Use pci_generic_prep_mwi on ia64
PATCH 1/1 Use pci_generic_prep_mwi on ia64
The pci_generic_prep_mwi() code does everything that pcibios_prep_mwi() does on ia64. All we need to do is be sure that pci_cache_line_size is set appropriately, and we can delete pcibios_prep_mwi().
Using SMP_CACHE_BYTES as the default was wrong on uniprocessor machines as it is only 8 bytes. The default in the generic code of L1_CACHE_BYTES is at least as good.
Signed-off-by Matthew Wilcox <[email protected]>
arch/ia64/pci/pci.c +20 -51
@@ -738,75 +738,44 @@ int ia64_pci_legacy_write(struct pci_busreturn ret; } +/* It's defined in drivers/pci/pci.c */ +extern u8 pci_cache_line_size; +/** -* pci_cacheline_size - determine cacheline size for PCI devices -* @dev: void +* set_pci_cacheline_size - determine cacheline size for PCI devices * * We want to use the line-size of the outer-most cache. We assume * that this line-size is the same for all CPUs. * * Code mostly taken from arch/ia64/kernel/palinfo.c:cache_info(). -* -* RETURNS: An appropriate -ERRNO error value on eror, or zero for success. */ -static unsigned long -pci_cacheline_size (void ) +static void __init set_pci_cacheline_size (void ) {u64 levels, unique_caches;s64 status;pal_cache_config_info_t cci; -static u8 cacheline_size; - -if (cacheline_size) -return cacheline_size; status =ia64_pal_cache_summary (&levels, &unique_caches);if (status !=0 ) { -printk (KERN_ERR "%s: ia64_pal_cache_summary() failed (status=%ld)\n" , - __FUNCTION__, status); -return SMP_CACHE_BYTES ; +printk (KERN_ERR "%s: ia64_pal_cache_summary() failed " +"(status=%ld)\n" , __FUNCTION__, status); +return ; } - status =ia64_pal_cache_config_info (levels -1 ,/* cache_type (data_or_unified)= */ 2 , - &cci); + status =ia64_pal_cache_config_info (levels -1 , +/* cache_type (data_or_unified)= */ 2 , &cci);if (status !=0 ) { -printk (KERN_ERR "%s: ia64_pal_cache_config_info() failed (status=%ld)\n" , - __FUNCTION__, status); -return SMP_CACHE_BYTES ; +printk (KERN_ERR "%s: ia64_pal_cache_config_info() failed " +"(status=%ld)\n" , __FUNCTION__, status); +return ; } - cacheline_size =1 << cci.pcci_line_size; -return cacheline_size; + pci_cache_line_size = (1 << cci.pcci_line_size) /4 ; } -/** -* pcibios_prep_mwi - helper function for drivers/pci/pci.c:pci_set_mwi() -* @dev: the PCI device for which MWI is enabled -* -* For ia64, we can get the cacheline sizes from PAL. -* -* RETURNS: An appropriate -ERRNO error value on eror, or zero for success. -*/ -int -pcibios_prep_mwi (struct pci_dev *dev) -{ -unsigned long desired_linesize, current_linesize; -int rc =0 ; -u8 pci_linesize; - - desired_linesize =pci_cacheline_size (); - -pci_read_config_byte (dev,PCI_CACHE_LINE_SIZE , &pci_linesize); - current_linesize =4 * pci_linesize; -if (desired_linesize != current_linesize) { -printk (KERN_WARNING "PCI: slot %s has incorrect PCI cache line size of %lu bytes," , -pci_name (dev), current_linesize); -if (current_linesize > desired_linesize) { -printk (" expected %lu bytes instead\n" , desired_linesize); - rc = -EINVAL ; - }else { -printk (" correcting to %lu\n" , desired_linesize); -pci_write_config_byte (dev,PCI_CACHE_LINE_SIZE , desired_linesize /4 ); - } - } -return rc; +static int __init pcibios_init (void ) +{ +set_pci_cacheline_size (); +return 0 ; } + +subsys_initcall (pcibios_init);
include/asm-ia64/pci.h +10 -11
@@ -26,16 +26,18 @@ void pcibios_config_init(void);struct pci_dev ;/* -* PCI_DMA_BUS_IS_PHYS should be set to 1 if there is _necessarily_ a direct correspondence -* between device bus addresses and CPU physical addresses. Platforms with a hardware I/O -* MMU _must_ turn this off to suppress the bounce buffer handling code in the block and -* network device layers. Platforms with separate bus address spaces _must_ turn this off -* and provide a device DMA mapping implementation that takes care of the necessary +* PCI_DMA_BUS_IS_PHYS should be set to 1 if there is _necessarily_ a direct +* correspondence between device bus addresses and CPU physical addresses. +* Platforms with a hardware I/O MMU _must_ turn this off to suppress the +* bounce buffer handling code in the block and network device layers. +* Platforms with separate bus address spaces _must_ turn this off and provide +* a device DMA mapping implementation that takes care of the necessary * address translation. * -* For now, the ia64 platforms which may have separate/multiple bus address spaces all -* have I/O MMUs which support the merging of physically discontiguous buffers, so we can -* use that as the sole factor to determine the setting of PCI_DMA_BUS_IS_PHYS. +* For now, the ia64 platforms which may have separate/multiple bus address +* spaces all have I/O MMUs which support the merging of physically +* discontiguous buffers, so we can use that as the sole factor to determine +* the setting of PCI_DMA_BUS_IS_PHYS. */ extern unsigned long ia64_max_iommu_merge_mask;#define PCI_DMA_BUS_IS_PHYS (ia64_max_iommu_merge_mask = ~0UL)
@@ -52,9 +54,6 @@ pcibios_penalize_isa_irq (int irq, int a/* We don't do dynamic PCI IRQ allocation */ } -#define HAVE_ARCH_PCI_MWI 1 -extern int pcibios_prep_mwi (struct pci_dev *); -#include <asm-generic/pci-dma-compat.h> /* pci_unmap_{single,page} is not a nop, thus... */