* Full disable of C++20 time zone support
@ 2023-06-28 18:42 Alexey Lapshin
2023-06-28 19:33 ` Jonathan Wakely
0 siblings, 1 reply; 9+ messages in thread
From: Alexey Lapshin @ 2023-06-28 18:42 UTC (permalink / raw)
To: gcc-help; +Cc: jwakely
Hi!
I faced an issue with compiling libstdc++-v3/src/c++20/tzdb.cc:
[ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: 'constinit' variable 'std::chrono::{anonymous}::list_mutex' does not have a constant initializer
[ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: call to non-'constexpr' function 'std::mutex::mutex()'
Any plans to create a configure option to disable tzdb from build? Or at least make it compilable?
Tried --with-libstdcxx-zoneinfo=no but it does not take effect, because it does not cover line with the bug.
For now using workaround with changes in libstdc++-v3/src/c++20/Makefile*
Thanks
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Full disable of C++20 time zone support
2023-06-28 18:42 Full disable of C++20 time zone support Alexey Lapshin
@ 2023-06-28 19:33 ` Jonathan Wakely
2023-06-28 19:40 ` Jonathan Wakely
0 siblings, 1 reply; 9+ messages in thread
From: Jonathan Wakely @ 2023-06-28 19:33 UTC (permalink / raw)
To: Alexey Lapshin; +Cc: gcc-help
[-- Attachment #1: Type: text/plain, Size: 880 bytes --]
On Wed, 28 Jun 2023 at 19:42, Alexey Lapshin wrote:
> Hi!
>
> I faced an issue with compiling libstdc++-v3/src/c++20/tzdb.cc:
>
>
> [ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: 'constinit'
> variable 'std::chrono::{anonymous}::list_mutex' does not have a constant
> initializer
> [ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: call to
> non-'constexpr' function 'std::mutex::mutex()'
>
>
Which target are you building for?
>
> Any plans to create a configure option to disable tzdb from build? Or at
> least make it compilable?
>
> Tried --with-libstdcxx-zoneinfo=no but it does not take effect, because it
> does not cover line with the bug.
>
Well then we should fix that. We don't want *another* configure option when
we already have two for tzdb.
>
> For now using workaround with changes in libstdc++-v3/src/c++20/Makefile*
>
> Thanks
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Full disable of C++20 time zone support
2023-06-28 19:33 ` Jonathan Wakely
@ 2023-06-28 19:40 ` Jonathan Wakely
2023-06-29 9:15 ` Alexey Lapshin
0 siblings, 1 reply; 9+ messages in thread
From: Jonathan Wakely @ 2023-06-28 19:40 UTC (permalink / raw)
To: Alexey Lapshin; +Cc: gcc-help
[-- Attachment #1: Type: text/plain, Size: 1376 bytes --]
On Wed, 28 Jun 2023 at 20:33, Jonathan Wakely <jwakely@redhat.com> wrote:
> On Wed, 28 Jun 2023 at 19:42, Alexey Lapshin wrote:
>
>> Hi!
>>
>> I faced an issue with compiling libstdc++-v3/src/c++20/tzdb.cc:
>>
>>
>> [ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: 'constinit'
>> variable 'std::chrono::{anonymous}::list_mutex' does not have a constant
>> initializer
>> [ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: call to
>> non-'constexpr' function 'std::mutex::mutex()'
>>
>>
> Which target are you building for?
>
>
>
>>
>> Any plans to create a configure option to disable tzdb from build? Or at
>> least make it compilable?
>>
>> Tried --with-libstdcxx-zoneinfo=no but it does not take effect, because
>> it does not cover line with the bug.
>>
>
> Well then we should fix that. We don't want *another* configure option
> when we already have two for tzdb.
>
I think this should fix it:
--- a/libstdc++-v3/src/c++20/tzdb.cc
+++ b/libstdc++-v3/src/c++20/tzdb.cc
@@ -43,8 +43,8 @@
#ifndef __GTHREADS
# define USE_ATOMIC_SHARED_PTR 0
-#elif _WIN32
-// std::mutex cannot be constinit, so Windows must use
atomic<shared_ptr<>>.
+#elif ! defined __GTHREAD_MUTEX_INIT
+// std::mutex cannot be constinit, so must use atomic<shared_ptr<>>.
