gallium: Fix llvmpipe on big-endian machines

Squashed commit of the following:

commit 0857a7e105bfcbc4d1431b2cc56612094c747ca3
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:07 2013 -0400

    gallivm: Fix lp_build_rgba8_to_fi32_soa for big endian

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 0d65131649a8aa140e2db228ba779d685c4333e3
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:07 2013 -0400

    gallivm: Fix big-endian machines

    This adds a bit-shift count to the format table, and adds the concept of
    vector or bitwise alignment on gathers.

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 9740bda9b7dc894b629ed38be9b51059ce90818f
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:07 2013 -0400

    llvmpipe: Fix convert_to_blend_type on big-endian

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit ae037c2de0f029e4e99371c0de25560484f0d8df
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    util: Convert color pack to packed formats

    This fixes them on big-endian.

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 5b05ac0c89ae092ea8ba5bba9f739708d7396b5c
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    graw-xlib: Convert to packed formats

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 51396e7d098cb6ff794391cf11afe4dbf86dbea0
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    format: Convert to packed formats

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 417b60bc66eb450e68a92ab0e47f76e292b385e6
Author: Adam Jackson <ajax@redhat.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    st/dri: Convert to packed formats

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 0934b2e022a5e0847d312c40734e2b44cac52fd8
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    st/xlib: Convert to packed formats

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit a307ea3c3716a706963acce7966b5e405ba11db9
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    gbm: Convert to packed formats

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 53eebdd253e1960a645ea278f31d7ef6a6cf4aeb
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    tests: Convert to packed formats

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 2f77fe3ee524945eacd546efcac34f7799fb3124
Author: Adam Jackson <ajax@redhat.com>
Date:   Tue Jun 18 13:07:37 2013 -0400

    gallium: Document packed formats

    Signed-off-by: Adam Jackson <ajax@redhat.com>

commit 1f1017159ce951f922210a430de9229f91f62714
Author: Richard Sandiford <r.sandiford@uk.ibm.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    gallium: Introduce 32-bit packed format names

    These are for interacting with buffers natively described in terms of
    bit shifts, like X11 visuals:

        uint32_t xyzw8888 = (x << 0) | (y << 8) | (z << 16) | (w << 24);

    Define these in terms of (endian-dependent) aliases to the array-style
    format names.

    Reviewed-by: Adam Jackson <ajax@redhat.com>
    Signed-off-by: Richard Sandiford <r.sandiford@uk.ibm.com>

commit 6cc7ab1ee66ed668da78c1d951dfd7782b4e786a
Author: Adam Jackson <ajax@redhat.com>
Date:   Mon Jun 3 12:10:32 2013 -0400

    gallium: Document format name conventions

    v2:
    - Fix a channel name thinko (Michel Dänzer)
    - Elaborate on SCALED versus INT
    - Add links to DirectX and FOURCC docs

    Signed-off-by: Adam Jackson <ajax@redhat.com>

commit df4d269e7fb62051a3c029b84147465001e5776e
Author: Adam Jackson <ajax@redhat.com>
Date:   Tue Jun 18 12:25:06 2013 -0400

    gallivm: Remove all notion of byte-swapping

    Signed-off-by: Adam Jackson <ajax@redhat.com>

Signed-off-by: Adam Jackson <ajax@redhat.com>

31 files changed, 297 insertions, 317 deletions

diff --git a/src/gallium/auxiliary/gallivm/lp_bld_conv.c b/src/gallium/auxiliary/gallivm/lp_bld_conv.c
index f11361a..cbea966 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_conv.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_conv.c
@@ -80,82 +80,6 @@
 
 
 /**
- * Byte swap on element. It will construct a call to intrinsic llvm.bswap
- * based on the type.
- *
- * @param res           element to byte swap.
- * @param type          int16_t, int32_t, int64_t, float or double
- * @param 
- */
-LLVMValueRef
-lp_build_bswap(struct gallivm_state *gallivm,
-               LLVMValueRef res,
-               struct lp_type type)
-{
-   LLVMTypeRef int_type = LLVMIntTypeInContext(gallivm->context,
-                                               type.width);
-   const char *intrinsic = NULL;
-   if (type.width == 8)
-      return res;
-   if (type.width == 16)
-      intrinsic = "llvm.bswap.i16";
-   else if (type.width == 32)
-     intrinsic = "llvm.bswap.i32";
-   else if (type.width == 64)
-      intrinsic = "llvm.bswap.i64";
-
-   assert (intrinsic != NULL);
-
-   /* In case of a floating-point type cast to a int of same size and then
-    * cast back to fp type.
-    */
-   if (type.floating)
-      res = LLVMBuildBitCast(gallivm->builder, res, int_type, "");
-   res = lp_build_intrinsic_unary(gallivm->builder, intrinsic, int_type, res);
-   if (type.floating)
-      res = LLVMBuildBitCast(gallivm->builder, res,
-                             lp_build_elem_type(gallivm, type), "");
-   return res;
-}
-
-
-/**
- * Byte swap every element in the vector.
- *
- * @param packed        <vector> to convert
- * @param src_type      <vector> type of int16_t, int32_t, int64_t, float or
- *                      double
- * @param dst_type      <vector> type to return
- */
-LLVMValueRef
-lp_build_bswap_vec(struct gallivm_state *gallivm,
-                   LLVMValueRef packed,
-                   struct lp_type src_type_vec,
-                   struct lp_type dst_type_vec)
-{
-   LLVMBuilderRef builder = gallivm->builder;
-   LLVMTypeRef dst_type = lp_build_elem_type(gallivm, dst_type_vec);
-   LLVMValueRef res;
-
-   if (src_type_vec.length == 1) {
-      res = lp_build_bswap(gallivm, packed, src_type_vec);
-      res = LLVMBuildBitCast(gallivm->builder, res, dst_type, "");
-   } else {
-      unsigned i;
-      res = LLVMGetUndef(lp_build_vec_type(gallivm, dst_type_vec));
-      for (i = 0; i < src_type_vec.length; ++i) {
-         LLVMValueRef index = lp_build_const_int32(gallivm, i);
-         LLVMValueRef elem = LLVMBuildExtractElement(builder, packed, index, "");
-         elem = lp_build_bswap(gallivm, elem, src_type_vec);
-         elem = LLVMBuildBitCast(gallivm->builder, elem, dst_type, "");
-         res = LLVMBuildInsertElement(gallivm->builder, res, elem, index, "");
-      }
-   }
-   return res;
-}
-
-
-/**
  * Converts int16 half-float to float32
  * Note this can be performed in 1 instruction if vcvtph2ps exists (f16c/cvt16)
  * [llvm.x86.vcvtph2ps / _mm_cvtph_ps]
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_conv.h b/src/gallium/auxiliary/gallivm/lp_bld_conv.h
index d7dfed8..42a1113 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_conv.h
+++ b/src/gallium/auxiliary/gallivm/lp_bld_conv.h
@@ -43,17 +43,6 @@
 struct lp_type;
 
 LLVMValueRef
-lp_build_bswap(struct gallivm_state *gallivm,
-               LLVMValueRef res,
-               struct lp_type type);
-
-LLVMValueRef
-lp_build_bswap_vec(struct gallivm_state *gallivm,
-                   LLVMValueRef packed,
-                   struct lp_type src_type,
-                   struct lp_type dst_type);
-
-LLVMValueRef
 lp_build_half_to_float(struct gallivm_state *gallivm,
                        LLVMValueRef src);
 
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c b/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c
index 6a1bf67..af755d4 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c
@@ -139,12 +139,12 @@ format_matches_type(const struct util_format_description *desc,
 
 
 /**
- * Unpack a single pixel into its RGBA components.
+ * Unpack a single pixel into its XYZW components.
  *
  * @param desc  the pixel format for the packed pixel value
  * @param packed integer pixel in a format such as PIPE_FORMAT_B8G8R8A8_UNORM
  *
- * @return RGBA in a float[4] or ubyte[4] or ushort[4] vector.
+ * @return XYZW in a float[4] or ubyte[4] or ushort[4] vector.
  */
 static INLINE LLVMValueRef
 lp_build_unpack_arith_rgba_aos(struct gallivm_state *gallivm,
@@ -159,7 +159,6 @@ lp_build_unpack_arith_rgba_aos(struct gallivm_state *gallivm,
 
    boolean normalized;
    boolean needs_uitofp;
-   unsigned shift;
    unsigned i;
 
    /* TODO: Support more formats */
@@ -172,10 +171,6 @@ lp_build_unpack_arith_rgba_aos(struct gallivm_state *gallivm,
     * matches floating point size */
    assert (LLVMTypeOf(packed) == LLVMInt32TypeInContext(gallivm->context));
 
