[PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()

[PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()

[Riyan Dhiman]

Change num_inbound_wins from u8 to int in brcm_pcie_setup() function to correctly 
handle potential negative error codes returned by brcm_pcie_get_inbound_wins().
The u8 type was inappropriate for capturing the function's return value,
which can include error codes.

Signed-off-by: Riyan Dhiman <riyandhiman14@gmail.com>
---
 drivers/pci/controller/pcie-brcmstb.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/controller/pcie-brcmstb.c b/drivers/pci/controller/pcie-brcmstb.c
index e8332fe5396e..b2859c4fd931 100644
--- a/drivers/pci/controller/pcie-brcmstb.c
+++ b/drivers/pci/controller/pcie-brcmstb.c
@@ -1030,7 +1030,8 @@ static int brcm_pcie_setup(struct brcm_pcie *pcie)
 	struct pci_host_bridge *bridge;
 	struct resource_entry *entry;
 	u32 tmp, burst, aspm_support;
-	u8 num_out_wins = 0, num_inbound_wins = 0;
+	u8 num_out_wins = 0
+	int num_inbound_wins = 0;
 	int memc, ret;
 
 	/* Reset the bridge */
-- 
2.46.0

Re: [PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()

[florian.fainelli]

On 9/4/24 09:19, Riyan Dhiman wrote:
> Change num_inbound_wins from u8 to int in brcm_pcie_setup() function to correctly
> handle potential negative error codes returned by brcm_pcie_get_inbound_wins().
> The u8 type was inappropriate for capturing the function's return value,
> which can include error codes.
> 
> Signed-off-by: Riyan Dhiman <riyandhiman14@gmail.com>

This looks fine, however it seems like we could either:

- update brcm_pcie_get_inbound_wins() to take a reference to an u8 and 
assign num_inbound_wins directly plus return a negative error code

or

- update brcm_pcie_get_inbound_wins() to return 0 when encountering an error

We should have at least 1 inbound window to operate this PCIe 
controller, so if we get 0, nothing useful is going to happen.

Deferring to Jim as to whether he prefers to take your patch or fix it 
in a different way. Thanks!
-- 
Florian

Re: [PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()

[Bjorn Helgaas]

On Wed, Sep 04, 2024 at 09:49:54PM +0530, Riyan Dhiman wrote:
> Change num_inbound_wins from u8 to int in brcm_pcie_setup() function to correctly 
> handle potential negative error codes returned by brcm_pcie_get_inbound_wins().
> The u8 type was inappropriate for capturing the function's return value,
> which can include error codes.
> 
> Signed-off-by: Riyan Dhiman <riyandhiman14@gmail.com>

Apparently a fix for 46c981fd60de ("PCI: brcmstb: Refactor for chips
with many regular inbound windows"), which is currently queued on the 
pci/controller/brcmstb branch?

I agree, this looks good, and we should squash it into 46c981fd60de.

> ---
>  drivers/pci/controller/pcie-brcmstb.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/pci/controller/pcie-brcmstb.c b/drivers/pci/controller/pcie-brcmstb.c
> index e8332fe5396e..b2859c4fd931 100644
> --- a/drivers/pci/controller/pcie-brcmstb.c
> +++ b/drivers/pci/controller/pcie-brcmstb.c
> @@ -1030,7 +1030,8 @@ static int brcm_pcie_setup(struct brcm_pcie *pcie)
>  	struct pci_host_bridge *bridge;
>  	struct resource_entry *entry;
>  	u32 tmp, burst, aspm_support;
> -	u8 num_out_wins = 0, num_inbound_wins = 0;
> +	u8 num_out_wins = 0
> +	int num_inbound_wins = 0;
>  	int memc, ret;
>  
>  	/* Reset the bridge */
> -- 
> 2.46.0
> 

Re: [PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()

[Krzysztof Wilczyński]

Hello,

> > Change num_inbound_wins from u8 to int in brcm_pcie_setup() function to correctly
> > handle potential negative error codes returned by brcm_pcie_get_inbound_wins().
> > The u8 type was inappropriate for capturing the function's return value,
> > which can include error codes.
[...]
> This looks fine, however it seems like we could either:
> 
> - update brcm_pcie_get_inbound_wins() to take a reference to an u8 and
> assign num_inbound_wins directly plus return a negative error code
> 
> or
> 
> - update brcm_pcie_get_inbound_wins() to return 0 when encountering an error
> 
> We should have at least 1 inbound window to operate this PCIe controller, so
> if we get 0, nothing useful is going to happen.
> 
> Deferring to Jim as to whether he prefers to take your patch or fix it in a
> different way. Thanks!

