Unable to open the clipboard

Unable to open the clipboard

[Marco Atzeri]

Hi Guys,

I noticed, from time to time, that on recent Cygwin

$ uname -svr
CYGWIN_NT-10.0 3.3.3(0.341/5/3) 2021-12-03 16:35

there seems to be a "failed race (?)" using the clipboard

$ putclip < Announce_geos
Unable to open the clipboard

$ putclip < Announce_geos


Usually never happend two time in the row and it is a relative
rare event. But I never had such issue in the past.

It is not caused by computer load, probably arised in
the 3.3.x times but I could be wrong.

Regards
Marco

Re: Unable to open the clipboard

[Ken Brown]

On 1/3/2022 9:34 AM, Marco Atzeri wrote:
> Hi Guys,
> 
> I noticed, from time to time, that on recent Cygwin
> 
> $ uname -svr
> CYGWIN_NT-10.0 3.3.3(0.341/5/3) 2021-12-03 16:35
> 
> there seems to be a "failed race (?)" using the clipboard
> 
> $ putclip < Announce_geos
> Unable to open the clipboard

Hi Marco,

There was a clipboard bug that Takashi fixed in commit 69ed8ca20.  I don't know 
if it's the same issue that you're seeing.  The fix will appear in Cygwin 3.3.4, 
or you could try the latest snapshot.

Ken

Re: Unable to open the clipboard

[Marco Atzeri]

On 03.01.2022 16:59, Ken Brown wrote:
> On 1/3/2022 9:34 AM, Marco Atzeri wrote:
>> Hi Guys,
>>
>> I noticed, from time to time, that on recent Cygwin
>>
>> $ uname -svr
>> CYGWIN_NT-10.0 3.3.3(0.341/5/3) 2021-12-03 16:35
>>
>> there seems to be a "failed race (?)" using the clipboard
>>
>> $ putclip < Announce_geos
>> Unable to open the clipboard
> 
> Hi Marco,
> 
> There was a clipboard bug that Takashi fixed in commit 69ed8ca20.  I 
> don't know if it's the same issue that you're seeing.  The fix will 
> appear in Cygwin 3.3.4, or you could try the latest snapshot.
> 
> Ken
> 

The last snapshot that includes that and another one still has the issue

$ uname -svr
CYGWIN_NT-10.0 3.4.0s(0.341/5/3) 2021-12-13 12:59

and it seems I am able to hit multiple failures in a row
very easily


$ putclip < COPYING

$ putclip < COPYING
Unable to open the clipboard

$ putclip < COPYING
Unable to open the clipboard

$ putclip < COPYING

$ putclip < COPYING
Unable to open the clipboard

$ putclip < COPYING
Unable to open the clipboard

$ putclip < COPYING
Unable to open the clipboard

$ putclip < COPYING
Unable to open the clipboard

$ putclip < COPYING

Regards
Marco

Re: Unable to open the clipboard

[Mark Geisert]

Hi Marco,

Marco Atzeri wrote:
> Hi Guys,
> 
> I noticed, from time to time, that on recent Cygwin
> 
> $ uname -svr
> CYGWIN_NT-10.0 3.3.3(0.341/5/3) 2021-12-03 16:35
> 
> there seems to be a "failed race (?)" using the clipboard
> 
> $ putclip < Announce_geos
> Unable to open the clipboard
> 
> $ putclip < Announce_geos
> 
> 
> Usually never happend two time in the row and it is a relative
> rare event. But I never had such issue in the past.
> 
> It is not caused by computer load, probably arised in
> the 3.3.x times but I could be wrong.

I'm assuming 64-bit Cygwin.  What is your cygutils version please?

..mark

Re: Unable to open the clipboard

[Marco Atzeri]

On 04.01.2022 01:25, Mark Geisert wrote:
> Hi Marco,
> 
> Marco Atzeri wrote:
>> Hi Guys,
>>
>> I noticed, from time to time, that on recent Cygwin
>>
>> $ uname -svr
>> CYGWIN_NT-10.0 3.3.3(0.341/5/3) 2021-12-03 16:35
>>
>> there seems to be a "failed race (?)" using the clipboard
>>
>> $ putclip < Announce_geos
>> Unable to open the clipboard
>>
>> $ putclip < Announce_geos
>>
>>
>> Usually never happend two time in the row and it is a relative
>> rare event. But I never had such issue in the past.
>>
>> It is not caused by computer load, probably arised in
>> the 3.3.x times but I could be wrong.
> 
> I'm assuming 64-bit Cygwin.  What is your cygutils version please?
> 
> ..mark
> 

