[Xcb] [PATCH 4/8] Add xcb_send_fd API

[Daniel Martin]

On 7 November 2013 13:14, Keith Packard <keithp at keithp.com> wrote:
> Uli Schlachter <psychon at znc.in> writes:
>
>> The above seems to be an unrelated change. If this is needed due to some
>> xcb-private header, then that header needs to be fixed. Otherwise this seems
>> unneeded (for this patch).
>
> Yup, I figured out how to make it compile without the network header
> edits; just needed to add sys/socket.h to xcbint.h and all is well.
>
>> This indents with tabs. I know that there is no coding style documented
>> anywhere, but "most" of libxcb only uses spaces.
>
> I'll fix that; xcb is 'outside' my xorg environment, and so doesn't
> inherit the usual xorg editor settings.
>
>>> +void
>>> +xcb_send_fd(xcb_connection_t *c, int fd)
>>> +{
>>> +#if HAVE_SENDMSG
>>
>>   if (c->has_error)
>>     return;
>>
>> This is needed before the pthread_mutex_lock(), because if
>> xcb_connect_to_display() fails, it returns a pointer to an integer. The
>> has_error 'field' is the only thing that might be accessed for such a connection
>> object.
>
> Ok, you just want an additional check for errors in case the display
> object doesn't even have the iolock initialized. Makes sense.
>
>>
>>> +    pthread_mutex_lock(&c->iolock);
>>> +    while (c->out.out_fd.nfd == XCB_MAX_PASS_FD) {
>>> +    _xcb_out_flush_to(c, c->out.request);
>>> +    if (c->has_error)
>>> +        break;
>>> +    }
>>> +    if (!c->has_error)
>>> +    c->out.out_fd.fd[c->out.out_fd.nfd++] = fd;
>>> +    pthread_mutex_unlock(&c->iolock);
>>> +#endif
>>> +}
>>
>> So, xcb_send_fd() hands FDs to libxcb which will all be included in the next
>> flush. The idea is that the user of the library calls (through the automatically
>> generated protocol functions) xcb_send_fd() and then
>> xcb_send_request(). Correct?
>
> Yes.
>
>> However, what happens if the connection is flushed between the xcb_send_fd() and
>> the xcb_send_request() calls, because another thread "interferes"? This can also
>> happen with a single thread: xcb_send_request() could have to insert sync
>> requests (GetInputFocus) because of sequence number 0 or sequence number wraps.
>> This could cause the output buffer to go full before the "real"
>> request is inserted.
>
> It doesn't matter as long as the FDs arrive before the full request in
> the X server. Given how FD passing works, there's no direct association
> between the FDs received and the requests within the X server
> anyways.
>
>> I guess that the server expects the FDs to be sent with the request that they
>> are meant to (otherwise I could feed the server lots of FDs which it would have
>> to keep around forever). So this function doesn't really work, does
>> it?
>
> That's the X server's problem, and it gets to deal with clients that
> don't play nice.
>
> I cannot get version 1.9.1 to build at all, which makes the prospects
> of releasing something on top of that pretty uncertain...
>
> Traceback (most recent call last):
>   File "./c_client.py", line 2925, in <module>
>     module.generate()
>   File "/local/xorg/lib/python2.7/site-packages/xcbgen/state.py", line 103, in generate
>     item.out(name)
>   File "./c_client.py", line 2788, in c_request
>     _c_type_setup(self, name, ('request',))
>   File "./c_client.py", line 338, in _c_type_setup
>     _c_type_setup(field.type, field.field_type, ())
>   File "./c_client.py", line 413, in _c_type_setup
>     _c_serialize('serialize', self)
>   File "./c_client.py", line 1171, in _c_serialize
>     count = _c_serialize_helper(context, self, code_lines, temp_vars, prefix=prefix)
>   File "./c_client.py", line 1045, in _c_serialize_helper
>     space, prefix)
>   File "./c_client.py", line 690, in _c_serialize_helper_switch
>     bitcase_expr = _c_accessor_get_expr(b.type.expr, None)
>   File "./c_client.py", line 1417, in _c_accessor_get_expr
>     lenexp = _c_accessor_get_length(expr, field_mapping)
>   File "./c_client.py", line 1394, in _c_accessor_get_length
>     lenfield_name = expr.lenfield_name
> AttributeError: 'list' object has no attribute 'lenfield_name'
> make[1]: *** [sync.c] Error 1

Sounds like this commit is missing:
    c_client.py: Handle multiple expr. in a bitcase
    http://cgit.freedesktop.org/xcb/libxcb/commit/?id=e602b65

which is the counterpart for:
    xcbgen: Handle multiple <enumref> in a <bitcase>
    http://cgit.freedesktop.org/xcb/proto/commit/?id=a9a892e

>
> --
> keith.packard at intel.com
>
> _______________________________________________
> Xcb mailing list
> Xcb at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/xcb