[PATCH] Implement nbsd_nat_target::thread_alive

[PATCH] Implement nbsd_nat_target::thread_alive

[Kamil Rytarowski]

gdb/ChangeLog:

	* nbsd-nat.h (nbsd_nat_target::thread_alive): Add.
	* nbsd-nat.c (nbsd_nat_target::thread_alive): Add.
---
 gdb/ChangeLog  |  5 +++++
 gdb/nbsd-nat.c | 30 ++++++++++++++++++++++++++++++
 gdb/nbsd-nat.h |  1 +
 3 files changed, 36 insertions(+)

diff --git a/gdb/ChangeLog b/gdb/ChangeLog
index 1ee58a541cb..d6789f4d7ca 100644
--- a/gdb/ChangeLog
+++ b/gdb/ChangeLog
@@ -1,3 +1,8 @@
+2020-03-25  Kamil Rytarowski  <n54@gmx.com>
+
+	* nbsd-nat.h (nbsd_nat_target::thread_alive): Add.
+	* nbsd-nat.c (nbsd_nat_target::thread_alive): Add.
+
 2020-03-25  Tom Tromey  <tom@tromey.com>

 	* compile/compile-object-load.c (get_out_value_type): Mention
diff --git a/gdb/nbsd-nat.c b/gdb/nbsd-nat.c
index 326bbe3aec3..aa0353309df 100644
--- a/gdb/nbsd-nat.c
+++ b/gdb/nbsd-nat.c
@@ -39,3 +39,33 @@ nbsd_nat_target::pid_to_exec_file (int pid)
     return NULL;
   return buf;
 }
+
+bool
+thread_alive (ptid_t ptid)
+{
+  pid_t pid = ptid.pid ();
+  int lwp = ptid.lwp ();
+  int mib[5] = {CTL_KERN, KERN_LWP, pid, sizeof (struct kinfo_lwp), 0};
+  size_t size;
+
+  if (sysctl (mib, 5, NULL, &size, NULL, 0) == -1 || size == 0)
+    perror_with_name (("sysctl"));
+
+  mib[4] = size / sizeof (size_t);
+
+  gdb::unique_xmalloc_ptr<struct kinfo_lwp> kl
+    ((struct kinfo_lwp *) xcalloc (size, 1));
+  if (kl == NULL)
+    perror_with_name (("calloc"));
+
+  if (sysctl (mib, 5, kl.get (), &size, NULL, 0) == -1 || size == 0)
+    perror_with_name (("sysctl"));
+
+  for (size_t i = 0; i < size / sizeof (struct kinfo_lwp); i++)
+    {
+      if (kl.get ()[i].l_lid == lwp)
+	 return true;
+    }
+
+  return false;
+}
diff --git a/gdb/nbsd-nat.h b/gdb/nbsd-nat.h
index a752fbe572d..11b49427a44 100644
--- a/gdb/nbsd-nat.h
+++ b/gdb/nbsd-nat.h
@@ -27,6 +27,7 @@
 struct nbsd_nat_target : public inf_ptrace_target
 {
   char *pid_to_exec_file (int pid) override;
+  bool thread_alive (ptid_t ptid) override;
 };

 #endif /* nbsd-nat.h */
--
2.25.0

Re: [PATCH] Implement nbsd_nat_target::thread_alive

[John Baldwin]

