Hi All,
On Wed, Nov 10, 2010 at 10:17 PM, Geert Uytterhoeven
<geert@xxxxxxxxxxxxxx> wrote:
Hi Michael,
On Wed, Nov 10, 2010 at 08:08, Michael Schmitz
<schmitzmic@xxxxxxxxxxxxxx> wrote:
seen this before?
Not this particular one...
LD .tmp_vmlinux1
fs/built-in.o: In function `alignfile':
binfmt_elf.c:(.text+0x307bc): undefined reference to `__divdi3'
fs/built-in.o: In function `elf_core_dump':
binfmt_elf.c:(.text+0x3139c): undefined reference to `__divdi3'
make: *** [.tmp_vmlinux1] Error 1
Apparently caused by the roundup macro in include/linux/kernel.h:
#define roundup(x, y) ( \
{ \
typeof(y) __y = y; \
(((x) + (__y - 1)) / __y) * __y; \
}
Toolchain issue?
... but depending on the version of gcc, we sometimes need things like
this, cfr.
arch/m68k/lib/. The fix is to add a __divdi3 there.
See attached patch, based on libgcc2.c code using asm inlines - please
review. The UDIV_NEEDS_NORMALIZATION value I am not sure about,
actually.
This patch makes the kernel compile with my toolchain but I am
uncertain on how exactly to test this code path. Write a test case?
Try and force a core dump? The kernel runs OK on my Sarge and Etch
disk images, that is. as good as any previous 2.3.35 or 2.3.36 ones.
Genrtc (which is obsolete) or rtc-generic (to be obsoleted by rtc-cmos
on Atari ;-)?
rtc-generic for now, with a view towards adding whatever is needed to
use rtc-cmos after.
rtc-generic tested OK at least on ARAnyM so far.
Cheers,
Michael
arch/m68k/lib/Makefile | 2 +-
arch/m68k/lib/divdi3.c | 554 ++++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 555 insertions(+), 1 deletions(-)
Signed-off-by: Michael Schmitz <schmitz@xxxxxxxxxx>
--
diff --git a/arch/m68k/lib/Makefile b/arch/m68k/lib/Makefile
index af9abf8..161297f 100644
--- a/arch/m68k/lib/Makefile
+++ b/arch/m68k/lib/Makefile
@@ -2,5 +2,5 @@
# Makefile for m68k-specific library files..
#
-lib-y := ashldi3.o ashrdi3.o lshrdi3.o muldi3.o \
+lib-y := ashldi3.o ashrdi3.o lshrdi3.o muldi3.o divdi3.o \
checksum.o string.o uaccess.o
diff --git a/arch/m68k/lib/divdi3.c b/arch/m68k/lib/divdi3.c
new file mode 100644
index 0000000..178c210
--- /dev/null
+++ b/arch/m68k/lib/divdi3.c
@@ -0,0 +1,554 @@
+/* divdi3.c extracted from gcc-2.95.3/libgcc2.c which is: */
+/* Copyright (C) 1989, 92-98, 1999 Free Software Foundation, Inc.
+
+This file is part of GNU CC.
+
+GNU CC is free software; you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation; either version 2, or (at your option)
+any later version.
+
+GNU CC 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 GNU CC; see the file COPYING. If not, write to
+the Free Software Foundation, 59 Temple Place - Suite 330,
+Boston, MA 02111-1307, USA. */
+
+/* As a special exception, if you link this library with other files,
+ some of which are compiled with GCC, to produce an executable,
+ this library does not by itself cause the resulting executable
+ to be covered by the GNU General Public License.
+ This exception does not however invalidate any other reasons why
+ the executable file might be covered by the GNU General Public License. */
+
+/* It is incorrect to include config.h here, because this file is being
+ compiled for the target, and hence definitions concerning only the host
+ do not apply. */
+
+
+/* longlong.h -- definitions for mixed size 32/64 bit arithmetic.
+ Copyright (C) 1991, 92, 94, 95, 96, 1997, 1998 Free Software Foundation, Inc.
+
+ This definition file is free software; you can redistribute it
+ and/or modify it under the terms of the GNU General Public
+ License as published by the Free Software Foundation; either
+ version 2, or (at your option) any later version.
