Re: [Xen-devel] [PATCH] tools/xenstore-watch: Add new timeout parameter

[Razvan Cojocaru <rcojocaru@xxxxxxxxxxxxxxx>]

On 03/16/2016 08:57 PM, Wei Liu wrote:
> On Wed, Mar 16, 2016 at 08:52:57PM +0200, Razvan Cojocaru wrote:
>> On 03/16/2016 08:46 PM, Wei Liu wrote:
>>> On Wed, Mar 16, 2016 at 05:50:46PM +0200, Razvan Cojocaru wrote:
>>> [...]
>>>>  }
>>>>  
>>>> @@ -273,27 +274,49 @@ do_chmod(char *path, struct xs_permissions *perms, 
>>>> int nperms, int upto,
>>>>  }
>>>>  
>>>>  static void
>>>> -do_watch(struct xs_handle *xsh, int max_events)
>>>> +do_watch(struct xs_handle *xsh, int max_events, int timeout)
>>>>  {
>>>> -    int count = 0;
>>>> +    int rc, ms_timeout = timeout * 1000;
>>>>      char **vec = NULL;
>>>> +    struct pollfd fd;
>>>>  
>>>> -    for ( count = 0; max_events == -1 || count < max_events; count++ ) {
>>>> -  unsigned int num;
>>>> +    fd.fd = xs_fileno(xsh);
>>>> +    fd.events = POLLIN | POLLERR;
>>>>  
>>>> -  vec = xs_read_watch(xsh, &num);
>>>> -  if (vec == NULL)
>>>> -      continue;
>>>> +    do {
>>>> +        rc = poll(&fd, 1, 100);
>>>>  
>>>> -  printf("%s\n", vec[XS_WATCH_PATH]);
>>>> -  fflush(stdout);
>>>> -  free(vec);
>>>> -    }
>>>> +        if (rc == 0 || (rc < 0 && errno == EINTR)) {
>>>> +            ms_timeout -= 100;
>>>
>>> Shouldn't you just exit the loop when getting EINTR?
>>
>> I don't think so, an EINTR means that a signal has been caught during
>> poll() - it's not really a failure case.
>>
> 
> What about Ctrl-C ? I think it is reasonable that user want to interrupt
> the process.

Fair enough, I'll re-spin V2.


Thanks,
Razvan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel