Skip to contentSkip to conversation

rockchip rk3399 port initialization fixes

PATCH Superseded Damien Le Moal 3 participants 7 messages Apr 12, 2024 – Apr 15, 2024

Subsystems (4)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v3 cover rockchip rk3399 port initialization fixes

A couple of patches to have the rockchip rk3399 host controller port initialization follow the PCI specifications around PERST# signal timing handling.

Changes from v2:
- Use PCIE_T_PVPERL_MS macro in patch 1 (and remove useless comments).
- Split second msleep() addition into patch 2 as suggested by Bjorn.

Changes from v1:
- Add more specification details to the commit message.
- Add missing msleep(100) after PERST# is deasserted.

Damien Le Moal (2):
PCI: rockchip-host: Fix rockchip_pcie_host_init_port() PERST# handling
PCI: rockchip-host: Wait 100ms after reset before starting configuration

 drivers/pci/controller/pcie-rockchip-host.c | 3 +++
 drivers/pci/pci.h                           | 7 +++++++
 2 files changed, 10 insertions(+)
Signature · 1 line
2.44.0
#1 lore raw

PATCH v3 1/2 PCI: rockchip-host: Fix rockchip_pcie_host_init_port() PERST# handling

The PCIe specifications (PCI Express Electromechanical Specification rev 2.0, section 2.6.2) mandate that the PERST# signal must remain asserted for at least 100 usec (Tperst-clk) after the PCIe reference clock becomes stable (if a reference clock is supplied), and for at least 100 msec after the power is stable (Tpvperl, defined by the macro PCIE_T_PVPERL_MS).

Modify rockchip_pcie_host_init_port() to satisfy these constraints by adding a sleep period before bringing back PESRT# signal to high using the ep_gpio GPIO. Since Tperst-clk is the shorter wait time, add an msleep() call for the longer PCIE_T_PVPERL_MS milliseconds to handle both timing requirements.

---
 drivers/pci/controller/pcie-rockchip-host.c | 1 +
 1 file changed, 1 insertion(+)
drivers/pci/controller/pcie-rockchip-host.c +1 -0
index 300b9dc85ecc..fc868251e570 100644
@@ -322,6 +322,7 @@ static int rockchip_pcie_host_init_port(struct rockchip_pcie *rockchip)
 	rockchip_pcie_write(rockchip, PCIE_CLIENT_LINK_TRAIN_ENABLE,
 			    PCIE_CLIENT_CONFIG);
 
+	msleep(PCIE_T_PVPERL_MS);
 	gpiod_set_value_cansleep(rockchip->ep_gpio, 1);
 
 	/* 500ms timeout value should be enough for Gen1/2 training */
Signature · 1 line
2.44.0
#2 lore raw

Re: [PATCH v3 1/2] PCI: rockchip-host: Fix rockchip_pcie_host_init_port() PERST# handling

Quoted 2 lines · Damien Le Moal

The PCIe specifications (PCI Express Electromechanical Specification rev 2.0, section 2.6.2) mandate that the PERST# signal must remain asserted

"PCIe CEM r5.1, sec 2.9.2"

Quoted 7 lines

for at least 100 usec (Tperst-clk) after the PCIe reference clock becomes stable (if a reference clock is supplied), and for at least 100 msec after the power is stable (Tpvperl, defined by the macro PCIE_T_PVPERL_MS).

Modify rockchip_pcie_host_init_port() to satisfy these constraints by adding a sleep period before bringing back PESRT# signal to high using

s/PESRT#/PERST#/
s/bringing back PERST# to high/deasserting PERST#/

Whoever applies this can probably fix these up for you.

Quoted 14 lines

the ep_gpio GPIO. Since Tperst-clk is the shorter wait time, add an msleep() call for the longer PCIE_T_PVPERL_MS milliseconds to handle both timing requirements.

---
 drivers/pci/controller/pcie-rockchip-host.c | 1 +
 1 file changed, 1 insertion(+)
drivers/pci/controller/pcie-rockchip-host.c +0 -0
index 300b9dc85ecc..fc868251e570 100644

drivers/pci/controller/pcie-rockchip-host.c static int rockchip_pcie_host_init_port(struct rockchip_pcie *rockchip) · line 325

Quoted 4 lines
drivers/pci/controller/pcie-rockchip-host.c +1 -0
@@ -322,6 +322,7 @@ static int rockchip_pcie_host_init_port(struct rockchip_pcie *rockchip)
 	rockchip_pcie_write(rockchip, PCIE_CLIENT_LINK_TRAIN_ENABLE,
 			    PCIE_CLIENT_CONFIG);
 
+	msleep(PCIE_T_PVPERL_MS);

Looks good, thanks!