Hi Mark,
I see in both versions 32 and 64

Windows 10 version 21H1 build 19043.1415

$ cygcheck -c cygutils
Cygwin Package Information
Package              Version        Status
cygutils             1.4.16-7       OK

Regards
Marco

Re: Unable to open the clipboard

[Mark Geisert]

Marco Atzeri wrote:
> On 04.01.2022 01:25, Mark Geisert wrote:
>> Hi Marco,
>>
>> Marco Atzeri wrote:
>>> Hi Guys,
>>>
>>> I noticed, from time to time, that on recent Cygwin
>>>
>>> $ uname -svr
>>> CYGWIN_NT-10.0 3.3.3(0.341/5/3) 2021-12-03 16:35
>>>
>>> there seems to be a "failed race (?)" using the clipboard
>>>
>>> $ putclip < Announce_geos
>>> Unable to open the clipboard
>>>
>>> $ putclip < Announce_geos
>>>
>>>
>>> Usually never happend two time in the row and it is a relative
>>> rare event. But I never had such issue in the past.
>>>
>>> It is not caused by computer load, probably arised in
>>> the 3.3.x times but I could be wrong.
>>
>> I'm assuming 64-bit Cygwin.  What is your cygutils version please?
>>
>> ..mark
>>
> 
> Hi Mark,
> I see in both versions 32 and 64
> 
> Windows 10 version 21H1 build 19043.1415
> 
> $ cygcheck -c cygutils
> Cygwin Package Information
> Package              Version        Status
> cygutils             1.4.16-7       OK

I haven't been able to reproduce the issue with -7 or the more recent -8 test 
version.  I suppose you might try -8 and see if it helps.  Puzzling...

..mark

Re: Unable to open the clipboard

[Mark Geisert]

Mark Geisert wrote:
> I haven't been able to reproduce the issue with -7 or the more recent -8 test 
> version.  I suppose you might try -8 and see if it helps.  Puzzling...

Sorry, the {put,get}clip.exe are in cygutils-extra package.  But -8 is correct.

..mark

Re: Unable to open the clipboard

[Marco Atzeri]

On 04.01.2022 06:15, Mark Geisert wrote:
> Mark Geisert wrote:
>> I haven't been able to reproduce the issue with -7 or the more recent 
>> -8 test version.  I suppose you might try -8 and see if it helps.  
>> Puzzling...
> 
> Sorry, the {put,get}clip.exe are in cygutils-extra package.  But -8 is 
> correct.
> 
> ..mark
> 

Thanks Mark,
-8 seems to work fine.

Re: Unable to open the clipboard

[Marco Atzeri]

[-- Attachment #1: Type: text/plain, Size: 746 bytes --]

On 04.01.2022 19:08, Marco Atzeri wrote:
> On 04.01.2022 06:15, Mark Geisert wrote:
>> Mark Geisert wrote:
>>> I haven't been able to reproduce the issue with -7 or the more recent 
>>> -8 test version.  I suppose you might try -8 and see if it helps. 
>>> Puzzling...
>>
>> Sorry, the {put,get}clip.exe are in cygutils-extra package.  But -8 is 
>> correct.
>>
>> ..mark
>>
> 
> Thanks Mark,
> -8 seems to work fine.
> 

Hi Mark,
just found another issue

$ putclip < Announce_octave
Aborted (core dumped)

It seems caused by the longest line (86 characters)

$ cat Announce_octave  | putclip
Aborted (core dumped)

$ cat Announce_octave | sed -e "/html$/ d" | putclip

of course the Clipboard is fine

$ cat Announce_octave  > /dev/clipboard

[-- Attachment #2: Announce_octave --]
[-- Type: text/plain, Size: 1544 bytes --]

To: cygwin-announce@cygwin.com
Subject: Updated: octave-6.4.0-1

New versions 6.4.0-1 of

  octave
  octave-devel
  octave-doc

are available in the Cygwin distribution:

CHANGES
Latest upstream  release. 
https://www.gnu.org/software/octave/news/release/2021/10/30/octave-6.4.0-released.html

GENERAL NOTE
there are 2 octave main binaries with multiple graphic properties
    
  octave          command line with full plotting capabilities

  octave --gui    graphical interface

  octave-cli      command line inteface with plotting
                  restricted to fltk and gnuplot toolkits                

Of the 3 graphics toolkit fltk, gnuplot and qt

  fltk is the worst performing on cygwin
  gnuplot the lightest
  qt the most graphic rich

OCTAVE-FORGE NOTE
You need to install with setup the specific forge package 
you are looking for.