The former would be my preference.

As such, I can make the change on the branch directly, if needed.  To avoid
reposting or sending a new patch.

	Krzysztof

Re: [PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()

[kernel test robot]

Hi Riyan,

kernel test robot noticed the following build errors:

[auto build test ERROR on next-20240904]

url:    https://github.com/intel-lab-lkp/linux/commits/Riyan-Dhiman/PCI-brmstb-Fix-type-mismatch-for-num_inbound_wins-in-brcm_pcie_setup/20240905-002339
base:   next-20240904
patch link:    https://lore.kernel.org/r/20240904161953.46790-2-riyandhiman14%40gmail.com
patch subject: [PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()
config: i386-buildonly-randconfig-006-20240907 (https://download.01.org/0day-ci/archive/20240907/202409072334.YP8Xi0bX-lkp@intel.com/config)
compiler: gcc-12 (Debian 12.2.0-14) 12.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240907/202409072334.YP8Xi0bX-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202409072334.YP8Xi0bX-lkp@intel.com/

All errors (new ones prefixed by >>):

   drivers/pci/controller/pcie-brcmstb.c: In function 'brcm_pcie_setup':
>> drivers/pci/controller/pcie-brcmstb.c:1034:9: error: expected ',' or ';' before 'int'
    1034 |         int num_inbound_wins = 0;
         |         ^~~
>> drivers/pci/controller/pcie-brcmstb.c:1093:9: error: 'num_inbound_wins' undeclared (first use in this function); did you mean 'inbound_wins'?
    1093 |         num_inbound_wins = brcm_pcie_get_inbound_wins(pcie, inbound_wins);
         |         ^~~~~~~~~~~~~~~~
         |         inbound_wins
   drivers/pci/controller/pcie-brcmstb.c:1093:9: note: each undeclared identifier is reported only once for each function it appears in


vim +1034 drivers/pci/controller/pcie-brcmstb.c

  1025	
  1026	static int brcm_pcie_setup(struct brcm_pcie *pcie)
  1027	{
  1028		struct inbound_win inbound_wins[PCIE_BRCM_MAX_INBOUND_WINS];
  1029		void __iomem *base = pcie->base;
  1030		struct pci_host_bridge *bridge;
  1031		struct resource_entry *entry;
  1032		u32 tmp, burst, aspm_support;
  1033		u8 num_out_wins = 0
> 1034		int num_inbound_wins = 0;
  1035		int memc, ret;
  1036	
  1037		/* Reset the bridge */
  1038		ret = pcie->bridge_sw_init_set(pcie, 1);
  1039		if (ret)
  1040			return ret;
  1041	
  1042		/* Ensure that PERST# is asserted; some bootloaders may deassert it. */
  1043		if (pcie->soc_base == BCM2711) {
  1044			ret = pcie->perst_set(pcie, 1);
  1045			if (ret) {
  1046				pcie->bridge_sw_init_set(pcie, 0);
  1047				return ret;
  1048			}
  1049		}
  1050	
  1051		usleep_range(100, 200);
  1052	
  1053		/* Take the bridge out of reset */
  1054		ret = pcie->bridge_sw_init_set(pcie, 0);
  1055		if (ret)
  1056			return ret;
  1057	
  1058		tmp = readl(base + HARD_DEBUG(pcie));
  1059		if (is_bmips(pcie))
  1060			tmp &= ~PCIE_BMIPS_MISC_HARD_PCIE_HARD_DEBUG_SERDES_IDDQ_MASK;
  1061		else
  1062			tmp &= ~PCIE_MISC_HARD_PCIE_HARD_DEBUG_SERDES_IDDQ_MASK;
  1063		writel(tmp, base + HARD_DEBUG(pcie));
  1064		/* Wait for SerDes to be stable */
  1065		usleep_range(100, 200);
  1066	
  1067		/*
  1068		 * SCB_MAX_BURST_SIZE is a two bit field.  For GENERIC chips it
  1069		 * is encoded as 0=128, 1=256, 2=512, 3=Rsvd, for BCM7278 it
  1070		 * is encoded as 0=Rsvd, 1=128, 2=256, 3=512.
  1071		 */
  1072		if (is_bmips(pcie))
  1073			burst = 0x1; /* 256 bytes */
  1074		else if (pcie->soc_base == BCM2711)
  1075			burst = 0x0; /* 128 bytes */
  1076		else if (pcie->soc_base == BCM7278)
  1077			burst = 0x3; /* 512 bytes */
  1078		else
  1079			burst = 0x2; /* 512 bytes */
  1080	
  1081		/*
  1082		 * Set SCB_MAX_BURST_SIZE, CFG_READ_UR_MODE, SCB_ACCESS_EN,
  1083		 * RCB_MPS_MODE, RCB_64B_MODE
  1084		 */
  1085		tmp = readl(base + PCIE_MISC_MISC_CTRL);
  1086		u32p_replace_bits(&tmp, 1, PCIE_MISC_MISC_CTRL_SCB_ACCESS_EN_MASK);
  1087		u32p_replace_bits(&tmp, 1, PCIE_MISC_MISC_CTRL_CFG_READ_UR_MODE_MASK);
  1088		u32p_replace_bits(&tmp, burst, PCIE_MISC_MISC_CTRL_MAX_BURST_SIZE_MASK);
  1089		u32p_replace_bits(&tmp, 1, PCIE_MISC_MISC_CTRL_PCIE_RCB_MPS_MODE_MASK);
  1090		u32p_replace_bits(&tmp, 1, PCIE_MISC_MISC_CTRL_PCIE_RCB_64B_MODE_MASK);
  1091		writel(tmp, base + PCIE_MISC_MISC_CTRL);
  1092	
> 1093		num_inbound_wins = brcm_pcie_get_inbound_wins(pcie, inbound_wins);
  1094		if (num_inbound_wins < 0)
  1095			return num_inbound_wins;
  1096	
  1097		set_inbound_win_registers(pcie, inbound_wins, num_inbound_wins);
  1098	
  1099		if (!brcm_pcie_rc_mode(pcie)) {
  1100			dev_err(pcie->dev, "PCIe RC controller misconfigured as Endpoint\n");
  1101			return -EINVAL;
  1102		}
  1103	
  1104		tmp = readl(base + PCIE_MISC_MISC_CTRL);
  1105		for (memc = 0; memc < pcie->num_memc; memc++) {
  1106			u32 scb_size_val = ilog2(pcie->memc_size[memc]) - 15;
  1107	
  1108			if (memc == 0)
  1109				u32p_replace_bits(&tmp, scb_size_val, SCB_SIZE_MASK(0));
  1110			else if (memc == 1)
  1111				u32p_replace_bits(&tmp, scb_size_val, SCB_SIZE_MASK(1));
  1112			else if (memc == 2)
  1113				u32p_replace_bits(&tmp, scb_size_val, SCB_SIZE_MASK(2));
  1114		}
  1115		writel(tmp, base + PCIE_MISC_MISC_CTRL);
  1116	
  1117		/*
  1118		 * We ideally want the MSI target address to be located in the 32bit
  1119		 * addressable memory area. Some devices might depend on it. This is
  1120		 * possible either when the inbound window is located above the lower
  1121		 * 4GB or when the inbound area is smaller than 4GB (taking into
  1122		 * account the rounding-up we're forced to perform).
  1123		 */
  1124		if (inbound_wins[2].pci_offset >= SZ_4G ||
  1125		    (inbound_wins[2].size + inbound_wins[2].pci_offset) < SZ_4G)
  1126			pcie->msi_target_addr = BRCM_MSI_TARGET_ADDR_LT_4GB;
  1127		else
  1128			pcie->msi_target_addr = BRCM_MSI_TARGET_ADDR_GT_4GB;
  1129	
  1130	
  1131		/* Don't advertise L0s capability if 'aspm-no-l0s' */
  1132		aspm_support = PCIE_LINK_STATE_L1;
  1133		if (!of_property_read_bool(pcie->np, "aspm-no-l0s"))
  1134			aspm_support |= PCIE_LINK_STATE_L0S;
  1135		tmp = readl(base + PCIE_RC_CFG_PRIV1_LINK_CAPABILITY);
  1136		u32p_replace_bits(&tmp, aspm_support,
  1137			PCIE_RC_CFG_PRIV1_LINK_CAPABILITY_ASPM_SUPPORT_MASK);
  1138		writel(tmp, base + PCIE_RC_CFG_PRIV1_LINK_CAPABILITY);
  1139	
  1140		/*
  1141		 * For config space accesses on the RC, show the right class for
  1142		 * a PCIe-PCIe bridge (the default setting is to be EP mode).
  1143		 */
  1144		tmp = readl(base + PCIE_RC_CFG_PRIV1_ID_VAL3);
  1145		u32p_replace_bits(&tmp, 0x060400,
  1146				  PCIE_RC_CFG_PRIV1_ID_VAL3_CLASS_CODE_MASK);
  1147		writel(tmp, base + PCIE_RC_CFG_PRIV1_ID_VAL3);
  1148	
  1149		bridge = pci_host_bridge_from_priv(pcie);
  1150		resource_list_for_each_entry(entry, &bridge->windows) {
  1151			struct resource *res = entry->res;
  1152	
  1153			if (resource_type(res) != IORESOURCE_MEM)
  1154				continue;
  1155	
  1156			if (num_out_wins >= BRCM_NUM_PCIE_OUT_WINS) {
  1157				dev_err(pcie->dev, "too many outbound wins\n");
  1158				return -EINVAL;
  1159			}
  1160	
  1161			if (is_bmips(pcie)) {
  1162				u64 start = res->start;
  1163				unsigned int j, nwins = resource_size(res) / SZ_128M;
  1164	
  1165				/* bmips PCIe outbound windows have a 128MB max size */
  1166				if (nwins > BRCM_NUM_PCIE_OUT_WINS)
  1167					nwins = BRCM_NUM_PCIE_OUT_WINS;
  1168				for (j = 0; j < nwins; j++, start += SZ_128M)
  1169					brcm_pcie_set_outbound_win(pcie, j, start,
  1170								   start - entry->offset,
  1171								   SZ_128M);
  1172				break;
  1173			}
  1174			brcm_pcie_set_outbound_win(pcie, num_out_wins, res->start,
  1175						   res->start - entry->offset,
  1176						   resource_size(res));
  1177			num_out_wins++;
  1178		}
  1179	
  1180		/* PCIe->SCB endian mode for inbound window */
  1181		tmp = readl(base + PCIE_RC_CFG_VENDOR_VENDOR_SPECIFIC_REG1);
  1182		u32p_replace_bits(&tmp, PCIE_RC_CFG_VENDOR_SPCIFIC_REG1_LITTLE_ENDIAN,
  1183			PCIE_RC_CFG_VENDOR_VENDOR_SPECIFIC_REG1_ENDIAN_MODE_BAR2_MASK);
  1184		writel(tmp, base + PCIE_RC_CFG_VENDOR_VENDOR_SPECIFIC_REG1);
  1185	
  1186		return 0;
  1187	}
  1188	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki

Re: [PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()

[kernel test robot]

Hi Riyan,

kernel test robot noticed the following build errors:

[auto build test ERROR on next-20240904]

url:    https://github.com/intel-lab-lkp/linux/commits/Riyan-Dhiman/PCI-brmstb-Fix-type-mismatch-for-num_inbound_wins-in-brcm_pcie_setup/20240905-002339
base:   next-20240904
patch link:    https://lore.kernel.org/r/20240904161953.46790-2-riyandhiman14%40gmail.com
patch subject: [PATCH next] PCI: brmstb: Fix type mismatch for num_inbound_wins in brcm_pcie_setup()
config: x86_64-allyesconfig (https://download.01.org/0day-ci/archive/20240908/202409080202.tevgFgI7-lkp@intel.com/config)
compiler: clang version 18.1.5 (https://github.com/llvm/llvm-project 617a15a9eac96088ae5e9134248d8236e34b91b1)
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240908/202409080202.tevgFgI7-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202409080202.tevgFgI7-lkp@intel.com/

All errors (new ones prefixed by >>):

>> drivers/pci/controller/pcie-brcmstb.c:1033:21: error: expected ';' at end of declaration
    1033 |         u8 num_out_wins = 0
         |                            ^
         |                            ;
   1 error generated.

Kconfig warnings: (for reference only)
   WARNING: unmet direct dependencies detected for OMAP2PLUS_MBOX
   Depends on [n]: MAILBOX [=y] && (ARCH_OMAP2PLUS || ARCH_K3)
   Selected by [y]:
   - TI_K3_M4_REMOTEPROC [=y] && REMOTEPROC [=y] && (ARCH_K3 || COMPILE_TEST [=y])


