Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / arch / parisc / kernel / cache.c
diff --git a/kernel/arch/parisc/kernel/cache.c b/kernel/arch/parisc/kernel/cache.c
new file mode 100644 (file)
index 0000000..f6448c7
--- /dev/null
@@ -0,0 +1,584 @@
+/*
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ *
+ * Copyright (C) 1999-2006 Helge Deller <deller@gmx.de> (07-13-1999)
+ * Copyright (C) 1999 SuSE GmbH Nuernberg
+ * Copyright (C) 2000 Philipp Rumpf (prumpf@tux.org)
+ *
+ * Cache and TLB management
+ *
+ */
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/seq_file.h>
+#include <linux/pagemap.h>
+#include <linux/sched.h>
+#include <asm/pdc.h>
+#include <asm/cache.h>
+#include <asm/cacheflush.h>
+#include <asm/tlbflush.h>
+#include <asm/page.h>
+#include <asm/pgalloc.h>
+#include <asm/processor.h>
+#include <asm/sections.h>
+#include <asm/shmparam.h>
+
+int split_tlb __read_mostly;
+int dcache_stride __read_mostly;
+int icache_stride __read_mostly;
+EXPORT_SYMBOL(dcache_stride);
+
+void flush_dcache_page_asm(unsigned long phys_addr, unsigned long vaddr);
+EXPORT_SYMBOL(flush_dcache_page_asm);
+void flush_icache_page_asm(unsigned long phys_addr, unsigned long vaddr);
+
+
+/* On some machines (e.g. ones with the Merced bus), there can be
+ * only a single PxTLB broadcast at a time; this must be guaranteed
+ * by software.  We put a spinlock around all TLB flushes  to
+ * ensure this.
+ */
+DEFINE_SPINLOCK(pa_tlb_lock);
+
+struct pdc_cache_info cache_info __read_mostly;
+#ifndef CONFIG_PA20
+static struct pdc_btlb_info btlb_info __read_mostly;
+#endif
+
+#ifdef CONFIG_SMP
+void
+flush_data_cache(void)
+{
+       on_each_cpu(flush_data_cache_local, NULL, 1);
+}
+void 
+flush_instruction_cache(void)
+{
+       on_each_cpu(flush_instruction_cache_local, NULL, 1);
+}
+#endif
+
+void
+flush_cache_all_local(void)
+{
+       flush_instruction_cache_local(NULL);
+       flush_data_cache_local(NULL);
+}
+EXPORT_SYMBOL(flush_cache_all_local);
+
+/* Virtual address of pfn.  */
+#define pfn_va(pfn)    __va(PFN_PHYS(pfn))
+
+void
+update_mmu_cache(struct vm_area_struct *vma, unsigned long address, pte_t *ptep)
+{
+       unsigned long pfn = pte_pfn(*ptep);
+       struct page *page;
+
+       /* We don't have pte special.  As a result, we can be called with
+          an invalid pfn and we don't need to flush the kernel dcache page.
+          This occurs with FireGL card in C8000.  */
+       if (!pfn_valid(pfn))
+               return;
+
+       page = pfn_to_page(pfn);
+       if (page_mapping(page) && test_bit(PG_dcache_dirty, &page->flags)) {
+               flush_kernel_dcache_page_addr(pfn_va(pfn));
+               clear_bit(PG_dcache_dirty, &page->flags);
+       } else if (parisc_requires_coherency())
+               flush_kernel_dcache_page_addr(pfn_va(pfn));
+}
+
+void
+show_cache_info(struct seq_file *m)
+{
+       char buf[32];
+
+       seq_printf(m, "I-cache\t\t: %ld KB\n", 
+               cache_info.ic_size/1024 );
+       if (cache_info.dc_loop != 1)
+               snprintf(buf, 32, "%lu-way associative", cache_info.dc_loop);
+       seq_printf(m, "D-cache\t\t: %ld KB (%s%s, %s)\n",
+               cache_info.dc_size/1024,
+               (cache_info.dc_conf.cc_wt ? "WT":"WB"),
+               (cache_info.dc_conf.cc_sh ? ", shared I/D":""),
+               ((cache_info.dc_loop == 1) ? "direct mapped" : buf));
+       seq_printf(m, "ITLB entries\t: %ld\n" "DTLB entries\t: %ld%s\n",
+               cache_info.it_size,
+               cache_info.dt_size,
+               cache_info.dt_conf.tc_sh ? " - shared with ITLB":""
+       );
+               
+#ifndef CONFIG_PA20
+       /* BTLB - Block TLB */
+       if (btlb_info.max_size==0) {
