* [PATCH] analyzer: Recognize __builtin_free as a matching deallocator
@ 2021-07-28 5:04 Siddhesh Poyarekar
2021-07-28 11:44 ` David Malcolm
0 siblings, 1 reply; 5+ messages in thread
From: Siddhesh Poyarekar @ 2021-07-28 5:04 UTC (permalink / raw)
To: gcc-patches
Recognize __builtin_free as being equivalent to free when passed into
__attribute__((malloc ())), similar to how it is treated when it is
encountered as a call. This fixes spurious warnings in glibc where
xmalloc family of allocators as well as reallocarray, memalign,
etc. are declared to have __builtin_free as the free function.
gcc/analyzer/ChangeLog:
* sm-malloc.cc
(malloc_state_machine::get_or_create_deallocator): Recognize
__builtin_free.
gcc/testsuite/ChangeLog:
* gcc.dg/analyzer/attr-malloc-1.c (compatible_alloc,
compatible_alloc2): New extern allocator declarations.
(test_9, test_10): New tests.
---
gcc/analyzer/sm-malloc.cc | 3 ++-
gcc/testsuite/gcc.dg/analyzer/attr-malloc-1.c | 19 +++++++++++++++++++
2 files changed, 21 insertions(+), 1 deletion(-)
diff --git a/gcc/analyzer/sm-malloc.cc b/gcc/analyzer/sm-malloc.cc
index 9707a6863ce..1d69d57df0e 100644
--- a/gcc/analyzer/sm-malloc.cc
+++ b/gcc/analyzer/sm-malloc.cc
@@ -1511,7 +1511,8 @@ malloc_state_machine::get_or_create_deallocator (tree deallocator_fndecl)
/* Reuse "free". */
deallocator *d;
if (is_named_call_p (deallocator_fndecl, "free")
- || is_std_named_call_p (deallocator_fndecl, "free"))
+ || is_std_named_call_p (deallocator_fndecl, "free")
+ || is_named_call_p (deallocator_fndecl, "__builtin_free"))
d = &m_free.m_deallocator;
else
{
diff --git a/gcc/testsuite/gcc.dg/analyzer/attr-malloc-1.c b/gcc/testsuite/gcc.dg/analyzer/attr-malloc-1.c
index 3de32b1b14b..a9a2a3dee85 100644
--- a/gcc/testsuite/gcc.dg/analyzer/attr-malloc-1.c
+++ b/gcc/testsuite/gcc.dg/analyzer/attr-malloc-1.c
@@ -11,6 +11,12 @@ extern struct foo *foo_acquire (void)
extern void use_foo (const struct foo *)
__attribute__((nonnull));
+extern struct foo *compatible_alloc (void)
+ __attribute__ ((malloc (__builtin_free)));
+
+extern struct foo *compatible_alloc2 (void)
+ __attribute__ ((malloc (free)));
+
void test_1 (void)
{
struct foo *p = foo_acquire ();
@@ -73,3 +79,16 @@ int test_8 (struct foo *p)
foo_release (p);
return p->m_int; /* { dg-warning "use after 'foo_release' of 'p'" } */
}
+
+/* Recognize that __builtin_free and free are the same thing. */
+void test_9 (void)
+{
+ struct foo *p = compatible_alloc ();
+ free (p);
+}
+
+void test_10 (void)
+{
+ struct foo *p = compatible_alloc2 ();
+ __builtin_free (p);
+}
--
2.31.1
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] analyzer: Recognize __builtin_free as a matching deallocator
2021-07-28 5:04 [PATCH] analyzer: Recognize __builtin_free as a matching deallocator Siddhesh Poyarekar
@ 2021-07-28 11:44 ` David Malcolm
2021-08-25 12:14 ` Matthias Klose
0 siblings, 1 reply; 5+ messages in thread
From: David Malcolm @ 2021-07-28 11:44 UTC (permalink / raw)
To: Siddhesh Poyarekar, gcc-patches
On Wed, 2021-07-28 at 10:34 +0530, Siddhesh Poyarekar wrote:
> Recognize __builtin_free as being equivalent to free when passed into
> __attribute__((malloc ())), similar to how it is treated when it is
> encountered as a call. This fixes spurious warnings in glibc where
> xmalloc family of allocators as well as reallocarray, memalign,
> etc. are declared to have __builtin_free as the free function.
>
> gcc/analyzer/ChangeLog:
> * sm-malloc.cc
> (malloc_state_machine::get_or_create_deallocator): Recognize
> __builtin_free.
>
> gcc/testsuite/ChangeLog:
> * gcc.dg/analyzer/attr-malloc-1.c (compatible_alloc,
> compatible_alloc2): New extern allocator declarations.
> (test_9, test_10): New tests.
Looks good to me, thanks
Dave
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] analyzer: Recognize __builtin_free as a matching deallocator
2021-07-28 11:44 ` David Malcolm
@ 2021-08-25 12:14 ` Matthias Klose
2021-08-25 15:16 ` Siddhesh Poyarekar
0 siblings, 1 reply; 5+ messages in thread
From: Matthias Klose @ 2021-08-25 12:14 UTC (permalink / raw)
To: David Malcolm, Siddhesh Poyarekar, gcc-patches
On 7/28/21 1:44 PM, David Malcolm via Gcc-patches wrote:
> On Wed, 2021-07-28 at 10:34 +0530, Siddhesh Poyarekar wrote:
>> Recognize __builtin_free as being equivalent to free when passed into
>> __attribute__((malloc ())), similar to how it is treated when it is
>> encountered as a call. This fixes spurious warnings in glibc where
>> xmalloc family of allocators as well as reallocarray, memalign,
>> etc. are declared to have __builtin_free as the free function.
