asm-generic based msgctl (was Re: [PATCH v3 3/3] sysv: linux: Pass 64-bit version of semctl syscall)

[Vineet Gupta]

On 3/10/20 5:29 PM, Vineet Gupta via Libc-alpha wrote:
> Hi Alistair,
> 
> On 3/4/20 5:26 PM, Alistair Francis wrote:
>> The semctl_syscall() function passes a union semun to the kernel. The
>> union includes struct semid_ds as a member. On 32-bit architectures the
>> Linux kernel provides a *_high version of the 32-bit sem_otime and
>> sem_ctime values. These can be combined to get a 64-bit version of the
>> time.
>>
>> This patch adjusts the struct semid_ds to support the *_high versions
>> of sem_otime and sem_ctime. For 32-bit systems with a 64-bit time_t
>> this can be used to get a 64-bit time from the two 32-bit values.
>>
>> We protect this new code via the __IPC_TIME64 marco, which is only true
>> for 32-bit architectures with a 64-bit time_t.

[snip...]

>>
>> diff --git a/bits/ipc.h b/bits/ipc.h
>> index e2981fd5c3..9ac8485193 100644
>> --- a/bits/ipc.h
>> +++ b/bits/ipc.h
>> @@ -29,7 +29,11 @@
>>  /* Control commands for `msgctl', `semctl', and `shmctl'.  */
>>  #define IPC_RMID	0		/* remove identifier */
>>  #define IPC_SET		1		/* set `ipc_perm' options */
>> -#define IPC_STAT	2		/* get `ipc_perm' options */
>> +#if __TIMESIZE == 64 && __WORDSIZE == 32
>> +# define IPC_STAT 0x102     /* Get `ipc_perm' options.  */
>> +#else
>> +# define IPC_STAT 2   /* Get `ipc_perm' options.  */
>> +#endif
> 
> Why is this needed. Linux kernel seems to be returning EINVAL for this cmd-id and
> following fail for ARC.
> 
> FAIL: sysvipc/test-sysvmsg
> FAIL: sysvipc/test-sysvsem
> FAIL: sysvipc/test-sysvshm
> 
> Shouldn't this use the default __IPC_64 value which is 0 not 0x100.
> 

>> diff --git a/sysdeps/unix/sysv/linux/semctl.c b/sysdeps/unix/sysv/linux/semctl.c
>> index 0c3eb0932f..3ac6d01b84 100644
>> --- a/sysdeps/unix/sysv/linux/semctl.c
>> +++ b/sysdeps/unix/sysv/linux/semctl.c
>> @@ -23,11 +23,16 @@
>>  #include <shlib-compat.h>
>>  #include <errno.h>
>>  
>> +#define __IPC_TIME64 (IPC_STAT & __IPC_64)

So __IPC_TIME64 can only happen if we have 0x102 based cmd ? But doesn't
asm-generic ABI preclude the 0x100 ?

>> +
>>  /* Define a `union semun' suitable for Linux here.  */
>>  union semun
>>  {
>>    int val;			/* value for SETVAL */
>>    struct semid_ds *buf;		/* buffer for IPC_STAT & IPC_SET */
>> +#if __WORDSIZE == 32
>> +  struct __semid_ds32 *buf32;   /* 32-bit buffer for IPC_STAT & IPC_SET */
>> +#endif
>>    unsigned short int *array;	/* array for GETALL & SETALL */
>>    struct seminfo *__buf;	/* buffer for IPC_INFO */
>>  };
>> @@ -43,13 +48,25 @@ union semun
>>  static int
>>  semctl_syscall (int semid, int semnum, int cmd, union semun arg)
>>  {
>> +  int ret;
>>  #ifdef __ASSUME_DIRECT_SYSVIPC_SYSCALLS
>> -  return INLINE_SYSCALL_CALL (semctl, semid, semnum, cmd | __IPC_64,
>> -			      arg.array);
>> +  ret = INLINE_SYSCALL_CALL (semctl, semid, semnum, cmd | __IPC_64,
>> +                             arg.array);
>>  #else
>> -  return INLINE_SYSCALL_CALL (ipc, IPCOP_semctl, semid, semnum, cmd | __IPC_64,
>> -			      SEMCTL_ARG_ADDRESS (arg));
>> +  ret = INLINE_SYSCALL_CALL (ipc, IPCOP_semctl, semid, semnum, cmd | __IPC_64,
>> +                             SEMCTL_ARG_ADDRESS (arg));
>> +#endif
>> +
>> +#if __IPC_TIME64
>> +  if (ret == 0 && (cmd & __IPC_TIME64))
>> +    {
>> +      arg.buf->sem_ctime = arg.buf32->sem_ctime |
>> +                               ((time_t) arg.buf32->sem_ctime_high << 32);
>> +      arg.buf->sem_otime = arg.buf32->sem_otime |
>> +                               ((time_t) arg.buf32->sem_otime_high << 32);
>> +    }
>>  #endif
>> +  return ret;
>>  }