vim +1033 drivers/pci/controller/pcie-brcmstb.c

  1025	
  1026	static int brcm_pcie_setup(struct brcm_pcie *pcie)
  1027	{
  1028		struct inbound_win inbound_wins[PCIE_BRCM_MAX_INBOUND_WINS];
  1029		void __iomem *base = pcie->base;
  1030		struct pci_host_bridge *bridge;
  1031		struct resource_entry *entry;
  1032		u32 tmp, burst, aspm_support;
> 1033		u8 num_out_wins = 0
  1034		int num_inbound_wins = 0;
  1035		int memc, ret;
  1036	
  1037		/* Reset the bridge */
  1038		ret = pcie->bridge_sw_init_set(pcie, 1);
  1039		if (ret)
  1040			return ret;
  1041	
  1042		/* Ensure that PERST# is asserted; some bootloaders may deassert it. */
  1043		if (pcie->soc_base == BCM2711) {
  1044			ret = pcie->perst_set(pcie, 1);
  1045			if (ret) {
  1046				pcie->bridge_sw_init_set(pcie, 0);
  1047				return ret;
  1048			}
  1049		}
  1050	
  1051		usleep_range(100, 200);
  1052	
  1053		/* Take the bridge out of reset */
  1054		ret = pcie->bridge_sw_init_set(pcie, 0);
  1055		if (ret)
  1056			return ret;
  1057	
  1058		tmp = readl(base + HARD_DEBUG(pcie));
  1059		if (is_bmips(pcie))
  1060			tmp &= ~PCIE_BMIPS_MISC_HARD_PCIE_HARD_DEBUG_SERDES_IDDQ_MASK;
  1061		else
  1062			tmp &= ~PCIE_MISC_HARD_PCIE_HARD_DEBUG_SERDES_IDDQ_MASK;
  1063		writel(tmp, base + HARD_DEBUG(pcie));
  1064		/* Wait for SerDes to be stable */
  1065		usleep_range(100, 200);
  1066	
  1067		/*
  1068		 * SCB_MAX_BURST_SIZE is a two bit field.  For GENERIC chips it
  1069		 * is encoded as 0=128, 1=256, 2=512, 3=Rsvd, for BCM7278 it
  1070		 * is encoded as 0=Rsvd, 1=128, 2=256, 3=512.
  1071		 */
  1072		if (is_bmips(pcie))
  1073			burst = 0x1; /* 256 bytes */
  1074		else if (pcie->soc_base == BCM2711)
  1075			burst = 0x0; /* 128 bytes */
  1076		else if (pcie->soc_base == BCM7278)
  1077			burst = 0x3; /* 512 bytes */
  1078		else
  1079			burst = 0x2; /* 512 bytes */
  1080	
  1081		/*
  1082		 * Set SCB_MAX_BURST_SIZE, CFG_READ_UR_MODE, SCB_ACCESS_EN,
  1083		 * RCB_MPS_MODE, RCB_64B_MODE
  1084		 */
  1085		tmp = readl(base + PCIE_MISC_MISC_CTRL);
  1086		u32p_replace_bits(&tmp, 1, PCIE_MISC_MISC_CTRL_SCB_ACCESS_EN_MASK);
  1087		u32p_replace_bits(&tmp, 1, PCIE_MISC_MISC_CTRL_CFG_READ_UR_MODE_MASK);
  1088		u32p_replace_bits(&tmp, burst, PCIE_MISC_MISC_CTRL_MAX_BURST_SIZE_MASK);
  1089		u32p_replace_bits(&tmp, 1, PCIE_MISC_MISC_CTRL_PCIE_RCB_MPS_MODE_MASK);
  1090		u32p_replace_bits(&tmp, 1, PCIE_MISC_MISC_CTRL_PCIE_RCB_64B_MODE_MASK);
  1091		writel(tmp, base + PCIE_MISC_MISC_CTRL);
  1092	
  1093		num_inbound_wins = brcm_pcie_get_inbound_wins(pcie, inbound_wins);
  1094		if (num_inbound_wins < 0)
  1095			return num_inbound_wins;
  1096	
  1097		set_inbound_win_registers(pcie, inbound_wins, num_inbound_wins);
  1098	
  1099		if (!brcm_pcie_rc_mode(pcie)) {
  1100			dev_err(pcie->dev, "PCIe RC controller misconfigured as Endpoint\n");
  1101			return -EINVAL;
  1102		}
  1103	
  1104		tmp = readl(base + PCIE_MISC_MISC_CTRL);
  1105		for (memc = 0; memc < pcie->num_memc; memc++) {
  1106			u32 scb_size_val = ilog2(pcie->memc_size[memc]) - 15;
  1107	
  1108			if (memc == 0)
  1109				u32p_replace_bits(&tmp, scb_size_val, SCB_SIZE_MASK(0));
  1110			else if (memc == 1)