>>
>> gcc/analyzer/ChangeLog:
>> * sm-malloc.cc
>> (malloc_state_machine::get_or_create_deallocator): Recognize
>> __builtin_free.
>>
>> gcc/testsuite/ChangeLog:
>> * gcc.dg/analyzer/attr-malloc-1.c (compatible_alloc,
>> compatible_alloc2): New extern allocator declarations.
>> (test_9, test_10): New tests.
>
> Looks good to me, thanks
> Dave
>
>
Please could this be backported to all active branches?
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] analyzer: Recognize __builtin_free as a matching deallocator
2021-08-25 12:14 ` Matthias Klose
@ 2021-08-25 15:16 ` Siddhesh Poyarekar
2021-08-25 23:08 ` Jeff Law
0 siblings, 1 reply; 5+ messages in thread
From: Siddhesh Poyarekar @ 2021-08-25 15:16 UTC (permalink / raw)
To: Matthias Klose, David Malcolm, gcc-patches
On 8/25/21 5:44 PM, Matthias Klose wrote:
> On 7/28/21 1:44 PM, David Malcolm via Gcc-patches wrote:
>> On Wed, 2021-07-28 at 10:34 +0530, Siddhesh Poyarekar wrote:
>>> Recognize __builtin_free as being equivalent to free when passed into
>>> __attribute__((malloc ())), similar to how it is treated when it is
>>> encountered as a call. This fixes spurious warnings in glibc where
>>> xmalloc family of allocators as well as reallocarray, memalign,
>>> etc. are declared to have __builtin_free as the free function.
>>>
>>> gcc/analyzer/ChangeLog:
>>> * sm-malloc.cc
>>> (malloc_state_machine::get_or_create_deallocator): Recognize
>>> __builtin_free.
>>>
>>> gcc/testsuite/ChangeLog:
>>> * gcc.dg/analyzer/attr-malloc-1.c (compatible_alloc,
>>> compatible_alloc2): New extern allocator declarations.
>>> (test_9, test_10): New tests.
>>
>> Looks good to me, thanks
>> Dave
>>
>>
>
>
> Please could this be backported to all active branches?
>
Sure, it looks like only gcc11 needs this since malloc attribute
matching seems recent. David, I've never done a backport before, may I
just cherry-pick, push and post a [committed] patch on list or does it
need to go through review?
Thanks,
Siddhesh
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] analyzer: Recognize __builtin_free as a matching deallocator
2021-08-25 15:16 ` Siddhesh Poyarekar
@ 2021-08-25 23:08 ` Jeff Law
0 siblings, 0 replies; 5+ messages in thread
From: Jeff Law @ 2021-08-25 23:08 UTC (permalink / raw)
To: Siddhesh Poyarekar, Matthias Klose, David Malcolm, gcc-patches
On 8/25/2021 9:16 AM, Siddhesh Poyarekar wrote:
> On 8/25/21 5:44 PM, Matthias Klose wrote:
>> On 7/28/21 1:44 PM, David Malcolm via Gcc-patches wrote:
>>> On Wed, 2021-07-28 at 10:34 +0530, Siddhesh Poyarekar wrote:
>>>> Recognize __builtin_free as being equivalent to free when passed into
>>>> __attribute__((malloc ())), similar to how it is treated when it is
>>>> encountered as a call. This fixes spurious warnings in glibc where
>>>> xmalloc family of allocators as well as reallocarray, memalign,
>>>> etc. are declared to have __builtin_free as the free function.
>>>>
>>>> gcc/analyzer/ChangeLog:
>>>> * sm-malloc.cc
>>>> (malloc_state_machine::get_or_create_deallocator): Recognize
>>>> __builtin_free.
>>>>
>>>> gcc/testsuite/ChangeLog:
>>>> * gcc.dg/analyzer/attr-malloc-1.c (compatible_alloc,
>>>> compatible_alloc2): New extern allocator declarations.
>>>> (test_9, test_10): New tests.
>>>
>>> Looks good to me, thanks
>>> Dave
>>>
>>>
>>
>>
>> Please could this be backported to all active branches?
>>
>
> Sure, it looks like only gcc11 needs this since malloc attribute
> matching seems recent. David, I've never done a backport before, may
> I just cherry-pick, push and post a [committed] patch on list or does
> it need to go through review?
In general it's fine to cherry-pick fixes from the mainline to a release
branch that fix regressions, incorrect code and the like.
This doesn't fall into one of those categories -- but given this is
limited to the analyzer, I think it's fine to cherry-pick into the
release branch as long as David is OK with it as well.
jeff
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2021-08-25 23:08 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-07-28 5:04 [PATCH] analyzer: Recognize __builtin_free as a matching deallocator Siddhesh Poyarekar
2021-07-28 11:44 ` David Malcolm
2021-08-25 12:14 ` Matthias Klose
2021-08-25 15:16 ` Siddhesh Poyarekar
2021-08-25 23:08 ` Jeff Law
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 read-only IMAP folder(s) and NNTP newsgroup(s).