diff mbox series

pthread: add PTHREAD_NULL

Message ID 20251101193213.684326-2-hi@alyssa.is (mailing list archive)
State Changes Requested
Headers
Series pthread: add PTHREAD_NULL |

Checks

Context Check Description
redhat-pt-bot/TryBot-apply_patch success Patch applied to master at the time it was sent
redhat-pt-bot/TryBot-32bit success Build for i686

Commit Message

Alyssa Ross Nov. 1, 2025, 7:32 p.m. UTC 
  New in POSIX.1-2024.

In htl, pthread_t values are integers starting at 1, so 0 is a safe
null value.

In nptl, pthread_t values are non-null pointers, so 0 is a safe null
value.
---
I wasn't sure whether this should be behind a __USE macro, but there
isn't one for POSIX.1-2024 yet:
https://inbox.sourceware.org/libc-alpha/d34488044ace7c01311fb1c1948e65a050e93252.1747972803.git.collin.funk1@gmail.com/

 manual/threads.texi    | 5 +++++
 sysdeps/htl/pthread.h  | 2 ++
 sysdeps/nptl/pthread.h | 3 +++
 3 files changed, 10 insertions(+)


base-commit: 720e89163702ffa1e921d926b6c36b53c3ccbee4

Comments

Collin Funk Nov. 1, 2025, 9:51 p.m. UTC | #1 
Hi Alyssa,

Alyssa Ross <hi@alyssa.is> writes:

> New in POSIX.1-2024.
>
> In htl, pthread_t values are integers starting at 1, so 0 is a safe
> null value.
>
> In nptl, pthread_t values are non-null pointers, so 0 is a safe null
> value.
> ---
> I wasn't sure whether this should be behind a __USE macro, but there
> isn't one for POSIX.1-2024 yet:
> https://inbox.sourceware.org/libc-alpha/d34488044ace7c01311fb1c1948e65a050e93252.1747972803.git.collin.funk1@gmail.com/
>
>  manual/threads.texi    | 5 +++++
>  sysdeps/htl/pthread.h  | 2 ++
>  sysdeps/nptl/pthread.h | 3 +++
>  3 files changed, 10 insertions(+)
>
> diff --git a/manual/threads.texi b/manual/threads.texi
> index 1e9be32d66..d3692e8939 100644
> --- a/manual/threads.texi
> +++ b/manual/threads.texi
> @@ -1099,6 +1099,11 @@ This function gets the kind of mutex @var{mutex} is.
>  @node POSIX Threads Other APIs
>  @subsection POSIX Threads Other APIs
>  
> +@deftypevr Macro pthread_t PTHREAD_NULL
> +@standards{POSIX.1-2024, pthread.h}
> +A value for pthread_t that is not considered equal to any thread ID.
> +@end deftypevr
> +
>  @deftypefun int pthread_equal (pthread_t @var{thread1}, pthread_t @var{thread2})
>  Compares two thread IDs.  If they are the same, returns nonzero, else returns zero.
>  @manpagefunctionstub{pthread_equal, 3}
> diff --git a/sysdeps/htl/pthread.h b/sysdeps/htl/pthread.h
> index a299fec278..d490f3cdf8 100644
> --- a/sysdeps/htl/pthread.h
> +++ b/sysdeps/htl/pthread.h
> @@ -55,6 +55,8 @@ __BEGIN_DECLS
>  
>  #include <bits/pthread.h>
>  
> +#define PTHREAD_NULL ((pthread_t)0)
> +
>  /* Possible values for the process shared attribute.  */
>  #define PTHREAD_PROCESS_PRIVATE __PTHREAD_PROCESS_PRIVATE
>  #define PTHREAD_PROCESS_SHARED __PTHREAD_PROCESS_SHARED
> diff --git a/sysdeps/nptl/pthread.h b/sysdeps/nptl/pthread.h
> index 92957a620d..b47b732a9f 100644
> --- a/sysdeps/nptl/pthread.h
> +++ b/sysdeps/nptl/pthread.h
> @@ -33,6 +33,9 @@
>  # include <bits/pthread_stack_min-dynamic.h>
>  #endif
>  
> +#define PTHREAD_NULL ((pthread_t)0)
> +
> +
>  /* Detach state.  */
>  enum
>  {
>
> base-commit: 720e89163702ffa1e921d926b6c36b53c3ccbee4

I don't think this will cause conform tests to fail since PTHREAD_* is
allowed. But it should probably be hidden under __USE_XOPEN2K24 which is
not yet introduced. I have an old patch that is still waiting on review
[1].