Quoted 6 lines
drivers/pci/controller/pcie-rockchip-host.c +0 -0
 	gpiod_set_value_cansleep(rockchip->ep_gpio, 1);
 
 	/* 500ms timeout value should be enough for Gen1/2 training */
Signature · 1 line
2.44.0
Reviewed-by on v3 1/2
#3 lore raw

Re: [PATCH v3 1/2] PCI: rockchip-host: Fix rockchip_pcie_host_init_port() PERST# handling

In reply to Damien Le Moal
Quoted 7 lines · Damien Le Moal

The PCIe specifications (PCI Express Electromechanical Specification rev 2.0, section 2.6.2) mandate that the PERST# signal must remain asserted for at least 100 usec (Tperst-clk) after the PCIe reference clock becomes stable (if a reference clock is supplied), and for at least 100 msec after the power is stable (Tpvperl, defined by the macro PCIE_T_PVPERL_MS).

Quoted 7 lines

Modify rockchip_pcie_host_init_port() to satisfy these constraints by adding a sleep period before bringing back PESRT# signal to high using the ep_gpio GPIO. Since Tperst-clk is the shorter wait time, add an msleep() call for the longer PCIE_T_PVPERL_MS milliseconds to handle both timing requirements.

- Mani

Quoted 20 lines
---
 drivers/pci/controller/pcie-rockchip-host.c | 1 +
 1 file changed, 1 insertion(+)
drivers/pci/controller/pcie-rockchip-host.c +1 -0
index 300b9dc85ecc..fc868251e570 100644
@@ -322,6 +322,7 @@ static int rockchip_pcie_host_init_port(struct rockchip_pcie *rockchip)
 	rockchip_pcie_write(rockchip, PCIE_CLIENT_LINK_TRAIN_ENABLE,
 			    PCIE_CLIENT_CONFIG);
 
+	msleep(PCIE_T_PVPERL_MS);
 	gpiod_set_value_cansleep(rockchip->ep_gpio, 1);
 
 	/* 500ms timeout value should be enough for Gen1/2 training */
Signature · 1 line
2.44.0
Signature · 1 line
மணிவண்ணன் சதாசிவம்
#4 lore raw

PATCH v3 2/2 PCI: rockchip-host: Wait 100ms after reset before starting configuration

In reply to Damien Le Moal

The PCI Express Base Specification r6.0, section 6.6.1, states that the host should wait for at least 100 msec from the end of a conventional reset (PERST# is de-asserted) before sending a configuration request to ensure that the device is able to respond with a "Request Retry Status" completion.

Add the PCIE_T_RRS_READY_MS macro to define this wait time and modify rockchip_pcie_host_init_port() to add this 100ms sleep after bringing back PESRT# signal to high using the ep_gpio GPIO.

---
 drivers/pci/controller/pcie-rockchip-host.c | 2 ++
 drivers/pci/pci.h                           | 7 +++++++
 2 files changed, 9 insertions(+)
drivers/pci/controller/pcie-rockchip-host.c +2 -0
index fc868251e570..cbec71114825 100644
@@ -325,6 +325,8 @@ static int rockchip_pcie_host_init_port(struct rockchip_pcie *rockchip)
 	msleep(PCIE_T_PVPERL_MS);
 	gpiod_set_value_cansleep(rockchip->ep_gpio, 1);
 
+	msleep(PCIE_T_RRS_READY_MS);
+
 	/* 500ms timeout value should be enough for Gen1/2 training */
 	err = readl_poll_timeout(rockchip->apb_base + PCIE_CLIENT_BASIC_STATUS1,
 				 status, PCIE_LINK_UP(status), 20,
drivers/pci/pci.h +7 -0
index 17fed1846847..c93ffc5e6e1f 100644
@@ -16,6 +16,13 @@
 /* Power stable to PERST# inactive from PCIe card Electromechanical Spec */
 #define PCIE_T_PVPERL_MS		100
 
+/*
+ * End of conventional reset (PERST# de-asserted) to first configuration
+ * request (device able to respond with a "Request Retry Status" completion),
+ * from PCI Express Base Specification r6.0, section 6.6.1.
+ */
+#define PCIE_T_RRS_READY_MS	100
+
 /*
  * PCIe r6.0, sec 5.3.3.2.1 <PME Synchronization>
  * Recommends 1ms to 10ms timeout to check L2 ready.
Signature · 1 line
2.44.0
#5 lore raw

Re: [PATCH v3 2/2] PCI: rockchip-host: Wait 100ms after reset before starting configuration

Quoted 9 lines · Damien Le Moal

The PCI Express Base Specification r6.0, section 6.6.1, states that the host should wait for at least 100 msec from the end of a conventional reset (PERST# is de-asserted) before sending a configuration request to ensure that the device is able to respond with a "Request Retry Status" completion.