# define USE_ATOMIC_SHARED_PTR 1
#elif ATOMIC_POINTER_LOCK_FREE < 2
# define USE_ATOMIC_SHARED_PTR 0
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Full disable of C++20 time zone support
2023-06-28 19:40 ` Jonathan Wakely
@ 2023-06-29 9:15 ` Alexey Lapshin
2023-06-29 9:23 ` Jonathan Wakely
0 siblings, 1 reply; 9+ messages in thread
From: Alexey Lapshin @ 2023-06-29 9:15 UTC (permalink / raw)
To: jwakely; +Cc: gcc-help
> Which target are you building for?
riscv32-esp-elf
build with gcc v4.9.2
> I think this should fix it:
This brings next errors:
gcc/libstdc++-v3/src/c++20/tzdb.cc:157:27: error: 'list_mutex' was not declared in this scope
gcc/libstdc++-v3/src/c++20/tzdb.cc:158:28: error: 'using std::chrono::tzdb_list::_Node::head_ptr = class std::atomic<std::shared_ptr<std::chrono::tzdb_list::_Node> >' {aka 'class std::atomic<std::shared_ptr<std::chrono::tzdb_list::_Node> >'} has no member named 'get'
Why it could not be disabled at all? like this:
--- a/libstdc++-v3/src/c++20/tzdb.cc
+++ b/libstdc++-v3/src/c++20/tzdb.cc
@@ -22,6 +22,8 @@
// see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
// <http://www.gnu.org/licenses/>.
+#ifndef TZDB_DISABLED
+
// The -Wabi warnings in this file are all for non-exported symbols.
#pragma GCC diagnostic ignored "-Wabi"
@@ -2037,3 +2039,5 @@ namespace std::chrono
} // namespace
#endif // TZDB_DISABLED
} // namespace std::chrono
+
+#endif
On Wed, 2023-06-28 at 20:40 +0100, Jonathan Wakely wrote:
> [External: This email originated outside Espressif]
>
>
> On Wed, 28 Jun 2023 at 20:33, Jonathan Wakely <jwakely@redhat.com> wrote:
> > On Wed, 28 Jun 2023 at 19:42, Alexey Lapshin wrote:
> > > Hi!
> > >
> > > I faced an issue with compiling libstdc++-v3/src/c++20/tzdb.cc:
> > >
> > >
> > > [ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: 'constinit' variable 'std::chrono::{anonymous}::list_mutex' does not have a constant initializer
> > > [ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: call to non-'constexpr' function 'std::mutex::mutex()'
> > >
> > >
> >
> >
> > Which target are you building for?
> >
> >
> > >
> > > Any plans to create a configure option to disable tzdb from build? Or at least make it compilable?
> > >
> > > Tried --with-libstdcxx-zoneinfo=no but it does not take effect, because it does not cover line with the bug.
> > >
> >
> >
> > Well then we should fix that. We don't want *another* configure option when we already have two for tzdb.
> >
>
>
> I think this should fix it:
>
> --- a/libstdc++-v3/src/c++20/tzdb.cc
> +++ b/libstdc++-v3/src/c++20/tzdb.cc
> @@ -43,8 +43,8 @@
>
> #ifndef __GTHREADS
> # define USE_ATOMIC_SHARED_PTR 0
> -#elif _WIN32
> -// std::mutex cannot be constinit, so Windows must use atomic<shared_ptr<>>.
> +#elif ! defined __GTHREAD_MUTEX_INIT
> +// std::mutex cannot be constinit, so must use atomic<shared_ptr<>>.
> # define USE_ATOMIC_SHARED_PTR 1
> #elif ATOMIC_POINTER_LOCK_FREE < 2
> # define USE_ATOMIC_SHARED_PTR 0
>
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Full disable of C++20 time zone support
2023-06-29 9:15 ` Alexey Lapshin
@ 2023-06-29 9:23 ` Jonathan Wakely
2023-06-29 9:27 ` Alexey Lapshin
0 siblings, 1 reply; 9+ messages in thread
From: Jonathan Wakely @ 2023-06-29 9:23 UTC (permalink / raw)
To: Alexey Lapshin; +Cc: jwakely, gcc-help
On Thu, 29 Jun 2023 at 10:16, Alexey Lapshin via Gcc-help
<gcc-help@gcc.gnu.org> wrote:
>
> > Which target are you building for?
>
> riscv32-esp-elf
>
> build with gcc v4.9.2
>
> > I think this should fix it:
>
> This brings next errors:
>
>
> gcc/libstdc++-v3/src/c++20/tzdb.cc:157:27: error: 'list_mutex' was not declared in this scope
> gcc/libstdc++-v3/src/c++20/tzdb.cc:158:28: error: 'using std::chrono::tzdb_list::_Node::head_ptr = class std::atomic<std::shared_ptr<std::chrono::tzdb_list::_Node> >' {aka 'class std::atomic<std::shared_ptr<std::chrono::tzdb_list::_Node> >'} has no member named 'get'
>
>
>
> Why it could not be disabled at all? like this:
>
>
> --- a/libstdc++-v3/src/c++20/tzdb.cc
> +++ b/libstdc++-v3/src/c++20/tzdb.cc
> @@ -22,6 +22,8 @@
> // see the files COPYING3 and COPYING.RUNTIME respectively. If not, see
> // <http://www.gnu.org/licenses/>.
