Merge Qualcom Hexagon architecture

This is the fifth version of the patchset (with one tiny whitespace fix)
to the Linux kernel to support the Qualcomm Hexagon architecture.

Between now and the next pull requests, Richard Kuo should have his key
signed, etc., and should be back on kernel.org.  In the meantime, this
got merged as a emailed patch-series.

* Hexagon: (36 commits)
  Add extra arch overrides to asm-generic/checksum.h
  Hexagon: Add self to MAINTAINERS
  Hexagon: Add basic stacktrace functionality for Hexagon architecture.
  Hexagon: Add configuration and makefiles for the Hexagon architecture.
  Hexagon: Comet platform support
  Hexagon: kgdb support files
  Hexagon: Add page-fault support.
  Hexagon: Add page table header files & etc.
  Hexagon: Add ioremap support
  Hexagon: Provide DMA implementation
  Hexagon: Implement basic TLB management routines for Hexagon.
  Hexagon: Implement basic cache-flush support
  Hexagon: Provide basic implementation and/or stubs for I/O routines.
  Hexagon: Add user access functions
  Hexagon: Add locking types and functions
  Hexagon: Add SMP support
  Hexagon: Provide basic debugging and system trap support.
  Hexagon: Add ptrace support
  Hexagon: Add time and timer functions
  Hexagon: Add interrupts
  ...

1 files changed, 139 insertions, 0 deletions

diff --git a/arch/hexagon/mm/strnlen_user.S b/arch/hexagon/mm/strnlen_user.S
new file mode 100644
index 0000000..5c6a16c
--- /dev/null
+++ b/arch/hexagon/mm/strnlen_user.S
@@ -0,0 +1,139 @@
+/*
+ * User string length functions for kernel
+ *
+ * Copyright (c) 2010-2011, Code Aurora Forum. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 and
+ * only version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ */
+
+#define isrc	r0
+#define max	r1	/*  Do not change!  */
+
+#define end	r2
+#define tmp1	r3
+
+#define obo	r6	/*  off-by-one  */
+#define start	r7
+#define mod8	r8
+#define dbuf    r15:14
+#define dcmp	r13:12
+
+/*
+ * The vector mask version of this turned out *really* badly.
+ * The hardware loop version also turned out *really* badly.
+ * Seems straight pointer arithmetic basically wins here.
+ */
+
+#define fname __strnlen_user
+
+	.text
+	.global fname
+	.type fname, @function
+	.p2align 5  /*  why?  */
+fname:
+	{
+		mod8 = and(isrc,#7);
+		end = add(isrc,max);
+		start = isrc;
+	}
+	{
+		P0 = cmp.eq(mod8,#0);
+		mod8 = and(end,#7);
+		dcmp = #0;
+		if (P0.new) jump:t dw_loop;	/*  fire up the oven  */
+	}
+
+alignment_loop:
+fail_1:	{
+		tmp1 = memb(start++#1);
+	}
+	{
+		P0 = cmp.eq(tmp1,#0);
+		if (P0.new) jump:nt exit_found;
+		P1 = cmp.gtu(end,start);
+		mod8 = and(start,#7);
+	}
+	{
+		if (!P1) jump exit_error;  /*  hit the end  */
+		P0 = cmp.eq(mod8,#0);
+	}
+	{
+		if (!P0) jump alignment_loop;
+	}
+
+
+
+dw_loop:
+fail_2:	{
+		dbuf = memd(start);
+		obo = add(start,#1);
+	}
+	{
+		P0 = vcmpb.eq(dbuf,dcmp);
+	}
+	{
+		tmp1 = P0;
+		P0 = cmp.gtu(end,start);
+	}
+	{
+		tmp1 = ct0(tmp1);
+		mod8 = and(end,#7);
+		if (!P0) jump end_check;
+	}
+	{
+		P0 = cmp.eq(tmp1,#32);
+		if (!P0.new) jump:nt exit_found;
+		if (!P0.new) start = add(obo,tmp1);
+	}
+	{
+		start = add(start,#8);
+		jump dw_loop;
+	}	/*  might be nice to combine these jumps...   */
+
+
+end_check:
+	{
+		P0 = cmp.gt(tmp1,mod8);
+		if (P0.new) jump:nt exit_error;	/*  neverfound!  */
+		start = add(obo,tmp1);
+	}
+
+exit_found:
+	{
+		R0 = sub(start,isrc);
+		jumpr R31;
+	}
+
+exit_error:
+	{
+		R0 = add(max,#1);
+		jumpr R31;
+	}
+
+	/*  Uh, what does the "fixup" return here?  */
+	.falign
+fix_1:
+	{
+		R0 = #0;
+		jumpr R31;
+	}
+
+	.size fname,.-fname
+
+
+.section __ex_table,"a"
+.long fail_1,fix_1
+.long fail_2,fix_1
+.previous