Patchwork nptl: Document AS-safe functions in cancellation.c.

login
register
mail settings
Submitter Carlos O'Donell
Date Oct. 7, 2019, 5:35 a.m.
Message ID <5ad2611d-bc28-524d-674e-9fa1f7a9add5@redhat.com>
Download mbox | patch
Permalink /patch/34837/
State New
Headers show

Comments

Carlos O'Donell - Oct. 7, 2019, 5:35 a.m.
On 10/4/19 4:45 PM, Zack Weinberg wrote:
> On Fri, Oct 4, 2019 at 4:01 PM Carlos O'Donell <carlos@redhat.com> wrote:
>>
>> In a recent conversation with Mathieu Desnoyer he pointed out that
>> because write is AS-safe all the wrappers around write should be
>> also AS-safe. We don't spell that out explicitly in the comments
>> for __pthread_enable_asynccancel and __pthread_disable_asynccancel
>> so I added them here.
>>
>>
>>  /* The next two functions are similar to pthread_setcanceltype() but
>>     more specialized for the use in the cancelable functions like write().
>> -   They do not need to check parameters etc.  */
>> +   They do not need to check parameters etc.  This function must be
>> +   AS-safe, with the exception of the actual cancellation, because they
>> +   are called by wrappers around AS-safe functions like write().*/
> 
> Grammar nit: Either "_These functions_ must be AS-safe, ..."  or
> "because _it is_ called."  The former is probably better because
> consistent with the first sentence in this comment...

Yes, I actually wrote "These" first, but then cut-and-pasted it into
both functions, and rewrote "These" to "This" and forgot to clean it up.
Thanks for the review! 
>> -
>> +/* This function must be AS-safe, with the exception of the actual
>> +   cancellation, because they are called by wrappers around AS-safe
>> +   functions like write().*/
> 
> ... but then maybe you _don't_ want to add this comment as well.

v2
- Rework comments based on Zack's review.

How about v2?

8< --- 8< --- 8<
Document in comments that __pthread_enable_asynccancel and
__pthread_disable_asynccancel must be AS-safe in general with
the exception of the act of cancellation.
---
 ChangeLog           | 5 +++++
 nptl/cancellation.c | 7 +++++--
 2 files changed, 10 insertions(+), 2 deletions(-)
Zack Weinberg - Oct. 7, 2019, 4:58 p.m.
On Mon, Oct 7, 2019 at 1:35 AM Carlos O'Donell <carlos@redhat.com> wrote:
>
> How about v2?

Revised patch LGTM.

zw
Carlos O'Donell - Oct. 18, 2019, 8:02 p.m.
On 10/7/19 12:58 PM, Zack Weinberg wrote:
> On Mon, Oct 7, 2019 at 1:35 AM Carlos O'Donell <carlos@redhat.com> wrote:
>>
>> How about v2?
> 
> Revised patch LGTM.

Pushed. Thanks.

Patch

diff --git a/ChangeLog b/ChangeLog
index cfabdaddf4..2ce48223f4 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,8 @@ 
+2019-10-07  Carlos O'Donell  <carlos@redhat.com>
+
+	* nptl/cancellation.c: Document that all functions here must be
+	AS-safe because they wrap AS-safe functions.
+
 2019-10-07  Leandro Pereira  <leandro.pereira@microsoft.com>
 
 	* elf/dl-load.c: Use __pread64_nocancel() instead of __lseek()+
diff --git a/nptl/cancellation.c b/nptl/cancellation.c
index 7712845561..23d7b2b34c 100644
--- a/nptl/cancellation.c
+++ b/nptl/cancellation.c
@@ -24,7 +24,9 @@ 
 
 /* The next two functions are similar to pthread_setcanceltype() but
    more specialized for the use in the cancelable functions like write().
-   They do not need to check parameters etc.  */
+   They do not need to check parameters etc.  These functions must be
+   AS-safe, with the exception of the actual cancellation, because they
+   are called by wrappers around AS-safe functions like write().*/
 int
 attribute_hidden
 __pthread_enable_asynccancel (void)
@@ -59,7 +61,8 @@  __pthread_enable_asynccancel (void)
   return oldval;
 }
 
-
+/* See the comment for __pthread_enable_asynccancel regarding
+   the AS-safety of this function.  */
 void
 attribute_hidden
 __pthread_disable_asynccancel (int oldtype)