;enum pci_barno peer_spad_reg_bar;enum pci_barno db_reg_bar; +enum pci_barno mw_bar;unsigned int mw_count;unsigned int spad_count;@@ -92,6 +92,8 @@ struct ntb_epf_data {enum pci_barno peer_spad_reg_bar;/* BAR that contains Doorbell region and Memory window '1' */ enum pci_barno db_reg_bar; +/* BAR that contains memory windows*/ +enum pci_barno mw_bar; }; static int ntb_epf_send_command(struct ntb_epf_dev *ndev, u32 command,@@ -411,7 +413,7 @@ static int ntb_epf_mw_set_trans(struct ntb_dev *ntb, int pidx, int idx,return -EINVAL ; } - bar = idx +NTB_EPF_MW_OFFSET ; + bar = idx + ndev->mw_bar; mw_size =pci_resource_len (ntb->pdev, bar);@@ -453,7 +455,7 @@ static int ntb_epf_peer_mw_get_addr(struct ntb_dev *ntb, int idx,if (idx ==0 ) offset =readl (ndev->ctrl_reg +NTB_EPF_MW1_OFFSET ); - bar = idx +NTB_EPF_MW_OFFSET ; + bar = idx + ndev->mw_bar;if (base) *base =pci_resource_start (ndev->ntb.pdev, bar) + offset;@@ -565,6 +567,7 @@ static int ntb_epf_init_pci(struct ntb_epf_dev *ndev, struct pci_dev *pdev) {struct device *dev = ndev->dev; +size_t spad_sz, spad_off;int ret;pci_set_drvdata (pdev, ndev);@@ -599,10 +602,16 @@ static int ntb_epf_init_pci(struct ntb_epf_dev *ndev, goto err_dma_mask; } - ndev->peer_spad_reg =pci_iomap (pdev, ndev->peer_spad_reg_bar,0 ); -if (!ndev->peer_spad_reg) { - ret = -EIO ; - goto err_dma_mask; +if (ndev->peer_spad_reg_bar) { + ndev->peer_spad_reg =pci_iomap (pdev, ndev->peer_spad_reg_bar,0 ); +if (!ndev->peer_spad_reg) { + ret = -EIO ; + goto err_dma_mask; + } + }else { + spad_sz =4 *readl (ndev->ctrl_reg +NTB_EPF_SPAD_COUNT ); + spad_off =readl (ndev->ctrl_reg +NTB_EPF_SPAD_OFFSET ); + ndev->peer_spad_reg = ndev->ctrl_reg + spad_off + spad_sz; } ndev->db_reg =pci_iomap (pdev, ndev->db_reg_bar,0 );@@ -657,6 +666,7 @@ static int ntb_epf_pci_probe(struct pci_dev *pdev,enum pci_barno peer_spad_reg_bar =BAR_1 ;enum pci_barno ctrl_reg_bar =BAR_0 ;enum pci_barno db_reg_bar =BAR_2 ; +enum pci_barno mw_bar =BAR_2 ;struct device *dev = &pdev->dev;struct ntb_epf_data *data;struct ntb_epf_dev *ndev;@@ -671,17 +681,16 @@ static int ntb_epf_pci_probe(struct pci_dev *pdev, data = (struct ntb_epf_data *)id->driver_data;if (data) { -if (data->peer_spad_reg_bar) - peer_spad_reg_bar = data->peer_spad_reg_bar; -if (data->ctrl_reg_bar) - ctrl_reg_bar = data->ctrl_reg_bar; -if (data->db_reg_bar) - db_reg_bar = data->db_reg_bar; + peer_spad_reg_bar = data->peer_spad_reg_bar; + ctrl_reg_bar = data->ctrl_reg_bar; + db_reg_bar = data->db_reg_bar; + mw_bar = data->mw_bar; } ndev->peer_spad_reg_bar = peer_spad_reg_bar; ndev->ctrl_reg_bar = ctrl_reg_bar; ndev->db_reg_bar = db_reg_bar; + ndev->mw_bar = mw_bar; ndev->dev = dev;ntb_epf_init_struct (ndev, pdev);@@ -729,6 +738,14 @@ static const struct ntb_epf_data j721e_data = { .ctrl_reg_bar = BAR_0, .peer_spad_reg_bar = BAR_1, .db_reg_bar = BAR_2, + .mw_bar = BAR_2, +}; + +static const struct ntb_epf_data mx8_data = { + .ctrl_reg_bar =BAR_0 , + .peer_spad_reg_bar =BAR_0 , + .db_reg_bar =BAR_2 , + .mw_bar =BAR_4 , };static const struct pci_device_id ntb_epf_pci_tbl[] = {@@ -737,6 +754,11 @@ static const struct pci_device_id ntb_epf_pci_tbl[] = { .class = PCI_CLASS_MEMORY_RAM << 8, .class_mask = 0xffff00, .driver_data = (kernel_ulong_t)&j721e_data, }, + { + PCI_DEVICE(PCI_VENDOR_ID_FREESCALE, 0x0809), + .class = PCI_CLASS_MEMORY_RAM << 8, .class_mask = 0xffff00, + .driver_data = (kernel_ulong_t)&mx8_data, + }, { }, }; --2.24.0.rc1
PATCH v2 3/4 PCI: endpoint: Support NTB transfer between RC and EP
In reply to Frank LiAdd NTB function driver and virtual PCI Bus and Virtual NTB driver to implement communication between PCIe Root Port and PCIe EP devices
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI Bus │ ◄─────► │ PCI EP Bus │ │ Virtual PCI │ │ │ PCI │ │ │ Bus │ └────────────┘ └───────────────┴──────┴──────────────┘ PCIe Root Port PCI EP
This driver includes 3 parts:
1 PCI EP NTB function driver
2 Virtual PCI bus
3 PCI virtual NTB driver, which is loaded only by above virtual PCI bus
---
Change from v1:
- Fix code style problem according to Bjorn's feedback
- Remove hardcode VPCI_BUS_NUM, let user choose free number by configfs
- Remove hardcode vid pid for virtual pci ntb driver, change by configfs
drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 3 files changed, 1436 insertions(+) create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c
drivers/pci/endpoint/functions/Kconfig +11 -0
@@ -25,3 +25,14 @@ config PCI_EPF_NTB device tree. If in doubt, say "N" to disable Endpoint NTB driver. + +config PCI_EPF_VNTB + tristate "PCI Endpoint NTB driver" + depends on PCI_ENDPOINT + select CONFIGFS_FS + help + Select this configuration option to enable the Non-Transparent + Bridge (NTB) driver for PCIe Endpoint. NTB driver implements NTB + between PCI Root Port and PCIe Endpoint. + + If in doubt, say "N" to disable Endpoint NTB driver.
drivers/pci/endpoint/functions/Makefile +1 -0
@@ -5,3 +5,4 @@ obj-$(CONFIG_PCI_EPF_TEST) += pci-epf-test.o obj-$(CONFIG_PCI_EPF_NTB) += pci-epf-ntb.o +obj-$(CONFIG_PCI_EPF_VNTB) += pci-epf-vntb.o
drivers/pci/endpoint/functions/pci-epf-vntb.c new file +1424 -0
1,424 lines not on this page · raw message
Signature · 1 line
2.24.0.rc1
Re: [PATCH v2 3/4] PCI: endpoint: Support NTB transfer between RC and EP
Quoted 27 lines · Frank Li
Add NTB function driver and virtual PCI Bus and Virtual NTB driver to implement communication between PCIe Root Port and PCIe EP devices
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI Bus │ ◄─────► │ PCI EP Bus │ │ Virtual PCI │ │ │ PCI │ │ │ Bus │ └────────────┘ └───────────────┴──────┴──────────────┘ PCIe Root Port PCI EP
Quoted 7 lines
This driver includes 3 parts:
1 PCI EP NTB function driver
2 Virtual PCI bus
3 PCI virtual NTB driver, which is loaded only by above virtual PCI busSigned-off-by Frank Li <[email protected]>---
Update ntb mail list.
Quoted 1,473 lines
Change from v1:
- Fix code style problem according to Bjorn's feedback
- Remove hardcodeVPCI_BUS_NUM, let user choose free number by configfs
- Remove hardcode vid pid for virtual pci ntb driver, change by configfsdrivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 3 files changed, 1436 insertions(+) create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.cdrivers/pci/endpoint/functions/Kconfig +11
-0@@ -25,3 +25,14 @@ config PCI_EPF_NTB device tree. If in doubt, say "N" to disable Endpoint NTB driver. + +config PCI_EPF_VNTB + tristate "PCI Endpoint NTB driver" + depends on PCI_ENDPOINT + select CONFIGFS_FS + help + Select this configuration option to enable the Non-Transparent + Bridge (NTB) driver for PCIe Endpoint. NTB driver implements NTB + between PCI Root Port and PCIe Endpoint. + + If in doubt, say "N" to disable Endpoint NTB driver.drivers/pci/endpoint/functions/Makefile +1
-0@@ -5,3 +5,4 @@ obj-$(CONFIG_PCI_EPF_TEST) += pci-epf-test.o obj-$(CONFIG_PCI_EPF_NTB) += pci-epf-ntb.o +obj-$(CONFIG_PCI_EPF_VNTB) += pci-epf-vntb.odrivers/pci/endpoint/functions/pci-epf-vntb.c new file +1424
-1@@ -0,0 +1,1424 @@ +// SPDX-License-Identifier: GPL-2.0 +/* +* Endpoint Function Driver to implement Non-Transparent Bridge functionality +* Between PCI RC and EP +* +* Copyright (C) 2020 Texas Instruments +* Copyright (C) 2022 NXP +* +* Based on pci-epf-ntb.c +* Author: Frank Li <[email protected]> +* Author: Kishon Vijay Abraham I <[email protected]> +*/ + +/** +* +------------+ +---------------------------------------+ +* | | | | +* +------------+ | +--------------+ +* | NTB | | | NTB | +* | NetDev | | | NetDev | +* +------------+ | +--------------+ +* | NTB | | | NTB | +* | Transfer | | | Transfer | +* +------------+ | +--------------+ +* | | | | | +* | PCI NTB | | | | +* | EPF | | | | +* | Driver | | | PCI Virtual | +* | | +---------------+ | NTB Driver | +* | | | PCI EP NTB |<------>| | +* | | | FN Driver | | | +* +------------+ +---------------+ +--------------+ +* | | | | | | +* | PCI Bus | <-----> | PCI EP Bus | | Virtual PCI | +* | | PCI | | | Bus | +* +------------+ +---------------+--------+--------------+ +* PCIe Root Port PCI EP +*/ + +#include <linux/delay.h> +#include <linux/io.h> +#include <linux/module.h> +#include <linux/slab.h> + +#include <linux/pci-epc.h> +#include <linux/pci-epf.h> +#include <linux/ntb.h> + +static struct workqueue_struct *kpcintb_workqueue; + +#define COMMAND_CONFIGURE_DOORBELL 1 +#define COMMAND_TEARDOWN_DOORBELL 2 +#define COMMAND_CONFIGURE_MW 3 +#define COMMAND_TEARDOWN_MW 4 +#define COMMAND_LINK_UP 5 +#define COMMAND_LINK_DOWN 6 + +#define COMMAND_STATUS_OK 1 +#define COMMAND_STATUS_ERROR 2 + +#define LINK_STATUS_UP BIT(0) + +#define SPAD_COUNT 64 +#define DB_COUNT 4 +#define NTB_MW_OFFSET 2 +#define DB_COUNT_MASK GENMASK(15, 0) +#define MSIX_ENABLE BIT(16) +#define MAX_DB_COUNT 32 +#define MAX_MW 4 + +enum epf_ntb_bar { +BAR_CONFIG , +BAR_DB , +BAR_MW0 , +BAR_MW1 , +BAR_MW2 , +}; + +/* +* +--------------------------------------------------+ Base +* | | +* | | +* | | +* | Common Control Register | +* | | +* | | +* | | +* +-----------------------+--------------------------+ Base+span_offset +* | | | +* | Peer Span Space | Span Space | +* | | | +* | | | +* +-----------------------+--------------------------+ Base+span_offset +* | | | +span_count * 4 +* | | | +* | Span Space | Peer Span Space | +* | | | +* +-----------------------+--------------------------+ +* Virtual PCI PCIe Endpoint +* NTB Driver NTB Driver +*/ +struct epf_ntb_ctrl { +u32 command; +u32 argument; +u16 command_status; +u16 link_status; +u32 topology; +u64 addr; +u64 size; +u32 num_mws; +u32 reserved; +u32 spad_offset; +u32 spad_count; +u32 db_entry_size; +u32 db_data[MAX_DB_COUNT ]; +u32 db_offset[MAX_DB_COUNT ]; +} __packed; + +struct epf_ntb { +struct ntb_dev ntb; +struct pci_epf *epf; +struct config_group group; + +u32 num_mws; +u32 db_count; +u32 spad_count; +u64 mws_size[MAX_MW ]; +u64 db; +u32 vbus_number; +u16 vntb_pid; +u16 vntb_vid; + +bool linkup; +u32 spad_size; + +enum pci_barno epf_ntb_bar[6 ]; + +struct epf_ntb_ctrl *reg; + +phys_addr_t epf_db_phy; +void __iomem *epf_db; + +phys_addr_t vpci_mw_phy[MAX_MW ]; +void __iomem *vpci_mw_addr[MAX_MW ]; + +struct delayed_work cmd_handler; +}; + +#define to_epf_ntb (epf_group) container_of((epf_group), struct epf_ntb, group) +#define ntb_ndev (__ntb) container_of(__ntb, struct epf_ntb, ntb) + +static struct pci_epf_header epf_ntb_header = { + .vendorid =PCI_ANY_ID , + .deviceid =PCI_ANY_ID , + .baseclass_code =PCI_BASE_CLASS_MEMORY , + .interrupt_pin =PCI_INTERRUPT_INTA , +}; + +/** +* epf_ntb_link_up() - Raise link_up interrupt to Virtual Host +* @ntb: NTB device that facilitates communication between HOST and VHOST +* @link_up: true or false indicating Link is UP or Down +* +* Once NTB function in HOST invoke ntb_link_enable(), +* this NTB function driver will trigger a link event to vhost. +*/ +static int epf_ntb_link_up (struct epf_ntb *ntb,bool link_up) +{ +if (link_up) + ntb->reg->link_status |=LINK_STATUS_UP ; +else + ntb->reg->link_status &= ~LINK_STATUS_UP ; + +ntb_link_event (&ntb->ntb); +return 0 ; +} + +/** +* epf_ntb_configure_mw() - Configure the Outbound Address Space for vhost +* to access the memory window of host +* @ntb: NTB device that facilitates communication between host and vhost +* @mw: Index of the memory window (either 0, 1, 2 or 3) +* +* EP Outbound Window +* +--------+ +-----------+ +* | | | | +* | | | | +* | | | | +* | | | | +* | | +-----------+ +* | Virtual| | Memory Win| +* | NTB | -----------> | | +* | Driver | | | +* | | +-----------+ +* | | | | +* | | | | +* +--------+ +-----------+ +* VHost PCI EP +*/ +static int epf_ntb_configure_mw (struct epf_ntb *ntb,u32 mw) +{ +phys_addr_t phys_addr; +u8 func_no, vfunc_no; +u64 addr, size; +int ret =0 ; + + phys_addr = ntb->vpci_mw_phy[mw]; + addr = ntb->reg->addr; + size = ntb->reg->size; + + func_no = ntb->epf->func_no; + vfunc_no = ntb->epf->vfunc_no; + + ret =pci_epc_map_addr (ntb->epf->epc, func_no, vfunc_no, phys_addr, addr, size); +if (ret) +dev_err (&ntb->epf->epc->dev, +"Failed to map memory window %d address\n" , mw); +return ret; +} + +/** +* epf_ntb_teardown_mw() - Teardown the configured OB ATU +* @ntb: NTB device that facilitates communication between HOST and vHOST +* @mw: Index of the memory window (either 0, 1, 2 or 3) +* +* Teardown the configured OB ATU configured in epf_ntb_configure_mw() using +* pci_epc_unmap_addr() +*/ +static void epf_ntb_teardown_mw (struct epf_ntb *ntb,u32 mw) +{ +pci_epc_unmap_addr (ntb->epf->epc, + ntb->epf->func_no, + ntb->epf->vfunc_no, + ntb->vpci_mw_phy[mw]); +} + +/** +* epf_ntb_cmd_handler() - Handle commands provided by the NTB Host +* @work: work_struct for the epf_ntb_epc +* +* Workqueue function that gets invoked for the two epf_ntb_epc +* periodically (once every 5ms) to see if it has received any commands +* from NTB host. The host can send commands to configure doorbell or +* configure memory window or to update link status. +*/ +static void epf_ntb_cmd_handler (struct work_struct *work) +{ +struct epf_ntb_ctrl *ctrl; +u32 command, argument; +struct epf_ntb *ntb; +struct device *dev; +int ret; +int i; + + ntb = container_of(work, struct epf_ntb, cmd_handler.work); + +for (i =1 ; i < ntb->db_count; i++) { +if (readl (ntb->epf_db + i *4 )) { +if (readl (ntb->epf_db + i *4 )) + ntb->db |=1 << (i -1 ); + +ntb_db_event (&ntb->ntb, i); +writel (0 , ntb->epf_db + i *4 ); + } + } + + ctrl = ntb->reg; + command = ctrl->command; +if (!command) + goto reset_handler; + argument = ctrl->argument; + + ctrl->command =0 ; + ctrl->argument =0 ; + + ctrl = ntb->reg; + dev = &ntb->epf->dev; + +switch (command) { +case COMMAND_CONFIGURE_DOORBELL : + ctrl->command_status =COMMAND_STATUS_OK ; +break ; +case COMMAND_TEARDOWN_DOORBELL : + ctrl->command_status =COMMAND_STATUS_OK ; +break ; +case COMMAND_CONFIGURE_MW : + ret =epf_ntb_configure_mw (ntb, argument); +if (ret <0 ) + ctrl->command_status =COMMAND_STATUS_ERROR ; +else + ctrl->command_status =COMMAND_STATUS_OK ; +break ; +case COMMAND_TEARDOWN_MW : +epf_ntb_teardown_mw (ntb, argument); + ctrl->command_status =COMMAND_STATUS_OK ; +break ; +case COMMAND_LINK_UP : + ntb->linkup = true; + ret =epf_ntb_link_up (ntb, true); +if (ret <0 ) + ctrl->command_status =COMMAND_STATUS_ERROR ; +else + ctrl->command_status =COMMAND_STATUS_OK ; + goto reset_handler; +case COMMAND_LINK_DOWN : + ntb->linkup = false; + ret =epf_ntb_link_up (ntb, false); +if (ret <0 ) + ctrl->command_status =COMMAND_STATUS_ERROR ; +else + ctrl->command_status =COMMAND_STATUS_OK ; +break ; +default : +dev_err (dev,"UNKNOWN command: %d\n" , command); +break ; + } + +reset_handler: +queue_delayed_work (kpcintb_workqueue, &ntb->cmd_handler, +msecs_to_jiffies (5 )); +} + +/** +* epf_ntb_config_sspad_bar_clear() - Clear Config + Self scratchpad BAR +* @ntb_epc: EPC associated with one of the HOST which holds peer's outbound +* address. +* +* Clear BAR0 of EP CONTROLLER 1 which contains the HOST1's config and +* self scratchpad region (removes inbound ATU configuration). While BAR0 is +* the default self scratchpad BAR, an NTB could have other BARs for self +* scratchpad (because of reserved BARs). This function can get the exact BAR +* used for self scratchpad from epf_ntb_bar[BAR_CONFIG]. +* +* Please note the self scratchpad region and config region is combined to +* a single region and mapped using the same BAR. Also note HOST2's peer +* scratchpad is HOST1's self scratchpad. +*/ +static void epf_ntb_config_sspad_bar_clear (struct epf_ntb *ntb) +{ +struct pci_epf_bar *epf_bar; +enum pci_barno barno; + + barno = ntb->epf_ntb_bar[BAR_CONFIG ]; + epf_bar = &ntb->epf->bar[barno]; + +pci_epc_clear_bar (ntb->epf->epc, ntb->epf->func_no, ntb->epf->vfunc_no, epf_bar); +} + +/** +* epf_ntb_config_sspad_bar_set() - Set Config + Self scratchpad BAR +* @ntb: NTB device that facilitates communication between HOST and vHOST +* +* Map BAR0 of EP CONTROLLER 1 which contains the HOST1's config and +* self scratchpad region. +* +* Please note the self scratchpad region and config region is combined to +* a single region and mapped using the same BAR. +*/ +static int epf_ntb_config_sspad_bar_set (struct epf_ntb *ntb) +{ +struct pci_epf_bar *epf_bar; +enum pci_barno barno; +u8 func_no, vfunc_no; +struct device *dev; +int ret; + + dev = &ntb->epf->dev; + func_no = ntb->epf->func_no; + vfunc_no = ntb->epf->vfunc_no; + barno = ntb->epf_ntb_bar[BAR_CONFIG ]; + epf_bar = &ntb->epf->bar[barno]; + + ret =pci_epc_set_bar (ntb->epf->epc, func_no, vfunc_no, epf_bar); +if (ret) { +dev_err (dev,"inft: Config/Status/SPAD BAR set failed\n" ); +return ret; + } +return 0 ; +} + +/** +* epf_ntb_config_spad_bar_free() - Free the physical memory associated with +* config + scratchpad region +* @ntb: NTB device that facilitates communication between HOST and vHOST +*/ +static void epf_ntb_config_spad_bar_free (struct epf_ntb *ntb) +{ +enum pci_barno barno; + + barno = ntb->epf_ntb_bar[BAR_CONFIG ]; +pci_epf_free_space (ntb->epf, ntb->reg, barno,0 ); +} + +/** +* epf_ntb_config_spad_bar_alloc() - Allocate memory for config + scratchpad +* region +* @ntb: NTB device that facilitates communication between HOST1 and HOST2 +* +* Allocate the Local Memory mentioned in the above diagram. The size of +* CONFIG REGION is sizeof(struct epf_ntb_ctrl) and size of SCRATCHPAD REGION +* is obtained from "spad-count" configfs entry. +*/ +static int epf_ntb_config_spad_bar_alloc (struct epf_ntb *ntb) +{ +size_t align; +enum pci_barno barno; +struct epf_ntb_ctrl *ctrl; +u32 spad_size, ctrl_size; +u64 size; +struct pci_epf *epf = ntb->epf; +struct device *dev = &epf->dev; +u32 spad_count; +void *base; +int i; +const struct pci_epc_features *epc_features =pci_epc_get_features (epf->epc, + epf->func_no, + epf->vfunc_no); + barno = ntb->epf_ntb_bar[BAR_CONFIG ]; + size = epc_features->bar_fixed_size[barno]; + align = epc_features->align; + +if ((!IS_ALIGNED (size, align))) +return -EINVAL ; + + spad_count = ntb->spad_count; + + ctrl_size =sizeof (struct epf_ntb_ctrl ); + spad_size =2 * spad_count *4 ; + +if (!align) { + ctrl_size =roundup_pow_of_two (ctrl_size); + spad_size =roundup_pow_of_two (spad_size); + }else { + ctrl_size =ALIGN (ctrl_size, align); + spad_size =ALIGN (spad_size, align); + } + +if (!size) + size = ctrl_size + spad_size; +else if (size < ctrl_size + spad_size) +return -EINVAL ; + + base =pci_epf_alloc_space (epf, size, barno, align,0 ); +if (!base) { +dev_err (dev,"Config/Status/SPAD alloc region fail\n" ); +return -ENOMEM ; + } + + ntb->reg = base; + + ctrl = ntb->reg; + ctrl->spad_offset = ctrl_size; + + ctrl->spad_count = spad_count; + ctrl->num_mws = ntb->num_mws; + ntb->spad_size = spad_size; + + ctrl->db_entry_size =4 ; + +for (i =0 ; i < ntb->db_count; i++) { + ntb->reg->db_data[i] =1 + i; + ntb->reg->db_offset[i] =0 ; + } + +return 0 ; +} + +/** +* epf_ntb_configure_interrupt() - Configure MSI/MSI-X capaiblity +* @ntb: NTB device that facilitates communication between HOST and vHOST +* +* Configure MSI/MSI-X capability for each interface with number of +* interrupts equal to "db_count" configfs entry. +*/ +static int epf_ntb_configure_interrupt (struct epf_ntb *ntb) +{ +const struct pci_epc_features *epc_features; +struct device *dev; +u32 db_count; +int ret; + + dev = &ntb->epf->dev; + + epc_features =pci_epc_get_features (ntb->epf->epc, ntb->epf->func_no, ntb->epf->vfunc_no); + +if (!(epc_features->msix_capable || epc_features->msi_capable)) { +dev_err (dev,"MSI or MSI-X is required for doorbell\n" ); +return -EINVAL ; + } + + db_count = ntb->db_count; +if (db_count >MAX_DB_COUNT ) { +dev_err (dev,"DB count cannot be more than %d\n" ,MAX_DB_COUNT ); +return -EINVAL ; + } + + ntb->db_count = db_count; + +if (epc_features->msi_capable) { + ret =pci_epc_set_msi (ntb->epf->epc, + ntb->epf->func_no, + ntb->epf->vfunc_no, +16 ); +if (ret) { +dev_err (dev,"MSI configuration failed\n" ); +return ret; + } + } + +return 0 ; +} + +/** +* epf_ntb_db_bar_init() - Configure Doorbell window BARs +* @ntb: NTB device that facilitates communication between HOST and vHOST +*/ +static int epf_ntb_db_bar_init (struct epf_ntb *ntb) +{ +const struct pci_epc_features *epc_features; +u32 align; +struct device *dev = &ntb->epf->dev; +int ret; +struct pci_epf_bar *epf_bar; +void __iomem *mw_addr; +enum pci_barno barno; +size_t size =4 * ntb->db_count; + + epc_features =pci_epc_get_features (ntb->epf->epc, + ntb->epf->func_no, + ntb->epf->vfunc_no); + align = epc_features->align; + +if (size <128 ) + size =128 ; + +if (align) + size =ALIGN (size, align); +else + size =roundup_pow_of_two (size); + + barno = ntb->epf_ntb_bar[BAR_DB ]; + + mw_addr =pci_epf_alloc_space (ntb->epf, size, barno, align,0 ); +if (!mw_addr) { +dev_err (dev,"Failed to allocate OB address\n" ); +return -ENOMEM ; + } + + ntb->epf_db = mw_addr; + + epf_bar = &ntb->epf->bar[barno]; + + ret =pci_epc_set_bar (ntb->epf->epc, ntb->epf->func_no, ntb->epf->vfunc_no, epf_bar); +if (ret) { +dev_err (dev,"Doorbell BAR set failed\n" ); + goto err_alloc_peer_mem; + } +return ret; + +err_alloc_peer_mem: +pci_epc_mem_free_addr (ntb->epf->epc, epf_bar->phys_addr, mw_addr, epf_bar->size); +return -1 ; +} + +/** +* epf_ntb_db_bar_clear() - Clear doorbell BAR and free memory +* allocated in peer's outbound address space +* @ntb: NTB device that facilitates communication between HOST and vHOST +*/ +static void epf_ntb_db_bar_clear (struct epf_ntb *ntb) +{ +enum pci_barno barno; + + barno = ntb->epf_ntb_bar[BAR_DB ]; +pci_epf_free_space (ntb->epf, ntb->epf_db, barno,0 ); +pci_epc_clear_bar (ntb->epf->epc, + ntb->epf->func_no, + ntb->epf->vfunc_no, + &ntb->epf->bar[barno]); +} + +/** +* epf_ntb_mw_bar_init() - Configure Memory window BARs +* @ntb: NTB device that facilitates communication between HOST and vHOST +* +*/ +static int epf_ntb_mw_bar_init (struct epf_ntb *ntb) +{ +int ret =0 ; +int i; +u64 size; +enum pci_barno barno; +struct device *dev = &ntb->epf->dev; + +for (i =0 ; i < ntb->num_mws; i++) { + size = ntb->mws_size[i]; + barno = ntb->epf_ntb_bar[BAR_MW0 + i]; + + ntb->epf->bar[barno].barno = barno; + ntb->epf->bar[barno].size = size; + ntb->epf->bar[barno].addr =0 ; + ntb->epf->bar[barno].phys_addr =0 ; + ntb->epf->bar[barno].flags |=upper_32_bits (size) ? +PCI_BASE_ADDRESS_MEM_TYPE_64 : +PCI_BASE_ADDRESS_MEM_TYPE_32 ; + + ret =pci_epc_set_bar (ntb->epf->epc, + ntb->epf->func_no, + ntb->epf->vfunc_no, + &ntb->epf->bar[barno]); +if (ret) { +dev_err (dev,"MW set failed\n" ); + goto err_alloc_mem; + } + +/* Allocate EPC outbound memory windows to vpci vntb device */ + ntb->vpci_mw_addr[i] =pci_epc_mem_alloc_addr (ntb->epf->epc, + &ntb->vpci_mw_phy[i], + size); +if (!ntb->vpci_mw_addr[i]) { +dev_err (dev,"Failed to allocate source address\n" ); + goto err_alloc_mem; + } + } + +return ret; +err_alloc_mem: +return ret; +} + +/** +* epf_ntb_mw_bar_clear() - Clear Memory window BARs +* @ntb: NTB device that facilitates communication between HOST and vHOST +*/ +static void epf_ntb_mw_bar_clear (struct epf_ntb *ntb) +{ +enum pci_barno barno; +int i; + +for (i =0 ; i < ntb->num_mws; i++) { + barno = ntb->epf_ntb_bar[BAR_MW0 + i]; +pci_epc_clear_bar (ntb->epf->epc, + ntb->epf->func_no, + ntb->epf->vfunc_no, + &ntb->epf->bar[barno]); + +pci_epc_mem_free_addr (ntb->epf->epc, + ntb->vpci_mw_phy[i], + ntb->vpci_mw_addr[i], + ntb->mws_size[i]); + } +} + +/** +* epf_ntb_epc_destroy() - Cleanup NTB EPC interface +* @ntb: NTB device that facilitates communication between HOST and vHOST +* +* Wrapper for epf_ntb_epc_destroy_interface() to cleanup all the NTB interfaces +*/ +static void epf_ntb_epc_destroy (struct epf_ntb *ntb) +{ +pci_epc_remove_epf (ntb->epf->epc, ntb->epf,0 ); +pci_epc_put (ntb->epf->epc); +} + +/** +* epf_ntb_init_epc_bar() - Identify BARs to be used for each of the NTB +* constructs (scratchpad region, doorbell, memorywindow) +* @ntb: NTB device that facilitates communication between HOST and vHOST +*/ +static int epf_ntb_init_epc_bar (struct epf_ntb *ntb) +{ +const struct pci_epc_features *epc_features; +enum pci_barno barno; +enum epf_ntb_bar bar; +struct device *dev; +u32 num_mws; +int i; + + barno =BAR_0 ; + num_mws = ntb->num_mws; + dev = &ntb->epf->dev; + epc_features =pci_epc_get_features (ntb->epf->epc, ntb->epf->func_no, ntb->epf->vfunc_no); + +/* These are required BARs which are mandatory for NTB functionality */ +for (bar =BAR_CONFIG ; bar <=BAR_MW0 ; bar++, barno++) { + barno =pci_epc_get_next_free_bar (epc_features, barno); +if (barno <0 ) { +dev_err (dev,"Fail to get NTB function BAR\n" ); +return barno; + } + ntb->epf_ntb_bar[bar] = barno; + } + +/* These are optional BARs which don't impact NTB functionality */ +for (bar =BAR_MW1 , i =1 ; i < num_mws; bar++, barno++, i++) { + barno =pci_epc_get_next_free_bar (epc_features, barno); +if (barno <0 ) { + ntb->num_mws = i; +dev_dbg (dev,"BAR not available for > MW%d\n" , i +1 ); + } + ntb->epf_ntb_bar[bar] = barno; + } + +return 0 ; +} + +/** +* epf_ntb_epc_init() - Initialize NTB interface +* @ntb: NTB device that facilitates communication between HOST and vHOST2 +* +* Wrapper to initialize a particular EPC interface and start the workqueue +* to check for commands from host. This function will write to the +* EP controller HW for configuring it. +*/ +static int epf_ntb_epc_init (struct epf_ntb *ntb) +{ +u8 func_no, vfunc_no; +struct pci_epc *epc; +struct pci_epf *epf; +struct device *dev; +int ret; + + epf = ntb->epf; + dev = &epf->dev; + epc = epf->epc; + func_no = ntb->epf->func_no; + vfunc_no = ntb->epf->vfunc_no; + + ret =epf_ntb_config_sspad_bar_set (ntb); +if (ret) { +dev_err (dev,"Config/self SPAD BAR init failed" ); +return ret; + } + + ret =epf_ntb_configure_interrupt (ntb); +if (ret) { +dev_err (dev,"Interrupt configuration failed\n" ); + goto err_config_interrupt; + } + + ret =epf_ntb_db_bar_init (ntb); +if (ret) { +dev_err (dev,"DB BAR init failed\n" ); + goto err_db_bar_init; + } + + ret =epf_ntb_mw_bar_init (ntb); +if (ret) { +dev_err (dev,"MW BAR init failed\n" ); + goto err_mw_bar_init; + } + +if (vfunc_no <=1 ) { + ret =pci_epc_write_header (epc, func_no, vfunc_no, epf->header); +if (ret) { +dev_err (dev,"Configuration header write failed\n" ); + goto err_write_header; + } + } + +INIT_DELAYED_WORK (&ntb->cmd_handler, epf_ntb_cmd_handler); +queue_work (kpcintb_workqueue, &ntb->cmd_handler.work); + +return 0 ; + +err_write_header: +epf_ntb_mw_bar_clear (ntb); +err_mw_bar_init: +epf_ntb_db_bar_clear (ntb); +err_db_bar_init: +err_config_interrupt: +epf_ntb_config_sspad_bar_clear (ntb); + +return ret; +} + + +/** +* epf_ntb_epc_cleanup() - Cleanup all NTB interfaces +* @ntb: NTB device that facilitates communication between HOST1 and HOST2 +* +* Wrapper to cleanup all NTB interfaces. +*/ +static void epf_ntb_epc_cleanup (struct epf_ntb *ntb) +{ +epf_ntb_db_bar_clear (ntb); +epf_ntb_mw_bar_clear (ntb); +} + +#define EPF_NTB_R (_name) \ +static ssize_t epf_ntb_##_name##_show(struct config_item *item, \ + char *page) \ +{ \ + struct config_group *group = to_config_group(item); \ + struct epf_ntb *ntb = to_epf_ntb(group); \ + \ + return sprintf(page, "%d\n", ntb->_name); \ +} + +#define EPF_NTB_W (_name) \ +static ssize_t epf_ntb_##_name##_store(struct config_item *item, \ + const char *page, size_t len) \ +{ \ + struct config_group *group = to_config_group(item); \ + struct epf_ntb *ntb = to_epf_ntb(group); \ + u32 val; \ + int ret; \ + \ + ret = kstrtou32(page, 0, &val); \ + if (ret) \ + return ret; \ + \ + ntb->_name = val; \ + \ + return len; \ +} + +#define EPF_NTB_MW_R (_name) \ +static ssize_t epf_ntb_##_name##_show(struct config_item *item, \ + char *page) \ +{ \ + struct config_group *group = to_config_group(item); \ + struct epf_ntb *ntb = to_epf_ntb(group); \ + int win_no; \ + \ + sscanf(#_name, "mw%d", &win_no); \ + \ + return