From: Zong Li <zongbox@gmail.com>
To: palmer@sifive.com, aou@eecs.berkeley.edu
Cc: vincentc@andestech.com, linux-riscv@lists.infradead.org,
linux-kernel@vger.kernel.org, zong@andestech.com,
Zong Li <zongbox@gmail.com>
Subject: [PATCH 3/5] lib: Add umoddi3 and udivmoddi4 of GCC library routines
Date: Tue, 18 Sep 2018 17:19:15 +0800 [thread overview]
Message-ID: <2ccc6c0758624dbb22f6fe451a3e476a000108ef.1537260207.git.zongbox@gmail.com> (raw)
In-Reply-To: <cover.1537260207.git.zongbox@gmail.com>
Add umoddi3 and udivmoddi4 support for 32-bit.
Signed-off-by: Zong Li <zong@andestech.com>
---
lib/Kconfig | 3 +
lib/Makefile | 1 +
lib/udivmoddi4.c | 291 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
lib/umoddi3.c | 16 +++
4 files changed, 311 insertions(+)
create mode 100644 lib/udivmoddi4.c
create mode 100644 lib/umoddi3.c
diff --git a/lib/Kconfig b/lib/Kconfig
index a3928d4..d82f206 100644
--- a/lib/Kconfig
+++ b/lib/Kconfig
@@ -621,3 +621,6 @@ config GENERIC_LIB_CMPDI2
config GENERIC_LIB_UCMPDI2
bool
+
+config GENERIC_LIB_UMODDI3
+ bool
diff --git a/lib/Makefile b/lib/Makefile
index ca3f7eb..51bf24d 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -271,3 +271,4 @@ obj-$(CONFIG_GENERIC_LIB_LSHRDI3) += lshrdi3.o
obj-$(CONFIG_GENERIC_LIB_MULDI3) += muldi3.o
obj-$(CONFIG_GENERIC_LIB_CMPDI2) += cmpdi2.o
obj-$(CONFIG_GENERIC_LIB_UCMPDI2) += ucmpdi2.o
+obj-$(CONFIG_GENERIC_LIB_UMODDI3) += umoddi3.o udivmoddi4.o
diff --git a/lib/udivmoddi4.c b/lib/udivmoddi4.c
new file mode 100644
index 0000000..69f2d36
--- /dev/null
+++ b/lib/udivmoddi4.c
@@ -0,0 +1,291 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/libgcc.h>
+
+#define count_leading_zeros(COUNT, X) ((COUNT) = __builtin_clz(X))
+
+#define W_TYPE_SIZE 32
+
+#define __ll_B ((unsigned long) 1 << (W_TYPE_SIZE / 2))
+#define __ll_lowpart(t) ((unsigned long) (t) & (__ll_B - 1))
+#define __ll_highpart(t) ((unsigned long) (t) >> (W_TYPE_SIZE / 2))
+
+/* If we still don't have umul_ppmm, define it using plain C. */
+#if !defined(umul_ppmm)
+#define umul_ppmm(w1, w0, u, v) \
+ do { \
+ unsigned long __x0, __x1, __x2, __x3; \
+ unsigned short __ul, __vl, __uh, __vh; \
+ \
+ __ul = __ll_lowpart(u); \
+ __uh = __ll_highpart(u); \
+ __vl = __ll_lowpart(v); \
+ __vh = __ll_highpart(v); \
+ \
+ __x0 = (unsigned long) __ul * __vl; \
+ __x1 = (unsigned long) __ul * __vh; \
+ __x2 = (unsigned long) __uh * __vl; \
+ __x3 = (unsigned long) __uh * __vh; \
+ \
+ __x1 += __ll_highpart(__x0); \
+ __x1 += __x2; \
+ if (__x1 < __x2) \
+ __x3 += __ll_B; \
+ \
+ (w1) = __x3 + __ll_highpart(__x1); \
+ (w0) = __ll_lowpart(__x1) * __ll_B + __ll_lowpart(__x0);\
+ } while (0)
+#endif
+
+#if !defined(sub_ddmmss)
+#define sub_ddmmss(sh, sl, ah, al, bh, bl) \
+ do { \
+ unsigned long __x; \
+ __x = (al) - (bl); \
+ (sh) = (ah) - (bh) - (__x > (al)); \
+ (sl) = __x; \
+ } while (0)
+#endif
+
+/* Define this unconditionally, so it can be used for debugging. */
+#define __udiv_qrnnd_c(q, r, n1, n0, d) \
+ do { \
+ unsigned long __d1, __d0, __q1, __q0; \
+ unsigned long __r1, __r0, __m; \
+ __d1 = __ll_highpart(d); \
+ __d0 = __ll_lowpart(d); \
+ \
+ __r1 = (n1) % __d1; \
+ __q1 = (n1) / __d1; \
+ __m = (unsigned long) __q1 * __d0; \
+ __r1 = __r1 * __ll_B | __ll_highpart(n0); \
+ if (__r1 < __m) { \
+ __q1--, __r1 += (d); \
+ if (__r1 >= (d)) \
+ if (__r1 < __m) \
+ __q1--, __r1 += (d); \
+ } \
+ __r1 -= __m; \
+ \
+ __r0 = __r1 % __d1; \
