[PATCH] libstdc++: eh_globals: gthreads: reset _S_init before deleting key

[PATCH] libstdc++: eh_globals: gthreads: reset _S_init before deleting key

[Alexandre Oliva]

Clear __eh_globals_init's _S_init in the dtor before deleting the
gthread key.

This ensures that, in case any code involved in deleting the key
interacts with eh_globals, the key that is being deleted won't be
used, and the non-thread-specific eh_globals fallback will.

Regstrapped on x86_64-linux-gnu, also tested with a cross to
aarch64-rtems6.  Ok to install?

PS: This is a fix for a theoretical issue, that came to mind while I
independently investigated the problem that I later found to be
PR105880.


for  libstdc++-v3/ChangeLog

	* libsupc++/eh_globals.cc [!_GLIBCXX_HAVE_TLS]
	(__eh_globals_init::~__eh_globals_init): Clear _S_init first.
---
 libstdc++-v3/libsupc++/eh_globals.cc |    9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/libstdc++-v3/libsupc++/eh_globals.cc b/libstdc++-v3/libsupc++/eh_globals.cc
index 768425c0f407a..0aadb692a9625 100644
--- a/libstdc++-v3/libsupc++/eh_globals.cc
+++ b/libstdc++-v3/libsupc++/eh_globals.cc
@@ -119,8 +119,13 @@ struct __eh_globals_init
   ~__eh_globals_init()
   {
     if (_S_init)
-      __gthread_key_delete(_M_key);
-    _S_init = false;
+      {
+	/* Set it before the call, so that, should
+	   __gthread_key_delete throw an exception, it won't rely on
+	   the key being deleted.  */
+	_S_init = false;
+	__gthread_key_delete(_M_key);
+      }
   }
 
   __eh_globals_init(const __eh_globals_init&) = delete;

-- 
Alexandre Oliva, happy hacker                https://FSFLA.org/blogs/lxo/
   Free Software Activist                       GNU Toolchain Engineer
Disinformation flourishes because many people care deeply about injustice
but very few check the facts.  Ask me about <https://stallmansupport.org>

Re: [PATCH] libstdc++: eh_globals: gthreads: reset _S_init before deleting key

[Jonathan Wakely]

On Tue, 21 Jun 2022 at 07:04, Alexandre Oliva via Libstdc++
<libstdc++@gcc.gnu.org> wrote:
>
>
> Clear __eh_globals_init's _S_init in the dtor before deleting the
> gthread key.
>
> This ensures that, in case any code involved in deleting the key
> interacts with eh_globals, the key that is being deleted won't be
> used, and the non-thread-specific eh_globals fallback will.
>
> Regstrapped on x86_64-linux-gnu, also tested with a cross to
> aarch64-rtems6.  Ok to install?

OK, thanks.

> PS: This is a fix for a theoretical issue, that came to mind while I
> independently investigated the problem that I later found to be
> PR105880.

It looks like a real problem though, if rare in practice.