sprintf(page, "%lld\n", ntb->mws_size[win_no - 1]); \ +} + +#define EPF_NTB_MW_W (_name) \ +static ssize_t epf_ntb_##_name##_store(struct config_item *item, \ + const char *page, size_t len) \ +{ \ + struct config_group *group = to_config_group(item); \ + struct epf_ntb *ntb = to_epf_ntb(group); \ + struct device *dev = &ntb->epf->dev; \ + int win_no; \ + u64 val; \ + int ret; \ + \ + ret = kstrtou64(page, 0, &val); \ + if (ret) \ + return ret; \ + \ + if (sscanf(#_name, "mw%d", &win_no) != 1) \ + return -EINVAL; \ + \ + if (ntb->num_mws < win_no) { \ + dev_err(dev, "Invalid num_nws: %d value\n", ntb->num_mws); \ + return -EINVAL; \ + } \ + \ + ntb->mws_size[win_no - 1] = val; \ + \ + return len; \ +} + +static ssize_t epf_ntb_num_mws_store (struct config_item *item, +const char *page,size_t len) +{ +struct config_group *group =to_config_group (item); +struct epf_ntb *ntb =to_epf_ntb (group); +u32 val; +int ret; + + ret =kstrtou32 (page,0 , &val); +if (ret) +return ret; + +if (val >MAX_MW ) +return -EINVAL ; + + ntb->num_mws = val; + +return len; +} + +EPF_NTB_R (spad_count) +EPF_NTB_W(spad_count) +EPF_NTB_R(db_count) +EPF_NTB_W(db_count) +EPF_NTB_R(num_mws) +EPF_NTB_R(vbus_number) +EPF_NTB_W(vbus_number) +EPF_NTB_R(vntb_pid) +EPF_NTB_W(vntb_pid) +EPF_NTB_R(vntb_vid) +EPF_NTB_W(vntb_vid) +EPF_NTB_MW_R(mw1) +EPF_NTB_MW_W(mw1) +EPF_NTB_MW_R(mw2) +EPF_NTB_MW_W(mw2) +EPF_NTB_MW_R(mw3) +EPF_NTB_MW_W(mw3) +EPF_NTB_MW_R(mw4) +EPF_NTB_MW_W(mw4) + +CONFIGFS_ATTR(epf_ntb_, spad_count); +CONFIGFS_ATTR (epf_ntb_, db_count); +CONFIGFS_ATTR (epf_ntb_, num_mws); +CONFIGFS_ATTR (epf_ntb_, mw1); +CONFIGFS_ATTR (epf_ntb_, mw2); +CONFIGFS_ATTR (epf_ntb_, mw3); +CONFIGFS_ATTR (epf_ntb_, mw4); +CONFIGFS_ATTR (epf_ntb_, vbus_number); +CONFIGFS_ATTR (epf_ntb_, vntb_pid); +CONFIGFS_ATTR (epf_ntb_, vntb_vid); + +static struct configfs_attribute *epf_ntb_attrs[] = { + &epf_ntb_attr_spad_count, + &epf_ntb_attr_db_count, + &epf_ntb_attr_num_mws, + &epf_ntb_attr_mw1, + &epf_ntb_attr_mw2, + &epf_ntb_attr_mw3, + &epf_ntb_attr_mw4, + &epf_ntb_attr_vbus_number, + &epf_ntb_attr_vntb_pid, + &epf_ntb_attr_vntb_vid, +NULL , +}; + +static const struct config_item_type ntb_group_type = { + .ct_attrs = epf_ntb_attrs, + .ct_owner =THIS_MODULE , +}; + +/** +* epf_ntb_add_cfs() - Add configfs directory specific to NTB +* @epf: NTB endpoint function device +* @group: A pointer to the config_group structure referencing a group of +* config_items of a specific type that belong to a specific sub-system. +* +* Add configfs directory specific to NTB. This directory will hold +* NTB specific properties like db_count, spad_count, num_mws etc., +*/ +static struct config_group *epf_ntb_add_cfs (struct pci_epf *epf, +struct config_group *group) +{ +struct epf_ntb *ntb =epf_get_drvdata (epf); +struct config_group *ntb_group = &ntb->group; +struct device *dev = &epf->dev; + +config_group_init_type_name (ntb_group,dev_name (dev), &ntb_group_type); + +return ntb_group; +} + +/*==== virtual PCI bus driver, which only load virtual NTB PCI driver ====*/ + +static u32 pci_space[] = { +0xffffffff ,/*DeviceID, Vendor ID*/ +0 ,/*Status, Command*/ +0xffffffff ,/*Class code, subclass, prog if, revision id*/ +0x40 ,/*bist, header type, latency Timer, cache line size*/ +0 ,/*BAR 0*/ +0 ,/*BAR 1*/ +0 ,/*BAR 2*/ +0 ,/*BAR 3*/ +0 ,/*BAR 4*/ +0 ,/*BAR 5*/ +0 ,/*Cardbus cis point*/ +0 ,/*Subsystem ID Subystem vendor id*/ +0 ,/*ROM Base Address*/ +0 ,/*Reserved, Cap. Point*/ +0 ,/*Reserved,*/ +0 ,/*Max Lat, Min Gnt, interrupt pin, interrupt line*/ +}; + +int pci_read (struct pci_bus *bus,unsigned int devfn,int where,int size,u32 *val) +{ +if (devfn ==0 ) { +memcpy (val, ((u8 *)pci_space) + where, size); +return PCIBIOS_SUCCESSFUL ; + } +return PCIBIOS_DEVICE_NOT_FOUND ; +} + +int pci_write (struct pci_bus *bus,unsigned int devfn,int where,int size,u32 val) +{ +return 0 ; +} + +struct pci_ops vpci_ops = { + .read = pci_read, + .write = pci_write, +}; + +static int vpci_scan_bus (void *sysdata) +{ +struct pci_bus *vpci_bus; +struct epf_ntb *ndev = sysdata; + + vpci_bus =pci_scan_bus (ndev->vbus_number, &vpci_ops, sysdata); +if (vpci_bus) +pr_err ("create pci bus\n" ); + +pci_bus_add_devices (vpci_bus); + +return 0 ; +} + +/*==================== Virtual PCIe NTB driver ==========================*/ + +static int vntb_epf_mw_count (struct ntb_dev *ntb,int pidx) +{ +struct epf_ntb *ndev =ntb_ndev (ntb); + +return ndev->num_mws; +} + +static int vntb_epf_spad_count (struct ntb_dev *ntb) +{ +return ntb_ndev (ntb)->spad_count; +} + +static int vntb_epf_peer_mw_count (struct ntb_dev *ntb) +{ +return ntb_ndev (ntb)->num_mws; +} + +static u64 vntb_epf_db_valid_mask (struct ntb_dev *ntb) +{ +return BIT_ULL (ntb_ndev (ntb)->db_count) -1 ; +} + +static int vntb_epf_db_set_mask (struct ntb_dev *ntb,u64 db_bits) +{ +return 0 ; +} + +static int vntb_epf_mw_set_trans (struct ntb_dev *ndev,int pidx,int idx, +dma_addr_t addr,resource_size_t size) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); +struct pci_epf_bar *epf_bar; +enum pci_barno barno; +int ret; +struct device *dev; + + dev = &ntb->ntb.dev; + barno = ntb->epf_ntb_bar[BAR_MW0 + idx]; + epf_bar = &ntb->epf->bar[barno]; + epf_bar->phys_addr = addr; + epf_bar->barno = barno; + epf_bar->size = size; + + ret =pci_epc_set_bar (ntb->epf->epc,0 ,0 , epf_bar); +if (ret) { +dev_err (dev,"failure set mw trans\n" ); +return ret; + } +return 0 ; +} + +static int vntb_epf_mw_clear_trans (struct ntb_dev *ntb,int pidx,int idx) +{ +return 0 ; +} + +static int vntb_epf_peer_mw_get_addr (struct ntb_dev *ndev,int idx, +phys_addr_t *base,resource_size_t *size) +{ + +struct epf_ntb *ntb =ntb_ndev (ndev); + +if (base) + *base = ntb->vpci_mw_phy[idx]; + +if (size) + *size = ntb->mws_size[idx]; + +return 0 ; +} + +static int vntb_epf_link_enable (struct ntb_dev *ntb, +enum ntb_speed max_speed, +enum ntb_width max_width) +{ +return 0 ; +} + +static u32 vntb_epf_spad_read (struct ntb_dev *ndev,int idx) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); +int off = ntb->reg->spad_offset, ct = ntb->reg->spad_count *4 ; +u32 val; +void __iomem *base = ntb->reg; + + val =readl (base + off + ct + idx *4 ); +return val; +} + +static int vntb_epf_spad_write (struct ntb_dev *ndev,int idx,u32 val) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); +struct epf_ntb_ctrl *ctrl = ntb->reg; +int off = ctrl->spad_offset, ct = ctrl->spad_count *4 ; +void __iomem *base = ntb->reg; + +writel (val, base + off + ct + idx *4 ); +return 0 ; +} + +static u32 vntb_epf_peer_spad_read (struct ntb_dev *ndev,int pidx,int idx) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); +struct epf_ntb_ctrl *ctrl = ntb->reg; +int off = ctrl->spad_offset; +void __iomem *base = ntb->reg; +u32 val; + + val =readl (base + off + idx *4 ); +return val; +} + +static int vntb_epf_peer_spad_write (struct ntb_dev *ndev,int pidx,int idx,u32 val) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); +struct epf_ntb_ctrl *ctrl = ntb->reg; +int off = ctrl->spad_offset; +void __iomem *base = ntb->reg; + +writel (val, base + off + idx *4 ); +return 0 ; +} + +static int vntb_epf_peer_db_set (struct ntb_dev *ndev,u64 db_bits) +{ +u32 interrupt_num =ffs (db_bits) +1 ; +struct epf_ntb *ntb =ntb_ndev (ndev); +u8 func_no, vfunc_no; +int ret; + + func_no = ntb->epf->func_no; + vfunc_no = ntb->epf->vfunc_no; + + ret =pci_epc_raise_irq (ntb->epf->epc, + func_no, + vfunc_no, +PCI_EPC_IRQ_MSI , + interrupt_num +1 ); +if (ret) +dev_err (&ntb->ntb.dev,"Failed to raise IRQ\n" ); + +return ret; +} + +static u64 vntb_epf_db_read (struct ntb_dev *ndev) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); + +return ntb->db; +} + +static int vntb_epf_mw_get_align (struct ntb_dev *ndev,int pidx,int idx, +resource_size_t *addr_align, +resource_size_t *size_align, +resource_size_t *size_max) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); + +if (addr_align) + *addr_align =SZ_4K ; + +if (size_align) + *size_align =1 ; + +if (size_max) + *size_max = ntb->mws_size[idx]; + +return 0 ; +} + +static u64 vntb_epf_link_is_up (struct ntb_dev *ndev, +enum ntb_speed *speed, +enum ntb_width *width) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); + +return ntb->reg->link_status; +} + +static int vntb_epf_db_clear_mask (struct ntb_dev *ndev,u64 db_bits) +{ +return 0 ; +} + +static int vntb_epf_db_clear (struct ntb_dev *ndev,u64 db_bits) +{ +struct epf_ntb *ntb =ntb_ndev (ndev); + + ntb->db &= ~db_bits; +return 0 ; +} + +static int vntb_epf_link_disable (struct ntb_dev *ntb) +{ +return 0 ; +} + +static const struct ntb_dev_ops vntb_epf_ops = { + .mw_count = vntb_epf_mw_count, + .spad_count = vntb_epf_spad_count, + .peer_mw_count = vntb_epf_peer_mw_count, + .db_valid_mask = vntb_epf_db_valid_mask, + .db_set_mask = vntb_epf_db_set_mask, + .mw_set_trans = vntb_epf_mw_set_trans, + .mw_clear_trans = vntb_epf_mw_clear_trans, + .peer_mw_get_addr = vntb_epf_peer_mw_get_addr, + .link_enable = vntb_epf_link_enable, + .spad_read = vntb_epf_spad_read, + .spad_write = vntb_epf_spad_write, + .peer_spad_read = vntb_epf_peer_spad_read, + .peer_spad_write = vntb_epf_peer_spad_write, + .peer_db_set = vntb_epf_peer_db_set, + .db_read = vntb_epf_db_read, + .mw_get_align = vntb_epf_mw_get_align, + .link_is_up = vntb_epf_link_is_up, + .db_clear_mask = vntb_epf_db_clear_mask, + .db_clear = vntb_epf_db_clear, + .link_disable = vntb_epf_link_disable, +}; + +static int pci_vntb_probe (struct pci_dev *pdev,const struct pci_device_id *id) +{ +int ret; +struct epf_ntb *ndev = (struct epf_ntb *)pdev->sysdata; +struct device *dev = &pdev->dev; + + ndev->ntb.pdev = pdev; + ndev->ntb.topo =NTB_TOPO_NONE ; + ndev->ntb.ops = &vntb_epf_ops; + + ret =dma_set_mask_and_coherent (dev,DMA_BIT_MASK (32 )); +if (ret) { +dev_err (dev,"Cannot set DMA mask\n" ); +return -EINVAL ; + } + + ret =ntb_register_device (&ndev->ntb); +if (ret) { +dev_err (dev,"Failed to register NTB device\n" ); + goto err_register_dev; + } + +dev_dbg (dev,"PCI Virtual NTB driver loaded\n" ); +return 0 ; + +err_register_dev: +return -EINVAL ; +} + +static struct pci_device_id pci_vntb_table[] = { + { +PCI_DEVICE (0xffff ,0xffff ), + }, + {}, +}; + +static struct pci_driver vntb_pci_driver = { + .name ="pci-vntb" , + .id_table = pci_vntb_table, + .probe = pci_vntb_probe, +}; + +/* ============ PCIe EPF Driver Bind ====================*/ + +/** +* epf_ntb_bind() - Initialize endpoint controller to provide NTB functionality +* @epf: NTB endpoint function device +* +* Initialize both the endpoint controllers associated with NTB function device. +* Invoked when a primary interface or secondary interface is bound to EPC +* device. This function will succeed only when EPC is bound to both the +* interfaces. +*/ +static int epf_ntb_bind (struct pci_epf *epf) +{ +struct epf_ntb *ntb =epf_get_drvdata (epf); +struct device *dev = &epf->dev; +int ret; + +if (!epf->epc) { +dev_dbg (dev,"PRIMARY EPC interface not yet bound\n" ); +return 0 ; + } + + ret =epf_ntb_init_epc_bar (ntb); +if (ret) { +dev_err (dev,"Failed to create NTB EPC\n" ); + goto err_bar_init; + } + + ret =epf_ntb_config_spad_bar_alloc (ntb); +if (ret) { +dev_err (dev,"Failed to allocate BAR memory\n" ); + goto err_bar_alloc; + } + + ret =epf_ntb_epc_init (ntb); +if (ret) { +dev_err (dev,"Failed to initialize EPC\n" ); + goto err_bar_alloc; + } + +epf_set_drvdata (epf, ntb); + + pci_space[0 ] = (ntb->vntb_pid <<16 ) | ntb->vntb_vid; + pci_vntb_table[0 ].vendor = ntb->vntb_vid; + pci_vntb_table[0 ].device = ntb->vntb_pid; + +if (pci_register_driver (&vntb_pci_driver)) { +dev_err (dev,"failure register vntb pci driver\n" ); + goto err_bar_alloc; + } + +vpci_scan_bus (ntb); + +return 0 ; + +err_bar_alloc: +epf_ntb_config_spad_bar_free (ntb); + +err_bar_init: +epf_ntb_epc_destroy (ntb); + +return ret; +} + +/** +* epf_ntb_unbind() - Cleanup the initialization from epf_ntb_bind() +* @epf: NTB endpoint function device +* +* Cleanup the initialization from epf_ntb_bind() +*/ +static void epf_ntb_unbind (struct pci_epf *epf) +{ +struct epf_ntb *ntb =epf_get_drvdata (epf); + +epf_ntb_epc_cleanup (ntb); +epf_ntb_config_spad_bar_free (ntb); +epf_ntb_epc_destroy (ntb); + +pci_unregister_driver (&vntb_pci_driver); +} + +// EPF driver probe +static struct pci_epf_ops epf_ntb_ops = { + .bind = epf_ntb_bind, + .unbind = epf_ntb_unbind, + .add_cfs = epf_ntb_add_cfs, +}; + +/** +* epf_ntb_probe() - Probe NTB function driver +* @epf: NTB endpoint function device +* +* Probe NTB function driver when endpoint function bus detects a NTB +* endpoint function. +*/ +static int epf_ntb_probe (struct pci_epf *epf) +{ +struct epf_ntb *ntb; +struct device *dev; + + dev = &epf->dev; + + ntb =devm_kzalloc (dev,sizeof (*ntb),GFP_KERNEL ); +if (!ntb) +return -ENOMEM ; + + epf->header = &epf_ntb_header; + ntb->epf = epf; + ntb->vbus_number =0xff ; +epf_set_drvdata (epf, ntb); + +dev_info (dev,"pci-ep epf driver loaded\n" ); +return 0 ; +} + +static const struct pci_epf_device_id epf_ntb_ids[] = { + { + .name ="pci_epf_vntb" , + }, + {}, +}; + +static struct pci_epf_driver epf_ntb_driver = { + .driver.name ="pci_epf_vntb" , + .probe = epf_ntb_probe, + .id_table = epf_ntb_ids, + .ops = &epf_ntb_ops, + .owner =THIS_MODULE , +}; + +static int __init epf_ntb_init (void ) +{ +int ret; + + kpcintb_workqueue =alloc_workqueue ("kpcintb" ,WQ_MEM_RECLAIM | +WQ_HIGHPRI ,0 ); + ret =pci_epf_register_driver (&epf_ntb_driver); +if (ret) { +destroy_workqueue (kpcintb_workqueue); +pr_err ("Failed to register pci epf ntb driver --> %d\n" , ret); +return ret; + } + +return 0 ; +} +module_init (epf_ntb_init); + +static void __exit epf_ntb_exit(void) +{ +pci_epf_unregister_driver (&epf_ntb_driver); +destroy_workqueue (kpcintb_workqueue); +} +module_exit (epf_ntb_exit); + +MODULE_DESCRIPTION ("PCI EPF NTB DRIVER" ); +MODULE_AUTHOR ("Frank Li <[email protected]>" ); +MODULE_LICENSE ("GPL v2" ); --2.24.0.rc1
Re: [PATCH v2 3/4] PCI: endpoint: Support NTB transfer between RC and EP
In reply to Frank LiQuoted 18 lines · Frank Li
+ * +--------------------------------------------------+ Base
+ * | | + * | | + * | | + * | Common Control Register | + * | | + * | | + * | | + * +-----------------------+--------------------------+ Base+span_offset + * | | | + * | Peer Span Space | Span Space | + * | | | + * | | | + * +-----------------------+--------------------------+ Base+span_offset + * | | | +span_count * 4 + * | | | + * | Span Space | Peer Span Space | + * | | |
Quoted 1 line
+ * +-----------------------+--------------------------+
Are these comments supposed to say *spad*, i.e., scratchpad space, instead of "span", to correspond with spad_offset and spad_count below?
