|
|
On Mon, May 04, 2009 at 04:27:09PM -0700, Tim Prouty wrote:
>
> On May 4, 2009, at 12:25 PM, Volker Lendecke wrote:
>
>> On Mon, May 04, 2009 at 12:22:48PM -0700, Tim Prouty wrote:
>>>> On Mon, May 04, 2009 at 11:58:19AM -0500, Tim Prouty wrote:
>>>>> - if (fnum1 != -1) cli_close(cli1, fnum1);
>>>>> - if (fnum2 != -1) cli_close(cli2, fnum2);
>>>>> + cli_close(cli1, fnum1);
>>>>> + cli_close(cli2, fnum2);
>>>>
>>>> I think that's wrong. We need a (fnumx 1= (uint16_t)-1).
>>>
>>> I was just mirroring the other usage in the file as the quick warning
>>> fix. It seems a little bit strange to me to make an fd uint, when
>>> the
>>> standard api uses -1 to indicate error.
>>
>> Which other usages do you mean?
>
> torture_denytest1 also closed them without checking. Fixing them the
> right way was easy, so I just pushed a fix that checks the status before
> closing.
Thanks, there are many cases in the old torture code where we're
not checking NTSTATUS returns when we probably should, but a general
cleanup of that torture code is a little out of scope at the moment.
I'm just making sure it compiles as I move the cli_XX API to NTSTATUS
returns.
Jeremy.
|
|