-#ifdef PIPE_ARCH_BIG_ENDIAN
-   packed = lp_build_bswap(gallivm, packed, lp_type_uint(32));
-#endif
-
    /* Broadcast the packed value to all four channels
     * before: packed = BGRA
     * after: packed = {BGRA, BGRA, BGRA, BGRA}
@@ -194,11 +189,11 @@ lp_build_unpack_arith_rgba_aos(struct gallivm_state *gallivm,
    /* Initialize vector constants */
    normalized = FALSE;
    needs_uitofp = FALSE;
-   shift = 0;
 
    /* Loop over 4 color components */
    for (i = 0; i < 4; ++i) {
       unsigned bits = desc->channel[i].size;
+      unsigned shift = desc->channel[i].shift;
 
       if (desc->channel[i].type == UTIL_FORMAT_TYPE_VOID) {
          shifts[i] = LLVMGetUndef(LLVMInt32TypeInContext(gallivm->context));
@@ -224,12 +219,10 @@ lp_build_unpack_arith_rgba_aos(struct gallivm_state *gallivm,
          else
             scales[i] =  lp_build_const_float(gallivm, 1.0);
       }
-
-      shift += bits;
    }
 
-   /* Ex: convert packed = {BGRA, BGRA, BGRA, BGRA}
-    * into masked = {B, G, R, A}
+   /* Ex: convert packed = {XYZW, XYZW, XYZW, XYZW}
+    * into masked = {X, Y, Z, W}
     */
    shifted = LLVMBuildLShr(builder, packed, LLVMConstVector(shifts, 4), "");
    masked = LLVMBuildAnd(builder, shifted, LLVMConstVector(masks, 4), "");
@@ -276,7 +269,6 @@ lp_build_pack_rgba_aos(struct gallivm_state *gallivm,
    LLVMValueRef shifts[4];
    LLVMValueRef scales[4];
    boolean normalized;
-   unsigned shift;
    unsigned i, j;
 
    assert(desc->layout == UTIL_FORMAT_LAYOUT_PLAIN);
@@ -302,9 +294,9 @@ lp_build_pack_rgba_aos(struct gallivm_state *gallivm,
                                        LLVMConstVector(swizzles, 4), "");
 
    normalized = FALSE;
-   shift = 0;
    for (i = 0; i < 4; ++i) {
       unsigned bits = desc->channel[i].size;
+      unsigned shift = desc->channel[i].shift;
 
       if (desc->channel[i].type == UTIL_FORMAT_TYPE_VOID) {
          shifts[i] = LLVMGetUndef(LLVMInt32TypeInContext(gallivm->context));
@@ -325,8 +317,6 @@ lp_build_pack_rgba_aos(struct gallivm_state *gallivm,
          else
             scales[i] = lp_build_const_float(gallivm, 1.0);
       }
-
-      shift += bits;
    }
 
    if (normalized)
@@ -410,16 +400,11 @@ lp_build_fetch_rgba_aos(struct gallivm_state *gallivm,
 
       packed = lp_build_gather(gallivm, type.length/4,
                                format_desc->block.bits, type.width*4,
-                               base_ptr, offset);
+                               base_ptr, offset, TRUE);
 
       assert(format_desc->block.bits <= vec_len);
 
       packed = LLVMBuildBitCast(gallivm->builder, packed, dst_vec_type, "");
-#ifdef PIPE_ARCH_BIG_ENDIAN
-      if (type.floating)
-         packed = lp_build_bswap_vec(gallivm, packed, type,
-                                    lp_type_float_vec(type.width, vec_len));
-#endif
       return lp_build_format_swizzle_aos(format_desc, &bld, packed);
    }
 
@@ -453,7 +438,7 @@ lp_build_fetch_rgba_aos(struct gallivm_state *gallivm,
 
          packed = lp_build_gather_elem(gallivm, num_pixels,
                                        format_desc->block.bits, 32,
-                                       base_ptr, offset, k);
+                                       base_ptr, offset, k, FALSE);
 
          tmps[k] = lp_build_unpack_arith_rgba_aos(gallivm,
                                                   format_desc,
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_aos_array.c b/src/gallium/auxiliary/gallivm/lp_bld_format_aos_array.c
index 3402a0b..ee3ca86 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_format_aos_array.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_format_aos_array.c
@@ -40,58 +40,6 @@
 #include "pipe/p_state.h"
 
 
-#ifdef PIPE_ARCH_BIG_ENDIAN
-static LLVMValueRef
-lp_build_read_int_bswap(struct gallivm_state *gallivm,
-                        LLVMValueRef base_ptr,
-                        unsigned src_width,
-                        LLVMTypeRef src_type,
-                        unsigned i,
-                        LLVMTypeRef dst_type)
-{
-   LLVMBuilderRef builder = gallivm->builder;
-   LLVMValueRef index = lp_build_const_int32(gallivm, i);
-   LLVMValueRef ptr = LLVMBuildGEP(builder, base_ptr, &index, 1, "");
-   LLVMValueRef res = LLVMBuildLoad(builder, ptr, "");
-   res = lp_build_bswap(gallivm, res, lp_type_uint(src_width));
-   return LLVMBuildBitCast(builder, res, dst_type, "");
-}
-
-static LLVMValueRef
-lp_build_fetch_read_big_endian(struct gallivm_state *gallivm,
-                               struct lp_type src_type,
-                               LLVMValueRef base_ptr)
-{
-   LLVMBuilderRef builder = gallivm->builder;
-   unsigned src_width = src_type.width;
-   unsigned length = src_type.length;
-   LLVMTypeRef src_elem_type = LLVMIntTypeInContext(gallivm->context, src_width);
-   LLVMTypeRef dst_elem_type = lp_build_elem_type (gallivm, src_type);
-   LLVMTypeRef src_ptr_type = LLVMPointerType(src_elem_type, 0);
-   LLVMValueRef res;
-
-   base_ptr = LLVMBuildPointerCast(builder, base_ptr, src_ptr_type, "");
-   if (length == 1) {
-      /* Scalar */
-      res = lp_build_read_int_bswap(gallivm, base_ptr, src_width, src_elem_type,
-                                    0, dst_elem_type);
-   } else {
-      /* Vector */
-      LLVMTypeRef dst_vec_type = LLVMVectorType(dst_elem_type, length);
-      unsigned i;
-
-      res = LLVMGetUndef(dst_vec_type);
-      for (i = 0; i < length; ++i) {
-         LLVMValueRef index = lp_build_const_int32(gallivm, i);
-         LLVMValueRef elem = lp_build_read_int_bswap(gallivm, base_ptr, src_width,
-                                                     src_elem_type, i, dst_elem_type);
-         res = LLVMBuildInsertElement(builder, res, elem, index, "");
-      }
-   }
-
-   return res;
-}
-#endif
 
 /**
  * @brief lp_build_fetch_rgba_aos_array
@@ -124,13 +72,9 @@ lp_build_fetch_rgba_aos_array(struct gallivm_state *gallivm,
 
    /* Read whole vector from memory, unaligned */
    ptr = LLVMBuildGEP(builder, base_ptr, &offset, 1, "");
-#ifdef PIPE_ARCH_BIG_ENDIAN
-   res = lp_build_fetch_read_big_endian(gallivm, src_type, ptr);
-#else
    ptr = LLVMBuildPointerCast(builder, ptr, LLVMPointerType(src_vec_type, 0), "");
    res = LLVMBuildLoad(builder, ptr, "");
    lp_set_load_alignment(res, src_type.width / 8);
-#endif
 
    /* Truncate doubles to float */
    if (src_type.floating && src_type.width == 64) {
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_format_soa.c
index edc2442..4c6bd81 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_format_soa.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_format_soa.c
@@ -115,7 +115,6 @@ lp_build_unpack_rgba_soa(struct gallivm_state *gallivm,
    LLVMBuilderRef builder = gallivm->builder;
    struct lp_build_context bld;
    LLVMValueRef inputs[4];
-   unsigned start;
    unsigned chan;
 
    assert(format_desc->layout == UTIL_FORMAT_LAYOUT_PLAIN);
@@ -128,9 +127,9 @@ lp_build_unpack_rgba_soa(struct gallivm_state *gallivm,
    lp_build_context_init(&bld, gallivm, type);
 
    /* Decode the input vector components */
-   start = 0;
    for (chan = 0; chan < format_desc->nr_channels; ++chan) {
       const unsigned width = format_desc->channel[chan].size;
+      const unsigned start = format_desc->channel[chan].shift;
       const unsigned stop = start + width;
       LLVMValueRef input;
 
@@ -256,8 +255,6 @@ lp_build_unpack_rgba_soa(struct gallivm_state *gallivm,
       }
 
       inputs[chan] = input;
-
-      start = stop;
    }
 
    lp_build_format_swizzle_soa(format_desc, &bld, inputs, rgba_out);
@@ -291,7 +288,11 @@ lp_build_rgba8_to_fi32_soa(struct gallivm_state *gallivm,
 
