linux-sparse.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] testsuite: add testcase for bogus linearization of >>= & /=
@ 2020-07-05 12:38 Luc Van Oostenryck
  0 siblings, 0 replies; only message in thread
From: Luc Van Oostenryck @ 2020-07-05 12:38 UTC (permalink / raw)
  To: linux-sparse; +Cc: Luc Van Oostenryck

When doing a shift operation, both arguments are subjected to
integer promotion and the type of the result is simply the type
of the promoted left operand. Easy.

But for a shift-assignment, things are slightly more complex:
  -) 'a >>= n' should be equivalent to 'a = a >> n'
  -) but the type of the result must be the type of the left
     operand *before* integer promotion.

Currently, the linearization code use the type of the right
operand to infer of the type of the operation. But simply changing
the code to use the type of the left operand will also be wrong
(for example for signed/unsigned divisions). Nasty.

For example, the following C code:
	int s = ...;
	s >>= 11U;
is linearized as a logical shift:
	lsr.32      %r2 <- %arg1, $11
while, of course it's an arithmetic shift that is expected:
	asr.32      %r2 <- %arg1, $11

So, add a testcase for these.

Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
---
 validation/linear/bug-assign-op0.c | 115 +++++++++++++++++++++++++++++
 1 file changed, 115 insertions(+)
 create mode 100644 validation/linear/bug-assign-op0.c

diff --git a/validation/linear/bug-assign-op0.c b/validation/linear/bug-assign-op0.c
new file mode 100644
index 000000000000..0cabc6222b8a
--- /dev/null
+++ b/validation/linear/bug-assign-op0.c
@@ -0,0 +1,115 @@
+int asr(int s)
+{
+	s >>= 11U;
+	return s;
+}
+
+unsigned int lsr(unsigned int u)
+{
+	u >>= 11;
+	return u;
+}
+
+int divr(int s, unsigned long u)
+{
+	extern int use(int, unsigned);
+	int t = s;
+	s = s / u;
+	u = u / t;
+	return use(s, u);
+}
+
+int sdivul(int s, unsigned long u)
+{
+	s /= u;			// divu
+	return s;
+}
+
+unsigned int udivsl(unsigned int u, long s)
+{
+	u /= s;			// divs
+	return u;
+}
+
+int uldivs(int s, unsigned long u)
+{
+	u /= s;			// divu
+	return u;
+}
+
+unsigned int sldivu(unsigned int u, long s)
+{
+	s /= u;			// divs
+	return s;
+}
+
+/*
+ * check-name: bug-assign-op0
+ * check-command: test-linearize -Wno-decl $file
+ * check-known-to-fail
+ *
+ * check-output-start
+asr:
+.L0:
+	<entry-point>
+	asr.32      %r2 <- %arg1, $11
+	ret.32      %r2
+
+
+lsr:
+.L2:
+	<entry-point>
+	lsr.32      %r6 <- %arg1, $11
+	ret.32      %r6
+
+
+divr:
+.L4:
+	<entry-point>
+	sext.64     %r11 <- (32) %arg1
+	divu.64     %r13 <- %r11, %arg2
+	trunc.32    %r14 <- (64) %r13
+	divu.64     %r18 <- %arg2, %r11
+	trunc.32    %r21 <- (64) %r18
+	call.32     %r22 <- use, %r14, %r21
+	ret.32      %r22
+
+
+sdivul:
+.L6:
+	<entry-point>
+	sext.64     %r26 <- (32) %arg1
+	divu.64     %r27 <- %r26, %arg2
+	trunc.32    %r28 <- (64) %r27
+	ret.32      %r28
+
+
+udivsl:
+.L8:
+	<entry-point>
+	zext.64     %r33 <- (32) %arg1
+	divs.64     %r34 <- %r33, %arg2
+	trunc.32    %r35 <- (64) %r34
+	ret.32      %r35
+
+
+uldivs:
+.L10:
+	<entry-point>
+	sext.64     %r39 <- (32) %arg1
+	divu.64     %r41 <- %arg2, %r39
+	trunc.32    %r43 <- (64) %r41
+	ret.32      %r43
+
+
+sldivu:
+.L12:
+	<entry-point>
+	zext.64     %r46 <- (32) %arg1
+	divs.64     %r48 <- %arg2, %r46
+	trunc.32    %r50 <- (64) %r48
+	ret.32      %r50
+
+
+ * check-output-end
+ */
-- 
2.27.0

^ permalink raw reply related	[flat|nested] only message in thread

only message in thread, other threads:[~2020-07-05 12:38 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-07-05 12:38 [PATCH] testsuite: add testcase for bogus linearization of >>= & /= Luc Van Oostenryck

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).