[PATCH] rseq/selftests: Use weak symbol reference, not definition, to link with glibc

[PATCH] rseq/selftests: Use weak symbol reference, not definition, to link with glibc

[Sean Christopherson]

Add "extern" to the glibc-defined weak rseq symbols to convert the rseq
selftest's usage from weak symbol definitions to weak symbol _references_.
Effectively re-defining the glibc symbols wreaks havoc when building with
-fno-common, e.g. generates segfaults when running multi-threaded programs,
as dynamically linked applications end up with multiple versions of the
symbols.

Building with -fcommon, which until recently has the been the default for
GCC and clang, papers over the bug by allowing the linker to resolve the
weak/tentative definition to glibc's "real" definition.

Note, the symbol itself (or rather its address), not the value of the
symbol, is set to 0/NULL for unresolved weak symbol references, as the
symbol doesn't exist and thus can't have a value.  Check for a NULL rseq
size pointer to handle the scenario where the test is statically linked
against a libc that doesn't support rseq in any capacity.

Fixes: 3bcbc20942db ("selftests/rseq: Play nice with binaries statically linked against glibc 2.35+")
Cc: stable@vger.kernel.org
Suggested-by: Florian Weimer <fweimer@redhat.com>
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Closes: https://lore.kernel.org/all/87frdoybk4.ffs@tglx
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 tools/testing/selftests/rseq/rseq.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/rseq/rseq.c b/tools/testing/selftests/rseq/rseq.c
index 663a9cef1952..dcac5cbe7933 100644
--- a/tools/testing/selftests/rseq/rseq.c
+++ b/tools/testing/selftests/rseq/rseq.c
@@ -40,9 +40,9 @@
  * Define weak versions to play nice with binaries that are statically linked
  * against a libc that doesn't support registering its own rseq.
  */
-__weak ptrdiff_t __rseq_offset;
-__weak unsigned int __rseq_size;
-__weak unsigned int __rseq_flags;
+extern __weak ptrdiff_t __rseq_offset;
+extern __weak unsigned int __rseq_size;
+extern __weak unsigned int __rseq_flags;
 
 static const ptrdiff_t *libc_rseq_offset_p = &__rseq_offset;
 static const unsigned int *libc_rseq_size_p = &__rseq_size;
@@ -209,7 +209,7 @@ void rseq_init(void)
 	 * libc not having registered a restartable sequence.  Try to find the
 	 * symbols if that's the case.
 	 */
-	if (!*libc_rseq_size_p) {
+	if (!libc_rseq_size_p || !*libc_rseq_size_p) {
 		libc_rseq_offset_p = dlsym(RTLD_NEXT, "__rseq_offset");
 		libc_rseq_size_p = dlsym(RTLD_NEXT, "__rseq_size");
 		libc_rseq_flags_p = dlsym(RTLD_NEXT, "__rseq_flags");

base-commit: c17b750b3ad9f45f2b6f7e6f7f4679844244f0b9
-- 
2.51.0.rc1.167.g924127e9c0-goog

Re: [PATCH] rseq/selftests: Use weak symbol reference, not definition, to link with glibc

[Mathieu Desnoyers]

On 2025-08-19 18:29, Sean Christopherson wrote:
> Add "extern" to the glibc-defined weak rseq symbols to convert the rseq
> selftest's usage from weak symbol definitions to weak symbol _references_.
> Effectively re-defining the glibc symbols wreaks havoc when building with
> -fno-common, e.g. generates segfaults when running multi-threaded programs,
> as dynamically linked applications end up with multiple versions of the
> symbols.
> 
> Building with -fcommon, which until recently has the been the default for
> GCC and clang, papers over the bug by allowing the linker to resolve the
> weak/tentative definition to glibc's "real" definition.
> 
> Note, the symbol itself (or rather its address), not the value of the
> symbol, is set to 0/NULL for unresolved weak symbol references, as the
> symbol doesn't exist and thus can't have a value.  Check for a NULL rseq
> size pointer to handle the scenario where the test is statically linked
> against a libc that doesn't support rseq in any capacity.
> 
> Fixes: 3bcbc20942db ("selftests/rseq: Play nice with binaries statically linked against glibc 2.35+")
> Cc: stable@vger.kernel.org
> Suggested-by: Florian Weimer <fweimer@redhat.com>
> Reported-by: Thomas Gleixner <tglx@linutronix.de>
> Closes: https://lore.kernel.org/all/87frdoybk4.ffs@tglx
> Signed-off-by: Sean Christopherson <seanjc@google.com>

+ CC: Michael Jeanson <mjeanson@efficios.com>

> ---
>   tools/testing/selftests/rseq/rseq.c | 8 ++++----
>   1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/tools/testing/selftests/rseq/rseq.c b/tools/testing/selftests/rseq/rseq.c
> index 663a9cef1952..dcac5cbe7933 100644
> --- a/tools/testing/selftests/rseq/rseq.c
> +++ b/tools/testing/selftests/rseq/rseq.c
> @@ -40,9 +40,9 @@
>    * Define weak versions to play nice with binaries that are statically linked
>    * against a libc that doesn't support registering its own rseq.

