| Message ID | 1381853200-5534-7-git-send-email-marex@denx.de (mailing list archive) |
|---|---|
| State | New, archived |
| Delegated to: | Bjorn Helgaas |
| Headers | show |
On Tue, Oct 15, 2013 at 9:06 AM, Marek Vasut <marex@denx.de> wrote: > > The resource retrieval from the DT was not done right. This resulted > in bogus values in the IO/MEM window configuration. I re-used the code > implementation from pci-tegra.c to do the resource retrieval correctly. > > This in turn resulted in very strange behavior on two different kinds > of setup: > > Setup #1: ,-------> Intel i210 (igb driver) > i.MX6 root port -----> PCIe switch > `-------> Empty slot > > Setup #2: ,-------> Marvell Yukon (sky2 driver) > i.MX6 root port -----> PCIe switch > `-------> Empty slot > > Both setups expressed the same behavior. The system booted and the > ethernet controllers were properly detected. After the controllers > were configured and just after they reported "Ethernet Link Up", the > entire system froze. It is now clear this happened because the resources > were incorrectly configured, which in turn means the iATU was also not > configured correctly. > > Signed-off-by: Marek Vasut <marex@denx.de> > Cc: Bjorn Helgaas <bhelgaas@google.com> > Cc: Frank Li <lznuaa@gmail.com> > Cc: Jingoo Han <jg1.han@samsung.com> > Cc: Mohit KUMAR <Mohit.KUMAR@st.com> > Cc: Pratyush Anand <pratyush.anand@st.com> > Cc: Richard Zhu <r65037@freescale.com> > Cc: Sascha Hauer <s.hauer@pengutronix.de> > Cc: Sean Cross <xobs@kosagi.com> > Cc: Shawn Guo <shawn.guo@linaro.org> > Cc: Siva Reddy Kallam <siva.kallam@samsung.com> > Cc: Srikanth T Shivanand <ts.srikanth@samsung.com> > Cc: Tim Harvey <tharvey@gateworks.com> > Cc: Troy Kisky <troy.kisky@boundarydevices.com> > Cc: Yinghai Lu <yinghai@kernel.org> > --- > drivers/pci/host/pcie-designware.c | 17 ++++++----------- > 1 file changed, 6 insertions(+), 11 deletions(-) > > diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c > index 5d183ae..97c5951 100644 > --- a/drivers/pci/host/pcie-designware.c > +++ b/drivers/pci/host/pcie-designware.c > @@ -370,6 +370,7 @@ int __init dw_pcie_host_init(struct pcie_port *pp) > struct device_node *np = pp->dev->of_node; > struct of_pci_range range; > struct of_pci_range_parser parser; > + struct resource res; > u32 val; > > struct irq_domain *irq_domain; > @@ -381,28 +382,22 @@ int __init dw_pcie_host_init(struct pcie_port *pp) > > /* Get the I/O and memory ranges from DT */ > for_each_of_pci_range(&parser, &range) { > - unsigned long restype = range.flags & IORESOURCE_TYPE_BITS; > + of_pci_range_to_resource(&range, np, &res); > + unsigned long restype = res.flags & IORESOURCE_TYPE_BITS; > if (restype == IORESOURCE_IO) { > - of_pci_range_to_resource(&range, np, &pp->io); > + memcpy(&pp->io, &res, sizeof(res)); > pp->io.name = "I/O"; > - pp->io.start = max_t(resource_size_t, > - PCIBIOS_MIN_IO, > - range.pci_addr + global_io_offset); > - pp->io.end = min_t(resource_size_t, > - IO_SPACE_LIMIT, > - range.pci_addr + range.size > - + global_io_offset); Nak - removing the adjustment to io.start/io.end breaks the io space mapping as the entire io-space allocation gets taken by the root complex. This causes any device that requests io space to fail to obtain it (your igb driver from Setup1 apparently doesn't use io space where my sky2 device from Setup2 above does). If I put back the adjustment to io.start/io.end above my sky2 device gets its io resource properly but still hangs the bus after a few interrupts. Are you