The update of the global pkg database is managed 
by cygwin setup during the postinstall phase.
A semaphore system is in place to update the database only when needed 
and just one time, also when multiple forge packages are added or removed.
 
DESCRIPTION
The GNU Octave language for numerical computations is a (mostly 
Matlab (R) compatible) high-level language, primarily intended 
for numerical computations.  It provides a convenient command-line 
interface for solving linear and nonlinear problems numerically.

HOMEPAGE
http://www.gnu.org/software/octave/

Marco Atzeri

If you have questions or comments, please send them to the 
cygwin mailing list at: cygwin (at) cygwin (dot) com .

Re: Unable to open the clipboard

[Mark Geisert]

Marco Atzeri wrote:
> On 04.01.2022 19:08, Marco Atzeri wrote:
>> On 04.01.2022 06:15, Mark Geisert wrote:
>>> Mark Geisert wrote:
>>>> I haven't been able to reproduce the issue with -7 or the more recent -8 test 
>>>> version.  I suppose you might try -8 and see if it helps. Puzzling...
>>>
>>> Sorry, the {put,get}clip.exe are in cygutils-extra package.  But -8 is correct.
>>>
>>> ..mark
>>>
>>
>> Thanks Mark,
>> -8 seems to work fine.
>>
> 
> Hi Mark,
> just found another issue
> 
> $ putclip < Announce_octave
> Aborted (core dumped)
> 
> It seems caused by the longest line (86 characters)
> 
> $ cat Announce_octave  | putclip
> Aborted (core dumped)
> 
> $ cat Announce_octave | sed -e "/html$/ d" | putclip
> 
> of course the Clipboard is fine
> 
> $ cat Announce_octave  > /dev/clipboard

Thanks Marco for the testcase.  I will investigate.

..mark

Re: Unable to open the clipboard

[Mark Geisert]

> Marco Atzeri wrote:
>> On 04.01.2022 19:08, Marco Atzeri wrote:
>>> On 04.01.2022 06:15, Mark Geisert wrote:
>>>> Mark Geisert wrote:
>>>>> I haven't been able to reproduce the issue with -7 or the more recent -8 test 
>>>>> version.  I suppose you might try -8 and see if it helps. Puzzling...
>>>>
>>>> Sorry, the {put,get}clip.exe are in cygutils-extra package.  But -8 is correct.
>>>>
>>>> ..mark
>>>>
>>>
>>> Thanks Mark,
>>> -8 seems to work fine.
>>>
>>
>> Hi Mark,
>> just found another issue
>>
>> $ putclip < Announce_octave
>> Aborted (core dumped)
>>
>> It seems caused by the longest line (86 characters)
>>
>> $ cat Announce_octave  | putclip
>> Aborted (core dumped)
>>
>> $ cat Announce_octave | sed -e "/html$/ d" | putclip
>>
>> of course the Clipboard is fine
>>
>> $ cat Announce_octave  > /dev/clipboard

I'm unable to reproduce.  If you happen to have a putclip.exe.stackdump left 
behind, could you post that please?

..mark

Re: Unable to open the clipboard

[Marco Atzeri]

On 10.01.2022 07:44, Mark Geisert wrote:
>> Marco Atzeri wrote:
>>> On 04.01.2022 19:08, Marco Atzeri wrote:
>>>> On 04.01.2022 06:15, Mark Geisert wrote:
>>>
>>> Hi Mark,
>>> just found another issue
>>>
>>> $ putclip < Announce_octave
>>> Aborted (core dumped)
>>>
>>> It seems caused by the longest line (86 characters)
>>>
>>> $ cat Announce_octave  | putclip
>>> Aborted (core dumped)
>>>
>>> $ cat Announce_octave | sed -e "/html$/ d" | putclip
>>>
>>> of course the Clipboard is fine
>>>
>>> $ cat Announce_octave  > /dev/clipboard
> 
> I'm unable to reproduce.  If you happen to have a putclip.exe.stackdump 
> left behind, could you post that please?
> 
> ..mark
> 