+
+ This definition file 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., 59 Temple Place - Suite 330,
+ Boston, MA 02111-1307, USA. */
+
+#ifndef SI_TYPE_SIZE
+#define SI_TYPE_SIZE 32
+#endif
+
+#define UDIV_NEEDS_NORMALIZATION 0
+
+#define __BITS4 (SI_TYPE_SIZE / 4)
+#define __ll_B (1L << (SI_TYPE_SIZE / 2))
+#define __ll_lowpart(t) ((USItype) (t) % __ll_B)
+#define __ll_highpart(t) ((USItype) (t) / __ll_B)
+
+typedef unsigned int UQItype __attribute__ ((mode (QI)));
+typedef int SItype __attribute__ ((mode (SI)));
+typedef unsigned int USItype __attribute__ ((mode (SI)));
+typedef int DItype __attribute__ ((mode (DI)));
+typedef unsigned int UDItype __attribute__ ((mode (DI)));
+
+typedef float SFtype __attribute__ ((mode (SF)));
+typedef float DFtype __attribute__ ((mode (DF)));
+
+typedef int word_type __attribute__ ((mode (__word__)));
+
+/* Define auxiliary asm macros.
+
+ 1) umul_ppmm(high_prod, low_prod, multipler, multiplicand)
+ multiplies two USItype integers MULTIPLER and MULTIPLICAND,
+ and generates a two-part USItype product in HIGH_PROD and
+ LOW_PROD.
+
+ 2) __umulsidi3(a,b) multiplies two USItype integers A and B,
+ and returns a UDItype product. This is just a variant of umul_ppmm.
+
+ 3) udiv_qrnnd(quotient, remainder, high_numerator, low_numerator,
+ denominator) divides a two-word unsigned integer, composed by the
+ integers HIGH_NUMERATOR and LOW_NUMERATOR, by DENOMINATOR and
+ places the quotient in QUOTIENT and the remainder in REMAINDER.
+ HIGH_NUMERATOR must be less than DENOMINATOR for correct operation.
+ If, in addition, the most significant bit of DENOMINATOR must be 1,
+ then the pre-processor symbol UDIV_NEEDS_NORMALIZATION is defined to 1.
+
+ 4) sdiv_qrnnd(quotient, remainder, high_numerator, low_numerator,
+ denominator). Like udiv_qrnnd but the numbers are signed. The
+ quotient is rounded towards 0.
+
+ 5) count_leading_zeros(count, x) counts the number of zero-bits from
+ the msb to the first non-zero bit. This is the number of steps X
+ needs to be shifted left to set the msb. Undefined for X == 0.
+
+ 6) add_ssaaaa(high_sum, low_sum, high_addend_1, low_addend_1,
+ high_addend_2, low_addend_2) adds two two-word unsigned integers,
+ composed by HIGH_ADDEND_1 and LOW_ADDEND_1, and HIGH_ADDEND_2 and
+ LOW_ADDEND_2 respectively. The result is placed in HIGH_SUM and
+ LOW_SUM. Overflow (i.e. carry out) is not stored anywhere, and is
+ lost.
+
+ 7) sub_ddmmss(high_difference, low_difference, high_minuend,
+ low_minuend, high_subtrahend, low_subtrahend) subtracts two
+ two-word unsigned integers, composed by HIGH_MINUEND_1 and
+ LOW_MINUEND_1, and HIGH_SUBTRAHEND_2 and LOW_SUBTRAHEND_2
+ respectively. The result is placed in HIGH_DIFFERENCE and
+ LOW_DIFFERENCE. Overflow (i.e. carry out) is not stored anywhere,
+ and is lost.
+
+ If any of these macros are left undefined for a particular CPU,
+ C macros are used. */
+
+/* The CPUs come in alphabetical order below.
+
+ Please add support for more CPUs here, or improve the current support
+ for the CPUs below!
+ (E.g. WE32100, IBM360.) */
+
+#if 1
+#define sub_ddmmss(sh, sl, ah, al, bh, bl) \
+ __asm__ ("\n" \
+ " sub%.l %5,%1\n" \
+ " subx%.l %3,%0" \
+ : "=d" ((USItype) (sh)), \
+ "=&d" ((USItype) (sl)) \
+ : "0" ((USItype) (ah)), \
+ "d" ((USItype) (bh)), \
+ "1" ((USItype) (al)), \
+ "g" ((USItype) (bl)))
+#else
+#define sub_ddmmss(sh, sl, ah, al, bh, bl) \
+ do { \
+ USItype __x; \
+ __x = (al) - (bl); \
+ (sh) = (ah) - (bh) - (__x > (al)); \
+ (sl) = __x; \
+ } while (0)
+#endif
+
+/* The '020, '030, '040 and CPU32 have 32x32->64 and 64/32->32q-32r. */
+
+#define umul_ppmm(w1, w0, u, v) \
+ __asm__ ("mulu%.l %3,%1:%0" \
+ : "=d" ((USItype) (w0)), \
+ "=d" ((USItype) (w1)) \
+ : "%0" ((USItype) (u)), \
+ "dmi" ((USItype) (v)))
+#define UMUL_TIME 45
+#define udiv_qrnnd(q, r, n1, n0, d) \
+ __asm__ ("divu%.l %4,%1:%0" \
+ : "=d" ((USItype) (q)), \
+ "=d" ((USItype) (r)) \
+ : "0" ((USItype) (n0)), \
+ "1" ((USItype) (n1)), \
+ "dmi" ((USItype) (d)))
+#define UDIV_TIME 90
+#define sdiv_qrnnd(q, r, n1, n0, d) \
+ __asm__ ("divs%.l %4,%1:%0" \
+ : "=d" ((USItype) (q)), \
+ "=d" ((USItype) (r)) \
+ : "0" ((USItype) (n0)), \
+ "1" ((USItype) (n1)), \
+ "dmi" ((USItype) (d)))
+
+/* The '020, '030, '040 and '060 have bitfield insns. */
+#define count_leading_zeros(count, x) \
+ __asm__ ("bfffo %1{%b2:%b2},%0" \
+ : "=d" ((USItype) (count)) \
+ : "od" ((USItype) (x)), "n" (0))
+
+struct DIstruct {SItype high, low;};
+
+typedef union
+{
+ struct DIstruct s;
+ DItype ll;
+} DIunion;
+
+
+static inline
+DItype
+__negdi2 (DItype u)
+{
+ DIunion w;
+ DIunion uu;
+
+ uu.ll = u;
+
+ w.s.low = -uu.s.low;
+ w.s.high = -uu.s.high - ((USItype) w.s.low > 0);
+
+ return w.ll;
+}
+
+USItype
+__udiv_w_sdiv (USItype *rp, USItype a1, USItype a0, USItype d)
+{
+ USItype q, r;
+ USItype c0, c1, b1;
+
+ if ((SItype) d >= 0)
+ {
+ if (a1 < d - a1 - (a0 >> (SI_TYPE_SIZE - 1)))
+ {
+ /* dividend, divisor, and quotient are nonnegative */
+ sdiv_qrnnd (q, r, a1, a0, d);
+ }
+ else
+ {
+ /* Compute c1*2^32 + c0 = a1*2^32 + a0 - 2^31*d */
+ sub_ddmmss (c1, c0, a1, a0, d >> 1, d << (SI_TYPE_SIZE - 1));
+ /* Divide (c1*2^32 + c0) by d */
+ sdiv_qrnnd (q, r, c1, c0, d);
+ /* Add 2^31 to quotient */
+ q += (USItype) 1 << (SI_TYPE_SIZE - 1);
+ }
+ }
+ else
+ {
+ b1 = d >> 1; /* d/2, between 2^30 and 2^31 - 1 */
+ c1 = a1 >> 1; /* A/2 */
+ c0 = (a1 << (SI_TYPE_SIZE - 1)) + (a0 >> 1);
+
+ if (a1 < b1) /* A < 2^32*b1, so A/2 < 2^31*b1 */
+ {
+ sdiv_qrnnd (q, r, c1, c0, b1); /* (A/2) / (d/2) */
+
+ r = 2*r + (a0 & 1); /* Remainder from A/(2*b1) */
+ if ((d & 1) != 0)
+ {
+ if (r >= q)
+ r = r - q;
+ else if (q - r <= d)
+ {
+ r = r - q + d;
+ q--;
+ }
+ else
+ {
+ r = r - q + 2*d;
+ q -= 2;
+ }
+ }
+ }
+ else if (c1 < b1) /* So 2^31 <= (A/2)/b1 < 2^32 */
+ {
+ c1 = (b1 - 1) - c1;
+ c0 = ~c0; /* logical NOT */
+
+ sdiv_qrnnd (q, r, c1, c0, b1); /* (A/2) / (d/2) */
+
+ q = ~q; /* (A/2)/b1 */
+ r = (b1 - 1) - r;
+
+ r = 2*r + (a0 & 1); /* A/(2*b1) */
+
+ if ((d & 1) != 0)
+ {
+ if (r >= q)
+ r = r - q;
+ else if (q - r <= d)
+ {
+ r = r - q + d;
+ q--;
+ }
+ else
+ {
+ r = r - q + 2*d;
+ q -= 2;
+ }
+ }
+ }
+ else /* Implies c1 = b1 */
+ { /* Hence a1 = d - 1 = 2*b1 - 1 */
+ if (a0 >= -d)
+ {
+ q = -1;
+ r = a0 + d;
+ }
+ else
+ {
+ q = -2;
+ r = a0 + 2*d;
+ }
+ }
+ }
+
+ *rp = r;
+ return q;
+}
+
+static const UQItype __clz_tab[] =
+{
+ 0,1,2,2,3,3,3,3,4,4,4,4,4,4,4,4,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,5,
+ 6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,6,
+ 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
+ 7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
+ 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
+};
+
+static inline
+UDItype
+__udivmoddi4 (UDItype n, UDItype d, UDItype *rp)
+{
+ DIunion ww;
+ DIunion nn, dd;
+ DIunion rr;
+ USItype d0, d1, n0, n1, n2;
+ USItype q0, q1;
+ USItype b, bm;
+
+ nn.ll = n;
+ dd.ll = d;
+
+ d0 = dd.s.low;
+ d1 = dd.s.high;
+ n0 = nn.s.low;
+ n1 = nn.s.high;
+
+#if !UDIV_NEEDS_NORMALIZATION
+ if (d1 == 0)
+ {
+ if (d0 > n1)
+ {
+ /* 0q = nn / 0D */
+
+ udiv_qrnnd (q0, n0, n1, n0, d0);
+ q1 = 0;
+
+ /* Remainder in n0. */
+ }
+ else
+ {
+ /* qq = NN / 0d */
+
+ if (d0 == 0)
+ d0 = 1 / d0; /* Divide intentionally by zero. */
+
+ udiv_qrnnd (q1, n1, 0, n1, d0);
+ udiv_qrnnd (q0, n0, n1, n0, d0);
+
+ /* Remainder in n0. */
+ }
+
+ if (rp != 0)
+ {
+ rr.s.low = n0;
+ rr.s.high = 0;
+ *rp = rr.ll;
+ }
+ }
+
+#else /* UDIV_NEEDS_NORMALIZATION */
+
+ if (d1 == 0)
+ {
+ if (d0 > n1)
+ {
+ /* 0q = nn / 0D */
+
+ count_leading_zeros (bm, d0);
+
+ if (bm != 0)
+ {
+ /* Normalize, i.e. make the most significant bit of the
+ denominator set. */
+
+ d0 = d0 << bm;
+ n1 = (n1 << bm) | (n0 >> (SI_TYPE_SIZE - bm));
+ n0 = n0 << bm;
+ }
+
+ udiv_qrnnd (q0, n0, n1, n0, d0);
+ q1 = 0;
+
+ /* Remainder in n0 >> bm. */
+ }
+ else
+ {
+ /* qq = NN / 0d */
+
+ if (d0 == 0)
+ d0 = 1 / d0; /* Divide intentionally by zero. */
+
+ count_leading_zeros (bm, d0);
+
+ if (bm == 0)
+ {
+ /* From (n1 >= d0) /\ (the most significant bit of d0 is set),
+ conclude (the most significant bit of n1 is set) /\ (the
+ leading quotient digit q1 = 1).
