nv50/ir/util: fix BitSet issues

BitSet::allocate() is being used with the expectation that it would
leave the bitfield untouched if its size hasn't changed, however,
the function always zeroed the last word, which led to obscure bugs
with live set computation.

This also fixes BitSet::resize(), which was broken, but luckily not
being used.

Cc: "10.2 10.3" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp
index 5ab6570..4b105b4 100644 (file)
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp
@@ -1657,6 +1657,10 @@ RegAlloc::execFunc()
            ret && i <= func->loopNestingBound;
            sequence = func->cfg.nextSequence(), ++i)
          ret = buildLiveSets(BasicBlock::get(func->cfg.getRoot()));
+      // reset marker
+      for (ArrayList::Iterator bi = func->allBBlocks.iterator();
+           !bi.end(); bi.next())
+         BasicBlock::get(bi)->liveSet.marker = false;
       if (!ret)
          break;
       func->orderInstructions(this->insns);

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_util.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_util.cpp
index 8959777..d26acb3 100644 (file)
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_util.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_util.cpp
@@ -254,7 +254,9 @@ bool BitSet::resize(unsigned int nBits)
       return false;
    }
    if (n > p)
-      memset(&data[4 * p + 4], 0, (n - p) * 4);
+      memset(&data[p], 0, (n - p) * 4);
+   if (nBits < size && (nBits % 32))
+      data[(nBits + 31) / 32 - 1] &= (1 << (nBits % 32)) - 1;
 
    size = nBits;
    return true;
@@ -274,8 +276,8 @@ bool BitSet::allocate(unsigned int nBits, bool zero)
    if (zero)
       memset(data, 0, (size + 7) / 8);
    else
-   if (nBits)
-      data[(size + 31) / 32 - 1] = 0; // clear unused bits (e.g. for popCount)
+   if (size % 32) // clear unused bits (e.g. for popCount)
+      data[(size + 31) / 32 - 1] &= (1 << (size % 32)) - 1;
 
    return data;
 }

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_util.h b/src/gallium/drivers/nouveau/codegen/nv50_ir_util.h
index a4ea9d9..fa2c480 100644 (file)
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_util.h
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_util.h
@@ -484,6 +484,7 @@ public:
          FREE(data);
    }
 
+   // allocate will keep old data iff size is unchanged
    bool allocate(unsigned int nBits, bool zero);
    bool resize(unsigned int nBits); // keep old data, zero additional bits