Add the PCIE_T_RRS_READY_MS macro to define this wait time and modify rockchip_pcie_host_init_port() to add this 100ms sleep after bringing back PESRT# signal to high using the ep_gpio GPIO.

s/PESRT#/PERST#/
s/bringing back PERST# signal to high/deasserting PERST#/

Quoted 25 lines
---
 drivers/pci/controller/pcie-rockchip-host.c | 2 ++
 drivers/pci/pci.h                           | 7 +++++++
 2 files changed, 9 insertions(+)
drivers/pci/controller/pcie-rockchip-host.c +2 -0
index fc868251e570..cbec71114825 100644
@@ -325,6 +325,8 @@ static int rockchip_pcie_host_init_port(struct rockchip_pcie *rockchip)
 	msleep(PCIE_T_PVPERL_MS);
 	gpiod_set_value_cansleep(rockchip->ep_gpio, 1);
 
+	msleep(PCIE_T_RRS_READY_MS);
+
 	/* 500ms timeout value should be enough for Gen1/2 training */
 	err = readl_poll_timeout(rockchip->apb_base + PCIE_CLIENT_BASIC_STATUS1,
 				 status, PCIE_LINK_UP(status), 20,
drivers/pci/pci.h +0 -0
index 17fed1846847..c93ffc5e6e1f 100644

drivers/pci/pci.h line 22

Quoted 7 lines
drivers/pci/pci.h +4 -0
@@ -16,6 +16,13 @@
 /* Power stable to PERST# inactive from PCIe card Electromechanical Spec */
 #define PCIE_T_PVPERL_MS		100
 
+/*
+ * End of conventional reset (PERST# de-asserted) to first configuration
+ * request (device able to respond with a "Request Retry Status" completion),
+ * from PCI Express Base Specification r6.0, section 6.6.1.

"PCIe r6.0, sec 6.6.1" to match typical style, e.g., the reference just below.

Whoever applies this can take care of this.

drivers/pci/pci.h line 24

Quoted 1 line
drivers/pci/pci.h +1 -0
+#define PCIE_T_RRS_READY_MS	100

Thanks a lot for doing this; there are many similar places we can update to use this #define.

Quoted 6 lines
drivers/pci/pci.h +0 -0
 /*
  * PCIe r6.0, sec 5.3.3.2.1 <PME Synchronization>
  * Recommends 1ms to 10ms timeout to check L2 ready.
Signature · 1 line
2.44.0
#6 lore raw

Re: [PATCH v3 2/2] PCI: rockchip-host: Wait 100ms after reset before starting configuration

Quoted 50 lines · Bjorn Helgaas
Quoted 9 lines · Damien Le Moal

The PCI Express Base Specification r6.0, section 6.6.1, states that the host should wait for at least 100 msec from the end of a conventional reset (PERST# is de-asserted) before sending a configuration request to ensure that the device is able to respond with a "Request Retry Status" completion.

Add the PCIE_T_RRS_READY_MS macro to define this wait time and modify rockchip_pcie_host_init_port() to add this 100ms sleep after bringing back PESRT# signal to high using the ep_gpio GPIO.

s/PESRT#/PERST#/
s/bringing back PERST# signal to high/deasserting PERST#/

Quoted 25 lines
---
 drivers/pci/controller/pcie-rockchip-host.c | 2 ++
 drivers/pci/pci.h                           | 7 +++++++
 2 files changed, 9 insertions(+)
drivers/pci/controller/pcie-rockchip-host.c +2 -0
index fc868251e570..cbec71114825 100644
@@ -325,6 +325,8 @@ static int rockchip_pcie_host_init_port(struct rockchip_pcie *rockchip)
 	msleep(PCIE_T_PVPERL_MS);
 	gpiod_set_value_cansleep(rockchip->ep_gpio, 1);
 
+	msleep(PCIE_T_RRS_READY_MS);
+
 	/* 500ms timeout value should be enough for Gen1/2 training */
 	err = readl_poll_timeout(rockchip->apb_base + PCIE_CLIENT_BASIC_STATUS1,
 				 status, PCIE_LINK_UP(status), 20,
drivers/pci/pci.h +0 -0
index 17fed1846847..c93ffc5e6e1f 100644

drivers/pci/pci.h line 22

Quoted 7 lines
drivers/pci/pci.h +4 -0
@@ -16,6 +16,13 @@
 /* Power stable to PERST# inactive from PCIe card Electromechanical Spec */
 #define PCIE_T_PVPERL_MS		100
 
+/*
+ * End of conventional reset (PERST# de-asserted) to first configuration
+ * request (device able to respond with a "Request Retry Status" completion),
+ * from PCI Express Base Specification r6.0, section 6.6.1.

"PCIe r6.0, sec 6.6.1" to match typical style, e.g., the reference just below.

Quoted 1 line

Whoever applies this can take care of this.

To make it easier to apply, I sent v4 with the corrections. Thanks.

Quoted 12 lines

drivers/pci/pci.h line 24

Quoted 2 lines
drivers/pci/pci.h +1 -0
+#define PCIE_T_RRS_READY_MS	100

Thanks a lot for doing this; there are many similar places we can update to use this #define.

Quoted 6 lines
drivers/pci/pci.h +0 -0
 /*
  * PCIe r6.0, sec 5.3.3.2.1 <PME Synchronization>
  * Recommends 1ms to 10ms timeout to check L2 ready.
Signature · 1 line
2.44.0
Signature · 2 lines
Damien Le Moal
Western Digital Research