diff mbox series

[v3] linux-user/s390x: Use the guest pointer for the sigreturn stub

Message ID 20210324185128.63971-1-krebbel@linux.ibm.com (mailing list archive)
State New, archived
Headers show
Series [v3] linux-user/s390x: Use the guest pointer for the sigreturn stub | expand

Commit Message

Andreas Krebbel March 24, 2021, 6:51 p.m. UTC
When setting up the pointer for the sigreturn stub in the return
address register (r14) we currently use the host frame address instead
of the guest frame address.

Note: This only caused problems if Qemu has been built with
--disable-pie (as it is in distros nowadays). Otherwise guest_base
defaults to 0 hiding the actual problem.

Signed-off-by: Andreas Krebbel <krebbel@linux.ibm.com>
---
 linux-user/s390x/signal.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

Comments

Laurent Vivier March 25, 2021, 10:42 a.m. UTC | #1
Le 24/03/2021 à 19:51, Andreas Krebbel a écrit :
> When setting up the pointer for the sigreturn stub in the return
> address register (r14) we currently use the host frame address instead
> of the guest frame address.
> 
> Note: This only caused problems if Qemu has been built with
> --disable-pie (as it is in distros nowadays). Otherwise guest_base
> defaults to 0 hiding the actual problem.
> 
> Signed-off-by: Andreas Krebbel <krebbel@linux.ibm.com>
> ---
>  linux-user/s390x/signal.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/linux-user/s390x/signal.c b/linux-user/s390x/signal.c
> index ecfa2a14a9..7107c5fb53 100644
> --- a/linux-user/s390x/signal.c
> +++ b/linux-user/s390x/signal.c
> @@ -211,9 +211,10 @@ void setup_rt_frame(int sig, struct target_sigaction *ka,
>      /* Set up to return from userspace.  If provided, use a stub
>         already in userspace.  */
>      if (ka->sa_flags & TARGET_SA_RESTORER) {
> -        env->regs[14] = (unsigned long) ka->sa_restorer | PSW_ADDR_AMODE;
> +        env->regs[14] = ka->sa_restorer | PSW_ADDR_AMODE;
>      } else {
> -        env->regs[14] = (unsigned long) frame->retcode | PSW_ADDR_AMODE;
> +        env->regs[14] = (frame_addr + offsetof(typeof(*frame), retcode))
> +                        | PSW_ADDR_AMODE;
>          __put_user(S390_SYSCALL_OPCODE | TARGET_NR_rt_sigreturn,
>                     (uint16_t *)(frame->retcode));
>      }
> 

Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Richard Henderson March 25, 2021, 12:46 p.m. UTC | #2
On 3/24/21 12:51 PM, Andreas Krebbel wrote:
> When setting up the pointer for the sigreturn stub in the return
> address register (r14) we currently use the host frame address instead
> of the guest frame address.
> 
> Note: This only caused problems if Qemu has been built with
> --disable-pie (as it is in distros nowadays). Otherwise guest_base
> defaults to 0 hiding the actual problem.
> 
> Signed-off-by: Andreas Krebbel<krebbel@linux.ibm.com>
> ---
>   linux-user/s390x/signal.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

r~
Laurent Vivier March 25, 2021, 7:55 p.m. UTC | #3
Le 24/03/2021 à 19:51, Andreas Krebbel a écrit :
> When setting up the pointer for the sigreturn stub in the return
> address register (r14) we currently use the host frame address instead
> of the guest frame address.
> 
> Note: This only caused problems if Qemu has been built with
> --disable-pie (as it is in distros nowadays). Otherwise guest_base
> defaults to 0 hiding the actual problem.
> 
> Signed-off-by: Andreas Krebbel <krebbel@linux.ibm.com>
> ---
>  linux-user/s390x/signal.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/linux-user/s390x/signal.c b/linux-user/s390x/signal.c
> index ecfa2a14a9..7107c5fb53 100644
> --- a/linux-user/s390x/signal.c
> +++ b/linux-user/s390x/signal.c
> @@ -211,9 +211,10 @@ void setup_rt_frame(int sig, struct target_sigaction *ka,
>      /* Set up to return from userspace.  If provided, use a stub
>         already in userspace.  */
>      if (ka->sa_flags & TARGET_SA_RESTORER) {
> -        env->regs[14] = (unsigned long) ka->sa_restorer | PSW_ADDR_AMODE;
> +        env->regs[14] = ka->sa_restorer | PSW_ADDR_AMODE;
>      } else {
> -        env->regs[14] = (unsigned long) frame->retcode | PSW_ADDR_AMODE;
> +        env->regs[14] = (frame_addr + offsetof(typeof(*frame), retcode))
> +                        | PSW_ADDR_AMODE;
>          __put_user(S390_SYSCALL_OPCODE | TARGET_NR_rt_sigreturn,
>                     (uint16_t *)(frame->retcode));
>      }
> 