On 3/25/20 12:36 PM, Kamil Rytarowski wrote:
> gdb/ChangeLog:
> 
> 	* nbsd-nat.h (nbsd_nat_target::thread_alive): Add.
> 	* nbsd-nat.c (nbsd_nat_target::thread_alive): Add.
> ---
>  gdb/ChangeLog  |  5 +++++
>  gdb/nbsd-nat.c | 30 ++++++++++++++++++++++++++++++
>  gdb/nbsd-nat.h |  1 +
>  3 files changed, 36 insertions(+)
> 
> diff --git a/gdb/ChangeLog b/gdb/ChangeLog
> index 1ee58a541cb..d6789f4d7ca 100644
> --- a/gdb/ChangeLog
> +++ b/gdb/ChangeLog
> @@ -1,3 +1,8 @@
> +2020-03-25  Kamil Rytarowski  <n54@gmx.com>
> +
> +	* nbsd-nat.h (nbsd_nat_target::thread_alive): Add.
> +	* nbsd-nat.c (nbsd_nat_target::thread_alive): Add.
> +
>  2020-03-25  Tom Tromey  <tom@tromey.com>
> 
>  	* compile/compile-object-load.c (get_out_value_type): Mention
> diff --git a/gdb/nbsd-nat.c b/gdb/nbsd-nat.c
> index 326bbe3aec3..aa0353309df 100644
> --- a/gdb/nbsd-nat.c
> +++ b/gdb/nbsd-nat.c
> @@ -39,3 +39,33 @@ nbsd_nat_target::pid_to_exec_file (int pid)
>      return NULL;
>    return buf;
>  }
> +
> +bool
> +thread_alive (ptid_t ptid)
> +{
> +  pid_t pid = ptid.pid ();
> +  int lwp = ptid.lwp ();
> +  int mib[5] = {CTL_KERN, KERN_LWP, pid, sizeof (struct kinfo_lwp), 0};
> +  size_t size;
> +
> +  if (sysctl (mib, 5, NULL, &size, NULL, 0) == -1 || size == 0)
> +    perror_with_name (("sysctl"));
> +
> +  mib[4] = size / sizeof (size_t);
> +
> +  gdb::unique_xmalloc_ptr<struct kinfo_lwp> kl
> +    ((struct kinfo_lwp *) xcalloc (size, 1));
> +  if (kl == NULL)
> +    perror_with_name (("calloc"));
> +
> +  if (sysctl (mib, 5, kl.get (), &size, NULL, 0) == -1 || size == 0)
> +    perror_with_name (("sysctl"));
> +
> +  for (size_t i = 0; i < size / sizeof (struct kinfo_lwp); i++)
> +    {
> +      if (kl.get ()[i].l_lid == lwp)
> +	 return true;
> +    }
> +
> +  return false;

Given that you have nearly identical code both here and in the thread_name
method in your other patch, perhaps it makes sense to have a single
"gdb::unique_xmalloc_ptr <struct kinfo_lwp> fetch_kinfo_lwp(ptid_t ptid)"
helper function that the other routines could share?  Perhaps instead of
just returning the pointer to the array of structures it could take a lamda
function that is invoked on the matching 'kl' instance with the correct lwp?

-- 
John Baldwin

Re: [PATCH] Implement nbsd_nat_target::thread_alive

[Kamil Rytarowski]

On 25.03.2020 21:42, John Baldwin wrote:
> On 3/25/20 12:36 PM, Kamil Rytarowski wrote:
>> gdb/ChangeLog:
>>
>> 	* nbsd-nat.h (nbsd_nat_target::thread_alive): Add.
>> 	* nbsd-nat.c (nbsd_nat_target::thread_alive): Add.
>> ---
>>  gdb/ChangeLog  |  5 +++++
>>  gdb/nbsd-nat.c | 30 ++++++++++++++++++++++++++++++
>>  gdb/nbsd-nat.h |  1 +
>>  3 files changed, 36 insertions(+)
>>
>> diff --git a/gdb/ChangeLog b/gdb/ChangeLog
>> index 1ee58a541cb..d6789f4d7ca 100644
>> --- a/gdb/ChangeLog
>> +++ b/gdb/ChangeLog
>> @@ -1,3 +1,8 @@
>> +2020-03-25  Kamil Rytarowski  <n54@gmx.com>
>> +
>> +	* nbsd-nat.h (nbsd_nat_target::thread_alive): Add.
>> +	* nbsd-nat.c (nbsd_nat_target::thread_alive): Add.
>> +
>>  2020-03-25  Tom Tromey  <tom@tromey.com>
>>
>>  	* compile/compile-object-load.c (get_out_value_type): Mention
>> diff --git a/gdb/nbsd-nat.c b/gdb/nbsd-nat.c
>> index 326bbe3aec3..aa0353309df 100644
>> --- a/gdb/nbsd-nat.c
>> +++ b/gdb/nbsd-nat.c
>> @@ -39,3 +39,33 @@ nbsd_nat_target::pid_to_exec_file (int pid)
>>      return NULL;
>>    return buf;
>>  }
>> +
>> +bool
>> +thread_alive (ptid_t ptid)
>> +{
>> +  pid_t pid = ptid.pid ();
>> +  int lwp = ptid.lwp ();
>> +  int mib[5] = {CTL_KERN, KERN_LWP, pid, sizeof (struct kinfo_lwp), 0};
>> +  size_t size;
>> +
>> +  if (sysctl (mib, 5, NULL, &size, NULL, 0) == -1 || size == 0)
>> +    perror_with_name (("sysctl"));
>> +
>> +  mib[4] = size / sizeof (size_t);
>> +
>> +  gdb::unique_xmalloc_ptr<struct kinfo_lwp> kl
>> +    ((struct kinfo_lwp *) xcalloc (size, 1));
>> +  if (kl == NULL)
>> +    perror_with_name (("calloc"));
>> +
>> +  if (sysctl (mib, 5, kl.get (), &size, NULL, 0) == -1 || size == 0)
>> +    perror_with_name (("sysctl"));
>> +
>> +  for (size_t i = 0; i < size / sizeof (struct kinfo_lwp); i++)
>> +    {
>> +      if (kl.get ()[i].l_lid == lwp)
>> +	 return true;
>> +    }
>> +
>> +  return false;
>
> Given that you have nearly identical code both here and in the thread_name
> method in your other patch, perhaps it makes sense to have a single
> "gdb::unique_xmalloc_ptr <struct kinfo_lwp> fetch_kinfo_lwp(ptid_t ptid)"
> helper function that the other routines could share?  Perhaps instead of
> just returning the pointer to the array of structures it could take a lamda
> function that is invoked on the matching 'kl' instance with the correct lwp?
>

