gmane.os.cygwin.patches

[PATCH] Export memrchr

Yaakov (Cygwin/X — 2012-05-09T08:18:35

Here are the patches for exporting memrchr, once my patches to newlib
are accepted.


Yaakov

Regenerate configures

Yaakov (Cygwin/X — 2012-04-23T22:17:10

cygwin/doc/configure and cygwin/testsuite/configure are now the only 
configure scripts in the winsup tree which were generated with 
autoconf-2.5x.  Any objections to regenerating them with 2.68?


Yaakov

[PATCH] Extended mouse coordinates

Thomas Wolff — 2012-04-22T19:07:50

This patch replaces my previous proposal 
(http://cygwin.com/ml/cygwin-patches/2012-q2/msg00005.html) with two 
modifications:

  * Fixed a bug that suppressed mouse reporting at large coordinates (in
    all modes actually:-\ )
  * Added mouse mode 1005 (total of 3 three new modes, so all reporting
    modes run by current terminal emulators would be implemented)

I would appreciate the patch to be applied this time, planned to be my 
last mouse patch :)

Kind regards,
Thomas
2012-04-03  Thomas Wolff  <towo< at >towo.net>

* fhandler.h (class dev_console): Two flags for extended mouse modes.
* fhandler_console.cc (fhandler_console::read): Implemented 
extended mouse modes 1015 (urxvt, mintty, xterm) and 1006 (xterm).
Not implemented extended mouse mode 1005 (xterm, mintty).
Supporting mouse coordinates greater than 222 (each axis).
Also: two { wrap formatting consistency fixes.
(fhandler_console::char_command) Initialization of enhanced 
mouse reporting modes.

diff -rup sav/fhandler.h ./fhandler.h
--- s

[PATCH] Mouse reporting modes 1006 and 1015 for extended coordinates

Thomas Wolff — 2012-04-03T19:42:10

This patch implements mouse modes 1006 and 1015 for the cygwin console 
(no experimental add-ons this time :) ).
They enable unlimited mouse coordinate reporting like recent xterm, 
mintty, urxvt. If someone is interested, I could also implement mouse 
mode 1005.
Thomas
2012-04-03  Thomas Wolff  <towo< at >towo.net>

* fhandler.h (class dev_console): Two flags for extended mouse modes.
* fhandler_console.cc (fhandler_console::read): Implemented 
extended mouse modes 1015 (urxvt, mintty, xterm) and 1006 (xterm).
Not implemented extended mouse mode 1005 (xterm, mintty).
Supporting mouse coordinates greater than 222 (each axis).
Also: two { wrap formatting consistency fixes.
(fhandler_console::char_command) Initialization of enhanced 
mouse reporting modes.

diff -rup sav/fhandler.h ./fhandler.h
--- sav/fhandler.h2012-04-01 19:46:04.000000000 +0200
+++ ./fhandler.h2012-04-03 15:52:07.893561600 +0200
< at >< at > -1282,6 +1282,8 < at >< at > class dev_console
 
   bool insert_mode;
   int use_mouse;
+  bool ext_mouse_mode6;
+

console: new mouse modes, request/response attempt

Thomas Wolff — 2012-04-02T18:46:31

This patch includes 2 things (to be fixed and separated anyway; change 
log still missing) for discussion:
* mouse modes 6 and 15 (numeric mouse coordinates)
* semi-fix for missing terminal status responses
The fix tries to detect the proper fhandler for CONIO, which is then 
used to queue the response.
Problem 1: I am not sure whether this detection is proper in all cases, 
what e.g. if /dev/tty is reopened etc. I don't know where else a 
relation between the handles for CONIN and CONOUT might be established.
Problem 2: While the response reaches the application with this patch, 
only the first byte is read right-away. Further bytes are delayed until 
other input is becoming present (typing a key). This may (or may not) be 
related to other issues with select(), so maybe it's worth analyzing it.

Thomas
diff -rup sav/fhandler.h ./fhandler.h
--- sav/fhandler.h2012-04-01 19:46:04.000000000 +0200
+++ ./fhandler.h2012-04-02 15:47:22.385727000 +0200
< at >< at > -1282,8 +1282,11 < at >< at > class dev_console
 
   bool insert_mod

Compiler warnings when building latest cygwin cvs with gcc-4.6 (0/2)

Ryan Johnson — 2012-03-29T14:36:35

Hi all

While trying to build the cygwin dll from source, I accidentally left my 
home-built gcc-4.6 in PATH... and it complains loudly about all kinds of 
things, some of which might actually be of interest. I'll follow up 
shortly with two patches that fix those problems in a backwards 
compatible way, in case this provides a improvement in code quality (and 
future-proofing for when we upgrade cygwin's gcc4).