Applied to my linux-user-for-6.0 branch

Thanks,
Laurent
Andreas Krebbel March 26, 2021, 6:51 a.m. UTC | #4
On 3/25/21 8:55 PM, Laurent Vivier wrote:
> Le 24/03/2021 à 19:51, Andreas Krebbel a écrit :
>> When setting up the pointer for the sigreturn stub in the return
>> address register (r14) we currently use the host frame address instead
>> of the guest frame address.
>>
>> Note: This only caused problems if Qemu has been built with
>> --disable-pie (as it is in distros nowadays). Otherwise guest_base
>> defaults to 0 hiding the actual problem.
>>
>> Signed-off-by: Andreas Krebbel <krebbel@linux.ibm.com>
>> ---
>>  linux-user/s390x/signal.c | 5 +++--
>>  1 file changed, 3 insertions(+), 2 deletions(-)
>>
>> diff --git a/linux-user/s390x/signal.c b/linux-user/s390x/signal.c
>> index ecfa2a14a9..7107c5fb53 100644
>> --- a/linux-user/s390x/signal.c
>> +++ b/linux-user/s390x/signal.c
>> @@ -211,9 +211,10 @@ void setup_rt_frame(int sig, struct target_sigaction *ka,
>>      /* Set up to return from userspace.  If provided, use a stub
>>         already in userspace.  */
>>      if (ka->sa_flags & TARGET_SA_RESTORER) {
>> -        env->regs[14] = (unsigned long) ka->sa_restorer | PSW_ADDR_AMODE;
>> +        env->regs[14] = ka->sa_restorer | PSW_ADDR_AMODE;
>>      } else {
>> -        env->regs[14] = (unsigned long) frame->retcode | PSW_ADDR_AMODE;
>> +        env->regs[14] = (frame_addr + offsetof(typeof(*frame), retcode))
>> +                        | PSW_ADDR_AMODE;
>>          __put_user(S390_SYSCALL_OPCODE | TARGET_NR_rt_sigreturn,
>>                     (uint16_t *)(frame->retcode));
>>      }
>>
> 
> Applied to my linux-user-for-6.0 branch

Thanks! Since this fixes a problem which prevents important applications (e.g. golang) to run
properly under Qemu for IBM Z right now I think it qualifies for stable branch.

The patch applies cleanly on 5.0 branch. Testsuite is clean and I've verfified that it fixes the
actual problem there as well.

Please consider inclusion into stable branch.

Andreas

> 
> Thanks,
> Laurent
>
diff mbox series

Patch

diff --git a/linux-user/s390x/signal.c b/linux-user/s390x/signal.c
index ecfa2a14a9..7107c5fb53 100644
--- a/linux-user/s390x/signal.c
+++ b/linux-user/s390x/signal.c
@@ -211,9 +211,10 @@  void setup_rt_frame(int sig, struct target_sigaction *ka,
     /* Set up to return from userspace.  If provided, use a stub
        already in userspace.  */
     if (ka->sa_flags & TARGET_SA_RESTORER) {
-        env->regs[14] = (unsigned long) ka->sa_restorer | PSW_ADDR_AMODE;
+        env->regs[14] = ka->sa_restorer | PSW_ADDR_AMODE;
     } else {
-        env->regs[14] = (unsigned long) frame->retcode | PSW_ADDR_AMODE;
+        env->regs[14] = (frame_addr + offsetof(typeof(*frame), retcode))
+                        | PSW_ADDR_AMODE;
         __put_user(S390_SYSCALL_OPCODE | TARGET_NR_rt_sigreturn,
                    (uint16_t *)(frame->retcode));
     }