We should therefore update the comment above to e.g.:

/*
  * Define weak symbol references to play nice with binaries that are
  * statically linked against a libc that doesn't support registering its
  * own rseq.
  */

For the rest:

Reviewed-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>

Michael, can you try it out ?

Kienan, we may want to add a configuration forcing "-fno-common" to our
CI, this problematic pattern may be hiding other issues elsewhere. I'm
thinking of LTTng-UST tracepoint headers and libside headers.

Sean, do you want to contribute the fix to librseq as well ?

Thanks everyone for looking into this. I'll be back from vacation next
week and will resume normal operations. :-)

Mathieu

>    */
> -__weak ptrdiff_t __rseq_offset;
> -__weak unsigned int __rseq_size;
> -__weak unsigned int __rseq_flags;
> +extern __weak ptrdiff_t __rseq_offset;
> +extern __weak unsigned int __rseq_size;
> +extern __weak unsigned int __rseq_flags;
>   
>   static const ptrdiff_t *libc_rseq_offset_p = &__rseq_offset;
>   static const unsigned int *libc_rseq_size_p = &__rseq_size;
> @@ -209,7 +209,7 @@ void rseq_init(void)
>   	 * libc not having registered a restartable sequence.  Try to find the
>   	 * symbols if that's the case.
>   	 */
> -	if (!*libc_rseq_size_p) {
> +	if (!libc_rseq_size_p || !*libc_rseq_size_p) {
>   		libc_rseq_offset_p = dlsym(RTLD_NEXT, "__rseq_offset");
>   		libc_rseq_size_p = dlsym(RTLD_NEXT, "__rseq_size");
>   		libc_rseq_flags_p = dlsym(RTLD_NEXT, "__rseq_flags");
> 
> base-commit: c17b750b3ad9f45f2b6f7e6f7f4679844244f0b9


-- 
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com

Re: [PATCH] rseq/selftests: Use weak symbol reference, not definition, to link with glibc

[Michael Jeanson]

On 2025-08-20 08:55, Mathieu Desnoyers wrote:
> 
> Michael, can you try it out ?

Will do.

> 
> Kienan, we may want to add a configuration forcing "-fno-common" to our
> CI, this problematic pattern may be hiding other issues elsewhere. I'm
> thinking of LTTng-UST tracepoint headers and libside headers.

AFAIK, -fno-common has been the default since GCC 10 so it's already 
well tested in CI, there are a few fixes from around 2020 in LTTng 
related to this.

> 
> Sean, do you want to contribute the fix to librseq as well ?

Librseq currently doesn't have the weak symbols like the selftests so 
the fix doesn't really apply.

> 
> Thanks everyone for looking into this. I'll be back from vacation next
> week and will resume normal operations. :-)
> 
> Mathieu
> 
>>     */
>> -__weak ptrdiff_t __rseq_offset;
>> -__weak unsigned int __rseq_size;
>> -__weak unsigned int __rseq_flags;
>> +extern __weak ptrdiff_t __rseq_offset;
>> +extern __weak unsigned int __rseq_size;
>> +extern __weak unsigned int __rseq_flags;
>>    
>>    static const ptrdiff_t *libc_rseq_offset_p = &__rseq_offset;
>>    static const unsigned int *libc_rseq_size_p = &__rseq_size;
>> @@ -209,7 +209,7 @@ void rseq_init(void)
>>    	 * libc not having registered a restartable sequence.  Try to find the
>>    	 * symbols if that's the case.
>>    	 */
>> -	if (!*libc_rseq_size_p) {
>> +	if (!libc_rseq_size_p || !*libc_rseq_size_p) {
>>    		libc_rseq_offset_p = dlsym(RTLD_NEXT, "__rseq_offset");
>>    		libc_rseq_size_p = dlsym(RTLD_NEXT, "__rseq_size");
>>    		libc_rseq_flags_p = dlsym(RTLD_NEXT, "__rseq_flags");
>>
>> base-commit: c17b750b3ad9f45f2b6f7e6f7f4679844244f0b9
> 
> 

Re: [PATCH] rseq/selftests: Use weak symbol reference, not definition, to link with glibc

[Michael Jeanson]

On 2025-08-20 08:55, Mathieu Desnoyers wrote:
> Michael, can you try it out ?

Tested on Debian 13 amd64 (Glibc 2.41 and GCC 14.2.0), no segfaults, all 
the rseq tests pass.