    /* Decode the input vector components */
    for (chan = 0; chan < 4; ++chan) {
+#ifdef PIPE_ARCH_LITTLE_ENDIAN
       unsigned start = chan*8;
+#else
+      unsigned start = (3-chan)*8;
+#endif
       unsigned stop = start + 8;
       LLVMValueRef input;
 
@@ -360,13 +361,14 @@ lp_build_fetch_rgba_soa(struct gallivm_state *gallivm,
 
       /*
        * gather the texels from the texture
-       * Ex: packed = {BGRA, BGRA, BGRA, BGRA}.
+       * Ex: packed = {XYZW, XYZW, XYZW, XYZW}
        */
+      assert(format_desc->block.bits <= type.width);
       packed = lp_build_gather(gallivm,
                                type.length,
                                format_desc->block.bits,
                                type.width,
-                               base_ptr, offset);
+                               base_ptr, offset, FALSE);
 
       /*
        * convert texels to float rgba
@@ -391,7 +393,8 @@ lp_build_fetch_rgba_soa(struct gallivm_state *gallivm,
 
       packed = lp_build_gather(gallivm, type.length,
                                format_desc->block.bits,
-                               type.width, base_ptr, offset);
+                               type.width, base_ptr, offset,
+			       FALSE);
       if (format_desc->format == PIPE_FORMAT_R11G11B10_FLOAT) {
          lp_build_r11g11b10_to_float(gallivm, packed, rgba_out);
       }
@@ -418,14 +421,14 @@ lp_build_fetch_rgba_soa(struct gallivm_state *gallivm,
          LLVMValueRef s_offset = lp_build_const_int_vec(gallivm, type, 4);
          offset = LLVMBuildAdd(builder, offset, s_offset, "");
          packed = lp_build_gather(gallivm, type.length,
-                                  32, type.width, base_ptr, offset);
+                                  32, type.width, base_ptr, offset, FALSE);
          packed = LLVMBuildAnd(builder, packed,
                                lp_build_const_int_vec(gallivm, type, mask), "");
       }
       else {
          assert (format_desc->format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT);
          packed = lp_build_gather(gallivm, type.length,
-                                  32, type.width, base_ptr, offset);
+                                  32, type.width, base_ptr, offset, TRUE);
          packed = LLVMBuildBitCast(builder, packed,
                                    lp_build_vec_type(gallivm, type), "");
       }
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_yuv.c b/src/gallium/auxiliary/gallivm/lp_bld_format_yuv.c
index e542abc..873f354 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_format_yuv.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_format_yuv.c
@@ -497,7 +497,7 @@ lp_build_fetch_subsampled_rgba_aos(struct gallivm_state *gallivm,
    assert(format_desc->block.width == 2);
    assert(format_desc->block.height == 1);
 
-   packed = lp_build_gather(gallivm, n, 32, 32, base_ptr, offset);
+   packed = lp_build_gather(gallivm, n, 32, 32, base_ptr, offset, FALSE);
 
    (void)j;
 
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_gather.c b/src/gallium/auxiliary/gallivm/lp_bld_gather.c
index eefb23a..9155d81 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_gather.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_gather.c
@@ -78,7 +78,8 @@ lp_build_gather_elem(struct gallivm_state *gallivm,
                      unsigned dst_width,
                      LLVMValueRef base_ptr,
                      LLVMValueRef offsets,
-                     unsigned i)
+                     unsigned i,
+                     boolean vector_justify)
 {
    LLVMTypeRef src_type = LLVMIntTypeInContext(gallivm->context, src_width);
    LLVMTypeRef src_ptr_type = LLVMPointerType(src_type, 0);
@@ -97,10 +98,12 @@ lp_build_gather_elem(struct gallivm_state *gallivm,
       res = LLVMBuildTrunc(gallivm->builder, res, dst_elem_type, "");
    } else if (src_width < dst_width) {
       res = LLVMBuildZExt(gallivm->builder, res, dst_elem_type, "");
+      if (vector_justify) {
 #ifdef PIPE_ARCH_BIG_ENDIAN
-      res = LLVMBuildShl(gallivm->builder, res,
-                         LLVMConstInt(dst_elem_type, dst_width - src_width, 0), "");
+         res = LLVMBuildShl(gallivm->builder, res,
+                            LLVMConstInt(dst_elem_type, dst_width - src_width, 0), "");
 #endif
+      }
    }
 
    return res;
@@ -112,11 +115,20 @@ lp_build_gather_elem(struct gallivm_state *gallivm,
  * Use for fetching texels from a texture.
  * For SSE, typical values are length=4, src_width=32, dst_width=32.
  *
+ * When src_width < dst_width, the return value can be justified in
+ * one of two ways:
+ * "integer justification" is used when the caller treats the destination
+ * as a packed integer bitmask, as described by the channels' "shift" and
+ * "width" fields;
+ * "vector justification" is used when the caller casts the destination
+ * to a vector and needs channel X to be in vector element 0.
+ *
  * @param length length of the offsets
  * @param src_width src element width in bits
  * @param dst_width result element width in bits (src will be expanded to fit)
  * @param base_ptr base pointer, should be a i8 pointer type.
  * @param offsets vector with offsets
+ * @param vector_justify select vector rather than integer justification
  */
 LLVMValueRef
 lp_build_gather(struct gallivm_state *gallivm,
@@ -124,7 +136,8 @@ lp_build_gather(struct gallivm_state *gallivm,
                 unsigned src_width,
                 unsigned dst_width,
                 LLVMValueRef base_ptr,
-                LLVMValueRef offsets)
+                LLVMValueRef offsets,
+                boolean vector_justify)
 {
    LLVMValueRef res;
 
@@ -132,7 +145,7 @@ lp_build_gather(struct gallivm_state *gallivm,
       /* Scalar */
       return lp_build_gather_elem(gallivm, length,
                                   src_width, dst_width,
-                                  base_ptr, offsets, 0);
+                                  base_ptr, offsets, 0, vector_justify);
    } else {
       /* Vector */
 
@@ -146,7 +159,7 @@ lp_build_gather(struct gallivm_state *gallivm,
          LLVMValueRef elem;
          elem = lp_build_gather_elem(gallivm, length,
                                      src_width, dst_width,
-                                     base_ptr, offsets, i);
+                                     base_ptr, offsets, i, vector_justify);
          res = LLVMBuildInsertElement(gallivm->builder, res, elem, index, "");
       }
    }
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_gather.h b/src/gallium/auxiliary/gallivm/lp_bld_gather.h
index 8e4c07d..ee69473 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_gather.h
+++ b/src/gallium/auxiliary/gallivm/lp_bld_gather.h
@@ -47,7 +47,8 @@ lp_build_gather_elem(struct gallivm_state *gallivm,
                      unsigned dst_width,
                      LLVMValueRef base_ptr,
                      LLVMValueRef offsets,
-                     unsigned i);
+                     unsigned i,
+                     boolean vector_justify);
 
 LLVMValueRef
 lp_build_gather(struct gallivm_state *gallivm,
@@ -55,7 +56,8 @@ lp_build_gather(struct gallivm_state *gallivm,
                 unsigned src_width,
                 unsigned dst_width,
                 LLVMValueRef base_ptr,
-                LLVMValueRef offsets);
+                LLVMValueRef offsets,
+                boolean vector_justify);
 
 LLVMValueRef
 lp_build_gather_values(struct gallivm_state * gallivm,
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c b/src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c
index c31b05d..104c24d 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c
@@ -531,7 +531,7 @@ lp_build_sample_fetch_image_nearest(struct lp_build_sample_context *bld,
                               bld->texel_type.length,
                               bld->format_desc->block.bits,
                               bld->texel_type.width,
-                              data_ptr, offset);
+                              data_ptr, offset, TRUE);
 
       rgba8 = LLVMBuildBitCast(builder, rgba8, u8n_vec_type, "");
    }
@@ -893,7 +893,7 @@ lp_build_sample_fetch_image_linear(struct lp_build_sample_context *bld,
                                        bld->texel_type.length,
                                        bld->format_desc->block.bits,
                                        bld->texel_type.width,
-                                       data_ptr, offset[k][j][i]);
+                                       data_ptr, offset[k][j][i], TRUE);
 
                rgba8 = LLVMBuildBitCast(builder, rgba8, u8n_vec_type, "");
             }
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_swizzle.c b/src/gallium/auxiliary/gallivm/lp_bld_swizzle.c
index 08d817a..79116bc 100644
--- a/src/gallium/auxiliary/gallivm/lp_bld_swizzle.c
+++ b/src/gallium/auxiliary/gallivm/lp_bld_swizzle.c
@@ -217,6 +217,20 @@ lp_build_swizzle_scalar_aos(struct lp_build_context *bld,
 
       a = LLVMBuildBitCast(builder, a, lp_build_vec_type(bld->gallivm, type2), "");
 