+
+ This special case is necessary, not an optimization.
+ (Shifts counts of SI_TYPE_SIZE are undefined.) */
+
+ n1 -= d0;
+ q1 = 1;
+ }
+ else
+ {
+ /* Normalize. */
+
+ b = SI_TYPE_SIZE - bm;
+
+ d0 = d0 << bm;
+ n2 = n1 >> b;
+ n1 = (n1 << bm) | (n0 >> b);
+ n0 = n0 << bm;
+
+ udiv_qrnnd (q1, n1, n2, n1, d0);
+ }
+
+ /* n1 != d0... */
+
+ udiv_qrnnd (q0, n0, n1, n0, d0);
+
+ /* Remainder in n0 >> bm. */
+ }
+
+ if (rp != 0)
+ {
+ rr.s.low = n0 >> bm;
+ rr.s.high = 0;
+ *rp = rr.ll;
+ }
+ }
+#endif /* UDIV_NEEDS_NORMALIZATION */
+
+ else
+ {
+ if (d1 > n1)
+ {
+ /* 00 = nn / DD */
+
+ q0 = 0;
+ q1 = 0;
+
+ /* Remainder in n1n0. */
+ if (rp != 0)
+ {
+ rr.s.low = n0;
+ rr.s.high = n1;
+ *rp = rr.ll;
+ }
+ }
+ else
+ {
+ /* 0q = NN / dd */
+
+ count_leading_zeros (bm, d1);
+ if (bm == 0)
+ {
+ /* From (n1 >= d1) /\ (the most significant bit of d1 is set),
+ conclude (the most significant bit of n1 is set) /\ (the
+ quotient digit q0 = 0 or 1).
+
+ This special case is necessary, not an optimization. */
+
+ /* The condition on the next line takes advantage of that
+ n1 >= d1 (true due to program flow). */
+ if (n1 > d1 || n0 >= d0)
+ {
+ q0 = 1;
+ sub_ddmmss (n1, n0, n1, n0, d1, d0);
+ }
+ else
+ q0 = 0;
+
+ q1 = 0;
+
+ if (rp != 0)
+ {
+ rr.s.low = n0;
+ rr.s.high = n1;
+ *rp = rr.ll;
+ }
+ }
+ else
+ {
+ USItype m1, m0;
+ /* Normalize. */
+
+ b = SI_TYPE_SIZE - bm;
+
+ d1 = (d1 << bm) | (d0 >> b);
+ d0 = d0 << bm;
+ n2 = n1 >> b;
+ n1 = (n1 << bm) | (n0 >> b);
+ n0 = n0 << bm;
+
+ udiv_qrnnd (q0, n1, n2, n1, d1);
+ umul_ppmm (m1, m0, q0, d0);
+
+ if (m1 > n1 || (m1 == n1 && m0 > n0))
+ {
+ q0--;
+ sub_ddmmss (m1, m0, m1, m0, d1, d0);
+ }
+
+ q1 = 0;
+
+ /* Remainder in (n1n0 - m1m0) >> bm. */
+ if (rp != 0)
+ {
+ sub_ddmmss (n1, n0, n1, n0, m1, m0);
+ rr.s.low = (n1 << b) | (n0 >> bm);
+ rr.s.high = n1 >> bm;
+ *rp = rr.ll;
+ }
+ }
+ }
+ }
+
+ ww.s.low = q0;
+ ww.s.high = q1;
+ return ww.ll;
+}
+
+#if 0
+UDItype __udivmoddi4 ();
+#endif
+
+DItype
+__divdi3 (DItype u, DItype v)
+{
+ word_type c = 0;
+ DIunion uu, vv;
+ DItype w;
+
+ uu.ll = u;
+ vv.ll = v;
+
+ if (uu.s.high < 0)
+ c = ~c,
+ uu.ll = __negdi2 (uu.ll);
+ if (vv.s.high < 0)
+ c = ~c,
+ vv.ll = __negdi2 (vv.ll);
+
+ w = __udivmoddi4 (uu.ll, vv.ll, (UDItype *) 0);
+ if (c)
+ w = __negdi2 (w);
+
+ return w;
+}
+
+