+               seq_printf(m, "BTLB\t\t: not supported\n" );
+       } else {
+               seq_printf(m, 
+               "BTLB fixed\t: max. %d pages, pagesize=%d (%dMB)\n"
+               "BTLB fix-entr.\t: %d instruction, %d data (%d combined)\n"
+               "BTLB var-entr.\t: %d instruction, %d data (%d combined)\n",
+               btlb_info.max_size, (int)4096,
+               btlb_info.max_size>>8,
+               btlb_info.fixed_range_info.num_i,
+               btlb_info.fixed_range_info.num_d,
+               btlb_info.fixed_range_info.num_comb, 
+               btlb_info.variable_range_info.num_i,
+               btlb_info.variable_range_info.num_d,
+               btlb_info.variable_range_info.num_comb
+               );
+       }
+#endif
+}
+
+void __init 
+parisc_cache_init(void)
+{
+       if (pdc_cache_info(&cache_info) < 0)
+               panic("parisc_cache_init: pdc_cache_info failed");
+
+#if 0
+       printk("ic_size %lx dc_size %lx it_size %lx\n",
+               cache_info.ic_size,
+               cache_info.dc_size,
+               cache_info.it_size);
+
+       printk("DC  base 0x%lx stride 0x%lx count 0x%lx loop 0x%lx\n",
+               cache_info.dc_base,
+               cache_info.dc_stride,
+               cache_info.dc_count,
+               cache_info.dc_loop);
+
+       printk("dc_conf = 0x%lx  alias %d blk %d line %d shift %d\n",
+               *(unsigned long *) (&cache_info.dc_conf),
+               cache_info.dc_conf.cc_alias,
+               cache_info.dc_conf.cc_block,
+               cache_info.dc_conf.cc_line,
+               cache_info.dc_conf.cc_shift);
+       printk("        wt %d sh %d cst %d hv %d\n",
+               cache_info.dc_conf.cc_wt,
+               cache_info.dc_conf.cc_sh,
+               cache_info.dc_conf.cc_cst,
+               cache_info.dc_conf.cc_hv);
+
+       printk("IC  base 0x%lx stride 0x%lx count 0x%lx loop 0x%lx\n",
+               cache_info.ic_base,
+               cache_info.ic_stride,
+               cache_info.ic_count,
+               cache_info.ic_loop);
+
+       printk("ic_conf = 0x%lx  alias %d blk %d line %d shift %d\n",
+               *(unsigned long *) (&cache_info.ic_conf),
+               cache_info.ic_conf.cc_alias,
+               cache_info.ic_conf.cc_block,
+               cache_info.ic_conf.cc_line,
+               cache_info.ic_conf.cc_shift);
+       printk("        wt %d sh %d cst %d hv %d\n",
+               cache_info.ic_conf.cc_wt,
+               cache_info.ic_conf.cc_sh,
+               cache_info.ic_conf.cc_cst,
+               cache_info.ic_conf.cc_hv);
+
+       printk("D-TLB conf: sh %d page %d cst %d aid %d pad1 %d\n",
+               cache_info.dt_conf.tc_sh,
+               cache_info.dt_conf.tc_page,
+               cache_info.dt_conf.tc_cst,
+               cache_info.dt_conf.tc_aid,
+               cache_info.dt_conf.tc_pad1);
+
+       printk("I-TLB conf: sh %d page %d cst %d aid %d pad1 %d\n",
+               cache_info.it_conf.tc_sh,
+               cache_info.it_conf.tc_page,
+               cache_info.it_conf.tc_cst,
+               cache_info.it_conf.tc_aid,
+               cache_info.it_conf.tc_pad1);
+#endif
+
+       split_tlb = 0;
+       if (cache_info.dt_conf.tc_sh == 0 || cache_info.dt_conf.tc_sh == 2) {
+               if (cache_info.dt_conf.tc_sh == 2)
+                       printk(KERN_WARNING "Unexpected TLB configuration. "
+                       "Will flush I/D separately (could be optimized).\n");
+
+               split_tlb = 1;
+       }
+
+       /* "New and Improved" version from Jim Hull 
+        *      (1 << (cc_block-1)) * (cc_line << (4 + cnf.cc_shift))
+        * The following CAFL_STRIDE is an optimized version, see
+        * http://lists.parisc-linux.org/pipermail/parisc-linux/2004-June/023625.html
+        * http://lists.parisc-linux.org/pipermail/parisc-linux/2004-June/023671.html
+        */
+#define CAFL_STRIDE(cnf) (cnf.cc_line << (3 + cnf.cc_block + cnf.cc_shift))
+       dcache_stride = CAFL_STRIDE(cache_info.dc_conf);
+       icache_stride = CAFL_STRIDE(cache_info.ic_conf);
+#undef CAFL_STRIDE
+
+#ifndef CONFIG_PA20
+       if (pdc_btlb_info(&btlb_info) < 0) {
+               memset(&btlb_info, 0, sizeof btlb_info);
+       }
+#endif
+
+       if ((boot_cpu_data.pdc.capabilities & PDC_MODEL_NVA_MASK) ==
+                                               PDC_MODEL_NVA_UNSUPPORTED) {
+               printk(KERN_WARNING "parisc_cache_init: Only equivalent aliasing supported!\n");
+#if 0
+               panic("SMP kernel required to avoid non-equivalent aliasing");
+#endif
+       }
+}
+
+void disable_sr_hashing(void)
+{
+       int srhash_type, retval;
+       unsigned long space_bits;
+
+       switch (boot_cpu_data.cpu_type) {
+       case pcx: /* We shouldn't get this far.  setup.c should prevent it. */
+               BUG();
+               return;
+
+       case pcxs:
+       case pcxt:
+       case pcxt_:
+               srhash_type = SRHASH_PCXST;
+               break;
+
+       case pcxl:
+               srhash_type = SRHASH_PCXL;
+               break;
+
+       case pcxl2: /* pcxl2 doesn't support space register hashing */
+               return;
+
+       default: /* Currently all PA2.0 machines use the same ins. sequence */
+               srhash_type = SRHASH_PA20;
+               break;
+       }
+
+       disable_sr_hashing_asm(srhash_type);
+
+       retval = pdc_spaceid_bits(&space_bits);
+       /* If this procedure isn't implemented, don't panic. */
+       if (retval < 0 && retval != PDC_BAD_OPTION)
+               panic("pdc_spaceid_bits call failed.\n");
+       if (space_bits != 0)
+               panic("SpaceID hashing is still on!\n");
+}
+
+static inline void
+__flush_cache_page(struct vm_area_struct *vma, unsigned long vmaddr,
+                  unsigned long physaddr)
+{
+       preempt_disable();
+       flush_dcache_page_asm(physaddr, vmaddr);
+       if (vma->vm_flags & VM_EXEC)
+               flush_icache_page_asm(physaddr, vmaddr);
+       preempt_enable();
+}
+
+void flush_dcache_page(struct page *page)
+{
+       struct address_space *mapping = page_mapping(page);
+       struct vm_area_struct *mpnt;
+       unsigned long offset;
+       unsigned long addr, old_addr = 0;
+       pgoff_t pgoff;
+
+       if (mapping && !mapping_mapped(mapping)) {
+               set_bit(PG_dcache_dirty, &page->flags);
+               return;
+       }
+
+       flush_kernel_dcache_page(page);
+
+       if (!mapping)
+               return;
+
+       pgoff = page->index << (PAGE_CACHE_SHIFT - PAGE_SHIFT);
+
+       /* We have carefully arranged in arch_get_unmapped_area() that
+        * *any* mappings of a file are always congruently mapped (whether
+        * declared as MAP_PRIVATE or MAP_SHARED), so we only need
+        * to flush one address here for them all to become coherent */
+
+       flush_dcache_mmap_lock(mapping);
+       vma_interval_tree_foreach(mpnt, &mapping->i_mmap, pgoff, pgoff) {
+               offset = (pgoff - mpnt->vm_pgoff) << PAGE_SHIFT;
+               addr = mpnt->vm_start + offset;
+
+               /* The TLB is the engine of coherence on parisc: The
+                * CPU is entitled to speculate any page with a TLB
+                * mapping, so here we kill the mapping then flush the
+                * page along a special flush only alias mapping.
+                * This guarantees that the page is no-longer in the
+                * cache for any process and nor may it be
+                * speculatively read in (until the user or kernel
+                * specifically accesses it, of course) */
+
+               flush_tlb_page(mpnt, addr);
+               if (old_addr == 0 || (old_addr & (SHM_COLOUR - 1))
+                                     != (addr & (SHM_COLOUR - 1))) {
+                       __flush_cache_page(mpnt, addr, page_to_phys(page));
+                       if (old_addr)
+                               printk(KERN_ERR "INEQUIVALENT ALIASES 0x%lx and 0x%lx in file %s\n", old_addr, addr, mpnt->vm_file ? (char *)mpnt->vm_file->f_path.dentry->d_name.name : "(null)");
+                       old_addr = addr;
+               }
+       }
+       flush_dcache_mmap_unlock(mapping);
+}
+EXPORT_SYMBOL(flush_dcache_page);
+
+/* Defined in arch/parisc/kernel/pacache.S */
+EXPORT_SYMBOL(flush_kernel_dcache_range_asm);
+EXPORT_SYMBOL(flush_kernel_dcache_page_asm);
+EXPORT_SYMBOL(flush_data_cache_local);
+EXPORT_SYMBOL(flush_kernel_icache_range_asm);
+
+#define FLUSH_THRESHOLD 0x80000 /* 0.5MB */
+int parisc_cache_flush_threshold __read_mostly = FLUSH_THRESHOLD;
+
+void __init parisc_setup_cache_timing(void)
+{
+       unsigned long rangetime, alltime;
+       unsigned long size;
+
+       alltime = mfctl(16);
+       flush_data_cache();
+       alltime = mfctl(16) - alltime;
+
+       size = (unsigned long)(_end - _text);
+       rangetime = mfctl(16);
+       flush_kernel_dcache_range((unsigned long)_text, size);
+       rangetime = mfctl(16) - rangetime;
+
+       printk(KERN_DEBUG "Whole cache flush %lu cycles, flushing %lu bytes %lu cycles\n",
+               alltime, size, rangetime);
+
+       /* Racy, but if we see an intermediate value, it's ok too... */
+       parisc_cache_flush_threshold = size * alltime / rangetime;
+
+       parisc_cache_flush_threshold = (parisc_cache_flush_threshold + L1_CACHE_BYTES - 1) &~ (L1_CACHE_BYTES - 1); 
+       if (!parisc_cache_flush_threshold)
+               parisc_cache_flush_threshold = FLUSH_THRESHOLD;
+
+       if (parisc_cache_flush_threshold > cache_info.dc_size)
+               parisc_cache_flush_threshold = cache_info.dc_size;
+
+       printk(KERN_INFO "Setting cache flush threshold to %x (%d CPUs online)\n", parisc_cache_flush_threshold, num_online_cpus());
+}
+
+extern void purge_kernel_dcache_page_asm(unsigned long);
+extern void clear_user_page_asm(void *, unsigned long);
+extern void copy_user_page_asm(void *, void *, unsigned long);
+
+void flush_kernel_dcache_page_addr(void *addr)
+{
+       unsigned long flags;
+
+       flush_kernel_dcache_page_asm(addr);
+       purge_tlb_start(flags);
+       pdtlb_kernel(addr);
+       purge_tlb_end(flags);
+}
+EXPORT_SYMBOL(flush_kernel_dcache_page_addr);
+
+void copy_user_page(void *vto, void *vfrom, unsigned long vaddr,
+       struct page *pg)
+{
+       /* Copy using kernel mapping.  No coherency is needed (all in
+         kunmap) for the `to' page.  However, the `from' page needs to
+         be flushed through a mapping equivalent to the user mapping
+         before it can be accessed through the kernel mapping. */
+       preempt_disable();
+       flush_dcache_page_asm(__pa(vfrom), vaddr);
+       preempt_enable();
+       copy_page_asm(vto, vfrom);
+}
+EXPORT_SYMBOL(copy_user_page);
+
+void purge_tlb_entries(struct mm_struct *mm, unsigned long addr)
+{
+       unsigned long flags;
+
+       /* Note: purge_tlb_entries can be called at startup with
+          no context.  */
+
+       purge_tlb_start(flags);
+       mtsp(mm->context, 1);
+       pdtlb(addr);
+       pitlb(addr);
+       purge_tlb_end(flags);
+}
+EXPORT_SYMBOL(purge_tlb_entries);
+
+void __flush_tlb_range(unsigned long sid, unsigned long start,
+                      unsigned long end)
+{
+       unsigned long npages;
+
+       npages = ((end - (start & PAGE_MASK)) + (PAGE_SIZE - 1)) >> PAGE_SHIFT;
+       if (npages >= 512)  /* 2MB of space: arbitrary, should be tuned */
+               flush_tlb_all();
+       else {
+               unsigned long flags;
+
+               purge_tlb_start(flags);
+               mtsp(sid, 1);
+               if (split_tlb) {
+                       while (npages--) {
+                               pdtlb(start);
+                               pitlb(start);
+                               start += PAGE_SIZE;
+                       }
+               } else {
+                       while (npages--) {
+                               pdtlb(start);
+                               start += PAGE_SIZE;
+                       }
+               }
+               purge_tlb_end(flags);
+       }
+}
+
+static void cacheflush_h_tmp_function(void *dummy)
+{
+       flush_cache_all_local();
+}
+
+void flush_cache_all(void)
+{
+       on_each_cpu(cacheflush_h_tmp_function, NULL, 1);
+}
+
+static inline unsigned long mm_total_size(struct mm_struct *mm)
+{
+       struct vm_area_struct *vma;
+       unsigned long usize = 0;
+
+       for (vma = mm->mmap; vma; vma = vma->vm_next)
+               usize += vma->vm_end - vma->vm_start;
+       return usize;
+}
+
+static inline pte_t *get_ptep(pgd_t *pgd, unsigned long addr)
+{
+       pte_t *ptep = NULL;
+
+       if (!pgd_none(*pgd)) {
+               pud_t *pud = pud_offset(pgd, addr);
+               if (!pud_none(*pud)) {
+                       pmd_t *pmd = pmd_offset(pud, addr);
+                       if (!pmd_none(*pmd))
+                               ptep = pte_offset_map(pmd, addr);
+               }
+       }
+       return ptep;
+}
+
+void flush_cache_mm(struct mm_struct *mm)
+{
+       struct vm_area_struct *vma;
+       pgd_t *pgd;
+
+       /* Flushing the whole cache on each cpu takes forever on
+          rp3440, etc.  So, avoid it if the mm isn't too big.  */
+       if (mm_total_size(mm) >= parisc_cache_flush_threshold) {
+               flush_cache_all();
+               return;
+       }
+
+       if (mm->context == mfsp(3)) {
+               for (vma = mm->mmap; vma; vma = vma->vm_next) {
+                       flush_user_dcache_range_asm(vma->vm_start, vma->vm_end);
+                       if ((vma->vm_flags & VM_EXEC) == 0)
+                               continue;
+                       flush_user_icache_range_asm(vma->vm_start, vma->vm_end);
+               }
+               return;
+       }
+
+       pgd = mm->pgd;
+       for (vma = mm->mmap; vma; vma = vma->vm_next) {
+               unsigned long addr;
+
+               for (addr = vma->vm_start; addr < vma->vm_end;
+                    addr += PAGE_SIZE) {
+                       unsigned long pfn;
+                       pte_t *ptep = get_ptep(pgd, addr);
+                       if (!ptep)
+                               continue;
+                       pfn = pte_pfn(*ptep);
+                       if (!pfn_valid(pfn))
+                               continue;
+                       __flush_cache_page(vma, addr, PFN_PHYS(pfn));
+               }
+       }
+}
+
+void
+flush_user_dcache_range(unsigned long start, unsigned long end)
+{
+       if ((end - start) < parisc_cache_flush_threshold)
+               flush_user_dcache_range_asm(start,end);
+       else
+               flush_data_cache();
+}
+
+void
+flush_user_icache_range(unsigned long start, unsigned long end)
+{
+       if ((end - start) < parisc_cache_flush_threshold)
+               flush_user_icache_range_asm(start,end);
+       else
+               flush_instruction_cache();
+}
+
+void flush_cache_range(struct vm_area_struct *vma,
+               unsigned long start, unsigned long end)
+{
+       unsigned long addr;
+       pgd_t *pgd;
+
+       BUG_ON(!vma->vm_mm->context);
+
+       if ((end - start) >= parisc_cache_flush_threshold) {
+               flush_cache_all();
+               return;
+       }
+
+       if (vma->vm_mm->context == mfsp(3)) {
+               flush_user_dcache_range_asm(start, end);
+               if (vma->vm_flags & VM_EXEC)
+                       flush_user_icache_range_asm(start, end);
+               return;
+       }
+
+       pgd = vma->vm_mm->pgd;
+       for (addr = start & PAGE_MASK; addr < end; addr += PAGE_SIZE) {
+               unsigned long pfn;
+               pte_t *ptep = get_ptep(pgd, addr);
+               if (!ptep)
+                       continue;
+               pfn = pte_pfn(*ptep);
+               if (pfn_valid(pfn))
+                       __flush_cache_page(vma, addr, PFN_PHYS(pfn));
+       }
+}
+
+void
+flush_cache_page(struct vm_area_struct *vma, unsigned long vmaddr, unsigned long pfn)
+{
+       BUG_ON(!vma->vm_mm->context);
+
+       if (pfn_valid(pfn)) {
+               flush_tlb_page(vma, vmaddr);
+               __flush_cache_page(vma, vmaddr, PFN_PHYS(pfn));
+       }
+}