>
> +#ifndef TZDB_DISABLED
That macro isn't even defined until later in the file, so would never
be defined at this point.
But more importantly, that would give linker errors when including
<chrono>. The point of that macro is to disable trying to process
tzdata, not to create a broken libstdc++ with missing parts of the
C++20 library. It should always be possible to call
std::chrono::get_tzdb() even if the tzdb only contains the UTC time
zone.
> +
> // The -Wabi warnings in this file are all for non-exported symbols.
> #pragma GCC diagnostic ignored "-Wabi"
>
> @@ -2037,3 +2039,5 @@ namespace std::chrono
> } // namespace
> #endif // TZDB_DISABLED
> } // namespace std::chrono
> +
> +#endif
>
>
>
> On Wed, 2023-06-28 at 20:40 +0100, Jonathan Wakely wrote:
> > [External: This email originated outside Espressif]
> >
> >
> > On Wed, 28 Jun 2023 at 20:33, Jonathan Wakely <jwakely@redhat.com> wrote:
> > > On Wed, 28 Jun 2023 at 19:42, Alexey Lapshin wrote:
> > > > Hi!
> > > >
> > > > I faced an issue with compiling libstdc++-v3/src/c++20/tzdb.cc:
> > > >
> > > >
> > > > [ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: 'constinit' variable 'std::chrono::{anonymous}::list_mutex' does not have a constant initializer
> > > > [ERROR] gcc/libstdc++-v3/src/c++20/tzdb.cc:110:21: error: call to non-'constexpr' function 'std::mutex::mutex()'
> > > >
> > > >
> > >
> > >
> > > Which target are you building for?
> > >
> > >
> > > >
> > > > Any plans to create a configure option to disable tzdb from build? Or at least make it compilable?
> > > >
> > > > Tried --with-libstdcxx-zoneinfo=no but it does not take effect, because it does not cover line with the bug.
> > > >
> > >
> > >
> > > Well then we should fix that. We don't want *another* configure option when we already have two for tzdb.
> > >
> >
> >
> > I think this should fix it:
> >
> > --- a/libstdc++-v3/src/c++20/tzdb.cc
> > +++ b/libstdc++-v3/src/c++20/tzdb.cc
> > @@ -43,8 +43,8 @@
> >
> > #ifndef __GTHREADS
> > # define USE_ATOMIC_SHARED_PTR 0
> > -#elif _WIN32
> > -// std::mutex cannot be constinit, so Windows must use atomic<shared_ptr<>>.
> > +#elif ! defined __GTHREAD_MUTEX_INIT
> > +// std::mutex cannot be constinit, so must use atomic<shared_ptr<>>.
> > # define USE_ATOMIC_SHARED_PTR 1
> > #elif ATOMIC_POINTER_LOCK_FREE < 2
> > # define USE_ATOMIC_SHARED_PTR 0
> >
> >
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Full disable of C++20 time zone support
2023-06-29 9:23 ` Jonathan Wakely
@ 2023-06-29 9:27 ` Alexey Lapshin
2023-06-29 9:39 ` Jonathan Wakely
0 siblings, 1 reply; 9+ messages in thread
From: Alexey Lapshin @ 2023-06-29 9:27 UTC (permalink / raw)
To: jwakely.gcc; +Cc: jwakely, gcc-help
On Thu, 2023-06-29 at 10:23 +0100, Jonathan Wakely wrote:
> That macro isn't even defined until later in the file, so would never
> be defined at this point.
Was a simplified example.
Why do you still need to have some mutexes or whatever in case TZDB_DISABLED?
> It should always be possible to call
std::chrono::get_tzdb() even if the tzdb only contains the UTC time
zone.
Just return NULL in case TZDB_DISABLED?
Thanks
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Full disable of C++20 time zone support
2023-06-29 9:27 ` Alexey Lapshin
@ 2023-06-29 9:39 ` Jonathan Wakely
2023-06-29 12:11 ` Jonathan Wakely
0 siblings, 1 reply; 9+ messages in thread
From: Jonathan Wakely @ 2023-06-29 9:39 UTC (permalink / raw)
To: Alexey Lapshin; +Cc: jwakely, gcc-help
On Thu, 29 Jun 2023 at 10:27, Alexey Lapshin
<alexey.lapshin@espressif.com> wrote:
>
> On Thu, 2023-06-29 at 10:23 +0100, Jonathan Wakely wrote:
> > That macro isn't even defined until later in the file, so would never
> > be defined at this point.