+      /*
+       * Vector element 0 is always channel X.
+       *
+       *                        76 54 32 10 (array numbering)
+       * Little endian reg in:  YX YX YX YX
+       * Little endian reg out: YY YY YY YY if shift right (shift == -1)
+       *                        XX XX XX XX if shift left (shift == 1)
+       *
+       *                        01 23 45 67 (array numbering)
+       * Big endian reg in:     XY XY XY XY
+       * Big endian reg out:    YY YY YY YY if shift left (shift == 1)
+       *                        XX XX XX XX if shift right (shift == -1)
+       *
+       */
 #ifdef PIPE_ARCH_LITTLE_ENDIAN
       shift = channel == 0 ? 1 : -1;
 #else
@@ -240,10 +254,23 @@ lp_build_swizzle_scalar_aos(struct lp_build_context *bld,
       /*
        * Bit mask and recursive shifts
        *
+       * Little-endian registers:
+       *
+       *   7654 3210
+       *   WZYX WZYX .... WZYX  <= input
+       *   00Y0 00Y0 .... 00Y0  <= mask
+       *   00YY 00YY .... 00YY  <= shift right 1 (shift amount -1)
+       *   YYYY YYYY .... YYYY  <= shift left 2 (shift amount 2)
+       *
+       * Big-endian registers:
+       *
+       *   0123 4567
        *   XYZW XYZW .... XYZW  <= input
-       *   0Y00 0Y00 .... 0Y00
-       *   YY00 YY00 .... YY00
-       *   YYYY YYYY .... YYYY  <= output
+       *   0Y00 0Y00 .... 0Y00  <= mask
+       *   YY00 YY00 .... YY00  <= shift left 1 (shift amount 1)
+       *   YYYY YYYY .... YYYY  <= shift right 2 (shift amount -2)
+       *
+       * shifts[] gives little-endian shift amounts; we need to negate for big-endian.
        */
       struct lp_type type4;
       const int shifts[4][2] = {
@@ -274,14 +301,15 @@ lp_build_swizzle_scalar_aos(struct lp_build_context *bld,
          LLVMValueRef tmp = NULL;
          int shift = shifts[channel][i];
 
-#ifdef PIPE_ARCH_LITTLE_ENDIAN
+         /* See endianness diagram above */
+#ifdef PIPE_ARCH_BIG_ENDIAN
          shift = -shift;
 #endif
 
          if(shift > 0)
-            tmp = LLVMBuildLShr(builder, a, lp_build_const_int_vec(bld->gallivm, type4, shift*type.width), "");
+            tmp = LLVMBuildShl(builder, a, lp_build_const_int_vec(bld->gallivm, type4, shift*type.width), "");
          if(shift < 0)
-            tmp = LLVMBuildShl(builder, a, lp_build_const_int_vec(bld->gallivm, type4, -shift*type.width), "");
+            tmp = LLVMBuildLShr(builder, a, lp_build_const_int_vec(bld->gallivm, type4, -shift*type.width), "");
 
          assert(tmp);
          if(tmp)
@@ -474,21 +502,39 @@ lp_build_swizzle_aos(struct lp_build_context *bld,
 
       /*
        * Mask and shift the channels, trying to group as many channels in the
-       * same shift as possible
+       * same shift as possible.  The shift amount is positive for shifts left
+       * and negative for shifts right.
        */
       for (shift = -3; shift <= 3; ++shift) {
          uint64_t mask = 0;
 
          assert(type4.width <= sizeof(mask)*8);
 
+         /*
+          * Vector element numbers follow the XYZW order, so 0 is always X, etc.
+          * After widening 4 times we have:
+          *
+          *                                3210
+          * Little-endian register layout: WZYX
+          *
+          *                                0123
+          * Big-endian register layout:    XYZW
+          *
+          * For little-endian, higher-numbered channels are obtained by a shift right
+          * (negative shift amount) and lower-numbered channels by a shift left
+          * (positive shift amount).  The opposite is true for big-endian.
+          */
          for (chan = 0; chan < 4; ++chan) {
-            /* FIXME: big endian */
-            if (swizzles[chan] < 4 &&
-                chan - swizzles[chan] == shift) {
+            if (swizzles[chan] < 4) {
+               /* We need to move channel swizzles[chan] into channel chan */
 #ifdef PIPE_ARCH_LITTLE_ENDIAN
-               mask |= ((1ULL << type.width) - 1) << (swizzles[chan] * type.width);
+               if (swizzles[chan] - chan == -shift) {
+                  mask |= ((1ULL << type.width) - 1) << (swizzles[chan] * type.width);
+               }
 #else
-               mask |= ((1ULL << type.width) - 1) << (type4.width - type.width) >> (swizzles[chan] * type.width);
+               if (swizzles[chan] - chan == shift) {
+                  mask |= ((1ULL << type.width) - 1) << (type4.width - type.width) >> (swizzles[chan] * type.width);
+               }
 #endif
             }
          }
@@ -502,21 +548,11 @@ lp_build_swizzle_aos(struct lp_build_context *bld,
             masked = LLVMBuildAnd(builder, a,
                                   lp_build_const_int_vec(bld->gallivm, type4, mask), "");
             if (shift > 0) {
-#ifdef PIPE_ARCH_LITTLE_ENDIAN
                shifted = LLVMBuildShl(builder, masked,
                                       lp_build_const_int_vec(bld->gallivm, type4, shift*type.width), "");
-#else
-               shifted = LLVMBuildLShr(builder, masked,
-                                       lp_build_const_int_vec(bld->gallivm, type4, shift*type.width), "");
-#endif
             } else if (shift < 0) {
-#ifdef PIPE_ARCH_LITTLE_ENDIAN
                shifted = LLVMBuildLShr(builder, masked,
                                        lp_build_const_int_vec(bld->gallivm, type4, -shift*type.width), "");
-#else
-               shifted = LLVMBuildShl(builder, masked,
-                                      lp_build_const_int_vec(bld->gallivm, type4, -shift*type.width), "");
-#endif
             } else {
                shifted = masked;
             }
diff --git a/src/gallium/auxiliary/util/u_format.h b/src/gallium/auxiliary/util/u_format.h
index 4cace6a..9774a2b 100644
--- a/src/gallium/auxiliary/util/u_format.h
+++ b/src/gallium/auxiliary/util/u_format.h
@@ -133,6 +133,7 @@ struct util_format_channel_description
    unsigned normalized:1;
    unsigned pure_integer:1;
    unsigned size:9;        /**< bits per channel */
+   unsigned shift:16;      /** number of bits from lsb */
 };
 
 
@@ -179,9 +180,31 @@ struct util_format_description
    unsigned is_mixed:1;
 
    /**
-    * Input channel description.
+    * Input channel description, in the order XYZW.
     *
     * Only valid for UTIL_FORMAT_LAYOUT_PLAIN formats.
+    *
+    * If each channel is accessed as an individual N-byte value, X is always
+    * at the lowest address in memory, Y is always next, and so on.  For all
+    * currently-defined formats, the N-byte value has native endianness.
+    *
+    * If instead a group of channels is accessed as a single N-byte value,
+    * the order of the channels within that value depends on endianness.
+    * For big-endian targets, X is the most significant subvalue,
+    * otherwise it is the least significant one.
+    *
+    * For example, if X is 8 bits and Y is 24 bits, the memory order is:
+    *
+    *                 0  1  2  3
+    *  little-endian: X  Yl Ym Yu    (l = lower, m = middle, u = upper)
+    *  big-endian:    X  Yu Ym Yl
+    *
+    * If X is 5 bits, Y is 5 bits, Z is 5 bits and W is 1 bit, the layout is:
+    *
+    *                        0        1
+    *                 msb  lsb msb  lsb
+    *  little-endian: YYYXXXXX WZZZZZYY
+    *  big-endian:    XXXXXYYY YYZZZZZW
     */
    struct util_format_channel_description channel[4];
 
diff --git a/src/gallium/auxiliary/util/u_format_pack.py b/src/gallium/auxiliary/util/u_format_pack.py
index 565d059..d1f68c8 100644
--- a/src/gallium/auxiliary/util/u_format_pack.py
+++ b/src/gallium/auxiliary/util/u_format_pack.py
@@ -99,15 +99,6 @@ def generate_format_type(format):
     print
 