The patched code compiles cleanly under both cygwin's gcc4 and my 
gcc-4.6.2, but the latter produces a broken cygwin1.dll. I haven't tried 
to figure out what goes wrong, since I don't know what changes went into 
cygwin's gcc4 to make it work properly in the first place. It could be 
something as simple as PATH/LD_LIBRARY_PATH...

Patch 1: fix function attribute conflicts
Patch 2: fix compiler misc. warnings

NOTE: the warnings patch applies safely, but with fuzz, if the 
attributes patch has already been applied. If you're paranoid, apply the 
warnings patch first.

Overview of changes:

1a. Conf

[PATCH] Ctrl-C and non-Cygwin programs

Ein Terakawa — 2012-03-29T02:55:51

This is a proof of concept demonstration which
makes Ctrl-C behave in a way a lot of people expect
concerning non-Cygwin console programs.

What it does actually is it generates CTRL_BREAK_EVENT with 
Windows Console API GenerateConsoleCtrlEvent on the arrival of SIGINT.
And to make this scheme to be functional it is required to specify
CREATE_NEW_PROCESS_GROUP when creating new non-Cygwin processes.

To my surprise there seem to be no way to generate CTRL_C_EVENT using API.

I must also point out that virtually all of the terminal emulators
are sneakily keeping hidden Windows Console in the background.
Thus several features of the Windows Console is still available to
processes running in the cygwin environment.
One of such features is this 'process group' and the other one is
'code page' which you manipulate with chcp.com utility.


Following is a bunch of random posts, I picked up from this list, 
talking about the same topic. Ordered by its significance under my judge.
These should help you understand (o

[PATCH] Fix possible infinite loop in hires_ms::timeGetTime_ns()

Christian Franke — 2012-03-20T17:48:10

ntdll.h:SharedUserData misses a volatile qualifier. This (at least) may 
result in an infinite loop in hires_ms::timeGetTime_ns(). Fortunately 
this could only happen if LowPart wraps around during the function call.

Generated code:

$ objdump -d -C times.o
...
1160 <hires_ms::timeGetTime_ns()>:
1160: 55                 push   %ebp
1161: 8b 15 0c 00 fe 7f  mov    0x7ffe000c,%edx
1167: 3b 15 10 00 fe 7f  cmp    0x7ffe0010,%edx
116d: 89 e5              mov    %esp,%ebp
116f: a1 08 00 fe 7f     mov    0x7ffe0008,%eax
1174: 75 02              jne    1178 <hires_ms::timeGetTime_ns()+0x18>
1176: 5d                 pop    %ebp
1177: c3                 ret
1178: eb fe              jmp    1178 <hires_ms::timeGetTime_ns()+0x18>
...


This function results in the same code:

LONGLONG hires_ms::timeGetTime_ns ()
{
   LARGE_INTEGER t;
   t.HighPart = SharedUserData.InterruptTime.High1Time;
   t.LowPart = SharedUserData.InterruptTime.LowPart;
   if (t.HighPart == SharedUserData.InterruptTime.High2Time)
     return t.Quad

1.7.10/1.7.11: .Net programs started from a cygwin console may fail.

Greg Gudenburr — 2012-03-17T13:53:07

http://cygwin.com/ml/cygwin/2012-03/msg00375.html

I am seeing a very similar bug after upgrading.   I see reference to a "test" patch. I am willing to test the update. How can I get a copy?

Greg

avoid calling strlen() twice in readlink()

Václav Zeman — 2012-03-08T13:37:09

Hi.

Here is a tiny patch to avoid calling strlen() twice in readlink().

ChangeLog:

2012-03-08  Václav Zeman  <vhaisman< at >gmail.com>

        * path.cc (readlink): Avoid calling strlen() twice.

Changing MAXPATHLEN in glob.cc back to 16384

Piotr Foltyn — 2012-03-06T17:06:10

Hi,

Revision 1.7 of src/winsup/cygwin/glob.cc reduced maximum allowed path length from 16384 to 4096 characters. This is unfortunate because some paths used in my build system can reach 6k characters in length. Attached patch reinstates 16384 characters limit.

fix tcgetsid return type

Eric Blake — 2012-02-27T23:25:36

Detected by gnulib's unit tests.  POSIX requires tcgetsid to return
pid_t, not int.