>
> Was a simplified example.
>
>
> Why do you still need to have some mutexes or whatever in case TZDB_DISABLED?
Because there is still a global object that needs to be lazily
constructed, and multiple threads can try to access it concurrently.
> > It should always be possible to call
> std::chrono::get_tzdb() even if the tzdb only contains the UTC time
> zone.
>
> Just return NULL in case TZDB_DISABLED?
It returns a reference.
As I said (and you quoted above), the intention is to return a valid
object that contains the UTC time zone. That is a graceful failure for
missing tzdata, not a crash.
>
> Thanks
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Full disable of C++20 time zone support
2023-06-29 9:39 ` Jonathan Wakely
@ 2023-06-29 12:11 ` Jonathan Wakely
2023-06-29 13:51 ` Alexey Lapshin
0 siblings, 1 reply; 9+ messages in thread
From: Jonathan Wakely @ 2023-06-29 12:11 UTC (permalink / raw)
To: Alexey Lapshin; +Cc: jwakely, gcc-help
[-- Attachment #1: Type: text/plain, Size: 627 bytes --]
On Thu, 29 Jun 2023 at 10:39, Jonathan Wakely wrote:
>
> On Thu, 29 Jun 2023 at 10:27, Alexey Lapshin
> <alexey.lapshin@espressif.com> wrote:
> >
> > On Thu, 2023-06-29 at 10:23 +0100, Jonathan Wakely wrote:
> > > That macro isn't even defined until later in the file, so would never
> > > be defined at this point.
> >
> > Was a simplified example.
> >
> >
> > Why do you still need to have some mutexes or whatever in case TZDB_DISABLED?
>
> Because there is still a global object that needs to be lazily
> constructed, and multiple threads can try to access it concurrently.
The attached patch should work for your target.
[-- Attachment #2: patch.txt --]
[-- Type: text/plain, Size: 5030 bytes --]
commit 881299967d8778e29f803d8801af9a0b1a5bd1bd
Author: Jonathan Wakely <jwakely@redhat.com>
Date: Thu Jun 29 11:40:32 2023
libstdc++: Fix src/c++20/tzdb.cc for non-constexpr std::mutex
Building libstdc++ reportedly fails for targets without lock-free
std::atomic<T*> which don't define __GTHREAD_MUTEX_INIT:
src/c++20/tzdb.cc:110:21: error: 'constinit' variable 'std::chrono::{anonymous}::list_mutex' does not have a constant initializer
src/c++20/tzdb.cc:110:21: error: call to non-'constexpr' function 'std::mutex::mutex()'
The solution implemented by this commit is to use a local static mutex
when it can't be constinit, so that it's constructed on first use.
With this change, we can also simplify the preprocessor logic for
defining USE_ATOMIC_SHARED_PTR. It now depends on the same conditions as
USE_ATOMIC_LIST_HEAD, so in theory we could have a single macro. Keeping
them separate would allow us to replace the use of atomic<shared_ptr<T>>
with a mutex if that performs better, without having to give up on the
lock-free cache for fast access to the list head.
libstdc++-v3/ChangeLog:
* src/c++20/tzdb.cc (USE_ATOMIC_SHARED_PTR): Define consistently
with USE_ATOMIC_LIST_HEAD.
(list_mutex): Replace global object with function. Use local
static object when std::mutex constructor isn't constexpr.
diff --git a/libstdc++-v3/src/c++20/tzdb.cc b/libstdc++-v3/src/c++20/tzdb.cc
index a43b4f33eba..8d27726016e 100644
--- a/libstdc++-v3/src/c++20/tzdb.cc
+++ b/libstdc++-v3/src/c++20/tzdb.cc
@@ -41,22 +41,17 @@
# include <cstdlib> // getenv
#endif
-#ifndef __GTHREADS
-# define USE_ATOMIC_SHARED_PTR 0
-#elif _WIN32
-// std::mutex cannot be constinit, so Windows must use atomic<shared_ptr<>>.
-# define USE_ATOMIC_SHARED_PTR 1
-#elif ATOMIC_POINTER_LOCK_FREE < 2
-# define USE_ATOMIC_SHARED_PTR 0
-#else
-// TODO benchmark atomic<shared_ptr<>> vs mutex.