 
-def bswap_format(format):
-    '''Generate a structure that describes the format.'''
-
-    if format.is_bitmask() and not format.is_array() and format.block_size() > 8:
-        print '#ifdef PIPE_ARCH_BIG_ENDIAN'
-        print '   pixel.value = util_bswap%u(pixel.value);' % format.block_size()
-        print '#endif'
-
-
 def is_format_supported(format):
     '''Determines whether we actually have the plumbing necessary to generate the 
     to read/write to/from this format.'''
@@ -423,16 +414,11 @@ def generate_unpack_kernel(format, dst_channel, dst_native_type):
             elif src_channel.type == SIGNED:
                 print '         int%u_t %s;' % (depth, src_channel.name)
 
-        if depth > 8:
-            print '#ifdef PIPE_ARCH_BIG_ENDIAN'
-            print '         value = util_bswap%u(value);' % depth
-            print '#endif'
-
         # Compute the intermediate unshifted values 
-        shift = 0
         for i in range(format.nr_channels()):
             src_channel = format.channels[i]
             value = 'value'
+            shift = src_channel.shift
             if src_channel.type == UNSIGNED:
                 if shift:
                     value = '%s >> %u' % (value, shift)
@@ -455,8 +441,6 @@ def generate_unpack_kernel(format, dst_channel, dst_native_type):
             if value is not None:
                 print '         %s = %s;' % (src_channel.name, value)
                 
-            shift += src_channel.size
-
         # Convert, swizzle, and store final values
         for i in range(4):
             swizzle = format.swizzles[i]
@@ -484,7 +468,6 @@ def generate_unpack_kernel(format, dst_channel, dst_native_type):
     else:
         print '         union util_format_%s pixel;' % format.short_name()
         print '         memcpy(&pixel, src, sizeof pixel);'
-        bswap_format(format)
     
         for i in range(4):
             swizzle = format.swizzles[i]
@@ -525,9 +508,9 @@ def generate_pack_kernel(format, src_channel, src_native_type):
         depth = format.block_size()
         print '         uint%u_t value = 0;' % depth 
 
-        shift = 0
         for i in range(4):
             dst_channel = format.channels[i]
+            shift = dst_channel.shift
             if inv_swizzle[i] is not None:
                 value ='src[%u]' % inv_swizzle[i]
                 dst_colorspace = format.colorspace
@@ -551,13 +534,6 @@ def generate_pack_kernel(format, src_channel, src_native_type):
                 if value is not None:
                     print '         value |= %s;' % (value)
                 
-            shift += dst_channel.size
-
-        if depth > 8:
-            print '#ifdef PIPE_ARCH_BIG_ENDIAN'
-            print '         value = util_bswap%u(value);' % depth
-            print '#endif'
-        
         print '         *(uint%u_t *)dst = value;' % depth 
 
     else:
@@ -579,7 +555,6 @@ def generate_pack_kernel(format, src_channel, src_native_type):
                                     dst_colorspace = dst_colorspace)
             print '         pixel.chan.%s = %s;' % (dst_channel.name, value)
     
-        bswap_format(format)
         print '         memcpy(dst, &pixel, sizeof pixel);'
     
 
diff --git a/src/gallium/auxiliary/util/u_format_parse.py b/src/gallium/auxiliary/util/u_format_parse.py
index 07052b9..e202099 100755
--- a/src/gallium/auxiliary/util/u_format_parse.py
+++ b/src/gallium/auxiliary/util/u_format_parse.py
@@ -30,6 +30,8 @@
 '''
 
 
+import sys
+
 VOID, UNSIGNED, SIGNED, FIXED, FLOAT = range(5)
 
 SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_W, SWIZZLE_0, SWIZZLE_1, SWIZZLE_NONE, = range(7)
@@ -42,6 +44,9 @@ YUV = 'yuv'
 ZS = 'zs'
 
 
+# Not cross-compiler friendly
+is_big_endian = sys.byteorder == 'big'
+
 def is_pot(x):
    return (x & (x - 1)) == 0
 
@@ -307,6 +312,11 @@ def parse(filename):
             channel = Channel(type, norm, pure, size, names[i])
             channels.append(channel)
 
+        shift = 0
+        for channel in channels[3::-1] if is_big_endian else channels:
+            channel.shift = shift
+            shift += channel.size
+
         format = Format(name, layout, block_width, block_height, channels, swizzles, colorspace)
         formats.append(format)
     return formats
diff --git a/src/gallium/auxiliary/util/u_format_table.py b/src/gallium/auxiliary/util/u_format_table.py
index 8edb505..9d44cf3 100755
--- a/src/gallium/auxiliary/util/u_format_table.py
+++ b/src/gallium/auxiliary/util/u_format_table.py
@@ -114,9 +114,9 @@ def write_format_table(formats):
             else:
                 sep = ""
             if channel.size:
-                print "      {%s, %s, %s, %u}%s\t/* %s = %s */" % (type_map[channel.type], bool_map(channel.norm), bool_map(channel.pure), channel.size, sep, "xyzw"[i], channel.name)
+                print "      {%s, %s, %s, %u, %u}%s\t/* %s = %s */" % (type_map[channel.type], bool_map(channel.norm), bool_map(channel.pure), channel.size, channel.shift, sep, "xyzw"[i], channel.name)
             else:
-                print "      {0, 0, 0, 0}%s" % (sep,)
+                print "      {0, 0, 0, 0, 0}%s" % (sep,)
         print "   },"
         print "   {"
         for i in range(4):
diff --git a/src/gallium/auxiliary/util/u_pack_color.h b/src/gallium/auxiliary/util/u_pack_color.h
index 1f6a56a..102ad60 100644
--- a/src/gallium/auxiliary/util/u_pack_color.h
+++ b/src/gallium/auxiliary/util/u_pack_color.h
@@ -65,32 +65,32 @@ util_pack_color_ub(ubyte r, ubyte g, ubyte b, ubyte a,
                    enum pipe_format format, union util_color *uc)
 {
    switch (format) {
-   case PIPE_FORMAT_A8B8G8R8_UNORM:
+   case PIPE_FORMAT_ABGR8888_UNORM:
       {
          uc->ui = (r << 24) | (g << 16) | (b << 8) | a;
       }
       return;
-   case PIPE_FORMAT_X8B8G8R8_UNORM:
+   case PIPE_FORMAT_XBGR8888_UNORM:
       {
          uc->ui = (r << 24) | (g << 16) | (b << 8) | 0xff;
       }
       return;
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
+   case PIPE_FORMAT_BGRA8888_UNORM:
       {
          uc->ui = (a << 24) | (r << 16) | (g << 8) | b;
       }
       return;
-   case PIPE_FORMAT_B8G8R8X8_UNORM:
+   case PIPE_FORMAT_BGRX8888_UNORM:
       {
          uc->ui = (0xff << 24) | (r << 16) | (g << 8) | b;
       }
       return;
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
+   case PIPE_FORMAT_ARGB8888_UNORM:
       {
          uc->ui = (b << 24) | (g << 16) | (r << 8) | a;
       }
       return;
-   case PIPE_FORMAT_X8R8G8B8_UNORM:
+   case PIPE_FORMAT_XRGB8888_UNORM:
       {
          uc->ui = (b << 24) | (g << 16) | (r << 8) | 0xff;
       }
@@ -166,7 +166,7 @@ util_unpack_color_ub(enum pipe_format format, union util_color *uc,
                      ubyte *r, ubyte *g, ubyte *b, ubyte *a)
 {
    switch (format) {
-   case PIPE_FORMAT_A8B8G8R8_UNORM:
+   case PIPE_FORMAT_ABGR8888_UNORM:
       {
          uint p = uc->ui;
          *r = (ubyte) ((p >> 24) & 0xff);
@@ -175,7 +175,7 @@ util_unpack_color_ub(enum pipe_format format, union util_color *uc,
          *a = (ubyte) ((p >>  0) & 0xff);
       }
       return;
-   case PIPE_FORMAT_X8B8G8R8_UNORM:
+   case PIPE_FORMAT_XBGR8888_UNORM:
       {
          uint p = uc->ui;
          *r = (ubyte) ((p >> 24) & 0xff);
@@ -184,7 +184,7 @@ util_unpack_color_ub(enum pipe_format format, union util_color *uc,
          *a = (ubyte) 0xff;
       }
       return;
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
+   case PIPE_FORMAT_BGRA8888_UNORM:
       {
          uint p = uc->ui;
          *r = (ubyte) ((p >> 16) & 0xff);
@@ -193,7 +193,7 @@ util_unpack_color_ub(enum pipe_format format, union util_color *uc,
          *a = (ubyte) ((p >> 24) & 0xff);
       }
       return;
-   case PIPE_FORMAT_B8G8R8X8_UNORM:
+   case PIPE_FORMAT_BGRX8888_UNORM:
       {
          uint p = uc->ui;
          *r = (ubyte) ((p >> 16) & 0xff);
@@ -202,7 +202,7 @@ util_unpack_color_ub(enum pipe_format format, union util_color *uc,
          *a = (ubyte) 0xff;
       }
       return;
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
+   case PIPE_FORMAT_ARGB8888_UNORM:
       {
          uint p = uc->ui;
          *r = (ubyte) ((p >>  8) & 0xff);
@@ -211,7 +211,7 @@ util_unpack_color_ub(enum pipe_format format, union util_color *uc,
          *a = (ubyte) ((p >>  0) & 0xff);
       }
       return;
-   case PIPE_FORMAT_X8R8G8B8_UNORM:
+   case PIPE_FORMAT_XRGB8888_UNORM:
       {
          uint p = uc->ui;
          *r = (ubyte) ((p >>  8) & 0xff);
@@ -350,32 +350,32 @@ util_pack_color(const float rgba[4], enum pipe_format format, union util_color *
    }
 
