linux-sparse.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
To: linux-sparse@vger.kernel.org
Cc: Christopher Li <sparse@chrisli.org>,
	Luc Van Oostenryck <luc.vanoostenryck@gmail.com>,
	Dibyendu Majumdar <mobile@majumdar.org.uk>
Subject: [PATCH] llvm: fix getting type of values
Date: Thu,  2 Mar 2017 08:02:08 +0100	[thread overview]
Message-ID: <20170302070208.13478-1-luc.vanoostenryck@gmail.com> (raw)
In-Reply-To: <CACXZuxdFQD+6yPcisZ=hJssO5fyxm9LYJbvVHsvnjRz87cCjrQ@mail.gmail.com>

In sparse-llvm: there was the assumption that the type
of a PSEUDO_VAL was always one of the integer type.
But this is not always the case: constant pointers,
like NULL, are also of the PSEUDO_VAL kind.

Fix this by using the type associated with the concerned
instruction to retrieve and use the correct type.

Note: while this patch improve the situation, like for example
the test cases added here, it's still not correct since now
we're make the assumption that insn->type is the type we need
for the pseudo. This is maybe often true, but certainly not
always. For example this is not true for:
- OP_STORE/OP_LOAD's insn->src
- OP_SET{EQ,...}'s   insn->src[12]
- in general for any instructions the target have a different type
  than the operands (when we're insterested in the operands).
- probably some  others ones

CC: Dibyendu Majumdar <mobile@majumdar.org.uk>
Reported-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Some-parts-also-by: Dibyendu Majumdar <mobile@majumdar.org.uk>
Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@gmail.com>
---
 sparse-llvm.c             | 16 +++++++++++++++-
 validation/backend/null.c | 24 ++++++++++++++++++++++++
 2 files changed, 39 insertions(+), 1 deletion(-)
 create mode 100644 validation/backend/null.c

diff --git a/sparse-llvm.c b/sparse-llvm.c
index 9f362b3ed..ff66a96a7 100644
--- a/sparse-llvm.c
+++ b/sparse-llvm.c
@@ -305,6 +305,7 @@ static void pseudo_name(pseudo_t pseudo, char *buf)
 
 static LLVMValueRef pseudo_to_value(struct function *fn, struct instruction *insn, pseudo_t pseudo)
 {
+	LLVMTypeRef type, iptr_type;
 	LLVMValueRef result = NULL;
 
 	switch (pseudo->type) {
@@ -360,7 +361,20 @@ static LLVMValueRef pseudo_to_value(struct function *fn, struct instruction *ins
 		break;
 	}
 	case PSEUDO_VAL:
-		result = LLVMConstInt(insn_symbol_type(fn->module, insn), pseudo->value, 1);
+		type = insn_symbol_type(fn->module, insn);
+		switch (LLVMGetTypeKind(type)) {
+		case LLVMPointerTypeKind:
+			iptr_type = LLVMIntType(bits_in_pointer);
+			result = LLVMConstInt(iptr_type, pseudo->value, 1);
+			result = LLVMConstIntToPtr(result, type);
+			break;
+		case LLVMIntegerTypeKind:
+			result = LLVMConstInt(type, pseudo->value, 1);
+			break;
+		default:
+			assert(0);
+		}
+
 		break;
 	case PSEUDO_ARG: {
 		result = LLVMGetParam(fn->fn, pseudo->nr - 1);
diff --git a/validation/backend/null.c b/validation/backend/null.c
new file mode 100644
index 000000000..5c595c70b
--- /dev/null
+++ b/validation/backend/null.c
@@ -0,0 +1,24 @@
+extern int *ip[];
+
+void foo(void);
+void foo(void)
+{
+	ip[0] = (void *)0L;
+	ip[1] = (int *)0L;
+	ip[2] = (void *)0;
+	ip[3] = (int *)0;
+	ip[4] = (void *)(long)0;
+	ip[5] = (int *)(long)0;
+	ip[6] = (void *)123;
+	ip[7] = (int *)123;
+	ip[8] = (void *)123L;
+	ip[9] = (int *)123L;
+	ip[10] = (void *)(long)123;
+	ip[11] = (int *)(long)123;
+}
+
+/*
+ * check-name: store constants to pointer
+ * check-command: sparse-llvm $file
+ * check-output-ignore
+ */
-- 
2.11.1


  parent reply	other threads:[~2017-03-02  7:03 UTC|newest]

