gru: support instruction completion interrupts

Add support for interrupts generated by GRU instruction completion.
Previously, the only interrupts were for TLB misses.  The hardware also
supports interrupts on instruction completion.  This will be supported for
instructions issued by the kernel.

Signed-off-by: Jack Steiner <steiner@sgi.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

3 files changed, 29 insertions, 7 deletions

diff --git a/drivers/misc/sgi-gru/grufault.c b/drivers/misc/sgi-gru/grufault.c
index f15152165a9..3220e95be6b 100644
--- a/drivers/misc/sgi-gru/grufault.c
+++ b/drivers/misc/sgi-gru/grufault.c
@@ -166,7 +166,8 @@ static inline struct gru_state *irq_to_gru(int irq)
  * the GRU, atomic operations must be used to clear bits.
  */
 static void get_clear_fault_map(struct gru_state *gru,
-                                struct gru_tlb_fault_map *map)
+                                struct gru_tlb_fault_map *imap,
+                                struct gru_tlb_fault_map *dmap)
 {
         unsigned long i, k;
         struct gru_tlb_fault_map *tfm;
@@ -177,7 +178,11 @@ static void get_clear_fault_map(struct gru_state *gru,
                 k = tfm->fault_bits[i];
                 if (k)
                         k = xchg(&tfm->fault_bits[i], 0UL);
-                map->fault_bits[i] = k;
+                imap->fault_bits[i] = k;
+                k = tfm->done_bits[i];
+                if (k)
+                        k = xchg(&tfm->done_bits[i], 0UL);
+                dmap->fault_bits[i] = k;
         }
 
         /*
@@ -449,7 +454,7 @@ failactive:
 irqreturn_t gru_intr(int irq, void *dev_id)
 {
         struct gru_state *gru;
-        struct gru_tlb_fault_map map;
+        struct gru_tlb_fault_map imap, dmap;
         struct gru_thread_state *gts;
         struct gru_tlb_fault_handle *tfh = NULL;
         int cbrnum, ctxnum;
@@ -462,11 +467,19 @@ irqreturn_t gru_intr(int irq, void *dev_id)
                         raw_smp_processor_id(), irq);
                 return IRQ_NONE;
         }
-        get_clear_fault_map(gru, &map);
-        gru_dbg(grudev, "irq %d, gru %x, map 0x%lx\n", irq, gru->gs_gid,
-                map.fault_bits[0]);
+        get_clear_fault_map(gru, &imap, &dmap);
+        gru_dbg(grudev,
+                "irq %d, gid %d, imap %016lx %016lx, dmap %016lx %016lx\n",
+                irq, gru->gs_gid, dmap.fault_bits[0], dmap.fault_bits[1],
+                dmap.fault_bits[0], dmap.fault_bits[1]);
+
+        for_each_cbr_in_tfm(cbrnum, dmap.fault_bits) {
+                complete(gru->gs_blade->bs_async_wq);
+                gru_dbg(grudev, "gid %d, cbr_done %d, done %d\n",
+                        gru->gs_gid, cbrnum, gru->gs_blade->bs_async_wq->done);
+        }
 
-        for_each_cbr_in_tfm(cbrnum, map.fault_bits) {
+        for_each_cbr_in_tfm(cbrnum, imap.fault_bits) {
                 tfh = get_tfh_by_index(gru, cbrnum);
                 prefetchw(tfh); /* Helps on hdw, required for emulator */
 
diff --git a/drivers/misc/sgi-gru/grumain.c b/drivers/misc/sgi-gru/grumain.c
index afc4c473c79..e38a0f1775f 100644
--- a/drivers/misc/sgi-gru/grumain.c
+++ b/drivers/misc/sgi-gru/grumain.c
@@ -572,8 +572,12 @@ void gru_load_context(struct gru_thread_state *gts)
 
         if (is_kernel_context(gts)) {
                 cch->unmap_enable = 1;
+                cch->tfm_done_bit_enable = 1;
+                cch->cb_int_enable = 1;
         } else {
                 cch->unmap_enable = 0;
+                cch->tfm_done_bit_enable = 0;
+                cch->cb_int_enable = 0;
                 asid = gru_load_mm_tracker(gru, gts);
                 for (i = 0; i < 8; i++) {
                         cch->asid[i] = asid + i;
diff --git a/drivers/misc/sgi-gru/grutables.h b/drivers/misc/sgi-gru/grutables.h
index 5f8f3bda2fa..ca81800146f 100644
--- a/drivers/misc/sgi-gru/grutables.h
+++ b/drivers/misc/sgi-gru/grutables.h
@@ -462,6 +462,11 @@ struct gru_blade_state {
         struct rw_semaphore     bs_kgts_sema;           /* lock for kgts */
         struct gru_thread_state *bs_kgts;               /* GTS for kernel use */
 
+        /* ---- the following are used for managing kernel async GRU CBRs --- */
+        int                     bs_async_dsr_bytes;     /* DSRs for async */
+        int                     bs_async_cbrs;          /* CBRs AU for async */
+        struct completion       *bs_async_wq;
+
         /* ---- the following are protected by the bs_lock spinlock ---- */
         spinlock_t              bs_lock;                /* lock used for
                                                            stealing contexts */