    switch (format) {
-   case PIPE_FORMAT_A8B8G8R8_UNORM:
+   case PIPE_FORMAT_ABGR8888_UNORM:
       {
          uc->ui = (r << 24) | (g << 16) | (b << 8) | a;
       }
       return;
-   case PIPE_FORMAT_X8B8G8R8_UNORM:
+   case PIPE_FORMAT_XBGR8888_UNORM:
       {
          uc->ui = (r << 24) | (g << 16) | (b << 8) | 0xff;
       }
       return;
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
+   case PIPE_FORMAT_BGRA8888_UNORM:
       {
          uc->ui = (a << 24) | (r << 16) | (g << 8) | b;
       }
       return;
-   case PIPE_FORMAT_B8G8R8X8_UNORM:
+   case PIPE_FORMAT_BGRX8888_UNORM:
       {
          uc->ui = (0xff << 24) | (r << 16) | (g << 8) | b;
       }
       return;
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
+   case PIPE_FORMAT_ARGB8888_UNORM:
       {
          uc->ui = (b << 24) | (g << 16) | (r << 8) | a;
       }
       return;
-   case PIPE_FORMAT_X8R8G8B8_UNORM:
+   case PIPE_FORMAT_XRGB8888_UNORM:
       {
          uc->ui = (b << 24) | (g << 16) | (r << 8) | 0xff;
       }
diff --git a/src/gallium/docs/source/format.rst b/src/gallium/docs/source/format.rst
new file mode 100644
index 0000000..f6d77d6
--- /dev/null
+++ b/src/gallium/docs/source/format.rst
@@ -0,0 +1,61 @@
+Formats in gallium
+==================
+
+Gallium format names mostly follow D3D10 conventions, with some extensions.
+
+Format names like XnYnZnWn have the X component in the lowest-address n bits
+and the W component in the highest-address n bits; for B8G8R8A8, byte 0 is
+blue and byte 3 is alpha.  Note that platform endianness is not considered
+in this definition.  In C:
+
+    struct x8y8z8w8 { uint8_t x, y, z, w; };
+
+Format aliases like XYZWstrq are (s+t+r+q)-bit integers in host endianness,
+with the X component in the s least-significant bits of the integer.  In C:
+
+    uint32_t xyzw8888 = (x << 0) | (y << 8) | (z << 16) | (w << 24);
+
+Format suffixes affect the interpretation of the channel:
+
+- ``SINT``:     N bit signed integer [-2^(N-1) ... 2^(N-1) - 1]
+- ``SNORM``:    N bit signed integer normalized to [-1 ... 1]
+- ``SSCALED``:  N bit signed integer [-2^(N-1) ... 2^(N-1) - 1]
+- ``FIXED``:    Signed fixed point integer, (N/2 - 1) bits of mantissa
+- ``FLOAT``:    N bit IEEE754 float
+- ``NORM``:     Normalized integers, signed or unsigned per channel
+- ``UINT``:     N bit unsigned integer [0 ... 2^N - 1]
+- ``UNORM``:    N bit unsigned integer normalized to [0 ... 1]
+- ``USCALED``:  N bit unsigned integer [0 ... 2^N - 1]
+
+The difference between ``SINT`` and ``SSCALED`` is that the former are pure
+integers in shaders, while the latter are floats; likewise for ``UINT`` versus
+``USCALED``.
+
+There are two exceptions for ``FLOAT``.  ``R9G9B9E5_FLOAT`` is nine bits
+each of red green and blue mantissa, with a shared five bit exponent.
+``R11G11B10_FLOAT`` is five bits of exponent and five or six bits of mantissa
+for each color channel.
+
+For the ``NORM`` suffix, the signedness of each channel is indicated with an
+S or U after the number of channel bits, as in ``R5SG5SB6U_NORM``.
+
+The ``SRGB`` suffix is like ``UNORM`` in range, but in the sRGB colorspace.
+
+Compressed formats are named first by the compression format string (``DXT1``,
+``ETC1``, etc), followed by a format-specific subtype.  Refer to the
+appropriate compression spec for details.
+
+Formats used in video playback are named by their FOURCC code.
+
+Format names with an embedded underscore are subsampled.  ``R8G8_B8G8`` is a
+single 32-bit block of two pixels, where the R and B values are repeated in
+both pixels.
+
+References
+----------
+
+DirectX Graphics Infrastructure documentation on DXGI_FORMAT enum:
+http://msdn.microsoft.com/en-us/library/windows/desktop/bb173059%28v=vs.85%29.aspx
+
+FOURCC codes for YUV formats:
+http://www.fourcc.org/yuv.php
diff --git a/src/gallium/drivers/llvmpipe/lp_bld_depth.c b/src/gallium/drivers/llvmpipe/lp_bld_depth.c
index 79891cf..06556dc 100644
--- a/src/gallium/drivers/llvmpipe/lp_bld_depth.c
+++ b/src/gallium/drivers/llvmpipe/lp_bld_depth.c
@@ -349,8 +349,6 @@ get_z_shift_and_mask(const struct util_format_description *format_desc,
 {
    unsigned total_bits;
    unsigned z_swizzle;
-   unsigned chan;
-   unsigned padding_left, padding_right;
 
    assert(format_desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS);
    assert(format_desc->block.width == 1);
@@ -365,25 +363,14 @@ get_z_shift_and_mask(const struct util_format_description *format_desc,
       return FALSE;
 
    *width = format_desc->channel[z_swizzle].size;
+   *shift = format_desc->channel[z_swizzle].shift;
 
-   padding_right = 0;
-   for (chan = 0; chan < z_swizzle; ++chan)
-      padding_right += format_desc->channel[chan].size;
-
-   padding_left =
-      total_bits - (padding_right + *width);
-
-   if (padding_left || padding_right) {
-      unsigned long long mask_left = (1ULL << (total_bits - padding_left)) - 1;
-      unsigned long long mask_right = (1ULL << (padding_right)) - 1;
-      *mask = mask_left ^ mask_right;
-   }
-   else {
+   if (*width == total_bits) {
       *mask = 0xffffffff;
+   } else {
+      *mask = ((1 << *width) - 1) << *shift;
    }
 
-   *shift = padding_right;
-
    return TRUE;
 }
 
@@ -398,7 +385,7 @@ get_s_shift_and_mask(const struct util_format_description *format_desc,
                      unsigned *shift, unsigned *mask)
 {
    unsigned s_swizzle;
-   unsigned chan, sz;
+   unsigned sz;
 
    s_swizzle = format_desc->swizzle[1];
 
@@ -407,16 +394,14 @@ get_s_shift_and_mask(const struct util_format_description *format_desc,
 
    /* just special case 64bit d/s format */
    if (format_desc->block.bits > 32) {
+      /* XXX big-endian? */
       assert(format_desc->format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT);
       *shift = 0;
       *mask = 0xff;
       return TRUE;
    }
 
-   *shift = 0;
-   for (chan = 0; chan < s_swizzle; chan++)
-      *shift += format_desc->channel[chan].size;
-
+   *shift = format_desc->channel[s_swizzle].shift;
    sz = format_desc->channel[s_swizzle].size;
    *mask = (1U << sz) - 1U;
 
diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c
index fc2ba5e..38e9fc7 100644
--- a/src/gallium/drivers/llvmpipe/lp_state_fs.c
+++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c
@@ -1015,12 +1015,17 @@ convert_to_blend_type(struct gallivm_state *gallivm,
    for (i = 0; i < num_srcs; ++i) {
       LLVMValueRef chans[4];
       LLVMValueRef res = NULL;
-      unsigned sa = 0;
 
       dst[i] = LLVMBuildZExt(builder, src[i], lp_build_vec_type(gallivm, src_type), "");
 
       for (j = 0; j < src_fmt->nr_channels; ++j) {
          unsigned mask = 0;
+         unsigned sa = src_fmt->channel[j].shift;
+#ifdef PIPE_ARCH_LITTLE_ENDIAN
+         unsigned from_lsb = j;
+#else
+         unsigned from_lsb = src_fmt->nr_channels - j - 1;
+#endif
 
          for (k = 0; k < src_fmt->channel[j].size; ++k) {
             mask |= 1 << k;
@@ -1046,11 +1051,9 @@ convert_to_blend_type(struct gallivm_state *gallivm,
          /* Insert bits into correct position */
          chans[j] = LLVMBuildShl(builder,
                                  chans[j],
-                                 lp_build_const_int_vec(gallivm, src_type, j * blend_type.width),
+                                 lp_build_const_int_vec(gallivm, src_type, from_lsb * blend_type.width),
                                  "");
 