Thread overview: 57+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-02-28  6:20 Sparse-LLVM issue compiling NULL pointers Dibyendu Majumdar
2017-02-28 15:09 ` Luc Van Oostenryck
2017-02-28 16:04   ` Dibyendu Majumdar
2017-02-28 16:47     ` Luc Van Oostenryck
2017-02-28 16:49     ` Dibyendu Majumdar
2017-03-02  6:48       ` Luc Van Oostenryck
2017-02-28 17:03   ` Luc Van Oostenryck
2017-02-28 17:35     ` Luc Van Oostenryck
2017-02-28 17:42       ` Dibyendu Majumdar
2017-02-28 18:08       ` Dibyendu Majumdar
2017-03-01  5:49         ` Luc Van Oostenryck
2017-03-02  7:02         ` Luc Van Oostenryck [this message]
2017-03-01 10:58     ` Dibyendu Majumdar
2017-03-01 14:45       ` Dibyendu Majumdar
2017-03-02  5:21         ` Luc Van Oostenryck
2017-03-02  5:41           ` Dibyendu Majumdar
2017-03-02 13:56             ` Luc Van Oostenryck
2017-03-02 14:05               ` Dibyendu Majumdar
2017-03-02 16:10                 ` Luc Van Oostenryck
2017-03-02 14:33               ` Dibyendu Majumdar
2017-03-02 16:04                 ` Luc Van Oostenryck
2017-03-02 16:29                   ` Dibyendu Majumdar
2017-03-02 16:30                     ` Dibyendu Majumdar
2017-03-02 17:18                       ` Luc Van Oostenryck
2017-03-02 17:36                         ` Dibyendu Majumdar
2017-03-02 20:09                         ` Luc Van Oostenryck
2017-03-03  2:52                           ` Dibyendu Majumdar
2017-03-03  3:01                             ` Dibyendu Majumdar
2017-03-03  4:03                               ` Dibyendu Majumdar
2017-03-03  5:24                                 ` [PATCH] llvm: fix output_op_[ptr]cast() Luc Van Oostenryck
2017-03-03  7:37                                   ` Dibyendu Majumdar
2017-03-03 18:06                                     ` Dibyendu Majumdar
2017-03-03 18:30                                       ` Dibyendu Majumdar
2017-03-03 19:55                                         ` Luc Van Oostenryck
2017-03-06  1:56                                           ` Christopher Li
2017-03-03 19:50                                       ` Luc Van Oostenryck
2017-03-03 19:54                                         ` Dibyendu Majumdar
2017-03-03 20:52                                           ` [PATCH] llvm: fix: do not mix pointers and floats when doing compares Luc Van Oostenryck
2017-03-03  4:16                             ` Sparse-LLVM issue compiling NULL pointers Luc Van Oostenryck
2017-03-03  4:27                             ` Luc Van Oostenryck
2017-03-03  4:38                               ` Dibyendu Majumdar
2017-03-03  7:50                                 ` Luc Van Oostenryck
2017-03-03 12:39                                   ` Dibyendu Majumdar
2017-03-02 17:03                     ` Dibyendu Majumdar
2017-03-02 17:18                       ` Dibyendu Majumdar
2017-03-02 17:43                         ` Luc Van Oostenryck
2017-03-02 18:58                           ` Dibyendu Majumdar
2017-03-02 19:34                             ` Luc Van Oostenryck
2017-03-02 17:50                       ` Luc Van Oostenryck
2017-03-02 17:57                         ` Luc Van Oostenryck
2017-03-02 18:02                           ` Dibyendu Majumdar
2017-03-03  4:21                             ` Luc Van Oostenryck
2017-03-02 17:27                   ` Luc Van Oostenryck
2017-03-02 18:41                   ` Dibyendu Majumdar
2017-03-03  5:35                     ` Luc Van Oostenryck
2017-03-02 16:39           ` Dibyendu Majumdar
2017-03-02 17:21             ` Luc Van Oostenryck

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=20170302070208.13478-1-luc.vanoostenryck@gmail.com \
    --to=luc.vanoostenryck@gmail.com \
    --cc=linux-sparse@vger.kernel.org \
    --cc=mobile@majumdar.org.uk \
    --cc=sparse@chrisli.org \
    /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).