  1111				u32p_replace_bits(&tmp, scb_size_val, SCB_SIZE_MASK(1));
  1112			else if (memc == 2)
  1113				u32p_replace_bits(&tmp, scb_size_val, SCB_SIZE_MASK(2));
  1114		}
  1115		writel(tmp, base + PCIE_MISC_MISC_CTRL);
  1116	
  1117		/*
  1118		 * We ideally want the MSI target address to be located in the 32bit
  1119		 * addressable memory area. Some devices might depend on it. This is
  1120		 * possible either when the inbound window is located above the lower
  1121		 * 4GB or when the inbound area is smaller than 4GB (taking into
  1122		 * account the rounding-up we're forced to perform).
  1123		 */
  1124		if (inbound_wins[2].pci_offset >= SZ_4G ||
  1125		    (inbound_wins[2].size + inbound_wins[2].pci_offset) < SZ_4G)
  1126			pcie->msi_target_addr = BRCM_MSI_TARGET_ADDR_LT_4GB;
  1127		else
  1128			pcie->msi_target_addr = BRCM_MSI_TARGET_ADDR_GT_4GB;
  1129	
  1130	
  1131		/* Don't advertise L0s capability if 'aspm-no-l0s' */
  1132		aspm_support = PCIE_LINK_STATE_L1;
  1133		if (!of_property_read_bool(pcie->np, "aspm-no-l0s"))
  1134			aspm_support |= PCIE_LINK_STATE_L0S;
  1135		tmp = readl(base + PCIE_RC_CFG_PRIV1_LINK_CAPABILITY);
  1136		u32p_replace_bits(&tmp, aspm_support,
  1137			PCIE_RC_CFG_PRIV1_LINK_CAPABILITY_ASPM_SUPPORT_MASK);
  1138		writel(tmp, base + PCIE_RC_CFG_PRIV1_LINK_CAPABILITY);
  1139	
  1140		/*
  1141		 * For config space accesses on the RC, show the right class for
  1142		 * a PCIe-PCIe bridge (the default setting is to be EP mode).
  1143		 */
  1144		tmp = readl(base + PCIE_RC_CFG_PRIV1_ID_VAL3);
  1145		u32p_replace_bits(&tmp, 0x060400,
  1146				  PCIE_RC_CFG_PRIV1_ID_VAL3_CLASS_CODE_MASK);
  1147		writel(tmp, base + PCIE_RC_CFG_PRIV1_ID_VAL3);
  1148	
  1149		bridge = pci_host_bridge_from_priv(pcie);
  1150		resource_list_for_each_entry(entry, &bridge->windows) {
  1151			struct resource *res = entry->res;
  1152	
  1153			if (resource_type(res) != IORESOURCE_MEM)
  1154				continue;
  1155	
  1156			if (num_out_wins >= BRCM_NUM_PCIE_OUT_WINS) {
  1157				dev_err(pcie->dev, "too many outbound wins\n");
  1158				return -EINVAL;
  1159			}
  1160	
  1161			if (is_bmips(pcie)) {
  1162				u64 start = res->start;
  1163				unsigned int j, nwins = resource_size(res) / SZ_128M;
  1164	
  1165				/* bmips PCIe outbound windows have a 128MB max size */
  1166				if (nwins > BRCM_NUM_PCIE_OUT_WINS)
  1167					nwins = BRCM_NUM_PCIE_OUT_WINS;
  1168				for (j = 0; j < nwins; j++, start += SZ_128M)
  1169					brcm_pcie_set_outbound_win(pcie, j, start,
  1170								   start - entry->offset,
  1171								   SZ_128M);
  1172				break;
  1173			}
  1174			brcm_pcie_set_outbound_win(pcie, num_out_wins, res->start,
  1175						   res->start - entry->offset,
  1176						   resource_size(res));
  1177			num_out_wins++;
  1178		}
  1179	
  1180		/* PCIe->SCB endian mode for inbound window */
  1181		tmp = readl(base + PCIE_RC_CFG_VENDOR_VENDOR_SPECIFIC_REG1);
  1182		u32p_replace_bits(&tmp, PCIE_RC_CFG_VENDOR_SPCIFIC_REG1_LITTLE_ENDIAN,
  1183			PCIE_RC_CFG_VENDOR_VENDOR_SPECIFIC_REG1_ENDIAN_MODE_BAR2_MASK);
  1184		writel(tmp, base + PCIE_RC_CFG_VENDOR_VENDOR_SPECIFIC_REG1);
  1185	
  1186		return 0;
  1187	}
  1188	

-- 
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki