[PATCH] ppc64 kexec: native hash clear

Add code to clear the hash table and invalidate the tlb for native (SMP,
non-LPAR) mode.  Supports 16M and 4k pages.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: R Sharada <sharada@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

2 files changed, 68 insertions, 1 deletions

diff --git a/arch/ppc64/mm/hash_native.c b/arch/ppc64/mm/hash_native.c
index 52b6b930534..4fec05817d6 100644
--- a/arch/ppc64/mm/hash_native.c
+++ b/arch/ppc64/mm/hash_native.c
@@ -304,6 +304,50 @@ static void native_hpte_invalidate(unsigned long slot, unsigned long va,
         local_irq_restore(flags);
 }
 
+/*
+ * clear all mappings on kexec.  All cpus are in real mode (or they will
+ * be when they isi), and we are the only one left.  We rely on our kernel
+ * mapping being 0xC0's and the hardware ignoring those two real bits.
+ *
+ * TODO: add batching support when enabled.  remember, no dynamic memory here,
+ * athough there is the control page available...
+ */
+static void native_hpte_clear(void)
+{
+        unsigned long slot, slots, flags;
+        HPTE *hptep = htab_address;
+        Hpte_dword0 dw0;
+        unsigned long pteg_count;
+
+        pteg_count = htab_hash_mask + 1;
+
+        local_irq_save(flags);
+
+        /* we take the tlbie lock and hold it.  Some hardware will
+         * deadlock if we try to tlbie from two processors at once.
+         */
+        spin_lock(&native_tlbie_lock);
+
+        slots = pteg_count * HPTES_PER_GROUP;
+
+        for (slot = 0; slot < slots; slot++, hptep++) {
+                /*
+                 * we could lock the pte here, but we are the only cpu
+                 * running,  right?  and for crash dump, we probably
+                 * don't want to wait for a maybe bad cpu.
+                 */
+                dw0 = hptep->dw0.dw0;
+
+                if (dw0.v) {
+                        hptep->dw0.dword0 = 0;
+                        tlbie(slot2va(dw0.avpn, dw0.l, dw0.h, slot), dw0.l);
+                }
+        }
+
+        spin_unlock(&native_tlbie_lock);
+        local_irq_restore(flags);
+}
+
 static void native_flush_hash_range(unsigned long context,
                                     unsigned long number, int local)
 {
@@ -415,7 +459,8 @@ void hpte_init_native(void)
         ppc_md.hpte_updatepp    = native_hpte_updatepp;
         ppc_md.hpte_updateboltedpp = native_hpte_updateboltedpp;
         ppc_md.hpte_insert      = native_hpte_insert;
-        ppc_md.hpte_remove      = native_hpte_remove;
+        ppc_md.hpte_remove      = native_hpte_remove;
+        ppc_md.hpte_clear_all   = native_hpte_clear;
         if (tlb_batching_enabled())
                 ppc_md.flush_hash_range = native_flush_hash_range;
         htab_finish_init();
diff --git a/include/asm-ppc64/mmu.h b/include/asm-ppc64/mmu.h
index 9d03a98a4fa..f373de5e3dd 100644
--- a/include/asm-ppc64/mmu.h
+++ b/include/asm-ppc64/mmu.h
@@ -181,6 +181,28 @@ static inline void tlbiel(unsigned long va)
         asm volatile("ptesync": : :"memory");
 }
 
+static inline unsigned long slot2va(unsigned long avpn, unsigned long large,
+                unsigned long secondary, unsigned long slot)
+{
+        unsigned long va;
+
+        va = avpn << 23;
+
+        if (!large) {
+                unsigned long vpi, pteg;
+
+                pteg = slot / HPTES_PER_GROUP;
+                if (secondary)
+                        pteg = ~pteg;
+
+                vpi = ((va >> 28) ^ pteg) & htab_hash_mask;
+
+                va |= vpi << PAGE_SHIFT;
+        }
+
+        return va;
+}
+
 /*
  * Handle a fault by adding an HPTE. If the address can't be determined
  * to be valid via Linux page tables, return 1. If handled return 0