I can do it. I will submit these two patches in one go. I will also add
a third similar function to list threads within a process.

Normally, I would go for PT_LWPSTATUS, but it is available since NetBSD
>10 and for the time being sysctl(3) is more portable.

Re: [PATCH] Implement nbsd_nat_target::thread_alive

[Simon Marchi]

On 2020-03-25 5:01 p.m., Kamil Rytarowski wrote:
> On 25.03.2020 21:42, John Baldwin wrote:
>> Given that you have nearly identical code both here and in the thread_name
>> method in your other patch, perhaps it makes sense to have a single
>> "gdb::unique_xmalloc_ptr <struct kinfo_lwp> fetch_kinfo_lwp(ptid_t ptid)"
>> helper function that the other routines could share?  Perhaps instead of
>> just returning the pointer to the array of structures it could take a lamda
>> function that is invoked on the matching 'kl' instance with the correct lwp?
>>
> 
> I can do it. I will submit these two patches in one go. I will also add
> a third similar function to list threads within a process.
> 
> Normally, I would go for PT_LWPSTATUS, but it is available since NetBSD
>> 10 and for the time being sysctl(3) is more portable.

And let me mention the same thing I've told John :).  Please add a meaningful
commit message, explaining why you do this change.  It's easy to see _what_
it does, but not _why_ you needed to do it.  Does it enable some feature, does
it fix some test case, something else?  What impact does it have?

Simon

Re: [PATCH] Implement nbsd_nat_target::thread_alive

[Kamil Rytarowski]

On 25.03.2020 22:08, Simon Marchi wrote:
> On 2020-03-25 5:01 p.m., Kamil Rytarowski wrote:
>> On 25.03.2020 21:42, John Baldwin wrote:
>>> Given that you have nearly identical code both here and in the thread_name
>>> method in your other patch, perhaps it makes sense to have a single
>>> "gdb::unique_xmalloc_ptr <struct kinfo_lwp> fetch_kinfo_lwp(ptid_t ptid)"
>>> helper function that the other routines could share?  Perhaps instead of
>>> just returning the pointer to the array of structures it could take a lamda
>>> function that is invoked on the matching 'kl' instance with the correct lwp?
>>>
>>
>> I can do it. I will submit these two patches in one go. I will also add
>> a third similar function to list threads within a process.
>>
>> Normally, I would go for PT_LWPSTATUS, but it is available since NetBSD
>>> 10 and for the time being sysctl(3) is more portable.
>
> And let me mention the same thing I've told John :).  Please add a meaningful
> commit message, explaining why you do this change.  It's easy to see _what_
> it does, but not _why_ you needed to do it.  Does it enable some feature, does
> it fix some test case, something else?  What impact does it have?
>
> Simon
>

Done in [PATCH v2] Implement basic threading support in the NetBSD target.

Please review.