Re: [4/6] msvcrt: Added _fread_nolock_s implementation

[Alexandre Julliard]

"Iván Matellanes"  writes:
> {
> - MSVCRT_size_t ret;
> -
> - MSVCRT__lock_file(stream);
> - ret = MSVCRT__fread_nolock_s(buf, buf_size, elem_size, count, stream);
> - MSVCRT__unlock_file(stream);

The parameter validation has to be done before locking the file.

[4/6] msvcrt: Added _fread_nolock_s implementation

[Iván Matellanes]

On 31/10/14 07:46, Alexandre Julliard wrote:
> "Iván Matellanes"  writes:
> > {
> > - MSVCRT_size_t ret;
> > -
> > - MSVCRT__lock_file(stream);
> > - ret = MSVCRT__fread_nolock_s(buf, buf_size, elem_size, count, stream);
> > - MSVCRT__unlock_file(stream);
>
> The parameter validation has to be done before locking the file.

Hi, thanks for the comment. Should all parameters be validated, or just stream != NULL?
Validating the rest seems redundant to me.

Regards,
Iván.

[4/6] msvcrt: Added _fread_nolock_s implementation

[Alexandre Julliard]

"Iván Matellanes"  writes:
> On 31/10/14 07:46, Alexandre Julliard wrote:
> > "Iván Matellanes"  writes:
> > > {
> > > - MSVCRT_size_t ret;
> > > -
> > > - MSVCRT__lock_file(stream);
> > > - ret = MSVCRT__fread_nolock_s(buf, buf_size, elem_size, count, stream);
> > > - MSVCRT__unlock_file(stream);
> >
> > The parameter validation has to be done before locking the file.
>
> Hi, thanks for the comment. Should all parameters be validated, or just stream != NULL?
> Validating the rest seems redundant to me.

I suspect you don't want to call the _s function here at all, since you
could then potentially call the invalid parameter handler with the file
locked. A few more tests for the error cases would probably be useful.