Re: [Qemu-devel] SH4 TARGET_NR_clone

[michael <trimarchi@gandalf.sssup.it>]

Hi,


Thiemo Seufer wrote:
> michael wrote:
>   
>> Hi,
>>
>> This patch fix the call of Clone in SH4 system
>>     
>
> What exactly does it fix? The kernel side of sys_clone seems to be no
> different to other architectures.
>   
Look at the libc code:

nptl/sysdeps/unix/sysv/linux/fork.c

the fork is implemente in this file. The ARCH_FORK is different in
i386 an sh code.

sh code:

/* TLS pointer argument is passed as the 5-th argument.  */
#define ARCH_FORK() \
  INLINE_SYSCALL (clone, 
5,                                                   \
                  CLONE_CHILD_SETTID | CLONE_CHILD_CLEARTID | SIGCHLD, 
0,     \
                  NULL, &THREAD_SELF->tid, NULL)

i386 code:

#define ARCH_FORK() \
  INLINE_SYSCALL (clone, 
5,                                                   \
                  CLONE_CHILD_SETTID | CLONE_CHILD_CLEARTID | SIGCHLD, 
0,     \
                  NULL, NULL, &THREAD_SELF->tid)


As you can see the 4th and 5th argument are inverted, and it maybe 
justify my problem.

>
> Thiemo
>
>
>   
>> This patch implements the correct TARGET_NR_clone for SH4
>> cpu.
>>
>> Signed-off-by: Michael Trimarchi <trimarchimichael@yahoo.it>
>>
>> Index: linux-user/syscall.c
>> ===================================================================
>> --- linux-user/syscall.c	(revision 4865)
>> +++ linux-user/syscall.c	(working copy)
>> @@ -53,6 +53,7 @@
>>  //#include <sys/user.h>
>>  #include <netinet/ip.h>
>>  #include <netinet/tcp.h>
>> +#include <qemu-common.h>
>>  
>>  #define termios host_termios
>>  #define winsize host_winsize
>> @@ -4657,7 +4658,11 @@
>>          ret = get_errno(fsync(arg1));
>>          break;
>>      case TARGET_NR_clone:
>> +#if !defined(TARGET_SH4)
>>          ret = get_errno(do_fork(cpu_env, arg1, arg2, arg3, arg4, arg5));
>> +#else
>> +        ret = get_errno(do_fork(cpu_env, arg1, arg2, arg3, arg5, arg4));
>> +#endif
>>          break;
>>  #ifdef __NR_exit_group
>>          /* new thread calls */
>>     
>
>
>
>   
Regards Michael