-         sa += src_fmt->channel[j].size;
-
          if (j == 0) {
             res = chans[j];
          } else {
@@ -1166,12 +1169,17 @@ convert_from_blend_type(struct gallivm_state *gallivm,
    for (i = 0; i < num_srcs; ++i) {
       LLVMValueRef chans[4];
       LLVMValueRef res = NULL;
-      unsigned sa = 0;
 
       dst[i] = LLVMBuildBitCast(builder, src[i], lp_build_vec_type(gallivm, src_type), "");
 
       for (j = 0; j < src_fmt->nr_channels; ++j) {
          unsigned mask = 0;
+         unsigned sa = src_fmt->channel[j].shift;
+#ifdef PIPE_ARCH_LITTLE_ENDIAN
+         unsigned from_lsb = j;
+#else
+         unsigned from_lsb = src_fmt->nr_channels - j - 1;
+#endif
 
          assert(blend_type.width > src_fmt->channel[j].size);
 
@@ -1182,7 +1190,7 @@ convert_from_blend_type(struct gallivm_state *gallivm,
          /* Extract bits */
          chans[j] = LLVMBuildLShr(builder,
                                   dst[i],
-                                  lp_build_const_int_vec(gallivm, src_type, j * blend_type.width),
+                                  lp_build_const_int_vec(gallivm, src_type, from_lsb * blend_type.width),
                                   "");
 
          chans[j] = LLVMBuildAnd(builder,
diff --git a/src/gallium/include/pipe/p_format.h b/src/gallium/include/pipe/p_format.h
index 098b25b..f181621 100644
--- a/src/gallium/include/pipe/p_format.h
+++ b/src/gallium/include/pipe/p_format.h
@@ -33,6 +33,7 @@
 extern "C" {
 #endif
 
+#include "p_config.h"
 
 enum pipe_type {
    PIPE_TYPE_UNORM = 0,
@@ -343,6 +344,27 @@ enum pipe_format {
    PIPE_FORMAT_COUNT
 };
 
+#if defined(PIPE_ARCH_LITTLE_ENDIAN)
+#define PIPE_FORMAT_RGBA8888_UNORM PIPE_FORMAT_R8G8B8A8_UNORM
+#define PIPE_FORMAT_RGBX8888_UNORM PIPE_FORMAT_R8G8B8X8_UNORM
+#define PIPE_FORMAT_BGRA8888_UNORM PIPE_FORMAT_B8G8R8A8_UNORM
+#define PIPE_FORMAT_BGRX8888_UNORM PIPE_FORMAT_B8G8R8X8_UNORM
+#define PIPE_FORMAT_ARGB8888_UNORM PIPE_FORMAT_A8R8G8B8_UNORM
+#define PIPE_FORMAT_XRGB8888_UNORM PIPE_FORMAT_X8R8G8B8_UNORM
+#define PIPE_FORMAT_ABGR8888_UNORM PIPE_FORMAT_A8B8G8R8_UNORM
+#define PIPE_FORMAT_XBGR8888_UNORM PIPE_FORMAT_X8B8G8R8_UNORM
+#elif defined(PIPE_ARCH_BIG_ENDIAN)
+#define PIPE_FORMAT_ABGR8888_UNORM PIPE_FORMAT_R8G8B8A8_UNORM
+#define PIPE_FORMAT_XBGR8888_UNORM PIPE_FORMAT_R8G8B8X8_UNORM
+#define PIPE_FORMAT_XRGB8888_UNORM PIPE_FORMAT_B8G8R8X8_UNORM
+#define PIPE_FORMAT_ARGB8888_UNORM PIPE_FORMAT_B8G8R8A8_UNORM
+#define PIPE_FORMAT_XRGB8888_UNORM PIPE_FORMAT_B8G8R8X8_UNORM
+#define PIPE_FORMAT_BGRA8888_UNORM PIPE_FORMAT_A8R8G8B8_UNORM
+#define PIPE_FORMAT_BGRX8888_UNORM PIPE_FORMAT_X8R8G8B8_UNORM
+#define PIPE_FORMAT_RGBA8888_UNORM PIPE_FORMAT_A8B8G8R8_UNORM
+#define PIPE_FORMAT_RGBX8888_UNORM PIPE_FORMAT_X8B8G8R8_UNORM
+#endif
+
 enum pipe_video_chroma_format
 {
    PIPE_VIDEO_CHROMA_FORMAT_420,
diff --git a/src/gallium/state_trackers/dri/common/dri_screen.c b/src/gallium/state_trackers/dri/common/dri_screen.c
index e09fe1d..6a037c4 100644
--- a/src/gallium/state_trackers/dri/common/dri_screen.c
+++ b/src/gallium/state_trackers/dri/common/dri_screen.c
@@ -87,8 +87,8 @@ dri_fill_in_modes(struct dri_screen *screen)
       MESA_FORMAT_RGB565,
    };
    static const enum pipe_format pipe_formats[3] = {
-      PIPE_FORMAT_B8G8R8A8_UNORM,
-      PIPE_FORMAT_B8G8R8X8_UNORM,
+      PIPE_FORMAT_BGRA8888_UNORM,
+      PIPE_FORMAT_BGRX8888_UNORM,
       PIPE_FORMAT_B5G6R5_UNORM,
    };
    gl_format format;
@@ -250,9 +250,9 @@ dri_fill_st_visual(struct st_visual *stvis, struct dri_screen *screen,
 
    if (mode->redBits == 8) {
       if (mode->alphaBits == 8)
-         stvis->color_format = PIPE_FORMAT_B8G8R8A8_UNORM;
+         stvis->color_format = PIPE_FORMAT_BGRA8888_UNORM;
       else
-         stvis->color_format = PIPE_FORMAT_B8G8R8X8_UNORM;
+         stvis->color_format = PIPE_FORMAT_BGRX8888_UNORM;
    } else {
       stvis->color_format = PIPE_FORMAT_B5G6R5_UNORM;
    }
diff --git a/src/gallium/state_trackers/gbm/gbm_drm.c b/src/gallium/state_trackers/gbm/gbm_drm.c
index 8490480..725f12f 100644
--- a/src/gallium/state_trackers/gbm/gbm_drm.c
+++ b/src/gallium/state_trackers/gbm/gbm_drm.c
@@ -45,9 +45,9 @@ gbm_format_to_gallium(enum gbm_bo_format format)
 {
    switch (format) {
    case GBM_BO_FORMAT_XRGB8888:
-      return PIPE_FORMAT_B8G8R8X8_UNORM;
+      return PIPE_FORMAT_BGRX8888_UNORM;
    case GBM_BO_FORMAT_ARGB8888:
-      return PIPE_FORMAT_B8G8R8A8_UNORM;
+      return PIPE_FORMAT_BGRA8888_UNORM;
    default:
       return PIPE_FORMAT_NONE;
    }
@@ -145,10 +145,10 @@ gbm_gallium_drm_bo_import(struct gbm_device *gbm,
    bo->base.base.height = resource->height0;
 