2012-02-27  Eric Blake  <eblake< at >redhat.com>

* include/sys/termios.h (tcgetsid): Fix return type.
* termios.cc (tcgetsid): Likewise.
* fhandler_termios.cc (fhandler_termios::tcgetsid): Likewise.
* fhandler.h (fhandler_base): Likewise.
* fhandler.cc (fhandler_base::tcgetsid): Likewise.

diff --git a/winsup/cygwin/fhandler.cc b/winsup/cygwin/fhandler.cc
index ef25a07..c3bca4f 100644
--- a/winsup/cygwin/fhandler.cc
+++ b/winsup/cygwin/fhandler.cc
< at >< at > -1464,7 +1464,7 < at >< at > fhandler_base::tcgetpgrp ()
   return -1;
 }

-int
+pid_t
 fhandler_base::tcgetsid ()
 {
   set_errno (ENOTTY);
diff --git a/winsup/cygwin/fhandler.h b/winsup/cygwin/fhandler.h
index 605f59b..3b186bd 100644
--- a/winsup/cygwin/fhandler.h
+++ b/winsup/cygwin/fhandler.h
< at >< at > -383,7 +383,7 < at >< at > public:
   virtual int tcgetattr (struct termios *t);
   virtual int tcsetpgrp (const pid_t pid);
   virtual int tcgetpgrp ();
-  virtual int tcgetsid ();
+  virtual pid_t tc

[PATCH] Add pldd(1)

Yaakov (Cygwin/X — 2012-02-24T08:38:40

The pldd(1) command apparently originates from Solaris and was added to
glibc-2.15[1].  Patches and new file attached.


Yaakov

[1] http://sourceware.org/git/?p=glibc.git;a=blob_plain;f=NEWS

[PATCH] Add pthread_getname_np, pthread_setname_np

Yaakov (Cygwin/X — 2012-02-24T03:38:15

This patchset adds pthread_getname_np and pthread_setname_np.  These
were added to glibc in 2.12[1] and are also present in some form on
NetBSD and several UNIXes.  IIUC recent versions of GDB can benefit from
this support.

The code is based on NetBSD's implementation with changes to better
match Linux behaviour.  It does differ from Linux in two points:

* The thread name is not affected by changing __progname (or
program_invocation_short_name on Linux).  I used the latter because it
is cheaper than the pinfo->progname dance (e.g. in
format_process_stat()).

* pthread_setname_np(thr, NULL) segfaults on Linux (and NetBSD), but our
snprintf is apparently more robust and treats it as an empty string.

I'll leave it up to you to decide if either of these matter.

I implemented this via class pthread_attr to make it easier to add
pthread_attr_[gs]etname_np (present in NetBSD and some UNIXes) should it
ever be added to Linux (or we decide we want it anyway).

Patches and test code attached.


Yaakov

[1] http://

[PATCH] Add scandirat(3)

Yaakov (Cygwin/X — 2012-02-21T22:44:58

scandirat(3) was added in glibc-2.15[1] and has supposedly been proposed
for addition to POSIX.1[2].  Patch attached.


Yaakov

[1] http://sourceware.org/git/?p=glibc.git;a=blob_plain;f=NEWS
[2] http://article.gmane.org/gmane.linux.man/2419

[PATCH] pthread.h: include

Yaakov (Cygwin/X — 2012-02-12T22:04:02

POSIX states:

The reason being that some pthread functions take a clockid_t argument,
and the CLOCK_* symbolic names are therein defined.

Patch attached.


Yaakov

[PATCH] Fix C++ compilation of wait(NULL)

Christian Franke — 2012-02-08T18:39:48

This fixes the regression I introduced in sys/wait.h.

Sorry, and thanks for the bug report.

Christian

[PATCH] ldd: support .oct and .so modules

Yaakov (Cygwin/X — 2012-01-24T05:53:23

Octave modules use the .oct extension, and several programs use .so for
modules even on Cygwin (e.g. Apache2, Mesa, OpenSSL, Ruby).  Currently,
running ldd(1) on any of these returns ENOEXEC.

The attached patch fixes ldd to treat these as DLLs and show their
runtime dependencies.


Yaakov

[PATCH] doc: document printf %m

Yaakov (Cygwin/X — 2012-01-13T09:26:01

New feature just added to newlib.  Patch for winsup/doc attached.


Yaakov

[PATCH] doc: tgmath.h

Yaakov (Cygwin/X — 2012-01-12T11:28:44

New header just committed to newlib.  Patch attached.


Yaakov

[PATCH] Add pthread_sigqueue(3)

Yaakov (Cygwin/X — 2012-01-01T18:59:00

This patchset adds pthread_sigqueue(3), a GNU extension:

http://www.kernel.org/doc/man-pages/online/pages/man3/pthread_sigqueue.3.html

The implementation is based on the existing sigqueue(2) and
pthread_kill(3) code.

Patches for winsup/cygwin and winsup/doc attached.


Yaakov