here it is

$ cat putclip.exe.stackdump
Stack trace:
Frame        Function    Args
000FFFFB980  001800620B7 (000FFFFBB88, 00000000002, 00000000002, 
000FFFFDE50)
00000000000  001800640F5 (00000000064, 00000000000, 00000000078, 
00000000000)
000FFFFC090  001801305E8 (00100000080, 000007E67F0, 00000000000, 
00000000000)
000000000C1  0018012BD1B (0000000001F, 00000000000, 00000000000, 
0018022DE20)
00000000609  0018012C125 (7FF84F2EBA68, 000FFFFC420, 7FF84FA9FB43, 
00000000000)
00000000609  00180212C08 (000FFFFC440, 00000000000, 00800000000, 
13920028A3BD5C75)
00000000609  00180213065 (00000400028, 7FF84FA958B0, 7FF84FA95990, 
00000400028)
00000000609  001800D7FE8 (00000000000, 00000400018, 00100000001, 
00800000BB0)
00000000609  0018018EFFB (00000000000, 00000400018, 00100000001, 
00800000BB0)
00000000609  0010040162C (37003600000002, EA8087F46D20, 7FF84EA653DC, 
00000000000)
00000000000  0010040261E (00180049B21, 00180048A70, 00000000002, 
00180322FC0)
000FFFFCD30  00180049B8D (00000000000, 00000000000, 00000000000, 
00000000000)
000FFFFFFF0  00180047746 (00000000000, 00000000000, 00000000000, 
00000000000)
000FFFFFFF0  001800477F4 (00000000000, 00000000000, 00000000000, 
00000000000)
End of stack trace

Re: Unable to open the clipboard

[Mark Geisert]

Hi Marco,

Marco Atzeri wrote:
> On 10.01.2022 07:44, Mark Geisert wrote:
>>> Marco Atzeri wrote:
>>>> On 04.01.2022 19:08, Marco Atzeri wrote:
>>>>> On 04.01.2022 06:15, Mark Geisert wrote:
>>>>
>>>> Hi Mark,
>>>> just found another issue
>>>>
>>>> $ putclip < Announce_octave
>>>> Aborted (core dumped)
>>>>
>>>> It seems caused by the longest line (86 characters)
>>>>
>>>> $ cat Announce_octave  | putclip
>>>> Aborted (core dumped)
>>>>
>>>> $ cat Announce_octave | sed -e "/html$/ d" | putclip
>>>>
>>>> of course the Clipboard is fine
>>>>
>>>> $ cat Announce_octave  > /dev/clipboard
>>
>> I'm unable to reproduce.  If you happen to have a putclip.exe.stackdump left 
>> behind, could you post that please?
>>
>> ..mark
>>
> 
> here it is
> 
> $ cat putclip.exe.stackdump
> Stack trace:
> Frame        Function    Args
> 000FFFFB980  001800620B7 (000FFFFBB88, 00000000002, 00000000002, 000FFFFDE50)
> 00000000000  001800640F5 (00000000064, 00000000000, 00000000078, 00000000000)
> 000FFFFC090  001801305E8 (00100000080, 000007E67F0, 00000000000, 00000000000)
> 000000000C1  0018012BD1B (0000000001F, 00000000000, 00000000000, 0018022DE20)
> 00000000609  0018012C125 (7FF84F2EBA68, 000FFFFC420, 7FF84FA9FB43, 00000000000)
> 00000000609  00180212C08 (000FFFFC440, 00000000000, 00800000000, 13920028A3BD5C75)
> 00000000609  00180213065 (00000400028, 7FF84FA958B0, 7FF84FA95990, 00000400028)
> 00000000609  001800D7FE8 (00000000000, 00000400018, 00100000001, 00800000BB0)
> 00000000609  0018018EFFB (00000000000, 00000400018, 00100000001, 00800000BB0)
> 00000000609  0010040162C (37003600000002, EA8087F46D20, 7FF84EA653DC, 00000000000)
> 00000000000  0010040261E (00180049B21, 00180048A70, 00000000002, 00180322FC0)
> 000FFFFCD30  00180049B8D (00000000000, 00000000000, 00000000000, 00000000000)
> 000FFFFFFF0  00180047746 (00000000000, 00000000000, 00000000000, 00000000000)
> 000FFFFFFF0  001800477F4 (00000000000, 00000000000, 00000000000, 00000000000)
> End of stack trace

Thanks for that.  Just wanted to say I haven't forgotten about this report. 
Unlike previous putclip|getclip error reports, this one is a Heisenbug.. difficult 
to reproduce under a debugger or with any change made to the source.  I do have a 
handle on the issue thanks to a malloc debugger: it's a buffer overrun.  Debugging 
continues...

..mark