sure you don't have other changes that you haven't posted yet that could have addressed the hang you get when you bring up your igb device? I agree that the the 'hang' we are seeing is a bus-hang caused by an improper iATU mapping, however I'm not convinced its a dts parsing issue. Tim > pp->config.io_size = resource_size(&pp->io); > pp->config.io_bus_addr = range.pci_addr; > } > if (restype == IORESOURCE_MEM) { > - of_pci_range_to_resource(&range, np, &pp->mem); > + memcpy(&pp->mem, &res, sizeof(res)); > pp->mem.name = "MEM"; > pp->config.mem_size = resource_size(&pp->mem); > pp->config.mem_bus_addr = range.pci_addr; > } > if (restype == 0) { > - of_pci_range_to_resource(&range, np, &pp->cfg); > + memcpy(&pp->cfg, &res, sizeof(res)); > pp->config.cfg0_size = resource_size(&pp->cfg)/2; > pp->config.cfg1_size = resource_size(&pp->cfg)/2; > } > -- > 1.8.4.rc3 > -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wednesday, October 16, 2013 9:15 AM, Tim Harvey wrote: > On Tue, Oct 15, 2013 at 9:06 AM, Marek Vasut <marex@denx.de> wrote: > > > > The resource retrieval from the DT was not done right. This resulted > > in bogus values in the IO/MEM window configuration. I re-used the code > > implementation from pci-tegra.c to do the resource retrieval correctly. > > > > This in turn resulted in very strange behavior on two different kinds > > of setup: > > > > Setup #1: ,-------> Intel i210 (igb driver) > > i.MX6 root port -----> PCIe switch > > `-------> Empty slot > > > > Setup #2: ,-------> Marvell Yukon (sky2 driver) > > i.MX6 root port -----> PCIe switch > > `-------> Empty slot > > > > Both setups expressed the same behavior. The system booted and the > > ethernet controllers were properly detected. After the controllers > > were configured and just after they reported "Ethernet Link Up", the > > entire system froze. It is now clear this happened because the resources > > were incorrectly configured, which in turn means the iATU was also not > > configured correctly. > > > > Signed-off-by: Marek Vasut <marex@denx.de> > > Cc: Bjorn Helgaas <bhelgaas@google.com> > > Cc: Frank Li <lznuaa@gmail.com> > > Cc: Jingoo Han <jg1.han@samsung.com> > > Cc: Mohit KUMAR <Mohit.KUMAR@st.com> > > Cc: Pratyush Anand <pratyush.anand@st.com> > > Cc: Richard Zhu <r65037@freescale.com> > > Cc: Sascha Hauer <s.hauer@pengutronix.de> > > Cc: Sean Cross <xobs@kosagi.com> > > Cc: Shawn Guo <shawn.guo@linaro.org> > > Cc: Siva Reddy Kallam <siva.kallam@samsung.com> > > Cc: Srikanth T Shivanand <ts.srikanth@samsung.com> > > Cc: Tim Harvey <tharvey@gateworks.com> > > Cc: Troy Kisky <troy.kisky@boundarydevices.com> > > Cc: Yinghai Lu <yinghai@kernel.org> > > --- > > drivers/pci/host/pcie-designware.c | 17 ++++++----------- > > 1 file changed, 6 insertions(+), 11 deletions(-) > > > > diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c > > index 5d183ae..97c5951 100644 > > --- a/drivers/pci/host/pcie-designware.c > > +++ b/drivers/pci/host/pcie-designware.c > > @@ -370,6 +370,7 @@ int __init dw_pcie_host_init(struct pcie_port *pp) > > struct device_node *np = pp->dev->of_node; > > struct of_pci_range range; > > struct of_pci_range_parser parser; > > + struct resource res; > > u32 val; > > > > struct irq_domain *irq_domain; > > @@ -381,28 +382,22 @@ int __init dw_pcie_host_init(struct pcie_port *pp) > > > > /* Get the I/O and memory ranges from DT */ > > for_each_of_pci_range(&parser, &range) { > > - unsigned long restype = range.flags & IORESOURCE_TYPE_BITS; > > + of_pci_range_to_resource(&range, np, &res); > > + unsigned long restype = res.flags & IORESOURCE_TYPE_BITS; This makes the following build warning. warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement] Please modify it as below: unsigned long restype; of_pci_range_to_resource(&range, np, &res); restype = res.flags & IORESOURCE_TYPE_BITS; > > if (restype == IORESOURCE_IO) { > > - of_pci_range_to_resource(&range, np, &pp->io); > > + memcpy(&pp->io, &res, sizeof(res)); > > pp->io.name = "I/O"; > > - pp->io.start = max_t(resource_size_t, > > - PCIBIOS_MIN_IO, > > - range.pci_addr + global_io_offset); > > - pp->io.end = min_t(resource_size_t, > > - IO_SPACE_LIMIT, > > - range.pci_addr + range.size > > - + global_io_offset); > > Nak - removing the adjustment to io.start/io.end breaks the io space > mapping as the entire io-space allocation gets taken by the root > complex. This causes any device that requests io space to fail to > obtain it (your igb driver from Setup1 apparently doesn't use io space > where my sky2 device from Setup2 above does). > > If I put back the adjustment to io.start/io.end above my sky2 device > gets its io resource properly but still hangs the bus after a few > interrupts. Are you sure you don't have other changes that you > haven't posted yet that could have addressed the hang you get when you > bring up your igb device? > > I agree that the the 'hang' we are seeing is a bus-hang caused by an > improper iATU mapping, however I'm not convinced its a dts parsing > issue. > > Tim +cc Kishon Vijay Abraham I, Arnd Bergmann I agree with Tim's opinion. Please, don't remove the adjustment to io.start/io.end mapping. It was already confirmed by Arnd Bergmann (ARM-SoC Maintainer). Best regards, Jingoo Han > > > pp->config.io_size = resource_size(&pp->io); > > pp->config.io_bus_addr = range.pci_addr; > > } > > if (restype == IORESOURCE_MEM) { > > - of_pci_range_to_resource(&range, np, &pp->mem); > > + memcpy(&pp->mem, &res, sizeof(res)); > > pp->mem.name = "MEM"; > > pp->config.mem_size = resource_size(&pp->mem); > > pp->config.mem_bus_addr = range.pci_addr; > > } > > if (restype == 0) { > > - of_pci_range_to_resource(&range, np, &pp->cfg); > > + memcpy(&pp->cfg, &res, sizeof(res)); > > pp->config.cfg0_size = resource_size(&pp->cfg)/2; > > pp->config.cfg1_size = resource_size(&pp->cfg)/2; > > } > > -- > > 1.8.4.rc3 > > -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Tim, > On Tue, Oct 15, 2013 at 9:06 AM, Marek Vasut <marex@denx.de> wrote: > > The resource retrieval from the DT was not done right. This resulted > > in bogus values in the IO/MEM window configuration. I re-used the code > > implementation from pci-tegra.c to do the resource retrieval correctly. > > > > This in turn resulted in very strange behavior on two different kinds > > of setup: > > > > Setup #1: ,-------> Intel i210 (igb driver) > > > > i.MX6 root port -----> PCIe switch > > > > `-------> Empty slot > > > > Setup #2: ,-------> Marvell Yukon (sky2 > > driver) > > > > i.MX6 root port -----> PCIe switch > > > > `-------> Empty slot > > > > Both setups expressed the same behavior. The system booted and the > > ethernet controllers were properly detected. After the controllers > > were configured and just after they reported "Ethernet Link Up", the > > entire system froze. It is now clear this happened because the resources > > were incorrectly configured, which in turn means the iATU was also not > > configured correctly. > > > > Signed-off-by: Marek Vasut <marex@denx.de> > > Cc: Bjorn