Quoted 16 lines
drivers/pci/endpoint/functions/pci-epf-vntb.c new file +16
-0+struct epf_ntb_ctrl { +u32 command; +u32 argument; +u16 command_status; +u16 link_status; +u32 topology; +u64 addr; +u64 size; +u32 num_mws; +u32 reserved; +u32 spad_offset; +u32 spad_count; +u32 db_entry_size; +u32 db_data[MAX_DB_COUNT ]; +u32 db_offset[MAX_DB_COUNT ]; +} __packed;
A message that is not in the archive (sent off-list or not yet received): CAHrpEqSGySHDET3YPu3czzoMBmCRJsgGgU4s3GWWbtruFLVHaA@mail.gmail.com
Re: [PATCH v2 3/4] PCI: endpoint: Support NTB transfer between RC and EP
Quoted 23 lines · Zhi Li
Quoted 22 lines · Bjorn Helgaas
Quoted 19 lines · Frank Li
+ * +--------------------------------------------------+ Base
+ * | | + * | | + * | | + * | Common Control Register | + * | | + * | | + * | | + * +-----------------------+--------------------------+ Base+span_offset + * | | | + * | Peer Span Space | Span Space | + * | | | + * | | | + * +-----------------------+--------------------------+ Base+span_offset + * | | | +span_count * 4 + * | | | + * | Span Space | Peer Span Space | + * | | |+ * +-----------------------+--------------------------+
Quoted 5 lines
Quoted 3 lines
Are these comments supposed to say *spad*, i.e., scratchpad space, instead of "span", to correspond with
spad_offsetandspad_countbelow?Strange, I received some of your comments on the very old patches.
What's strange about it? I went to the trouble to look up the patch that introduced the thing I'm asking about. I sent the email a few minutes ago. The question still applies to the current tree.
Please use plain text email on the Linux mailing lists.
Bjorn
RE: [EXT] Re: [PATCH v2 3/4] PCI: endpoint: Support NTB transfer between RC and EP
Quoted 46 lines
-----Original Message-----
From: Bjorn Helgaas <[email protected]>
Sent: Tuesday, December 13, 2022 6:29 PM
To: Zhi Li <[email protected]>
Cc: Bjorn Helgaas <[email protected]>; Frank Li <[email protected]>; [email protected]; [email protected]; [email protected]; Hongxing Zhu <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]
Subject: [EXT] Re: [PATCH v2 3/4] PCI: endpoint: Support NTB transfer between RC and EPCaution: EXT Email
Quoted 1 line · Zhi Li
On Tue, Dec 13, 2022 at 6:08 PM Bjorn Helgaas <[email protected]>
Quoted 26 lines · wrote:
Quoted 25 lines · Frank Li
Quoted 19 lines
+ * +--------------------------------------------------+ Base
+ * | | + * | | + * | | + * | Common Control Register | + * | | + * | | + * | | + * +-----------------------+--------------------------+ Base+span_offset + * | | | + * | Peer Span Space | Span Space | + * | | | + * | | | + * +-----------------------+--------------------------+ Base+span_offset + * | | | +span_count * 4 + * | | | + * | Span Space | Peer Span Space | + * | | |Quoted 1 line
+ * +-----------------------+--------------------------+
Are these comments supposed to say *spad*, i.e., scratchpad space, instead of "span", to correspond with
spad_offsetandspad_countbelow?
Quoted 5 lines
Quoted 1 line
Strange, I received some of your comments on the very old patches.
What's strange about it? I went to the trouble to look up the patch that introduced the thing I'm asking about. I sent the email a few minutes ago. The question still applies to the current tree.
Okay, I doesn't realized you are talking about current tree also. Let me send patch to fix this soon.
Best regards
Frank Li
Quoted 4 lines
Please use plain text email on the Linux mailing lists.
Bjorn
PATCH v2 4/4 Documentation: PCI: Add specification for the PCI vNTB function device
In reply to Frank LiAdd specification for the PCI vNTB function device. The endpoint function driver and the host PCI driver should be created based on this specification.
---
Change from v1:
- Add virtual pci ntb driver config
Documentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 +++++++++++++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++++++++++++++++++ 3 files changed, 295 insertions(+) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst
Documentation/PCI/endpoint/index.rst +2 -0
@@ -13,6 +13,8 @@ PCI Endpoint Framework pci-test-howto pci-ntb-function pci-ntb-howto + pci-vntb-function + pci-vntb-howto function/binding/pci-test function/binding/pci-ntb
Documentation/PCI/endpoint/pci-vntb-function.rst new file +126 -0
126 lines not on this page · raw message
Documentation/PCI/endpoint/pci-vntb-howto.rst new file +167 -0
167 lines not on this page · raw message
Signature · 1 line
2.24.0.rc1
Quoted 31 lines · Frank Li
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EP
Quoted 6 lines
Frank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function device
Ping
Quoted 16 lines
Documentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c--
2.24.0.rc1
Quoted 33 lines · Zhi Li
Quoted 31 lines · Frank Li
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EP
Quoted 7 lines
Quoted 6 lines
Frank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function device
Update ntb mail list
Quoted 17 lines
Quoted 17 lines
Documentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c--
2.24.0.rc1
Quoted 43 lines · Zhi Li
Quoted 33 lines · Zhi Li
Quoted 31 lines · Frank Li
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EPQuoted 7 lines
Quoted 6 lines
Frank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function device
Quoted 1 line
Update ntb mail list
Friendly ping!
Frank Li
Quoted 18 lines
Quoted 18 lines
Quoted 18 lines
Documentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c--
2.24.0.rc1
Hi Frank Li,
Quoted 2 lines · Frank Li
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
As I had earlier mentioned in [1], IMHO ideal solution would be build on virtio layer instead of trying to build on NTB layer (which is specific to RC<->RC communication).
Are there any specific reasons for not taking that path?
Thanks,
Kishon
[1] -> https://lore.kernel.org/r/[email protected]
Quoted 47 lines
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EPFrank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function deviceDocumentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c
Quoted 4 lines · Kishon Vijay Abraham I
Hi Frank Li,
On 22/02/22 9:53 pm, Frank Li wrote:
Quoted 8 lines
Quoted 2 lines
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
As I had earlier mentioned in [1], IMHO ideal solution would be build on virtio layer instead of trying to build on NTB layer (which is specific to RC<->RC communication).
Are there any specific reasons for not taking that path?
1. EP side work as vHOST mode. vHost suppose access all memory of virtual io. But there are only map windows on the EP side to access RC side memory. You have to move
map windows for each access. It is quite low efficiency.
