Re: [PATCH v5] gdb: ADI support

[Wei-min Pan <weimin.pan@oracle.com>]

On 7/27/2017 8:49 AM, Yao Qi wrote:
> Weimin Pan <weimin.pan@oracle.com> writes:
>
>> +
>> +/* ADI stat settings.  */
>> +typedef struct
>> +{
>> +  /* The ADI block size.  */
>> +  unsigned long blksize;
>> +
>> +  /* Number of bits used for an ADI version tag which can be
>> +   * used together with the shift value for an ADI version tag
>> +   * to encode or extract the ADI version value in a pointer.  */
>> +  unsigned long nbits;
>> +
>> +  /* The maximum ADI version tag value supported.  */
>> +  int max_version;
>> +
>> +  /* ADI version tag file.  */
>> +  int tag_fd;
>> +
>> +  /* ADI availability check has been done.  */
>> +  bool checked_avail;
>> +
>> +  /* ADI is available.  */
>> +  bool is_avail;
>> +
>> +} adi_stat_t;
>> +
>> +/* Per-process ADI stat info.  */
>> +
>> +typedef struct
>> +{
>> +  /* The process identifier.  */
>> +  pid_t pid;
>> +
>> +  /* The ADI stat.  */
>> +  adi_stat_t stat;
>> +} sparc64_adi_info;
>> +
>> +static std::forward_list<sparc64_adi_info *> a_proc_list;
> a_proc_list is not a good variable name.  How about adi_proc_list?

It was originally named "adi_proc_list". I shortened it so I don't run 
into those "long source lines" :)
Will change it back.

> Can you use "std::forward_list<sparc64_adi_info>" instead?  I am reading
> book "C++ Coding Standards: 101 Rules, Guidelines, and Best Practices",
> and know rule 79 is "Store only values and smart pointers in
> containers".  I know we have "std::forward_list<regcache *>", and I am
> trying to fix it.

Will take a look.

>> +
>> +/* Find ADI info for process PID.  */
>> +
>> +static sparc64_adi_info *
>> +find_adi_info (pid_t pid)
>> +{
>> +  sparc64_adi_info *proc;
>> +
>> +  for ( auto it = a_proc_list.begin(); it != a_proc_list.end(); ++it
>> )
> for (const auto &info : a_proc_list)
>
>> +    if ((*it)->pid == pid)
>> +      return (*it);
>> +
>> +  return NULL;
>> +}
>> +
>> +/* Add ADI info for process PID.  Returns newly allocated info
>> +   object.  */
>> +
>> +static sparc64_adi_info *
>> +add_adi_info (pid_t pid)
>> +{
>> +  sparc64_adi_info *proc = XCNEW (sparc64_adi_info);
>> +
>> +  proc->pid = pid;
>> +  proc->stat.is_avail = false;
>> +  proc->stat.checked_avail = false;
>> +  proc->stat.tag_fd = 0;
>> +  a_proc_list.push_front (proc);
>> +
>> +  return proc;
>> +}
>> +
>> +/* Get ADI info for process PID, creating one if it doesn't exist.  */
>> +
>> +static sparc64_adi_info *
>> +get_adi_info_proc (pid_t pid)
>> +{
>> +  sparc64_adi_info *proc;
>> +
>> +  proc = find_adi_info (pid);
>> +  if (proc == NULL)
>> +    proc = add_adi_info (pid);
>> +
>> +  return proc;
>> +}
>> +
>> +static adi_stat_t
>> +get_adi_info (pid_t pid)
>> +{
>> +  sparc64_adi_info *proc;
>> +
>> +  proc = get_adi_info_proc (pid);
>> +  return proc->stat;
>> +}
>> +
>> +/* Is called when GDB is no longer debugging process PID.  It
>> +   deletes data structure that keeps track of the ADI stat.  */
>> +
>> +void
>> +sparc64_forget_process (pid_t pid)
>> +{
>> +  sparc64_adi_info *proc;
>> +  int target_errno;
>> +
>> +  proc = find_adi_info (pid);
>> +  if (proc != NULL)
>> +    {
>> +      if (proc->stat.tag_fd > 0)
>> +        target_fileio_close (proc->stat.tag_fd, &target_errno);
>> +      a_proc_list.remove (proc);
>> +      xfree (proc);
>> +    }
>> +
>> +}
>> +
>> +
>> +int main ()
>> +{
>> +  char *haddr;
>> +  caddr_t vaddr;
>> +  int	version;
>> +
>> +  // test ISM
> We have mixed comment styles, // and "/**/".  Since it is a c code, we
> can use /**/?
>
>> +  int shmid = shmget (IPC_PRIVATE, SHMSIZE, IPC_CREAT | 0666);
>> +  if (shmid == -1)
>> +    exit(1);
>> +  char *shmaddr = (char *)shmat (shmid, NULL, 0x666 | SHM_RND);
>> +  if (shmaddr == (char *)-1)
>> +    {
>> +      shmctl (shmid, IPC_RMID, NULL);
>> +      exit(1);
>> +    }
>> +  // enable ADI on ISM segment
> It is a sentence, so Capitalize 'E', and put "." at the end.
>
> /* Enable ADI on ISM segment.  */
>

OK, will update the test with your comments.

Thanks.