Helgaas <bhelgaas@google.com> > > Cc: Frank Li <lznuaa@gmail.com> > > Cc: Jingoo Han <jg1.han@samsung.com> > > Cc: Mohit KUMAR <Mohit.KUMAR@st.com> > > Cc: Pratyush Anand <pratyush.anand@st.com> > > Cc: Richard Zhu <r65037@freescale.com> > > Cc: Sascha Hauer <s.hauer@pengutronix.de> > > Cc: Sean Cross <xobs@kosagi.com> > > Cc: Shawn Guo <shawn.guo@linaro.org> > > Cc: Siva Reddy Kallam <siva.kallam@samsung.com> > > Cc: Srikanth T Shivanand <ts.srikanth@samsung.com> > > Cc: Tim Harvey <tharvey@gateworks.com> > > Cc: Troy Kisky <troy.kisky@boundarydevices.com> > > Cc: Yinghai Lu <yinghai@kernel.org> > > --- > > > > drivers/pci/host/pcie-designware.c | 17 ++++++----------- > > 1 file changed, 6 insertions(+), 11 deletions(-) > > > > diff --git a/drivers/pci/host/pcie-designware.c > > b/drivers/pci/host/pcie-designware.c index 5d183ae..97c5951 100644 > > --- a/drivers/pci/host/pcie-designware.c > > +++ b/drivers/pci/host/pcie-designware.c > > @@ -370,6 +370,7 @@ int __init dw_pcie_host_init(struct pcie_port *pp) > > > > struct device_node *np = pp->dev->of_node; > > struct of_pci_range range; > > struct of_pci_range_parser parser; > > > > + struct resource res; > > > > u32 val; > > > > struct irq_domain *irq_domain; > > > > @@ -381,28 +382,22 @@ int __init dw_pcie_host_init(struct pcie_port *pp) > > > > /* Get the I/O and memory ranges from DT */ > > for_each_of_pci_range(&parser, &range) { > > > > - unsigned long restype = range.flags & > > IORESOURCE_TYPE_BITS; + of_pci_range_to_resource(&range, > > np, &res); > > + unsigned long restype = res.flags & IORESOURCE_TYPE_BITS; > > > > if (restype == IORESOURCE_IO) { > > > > - of_pci_range_to_resource(&range, np, &pp->io); > > + memcpy(&pp->io, &res, sizeof(res)); > > > > pp->io.name = "I/O"; > > > > - pp->io.start = max_t(resource_size_t, > > - PCIBIOS_MIN_IO, > > - range.pci_addr + > > global_io_offset); - pp->io.end = > > min_t(resource_size_t, > > - IO_SPACE_LIMIT, > > - range.pci_addr + range.size > > - + global_io_offset); > > Nak - removing the adjustment to io.start/io.end breaks the io space > mapping as the entire io-space allocation gets taken by the root > complex. This causes any device that requests io space to fail to > obtain it (your igb driver from Setup1 apparently doesn't use io space > where my sky2 device from Setup2 above does). OK. I still think this kind of restype comparison is bogus. The range.flags and IORESOURCE_.* do not match bit-wise. Do we agree on this part? I also agree -- as disussed with you -- that there might be bugs in the DT bindings. I will repost the V2s later tonight. --->8--- unsigned long restype = range.flags & IORESOURCE_TYPE_BITS; [...] if (restype == IORESOURCE_MEM) { ---8<--- > If I put back the adjustment to io.start/io.end above my sky2 device > gets its io resource properly but still hangs the bus after a few > interrupts. Are you sure you don't have other changes that you > haven't posted yet that could have addressed the hang you get when you > bring up your igb device? I'm absolutely positive. I double-checked this now. One more patch I have is this IRQ fixup, but that's all. pcie: correct the irq map when pcie switch is used > I agree that the the 'hang' we are seeing is a bus-hang caused by an > improper iATU mapping, however I'm not convinced its a dts parsing > issue. > > Tim > > > > - of_pci_range_to_resource(&range, np, &pp->mem); > > + memcpy(&pp->mem, &res, sizeof(res)); > > > > pp->mem.name = "MEM"; > > pp->config.mem_size = resource_size(&pp->mem); > > pp->config.mem_bus_addr = range.pci_addr; > > > > } > > if (restype == 0) { > > > > - of_pci_range_to_resource(&range, np, &pp->cfg); > > + memcpy(&pp->cfg, &res, sizeof(res)); > > > > pp->config.cfg0_size = resource_size(&pp->cfg)/2; > > pp->config.cfg1_size = resource_size(&pp->cfg)/2; > > > > } > > > > -- > > 1.8.4.rc3 -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c index 5d183ae..97c5951 100644 --- a/drivers/pci/host/pcie-designware.c +++ b/drivers/pci/host/pcie-designware.c @@ -370,6 +370,7 @@ int __init dw_pcie_host_init(struct pcie_port *pp) struct device_node *np = pp->dev->of_node; struct of_pci_range range; struct of_pci_range_parser parser; + struct resource res; u32 val; struct irq_domain *irq_domain; @@ -381,28 +382,22 @@ int __init dw_pcie_host_init(struct pcie_port *pp) /* Get the I/O and memory ranges from DT */ for_each_of_pci_range(&parser, &range) { - unsigned long restype = range.flags & IORESOURCE_TYPE_BITS; + of_pci_range_to_resource(&range, np, &res); + unsigned long restype = res.flags & IORESOURCE_TYPE_BITS; if (restype == IORESOURCE_IO) { - of_pci_range_to_resource(&range, np, &pp->io); + memcpy(&pp->io, &res, sizeof(res)); pp->io.name = "I/O"; - pp->io.start = max_t(resource_size_t, - PCIBIOS_MIN_IO, - range.pci_addr + global_io_offset); - pp->io.end = min_t(resource_size_t, - IO_SPACE_LIMIT, - range.pci_addr + range.size - + global_io_offset); pp->config.io_size = resource_size(&pp->io); pp->config.io_bus_addr = range.pci_addr; } if (restype == IORESOURCE_MEM) { - of_pci_range_to_resource(&range, np, &pp->mem); + memcpy(&pp->mem, &res, sizeof(res)); pp->mem.name = "MEM"; pp->config.mem_size = resource_size(&pp->mem); pp->config.mem_bus_addr = range.pci_addr; } if (restype == 0) { - of_pci_range_to_resource(&range, np, &pp->cfg); + memcpy(&pp->cfg, &res, sizeof(res)); pp->config.cfg0_size = resource_size(&pp->cfg)/2; pp->config.cfg1_size = resource_size(&pp->cfg)/2; }
The resource retrieval from the DT was not done right. This resulted in bogus values in the IO/MEM window configuration. I re-used the code implementation from pci-tegra.c to do the resource retrieval correctly. This in turn resulted in very strange behavior on two different kinds of setup: Setup #1: ,-------> Intel i210 (igb driver) i.MX6 root port -----> PCIe switch `-------> Empty slot Setup #2: ,-------> Marvell Yukon (sky2 driver) i.MX6 root port -----> PCIe switch `-------> Empty slot Both setups expressed the same behavior. The system booted and the ethernet controllers were properly detected. After the controllers were configured and just after they reported "Ethernet Link Up", the entire system froze. It is now clear this happened because the resources were incorrectly configured, which in turn means the iATU was also not configured correctly. Signed-off-by: Marek Vasut <marex@denx.de> Cc: Bjorn Helgaas <bhelgaas@google.com> Cc: Frank Li <lznuaa@gmail.com> Cc: Jingoo Han <jg1.han@samsung.com> Cc: Mohit KUMAR <Mohit.KUMAR@st.com> Cc: Pratyush Anand <pratyush.anand@st.com> Cc: Richard Zhu <r65037@freescale.com> Cc: Sascha Hauer <s.hauer@pengutronix.de> Cc: Sean Cross <xobs@kosagi.com> Cc: Shawn Guo <shawn.guo@linaro.org> Cc: Siva Reddy Kallam <siva.kallam@samsung.com> Cc: Srikanth T Shivanand <ts.srikanth@samsung.com> Cc: Tim Harvey <tharvey@gateworks.com> Cc: Troy Kisky <troy.kisky@boundarydevices.com> Cc: Yinghai Lu <yinghai@kernel.org> --- drivers/pci/host/pcie-designware.c | 17 ++++++----------- 1 file changed, 6 insertions(+), 11 deletions(-)