| Message ID | 20200205190726.15753-1-xypron.glpk@gmx.de |
|---|---|
| State | New |
| Headers | show |
| Series | [1/1] pci: definition of pci_addr_t and pci_size_t | expand |
Hi Heinrich, On Wed, 5 Feb 2020 at 12:07, Heinrich Schuchardt <xypron.glpk at gmx.de> wrote: > > Currently the size of pci_addr_t and pci_size_t depends on > CONFIG_SYS_PCI_64BIT. For qemu_arm64_defconfig with 4 GiB RAM this leads > to an error > > pci_hose_phys_to_bus: invalid physical address > > which is due to the truncation of the bus address in _dm_pci_phys_to_bus. > > Defining CONFIG_SYS_PCI_64BIT is not a solution as this results in an error > > PCI: Failed autoconfig bar 10 > > So let's use unsigned long for pci_addr_t and pci_size_t. But how will this work on x86 where we might have 32-bit U-Boot but need 64-bit PCI addresses? Regards, Simon
On 2/5/20 9:05 PM, Simon Glass wrote: > Hi Heinrich, > > On Wed, 5 Feb 2020 at 12:07, Heinrich Schuchardt <xypron.glpk at gmx.de> wrote: >> >> Currently the size of pci_addr_t and pci_size_t depends on >> CONFIG_SYS_PCI_64BIT. For qemu_arm64_defconfig with 4 GiB RAM this leads >> to an error >> >> pci_hose_phys_to_bus: invalid physical address >> >> which is due to the truncation of the bus address in _dm_pci_phys_to_bus. >> >> Defining CONFIG_SYS_PCI_64BIT is not a solution as this results in an error >> >> PCI: Failed autoconfig bar 10 >> >> So let's use unsigned long for pci_addr_t and pci_size_t. > > But how will this work on x86 where we might have 32-bit U-Boot but > need 64-bit PCI addresses? > > Regards, > Simon > So would you suggest to only change the code for CONFIG_SYS_PCI_64BIT=n? Best regards Heinrich
Hi Heinrich, On Wed, 5 Feb 2020 at 13:18, Heinrich Schuchardt <xypron.glpk at gmx.de> wrote: > > > > On 2/5/20 9:05 PM, Simon Glass wrote: > > Hi Heinrich, > > > > On Wed, 5 Feb 2020 at 12:07, Heinrich Schuchardt <xypron.glpk at gmx.de> wrote: > >> > >> Currently the size of pci_addr_t and pci_size_t depends on > >> CONFIG_SYS_PCI_64BIT. For qemu_arm64_defconfig with 4 GiB RAM this leads > >> to an error > >> > >> pci_hose_phys_to_bus: invalid physical address > >> > >> which is due to the truncation of the bus address in _dm_pci_phys_to_bus. > >> > >> Defining CONFIG_SYS_PCI_64BIT is not a solution as this results in an error > >> > >> PCI: Failed autoconfig bar 10 > >> > >> So let's use unsigned long for pci_addr_t and pci_size_t. > > > > But how will this work on x86 where we might have 32-bit U-Boot but > > need 64-bit PCI addresses? > > > > Regards, > > Simon > > > > So would you suggest to only change the code for CONFIG_SYS_PCI_64BIT=n? I think that is safe, since nothing has a ulong less than 32 bits. Regards, Simon
diff --git a/include/pci.h b/include/pci.h index 8c761d8da3..d8bce8ab80 100644 --- a/include/pci.h +++ b/include/pci.h @@ -484,13 +484,8 @@ #include <dm/pci.h> -#ifdef CONFIG_SYS_PCI_64BIT -typedef u64 pci_addr_t; -typedef u64 pci_size_t; -#else -typedef u32 pci_addr_t; -typedef u32 pci_size_t; -#endif +typedef unsigned long pci_addr_t; +typedef unsigned long pci_size_t; struct pci_region { pci_addr_t bus_start; /* Start on the bus */
Currently the size of pci_addr_t and pci_size_t depends on CONFIG_SYS_PCI_64BIT. For qemu_arm64_defconfig with 4 GiB RAM this leads to an error pci_hose_phys_to_bus: invalid physical address which is due to the truncation of the bus address in _dm_pci_phys_to_bus. Defining CONFIG_SYS_PCI_64BIT is not a solution as this results in an error PCI: Failed autoconfig bar 10 So let's use unsigned long for pci_addr_t and pci_size_t. Signed-off-by: Heinrich Schuchardt <xypron.glpk at gmx.de> --- include/pci.h | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-) -- 2.24.1