    switch (resource->format) {
-   case PIPE_FORMAT_B8G8R8X8_UNORM:
+   case PIPE_FORMAT_BGRX8888_UNORM:
       bo->base.base.format = GBM_BO_FORMAT_XRGB8888;
       break;
-   case PIPE_FORMAT_B8G8R8A8_UNORM:
+   case PIPE_FORMAT_BGRA8888_UNORM:
       bo->base.base.format = GBM_BO_FORMAT_ARGB8888;
       break;
    default:
diff --git a/src/gallium/state_trackers/glx/xlib/xm_api.c b/src/gallium/state_trackers/glx/xlib/xm_api.c
index daf5ad4..bb2dd8e 100644
--- a/src/gallium/state_trackers/glx/xlib/xm_api.c
+++ b/src/gallium/state_trackers/glx/xlib/xm_api.c
@@ -328,10 +328,10 @@ choose_pixel_format(XMesaVisual v)
        && v->BitsPerPixel == 32) {
       if (native_byte_order) {
          /* no byteswapping needed */
-         return PIPE_FORMAT_R8G8B8A8_UNORM;
+         return PIPE_FORMAT_RGBA8888_UNORM;
       }
       else {
-         return PIPE_FORMAT_A8B8G8R8_UNORM;
+         return PIPE_FORMAT_ABGR8888_UNORM;
       }
    }
    else if (   GET_REDMASK(v)   == 0xff0000
@@ -340,10 +340,10 @@ choose_pixel_format(XMesaVisual v)
             && v->BitsPerPixel == 32) {
       if (native_byte_order) {
          /* no byteswapping needed */
-         return PIPE_FORMAT_B8G8R8A8_UNORM;
+         return PIPE_FORMAT_BGRA8888_UNORM;
       }
       else {
-         return PIPE_FORMAT_A8R8G8B8_UNORM;
+         return PIPE_FORMAT_ARGB8888_UNORM;
       }
    }
    else if (   GET_REDMASK(v)   == 0x0000ff00
@@ -352,10 +352,10 @@ choose_pixel_format(XMesaVisual v)
             && v->BitsPerPixel == 32) {
       if (native_byte_order) {
          /* no byteswapping needed */
-         return PIPE_FORMAT_A8R8G8B8_UNORM;
+         return PIPE_FORMAT_ARGB8888_UNORM;
       }
       else {
-         return PIPE_FORMAT_B8G8R8A8_UNORM;
+         return PIPE_FORMAT_BGRA8888_UNORM;
       }
    }
    else if (   GET_REDMASK(v)   == 0xf800
diff --git a/src/gallium/targets/graw-xlib/graw_xlib.c b/src/gallium/targets/graw-xlib/graw_xlib.c
index 2827747..148837c 100644
--- a/src/gallium/targets/graw-xlib/graw_xlib.c
+++ b/src/gallium/targets/graw-xlib/graw_xlib.c
@@ -89,13 +89,13 @@ graw_create_window_and_screen( int x,
    if (visinfo->red_mask == 0xff0000 &&
        visinfo->green_mask == 0xff00 &&
        visinfo->blue_mask == 0xff) {
-      if (format != PIPE_FORMAT_B8G8R8A8_UNORM)
+      if (format != PIPE_FORMAT_BGRA8888_UNORM)
          goto fail;
    }
    else if (visinfo->red_mask == 0xff &&
             visinfo->green_mask == 0xff00 &&
             visinfo->blue_mask == 0xff0000) {
-      if (format != PIPE_FORMAT_R8G8B8A8_UNORM)
+      if (format != PIPE_FORMAT_RGBA8888_UNORM)
          goto fail;
    }
    else {
diff --git a/src/gallium/tests/graw/clear.c b/src/gallium/tests/graw/clear.c
index 6afdf40..77c59db 100644
--- a/src/gallium/tests/graw/clear.c
+++ b/src/gallium/tests/graw/clear.c
@@ -10,8 +10,8 @@
 #include "pipe/p_defines.h"
 
 enum pipe_format formats[] = {
-   PIPE_FORMAT_R8G8B8A8_UNORM,
-   PIPE_FORMAT_B8G8R8A8_UNORM,
+   PIPE_FORMAT_RGBA8888_UNORM,
+   PIPE_FORMAT_BGRA8888_UNORM,
    PIPE_FORMAT_NONE
 };
 
diff --git a/src/gallium/tests/graw/fs-test.c b/src/gallium/tests/graw/fs-test.c
index 685be92..38a2c4b 100644
--- a/src/gallium/tests/graw/fs-test.c
+++ b/src/gallium/tests/graw/fs-test.c
@@ -31,8 +31,8 @@ static void usage(char *name)
 
 
 enum pipe_format formats[] = {
-   PIPE_FORMAT_R8G8B8A8_UNORM,
-   PIPE_FORMAT_B8G8R8A8_UNORM,
+   PIPE_FORMAT_RGBA8888_UNORM,
+   PIPE_FORMAT_BGRA8888_UNORM,
    PIPE_FORMAT_NONE
 };
 
diff --git a/src/gallium/tests/graw/graw_util.h b/src/gallium/tests/graw/graw_util.h
index 84456b4..8557285 100644
--- a/src/gallium/tests/graw/graw_util.h
+++ b/src/gallium/tests/graw/graw_util.h
@@ -32,8 +32,8 @@ graw_util_create_window(struct graw_info *info,
                         int num_cbufs, bool zstencil_buf)
 {
    static const enum pipe_format formats[] = {
-      PIPE_FORMAT_R8G8B8A8_UNORM,
-      PIPE_FORMAT_B8G8R8A8_UNORM,
+      PIPE_FORMAT_RGBA8888_UNORM,
+      PIPE_FORMAT_BGRA8888_UNORM,
       PIPE_FORMAT_NONE
    };
    enum pipe_format format;
@@ -226,7 +226,7 @@ graw_util_create_tex2d(const struct graw_info *info,
    struct pipe_box box;
 
    temp.target = PIPE_TEXTURE_2D;
-   temp.format = PIPE_FORMAT_B8G8R8A8_UNORM;
+   temp.format = format;
    temp.width0 = width;
    temp.height0 = height;
    temp.depth0 = 1;
diff --git a/src/gallium/tests/graw/quad-sample.c b/src/gallium/tests/graw/quad-sample.c
index 9100272..969ffa7 100644
--- a/src/gallium/tests/graw/quad-sample.c
+++ b/src/gallium/tests/graw/quad-sample.c
@@ -18,8 +18,8 @@
 #include <stdio.h>
 
 enum pipe_format formats[] = {
-   PIPE_FORMAT_R8G8B8A8_UNORM,
-   PIPE_FORMAT_B8G8R8A8_UNORM,
+   PIPE_FORMAT_RGBA8888_UNORM,
+   PIPE_FORMAT_BGRA8888_UNORM,
    PIPE_FORMAT_NONE
 };
 
diff --git a/src/gallium/tests/graw/shader-leak.c b/src/gallium/tests/graw/shader-leak.c
index 014e0cc..3076210 100644
--- a/src/gallium/tests/graw/shader-leak.c
+++ b/src/gallium/tests/graw/shader-leak.c
@@ -18,8 +18,8 @@ static int num_iters = 100;
 
 
 enum pipe_format formats[] = {
-   PIPE_FORMAT_R8G8B8A8_UNORM,
-   PIPE_FORMAT_B8G8R8A8_UNORM,
+   PIPE_FORMAT_RGBA8888_UNORM,
+   PIPE_FORMAT_BGR8888_UNORM,
    PIPE_FORMAT_NONE
 };
 
diff --git a/src/gallium/tests/graw/tri-gs.c b/src/gallium/tests/graw/tri-gs.c
index 535825e..37323aa 100644
--- a/src/gallium/tests/graw/tri-gs.c
+++ b/src/gallium/tests/graw/tri-gs.c
@@ -14,8 +14,8 @@
 #include "util/u_inlines.h"
 
 enum pipe_format formats[] = {
-   PIPE_FORMAT_R8G8B8A8_UNORM,
-   PIPE_FORMAT_B8G8R8A8_UNORM,
+   PIPE_FORMAT_RGBA8888_UNORM,
+   PIPE_FORMAT_BGRA8888_UNORM,
    PIPE_FORMAT_NONE
 };
 
diff --git a/src/gallium/tests/graw/tri-instanced.c b/src/gallium/tests/graw/tri-instanced.c
index d00e7e9..f84463d 100644
--- a/src/gallium/tests/graw/tri-instanced.c
+++ b/src/gallium/tests/graw/tri-instanced.c
@@ -17,8 +17,8 @@
 
 
 enum pipe_format formats[] = {
-   PIPE_FORMAT_R8G8B8A8_UNORM,
-   PIPE_FORMAT_B8G8R8A8_UNORM,
+   PIPE_FORMAT_RGBA8888_UNORM,
+   PIPE_FORMAT_BGRA8888_UNORM,
    PIPE_FORMAT_NONE
 };
 
diff --git a/src/gallium/tests/graw/vs-test.c b/src/gallium/tests/graw/vs-test.c
index bfb40be..5a7d0a0 100644
--- a/src/gallium/tests/graw/vs-test.c
+++ b/src/gallium/tests/graw/vs-test.c
@@ -32,8 +32,8 @@ static void usage(char *name)
 
 
 enum pipe_format formats[] = {
-   PIPE_FORMAT_R8G8B8A8_UNORM,
-   PIPE_FORMAT_B8G8R8A8_UNORM,
+   PIPE_FORMAT_RGBA8888_UNORM,
+   PIPE_FORMAT_BGRA8888_UNORM,
    PIPE_FORMAT_NONE
 };