Collin

[1] https://inbox.sourceware.org/libc-alpha/d34488044ace7c01311fb1c1948e65a050e93252.1747972803.git.collin.funk1@gmail.com/
Carlos O'Donell Nov. 3, 2025, 2:10 p.m. UTC | #2 
On 11/1/25 5:51 PM, Collin Funk wrote:
> Hi Alyssa,
> 
> Alyssa Ross <hi@alyssa.is> writes:
> 
>> New in POSIX.1-2024.
>>
>> In htl, pthread_t values are integers starting at 1, so 0 is a safe
>> null value.
>>
>> In nptl, pthread_t values are non-null pointers, so 0 is a safe null
>> value.
>> ---
>> I wasn't sure whether this should be behind a __USE macro, but there
>> isn't one for POSIX.1-2024 yet:
>> https://inbox.sourceware.org/libc-alpha/d34488044ace7c01311fb1c1948e65a050e93252.1747972803.git.collin.funk1@gmail.com/
>>
>>   manual/threads.texi    | 5 +++++
>>   sysdeps/htl/pthread.h  | 2 ++
>>   sysdeps/nptl/pthread.h | 3 +++
>>   3 files changed, 10 insertions(+)
>>
>> diff --git a/manual/threads.texi b/manual/threads.texi
>> index 1e9be32d66..d3692e8939 100644
>> --- a/manual/threads.texi
>> +++ b/manual/threads.texi
>> @@ -1099,6 +1099,11 @@ This function gets the kind of mutex @var{mutex} is.
>>   @node POSIX Threads Other APIs
>>   @subsection POSIX Threads Other APIs
>>   
>> +@deftypevr Macro pthread_t PTHREAD_NULL
>> +@standards{POSIX.1-2024, pthread.h}
>> +A value for pthread_t that is not considered equal to any thread ID.
>> +@end deftypevr
>> +
>>   @deftypefun int pthread_equal (pthread_t @var{thread1}, pthread_t @var{thread2})
>>   Compares two thread IDs.  If they are the same, returns nonzero, else returns zero.
>>   @manpagefunctionstub{pthread_equal, 3}
>> diff --git a/sysdeps/htl/pthread.h b/sysdeps/htl/pthread.h
>> index a299fec278..d490f3cdf8 100644
>> --- a/sysdeps/htl/pthread.h
>> +++ b/sysdeps/htl/pthread.h
>> @@ -55,6 +55,8 @@ __BEGIN_DECLS
>>   
>>   #include <bits/pthread.h>
>>   
>> +#define PTHREAD_NULL ((pthread_t)0)
>> +
>>   /* Possible values for the process shared attribute.  */
>>   #define PTHREAD_PROCESS_PRIVATE __PTHREAD_PROCESS_PRIVATE
>>   #define PTHREAD_PROCESS_SHARED __PTHREAD_PROCESS_SHARED
>> diff --git a/sysdeps/nptl/pthread.h b/sysdeps/nptl/pthread.h
>> index 92957a620d..b47b732a9f 100644
>> --- a/sysdeps/nptl/pthread.h
>> +++ b/sysdeps/nptl/pthread.h
>> @@ -33,6 +33,9 @@
>>   # include <bits/pthread_stack_min-dynamic.h>
>>   #endif
>>   
>> +#define PTHREAD_NULL ((pthread_t)0)
>> +
>> +
>>   /* Detach state.  */
>>   enum
>>   {
>>
>> base-commit: 720e89163702ffa1e921d926b6c36b53c3ccbee4
> 
> I don't think this will cause conform tests to fail since PTHREAD_* is
> allowed. But it should probably be hidden under __USE_XOPEN2K24 which is
> not yet introduced. I have an old patch that is still waiting on review
> [1].

That's stuck in my queue for review and I haven't gotten to it yet,
but I agree it should likely go under some conditional e.g. __USE_XOPEN2K24.

I should finish the review for you for the additional constants. Let me see
if I can get that off my plate.

> Collin
> 
> [1] https://inbox.sourceware.org/libc-alpha/d34488044ace7c01311fb1c1948e65a050e93252.1747972803.git.collin.funk1@gmail.com/
>
diff mbox series

Patch

diff --git a/manual/threads.texi b/manual/threads.texi
index 1e9be32d66..d3692e8939 100644
--- a/manual/threads.texi
+++ b/manual/threads.texi
@@ -1099,6 +1099,11 @@  This function gets the kind of mutex @var{mutex} is.
 @node POSIX Threads Other APIs
 @subsection POSIX Threads Other APIs
 
+@deftypevr Macro pthread_t PTHREAD_NULL
+@standards{POSIX.1-2024, pthread.h}
+A value for pthread_t that is not considered equal to any thread ID.
+@end deftypevr
+
 @deftypefun int pthread_equal (pthread_t @var{thread1}, pthread_t @var{thread2})
 Compares two thread IDs.  If they are the same, returns nonzero, else returns zero.
 @manpagefunctionstub{pthread_equal, 3}
diff --git a/sysdeps/htl/pthread.h b/sysdeps/htl/pthread.h
index a299fec278..d490f3cdf8 100644
--- a/sysdeps/htl/pthread.h
+++ b/sysdeps/htl/pthread.h
@@ -55,6 +55,8 @@  __BEGIN_DECLS
 
 #include <bits/pthread.h>
 
+#define PTHREAD_NULL ((pthread_t)0)
+
 /* Possible values for the process shared attribute.  */
 #define PTHREAD_PROCESS_PRIVATE __PTHREAD_PROCESS_PRIVATE
 #define PTHREAD_PROCESS_SHARED __PTHREAD_PROCESS_SHARED
diff --git a/sysdeps/nptl/pthread.h b/sysdeps/nptl/pthread.h
index 92957a620d..b47b732a9f 100644
--- a/sysdeps/nptl/pthread.h
+++ b/sysdeps/nptl/pthread.h
@@ -33,6 +33,9 @@ 
 # include <bits/pthread_stack_min-dynamic.h>
 #endif
 
+#define PTHREAD_NULL ((pthread_t)0)
+
+
 /* Detach state.  */
 enum
 {