2. So far as I know, virtio is still not DMA yet. CPU access PCI can't generate longer PCI TLP,
So the speed is quite slow. NTB already has DMA support. If you use system level DMA,
no change is needed at NTB level. If we want to use a PCI controller embedded DMA, some small
changes need if based on my other Designware PCI eDMA patches, which are under review.
3. All the major data transfer of NTB is using write. Because TLP write needn't wait for complete, write
performance is better than reading. On our platform, write performance is about 10% better than read.
Frank
Quoted 52 lines
Thanks,
Kishon[1] -> https://lore.kernel.org/r/[email protected]
Quoted 47 lines
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EPFrank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function deviceDocumentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c
Quoted 29 lines · Zhi Li
Quoted 4 lines · Kishon Vijay Abraham I
Hi Frank Li,
On 22/02/22 9:53 pm, Frank Li wrote:
Quoted 8 lines
Quoted 2 lines
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
As I had earlier mentioned in [1], IMHO ideal solution would be build on virtio layer instead of trying to build on NTB layer (which is specific to RC<->RC communication).
Are there any specific reasons for not taking that path?
1. EP side work as vHOST mode. vHost suppose access all memory of virtual io. But there are only map windows on the EP side to access RC side memory. You have to move
map windows for each access. It is quite low efficiency.2. So far as I know, virtio is still not DMA yet. CPU access PCI can't generate longer PCI TLP,
So the speed is quite slow. NTB already has DMA support. If you use system level DMA,
no change is needed at NTB level. If we want to use a PCI controller embedded DMA, some small
changes need if based on my other Designware PCI eDMA patches, which are under review.
Quoted 6 lines
3. All the major data transfer of NTB is using write. Because TLP write needn't wait for complete, write
performance is better than reading. On our platform, write performance is about 10% better than read.Frank
Any Comments or rejection? @Kishon Vijay Abraham I
best regards
Frank Li
Quoted 53 lines
Quoted 53 lines
Thanks,
Kishon[1] -> https://lore.kernel.org/r/[email protected]
Quoted 47 lines
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EPFrank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function deviceDocumentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c
Hi Frank,
Quoted 14 lines · Zhi Li
Quoted 13 lines · Zhi Li
Quoted 11 lines · Kishon Vijay Abraham I
Hi Frank Li,
Quoted 2 lines · Frank Li
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
As I had earlier mentioned in [1], IMHO ideal solution would be build on virtio layer instead of trying to build on NTB layer (which is specific to RC<->RC communication).
Quoted 6 lines
Quoted 6 lines
Quoted 1 line
Are there any specific reasons for not taking that path?
1. EP side work as vHOST mode. vHost suppose access all memory of virtual io. But there are only map windows on the EP side to access RC side memory. You have to move
map windows for each access. It is quite low efficiency.
I'm not sure I quite get this. EP HW has limited outbound memory to access RC memory irrespective of how we implement it. This is not a SW framework limitation AFAICS.
Quoted 9 lines
Quoted 9 lines
2. So far as I know, virtio is still not DMA yet. CPU access PCI can't generate longer PCI TLP,
So the speed is quite slow. NTB already has DMA support. If you use system level DMA,
no change is needed at NTB level. If we want to use a PCI controller embedded DMA, some small
changes need if based on my other Designware PCI eDMA patches, which are under review.
Adding dmaengine API to do memcopy should be simple to add in vhost/virtio interface.
Quoted 9 lines
Quoted 7 lines
3. All the major data transfer of NTB is using write. Because TLP write needn't wait for complete, write
performance is better than reading. On our platform, write performance is about 10% better than read.Frank
Any Comments or rejection? @Kishon Vijay Abraham I
I'd strongly recommend going with virtio/vhost based approach and standardizing it IMO.
Thanks,
Kishon
Quoted 57 lines
best regards
Frank LiQuoted 53 lines
Quoted 53 lines
Thanks,
Kishon[1] -> https://lore.kernel.org/r/[email protected]
Quoted 47 lines
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EPFrank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function deviceDocumentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c
Quoted 20 lines · Kishon Vijay Abraham I
Hi Frank,
Quoted 16 lines · Zhi Li
Quoted 15 lines · Zhi Li
Quoted 12 lines · Kishon Vijay Abraham I
Hi Frank Li,
Quoted 2 lines · Frank Li
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
As I had earlier mentioned in [1], IMHO ideal solution would be build on virtio layer instead of trying to build on NTB layer (which is specific to RC<->RC communication).
Are there any specific reasons for not taking that path?
Quoted 8 lines
Quoted 4 lines
Quoted 4 lines
1. EP side work as vHOST mode. vHost suppose access all memory of virtual io. But there are only map windows on the EP side to access RC side memory. You have to move
map windows for each access. It is quite low efficiency.I'm not sure I quite get this. EP HW has limited outbound memory to access RC memory irrespective of how we implement it. This is not a SW framework limitation AFAICS.
Almost all EP HW have limited outbound memory windows to access RC. We face transfer efficiency problems if we stick into vhost.
Quoted 18 lines
Quoted 9 lines
Quoted 9 lines
2. So far as I know, virtio is still not DMA yet. CPU access PCI can't generate longer PCI TLP,
So the speed is quite slow. NTB already has DMA support. If you use system level DMA,
no change is needed at NTB level. If we want to use a PCI controller embedded DMA, some small
changes need if based on my other Designware PCI eDMA patches, which are under review.Adding dmaengine API to do memcopy should be simple to add in vhost/virtio interface.
Quoted 6 lines
Quoted 6 lines
3. All the major data transfer of NTB is using write. Because TLP write needn't wait for complete, write
performance is better than reading. On our platform, write performance is about 10% better than read.
Quoted 6 lines
Quoted 3 lines
Quoted 1 line
Frank
Any Comments or rejection? @Kishon Vijay Abraham I
I'd strongly recommend going with virtio/vhost based approach and standardizing it IMO.
But No progress in recent years on this path. At least, my patches make PCIe EP work as enet with minimized change.
And NTB don't conflict with virtio/vhost solution.
Frank
Quoted 61 lines
Thanks,
KishonQuoted 57 lines
best regards
Frank LiQuoted 53 lines
Quoted 53 lines
Thanks,
Kishon[1] -> https://lore.kernel.org/r/[email protected]
Quoted 47 lines
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EPFrank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function deviceDocumentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.c
Jon Mason
maintainerQuoted 36 lines · Frank Li
This implement NTB function for PCIe EP to RC connections. The existed ntb epf need two PCI EPs and two PCI Host.
This just need EP to RC connections.
┌────────────┐ ┌─────────────────────────────────────┐ │ │ │ │ ├────────────┤ │ ┌──────────────┤ │ NTB │ │ │ NTB │ │ NetDev │ │ │ NetDev │ ├────────────┤ │ ├──────────────┤ │ NTB │ │ │ NTB │ │ Transfer │ │ │ Transfer │ ├────────────┤ │ ├──────────────┤ │ │ │ │ │ │ PCI NTB │ │ │ │ │ EPF │ │ │ │ │ Driver │ │ │ PCI Virtual │ │ │ ├───────────────┐ │ NTB Driver │ │ │ │ PCI EP NTB │◄────►│ │ │ │ │ FN Driver │ │ │ ├────────────┤ ├───────────────┤ ├──────────────┤ │ │ │ │ │ │ │ PCI BUS │ ◄─────► │ PCI EP BUS │ │ Virtual PCI │ │ │ PCI │ │ │ BUS │ └────────────┘ └───────────────┴──────┴──────────────┘ PCI RC PCI EPFrank Li (4):
PCI: designware-ep: Allowpci_epc_set_bar()update inbound map address
NTB: epf: Allow more flexibility in the memory BAR map method
PCI: endpoint: Support NTB transfer between RC and EP
Documentation: PCI: Add specification for the PCI vNTB function device
Quoted 16 lines
Documentation/PCI/endpoint/index.rst | 2 + .../PCI/endpoint/pci-vntb-function.rst | 126 ++ Documentation/PCI/endpoint/pci-vntb-howto.rst | 167 ++ drivers/ntb/hw/epf/ntb_hw_epf.c | 48 +- .../pci/controller/dwc/pcie-designware-ep.c | 10 +- drivers/pci/endpoint/functions/Kconfig | 11 + drivers/pci/endpoint/functions/Makefile | 1 + drivers/pci/endpoint/functions/pci-epf-vntb.c | 1424 +++++++++++++++++ 8 files changed, 1775 insertions(+), 14 deletions(-) create mode 100644 Documentation/PCI/endpoint/pci-vntb-function.rst create mode 100644 Documentation/PCI/endpoint/pci-vntb-howto.rst create mode 100644 drivers/pci/endpoint/functions/pci-epf-vntb.cSignature · 1 line
2.24.0.rc1
Sorry for the extremely long delay in response. This series has been in my ntb-next branch for some time and will be in my pull request for v5.20 which should be going out later today.
Thanks,
Jon