+ __q0 = __r1 / __d1; \
+ __m = (unsigned long) __q0 * __d0; \
+ __r0 = __r0 * __ll_B | __ll_lowpart(n0); \
+ if (__r0 < __m) { \
+ __q0--, __r0 += (d); \
+ if (__r0 >= (d)) \
+ if (__r0 < __m) \
+ __q0--, __r0 += (d); \
+ } \
+ __r0 -= __m; \
+ \
+ (q) = (unsigned long) __q1 * __ll_B | __q0; \
+ (r) = __r0; \
+ } while (0)
+
+/* If udiv_qrnnd was not defined for this processor, use __udiv_qrnnd_c. */
+#if !defined(udiv_qrnnd)
+#define UDIV_NEEDS_NORMALIZATION 1
+#define udiv_qrnnd __udiv_qrnnd_c
+#endif
+
+unsigned long long __udivmoddi4(unsigned long long u, unsigned long long v,
+ unsigned long long *rp)
+{
+ const DWunion nn = {.ll = u };
+ const DWunion dd = {.ll = v };
+ DWunion rr, ww;
+ unsigned long d0, d1, n0, n1, n2;
+ unsigned long q0 = 0, q1 = 0;
+ unsigned long b, bm;
+
+ 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)
+ /* Divide intentionally by zero. */
+ d0 = 1 / d0;
+
+ 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 >> (W_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)
+ /* Divide intentionally by zero. */
+ d0 = 1 / d0;
+
+ 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) /\ (theleading quotient
+ * digit q1 = 1).
+ *
+ * This special case is necessary, not an
+ * optimization. (Shifts counts of W_TYPE_SIZE
+ * are undefined.)
+ */
+
+ n1 -= d0;
+ q1 = 1;
+ } else {
+ /* Normalize. */
+
+ b = W_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 {
+ unsigned long m1, m0;
+ /* Normalize. */
+
+ b = W_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;
+}
diff --git a/lib/umoddi3.c b/lib/umoddi3.c
new file mode 100644
index 0000000..85b2b86
--- /dev/null
+++ b/lib/umoddi3.c
@@ -0,0 +1,16 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/module.h>
+#include <linux/libgcc.h>
+
+extern unsigned long long __udivmoddi4(unsigned long long u,
+ unsigned long long v,
+ unsigned long long *rp);
+
+unsigned long long __umoddi3(unsigned long long u, unsigned long long v)
+{
+ unsigned long long w;
+ (void)__udivmoddi4(u, v, &w);
+ return w;
+}
+EXPORT_SYMBOL(__umoddi3);
--
2.7.4
next prev parent reply other threads:[~2018-09-18 9:20 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-09-18 9:19 [PATCH 0/5] Fix some bugs on RV32 build fail and issue Zong Li
2018-09-18 9:19 ` [PATCH 1/5] RISC-V: Build tishift only on 64-bit Zong Li
2018-09-21 6:58 ` Christoph Hellwig
2018-09-25 2:33 ` Zong Li
2018-09-18 9:19 ` [PATCH 2/5] RISC-V: Use swiotlb on RV64 only Zong Li
2018-09-21 6:59 ` Christoph Hellwig
2018-09-18 9:19 ` Zong Li [this message]
2018-09-21 7:00 ` [PATCH 3/5] lib: Add umoddi3 and udivmoddi4 of GCC library routines Christoph Hellwig
2018-09-25 2:19 ` Zong Li
2018-09-25 7:20 ` Andreas Schwab
2018-09-26 7:12 ` Zong Li
2018-09-25 15:25 ` Christoph Hellwig
2018-09-26 2:40 ` Zong Li
2018-09-18 9:19 ` [PATCH 4/5] RISC-V: Select GENERIC_LIB_UMODDI3 on RV32 Zong Li
2018-09-18 9:19 ` [PATCH 5/5] RISC-V: Avoid corrupting the upper 32-bit of phys_addr_t in ioremap Zong Li
2018-09-21 7:00 ` Christoph Hellwig
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=2ccc6c0758624dbb22f6fe451a3e476a000108ef.1537260207.git.zongbox@gmail.com \
--to=zongbox@gmail.com \
--cc=aou@eecs.berkeley.edu \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-riscv@lists.infradead.org \
--cc=palmer@sifive.com \
--cc=vincentc@andestech.com \
--cc=zong@andestech.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).