-# define USE_ATOMIC_SHARED_PTR 1
-#endif
-
#if defined __GTHREADS && ATOMIC_POINTER_LOCK_FREE == 2
# define USE_ATOMIC_LIST_HEAD 1
+// TODO benchmark atomic<shared_ptr<>> vs mutex.
+# define USE_ATOMIC_SHARED_PTR 1
#else
# define USE_ATOMIC_LIST_HEAD 0
+# define USE_ATOMIC_SHARED_PTR 0
+#endif
+
+#if USE_ATOMIC_SHARED_PTR && ! USE_ATOMIC_LIST_HEAD
+# error Unsupported combination
#endif
#if ! __cpp_constinit
@@ -106,9 +101,18 @@ namespace std::chrono
// Dummy no-op mutex type for single-threaded targets.
struct mutex { void lock() { } void unlock() { } };
#endif
- /// XXX std::mutex::mutex() not constexpr on Windows, so can't be constinit
- constinit mutex list_mutex;
+ inline mutex& list_mutex()
+ {
+#ifdef __GTHREAD_MUTEX_INIT
+ constinit static mutex m;
+#else
+ // Cannot use a constinit mutex, so use a local static.
+ alignas(mutex) constinit static char buf[sizeof(mutex)];
+ static mutex& m = *::new(buf) mutex();
#endif
+ return m;
+ }
+#endif // ! USE_ATOMIC_SHARED_PTR
struct Rule;
}
@@ -154,7 +158,7 @@ namespace std::chrono
static _Node*
_S_list_head(memory_order)
{
- lock_guard<mutex> l(list_mutex);
+ lock_guard<mutex> l(list_mutex());
return _S_head_owner.get();
}
@@ -1279,7 +1283,7 @@ namespace std::chrono
}
// XXX small window here where _S_head_cache still points to previous tzdb.
#else
- lock_guard<mutex> l(list_mutex);
+ lock_guard<mutex> l(list_mutex());
if (const _Node* h = _S_head_owner.get())
{
if (h->db.version == new_head_ptr->db.version)
@@ -1406,11 +1410,12 @@ namespace std::chrono
#else
if (Node::_S_list_head(memory_order::relaxed) != nullptr) [[likely]]
{
- lock_guard<mutex> l(list_mutex);
+ lock_guard<mutex> l(list_mutex());
const tzdb& current = Node::_S_head_owner->db;
if (current.version == version)
return current;
}
+ shared_ptr<Node> head; // Passed as unused arg to _S_replace_head.
#endif
auto [leaps, leaps_ok] = Node::_S_read_leap_seconds();
@@ -1499,9 +1504,6 @@ namespace std::chrono
ranges::sort(node->db.links, {}, &time_zone_link::name);
ranges::stable_sort(node->rules, {}, &Rule::name);
-#if ! USE_ATOMIC_SHARED_PTR
- shared_ptr<Node> head;
-#endif
return Node::_S_replace_head(std::move(head), std::move(node));
#else
__throw_disabled();
@@ -1526,7 +1528,7 @@ namespace std::chrono
#if USE_ATOMIC_SHARED_PTR
return const_iterator{_Node::_S_head_owner.load()};
#else
- lock_guard<mutex> l(list_mutex);
+ lock_guard<mutex> l(list_mutex());
return const_iterator{_Node::_S_head_owner};
#endif
}
@@ -1539,7 +1541,7 @@ namespace std::chrono
if (p._M_node) [[likely]]
{
#if ! USE_ATOMIC_SHARED_PTR
- lock_guard<mutex> l(list_mutex);
+ lock_guard<mutex> l(list_mutex());
#endif
if (auto next = p._M_node->next) [[likely]]
return const_iterator{p._M_node->next = std::move(next->next)};
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: Full disable of C++20 time zone support
2023-06-29 12:11 ` Jonathan Wakely
@ 2023-06-29 13:51 ` Alexey Lapshin
0 siblings, 0 replies; 9+ messages in thread
From: Alexey Lapshin @ 2023-06-29 13:51 UTC (permalink / raw)
To: jwakely.gcc; +Cc: jwakely, gcc-help
The patch does work. Thank you for a quick fix!
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2023-06-29 13:51 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-06-28 18:42 Full disable of C++20 time zone support Alexey Lapshin
2023-06-28 19:33 ` Jonathan Wakely
2023-06-28 19:40 ` Jonathan Wakely
2023-06-29 9:15 ` Alexey Lapshin
2023-06-29 9:23 ` Jonathan Wakely
2023-06-29 9:27 ` Alexey Lapshin
2023-06-29 9:39 ` Jonathan Wakely
2023-06-29 12:11 ` Jonathan Wakely
2023-06-29 13:51 ` Alexey Lapshin
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).