From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 29229 invoked by alias); 23 Aug 2013 02:20:54 -0000 Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org Received: (qmail 29212 invoked by uid 89); 23 Aug 2013 02:20:53 -0000 X-Spam-SWARE-Status: No, score=-4.6 required=5.0 tests=AWL,BAYES_00,KHOP_RCVD_UNTRUST,KHOP_THREADED,RCVD_IN_HOSTKARMA_W,RCVD_IN_HOSTKARMA_WL autolearn=ham version=3.3.2 Received: from relay1.mentorg.com (HELO relay1.mentorg.com) (192.94.38.131) by sourceware.org (qpsmtpd/0.84/v0.84-167-ge50287c) with ESMTP; Fri, 23 Aug 2013 02:20:25 +0000 Received: from svr-orw-exc-10.mgc.mentorg.com ([147.34.98.58]) by relay1.mentorg.com with esmtp id 1VCgz8-0006rE-CE from Yao_Qi@mentor.com ; Thu, 22 Aug 2013 19:20:22 -0700 Received: from SVR-ORW-FEM-02.mgc.mentorg.com ([147.34.96.206]) by SVR-ORW-EXC-10.mgc.mentorg.com with Microsoft SMTPSVC(6.0.3790.4675); Thu, 22 Aug 2013 19:20:18 -0700 Received: from qiyao.dyndns.org (147.34.91.1) by svr-orw-fem-02.mgc.mentorg.com (147.34.96.168) with Microsoft SMTP Server id 14.2.247.3; Thu, 22 Aug 2013 19:20:17 -0700 Message-ID: <5216C6B0.90906@codesourcery.com> Date: Fri, 23 Aug 2013 02:20:00 -0000 From: Yao Qi User-Agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130110 Thunderbird/17.0.2 MIME-Version: 1.0 To: Joel Brobecker CC: Pedro Alves , Eli Zaretskii , Subject: Re: [PATCH] Unbuffer stdout and stderr on windows References: <83eh9uonlg.fsf@gnu.org> <20130815175940.GD6955@ednor.casa.cgf.cx> <520E1109.7000304@redhat.com> <520E1C34.2000907@codesourcery.com> <520E2B13.8020706@redhat.com> <83r4dtn35q.fsf@gnu.org> <520E357E.6080803@redhat.com> <83mwohn0nj.fsf@gnu.org> <520E40CD.7080604@redhat.com> <5215AC28.3030506@codesourcery.com> <20130822141756.GB5221@adacore.com> In-Reply-To: <20130822141756.GB5221@adacore.com> Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8bit X-SW-Source: 2013-08/txt/msg00660.txt.bz2 Thanks for the review, Joel. On 08/22/2013 10:17 PM, Joel Brobecker wrote: >> +#ifdef __MINGW32__ >> >+ /* Ensure stderr is unbuffered. */ >> >+ setvbuf (stderr, NULL, _IONBF, BUFSIZ); >> >+#endif > Can you also document the reason why you make stderr unbuffered here? > If we capture most of what was discussed, it'll help anyone trying > to understand the code later on... > Sure, comment is added. >> > gdb_stdout = stdio_fileopen (stdout); >> > gdb_stderr = stdio_fileopen (stderr); >> >+#ifdef __MINGW32__ >> >+ /* There is no real line-buffering on Windows, see >> >+http://msdn.microsoft.com/en-us/library/86cebhfs%28v=vs.71%29.aspx >> >+ so the stdout is either fully-buffered or non-buffered. We can't >> >+ make stdout non-buffered, because of two concerns, >> >+ 1. non-buffering hurts performance, >> >+ 2. non-buffering may change GDB's behavior when it is interacting >> >+ with front-end, such as Emacs. >> >+ >> >+ We decide to leave stdout as fully buffered, but flush it first > decided Fixed. >> >+ when something is written to stderr. */ >> >+ update_stderr_ui_file (gdb_stderr); >> >+#endif > Just a question: Would it be possible to emulate line buffering > as well? In my experience, the size of the buffer is fairly big, > and causes the output to appear only at fairly large intervals, > and in fairly large blobs. > I am not sure. The goal of this patch is to keep the order of outputs from stdout and stderr correct, so I don't think much on emulating line buffering for stdout. IMO, it can be addressed by a separated patch. > >> >+ >> > gdb_stdlog = gdb_stderr; /* for moment */ >> > gdb_stdtarg = gdb_stderr; /* for moment */ >> > gdb_stdin = stdio_fileopen (stdin); >> >diff --git a/gdb/ui-file.c b/gdb/ui-file.c >> >index cf5a86d..649d892 100644 >> >--- a/gdb/ui-file.c >> >+++ b/gdb/ui-file.c >> >@@ -654,6 +654,32 @@ stdio_file_fseek (struct ui_file *file, long offset, int whence) >> > return fseek (stdio->file, offset, whence); >> > } >> > >> >+static void >> >+stderr_file_write (struct ui_file *file, const char *buf, long length_buf) > Please, document every new function, no matter how trivial. That way, > there is no discussion about whether the function should be documented > or not. OK, documented in the updated patch. -- Yao (齐尧) gdb: 2013-08-23 Yao Qi * main.c (captured_main) [__MINGW32__]: Set stderr unbuffered. and all update_stderr_ui_file. * ui-file.c (stderr_file_write): New function. (stderr_file_fputs): New function. (update_stderr_ui_file): New function. * ui-file.h (update_stderr_ui_file): Declare. --- gdb/main.c | 20 ++++++++++++++++++++ gdb/ui-file.c | 32 ++++++++++++++++++++++++++++++++ gdb/ui-file.h | 3 +++ 3 files changed, 55 insertions(+), 0 deletions(-) diff --git a/gdb/main.c b/gdb/main.c index 1c240e4..332df9e 100644 --- a/gdb/main.c +++ b/gdb/main.c @@ -375,8 +375,28 @@ captured_main (void *data) saved_command_line[0] = '\0'; instream = stdin; +#ifdef __MINGW32__ + /* Ensure stderr is unbuffered. A Cygwin pty or pipe is implemented + as a Windows pipe, and Windows buffers on pipes. */ + setvbuf (stderr, NULL, _IONBF, BUFSIZ); +#endif + gdb_stdout = stdio_fileopen (stdout); gdb_stderr = stdio_fileopen (stderr); +#ifdef __MINGW32__ + /* There is no real line-buffering on Windows, see + http://msdn.microsoft.com/en-us/library/86cebhfs%28v=vs.71%29.aspx + so the stdout is either fully-buffered or non-buffered. We can't + make stdout non-buffered, because of two concerns, + 1. non-buffering hurts performance, + 2. non-buffering may change GDB's behavior when it is interacting + with front-end, such as Emacs. + + We decided to leave stdout as fully buffered, but flush it first + when something is written to stderr. */ + update_stderr_ui_file (gdb_stderr); +#endif + gdb_stdlog = gdb_stderr; /* for moment */ gdb_stdtarg = gdb_stderr; /* for moment */ gdb_stdin = stdio_fileopen (stdin); diff --git a/gdb/ui-file.c b/gdb/ui-file.c index cf5a86d..4b26b78 100644 --- a/gdb/ui-file.c +++ b/gdb/ui-file.c @@ -654,6 +654,38 @@ stdio_file_fseek (struct ui_file *file, long offset, int whence) return fseek (stdio->file, offset, whence); } +/* This is the implementation of ui_file method to_write for stderr. + gdb_stdout is flushed before writing to gdb_stderr. */ + +static void +stderr_file_write (struct ui_file *file, const char *buf, long length_buf) +{ + gdb_flush (gdb_stdout); + stdio_file_write (file, buf, length_buf); +} + +/* This is the implementation of ui_file method to_fputs for stderr. + gdb_stdout is flushed before writing to gdb_stderr. */ + +static void +stderr_file_fputs (const char *linebuffer, struct ui_file *file) +{ + gdb_flush (gdb_stdout); + stdio_file_fputs (linebuffer, file); +} + +/* Overwrite UI_FILE's methods 'to_write' and 'to_fputs' by function + in which gdb_stdout is flushed. */ + +void +update_stderr_ui_file (struct ui_file *ui_file) +{ + /* Method 'to_write_async_safe' is not overwritten, because it is + only used on linux host. */ + set_ui_file_write (ui_file, stderr_file_write); + set_ui_file_fputs (ui_file, stderr_file_fputs); +} + /* Like fdopen(). Create a ui_file from a previously opened FILE. */ struct ui_file * diff --git a/gdb/ui-file.h b/gdb/ui-file.h index 9fef68c..6d96c0b 100644 --- a/gdb/ui-file.h +++ b/gdb/ui-file.h @@ -129,6 +129,9 @@ extern struct ui_file *mem_fileopen (void); /* Open/create a STDIO based UI_FILE using the already open FILE. */ extern struct ui_file *stdio_fileopen (FILE *file); +extern void update_stderr_ui_file (struct ui_file *ui_file); + + /* Open NAME returning an STDIO based UI_FILE. */ extern struct ui_file *gdb_fopen (const char *name, const char *mode); -- 1.7.7.6