This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Fix file descriptor existence of MinGW.

On 8/6/19 9:55 AM, Martin Liška wrote:
On 8/6/19 5:35 PM, Martin Sebor wrote:
On 8/6/19 6:04 AM, Martin Liška wrote:

The patch is about proper checking of file descriptors on Windows.

Patch can bootstrap on x86_64-linux-gnu and survives regression tests.

Is there a way to share the definition of the new function so
it doesn't have to be duplicated?

I would like to, but I'm wondering which header and source file
should I use for it?

I'm not sure.  It's a general purpose function that could go
in some utility library like libiberty but I don't know what
the process for enhancing it is to tell if that's appropriate.

Other than that, I'm wondering if the guard for F_GETFD is
necessary and when (the original code didn't guard its use).

Because the value is not defined on MinGW targets where _get_osfhandle
call must be used ;)

I was unclear -- sorry.  What I meant was: since there is
a check for MinGW, is the check for F_GETFD necessary, or
would this be sufficient and preferable?

  #if defined(_WIN32)
    return _get_osfhandle (fd) != -1;
    return fcntl (fd, F_GETFD) >= 0;

That way it's clear that only two possibilities are expected.




@Pekka: Can you please test it on Windows?

Ready to be installed?


2019-08-06  Martin Liska  <>

     PR bootstrap/91352
     * gcc.c (fd_exists): New.
     (driver::detect_jobserver): Use fd_exists.
     * lto-wrapper.c (fd_exists): New.
     (jobserver_active_p): Use fd_exists.
   gcc/gcc.c         | 19 +++++++++++++++++--
   gcc/lto-wrapper.c | 19 +++++++++++++++++--
   2 files changed, 34 insertions(+), 4 deletions(-)

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]