* Race condition hangs on multiple mintty/tcsh? Brad Wetmore
@ 2020-08-03 22:13 Brad Wetmore
2020-08-04 10:02 ` Thomas Wolff
2020-08-06 18:07 ` Brad Wetmore
0 siblings, 2 replies; 20+ messages in thread
From: Brad Wetmore @ 2020-08-03 22:13 UTC (permalink / raw)
To: cygwin
Hi,
I generally kick off multiple (10) mintty sessions, and place them around the screen.
C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
--position @3 --position 120,0 --size 80x71 /bin/tcsh &
C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
--position @3 --position 715,0 --size 80x45 /bin/tcsh &
C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
--position @3 --position 715,660 --size 80x24 /bin/tcsh &
Within the last 6 months or so, about 2-3 of them would hang and either mintty/tcsh would not start. I put a "sleep 1" in between each invocation and that seemed to take care of it.
With the latest cygwin update, about 8 of them just hang even with the sleep 1. I put in a "sleep 2", and now everything is coming up again.
Not sure if this is a mintty or tcsh issue, but just wondering if others are seeing this before I start trying to debug this.
Thanks,
Brad
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-03 22:13 Race condition hangs on multiple mintty/tcsh? Brad Wetmore Brad Wetmore
@ 2020-08-04 10:02 ` Thomas Wolff
2020-08-05 22:56 ` Thomas Wolff
2020-08-06 18:07 ` Brad Wetmore
1 sibling, 1 reply; 20+ messages in thread
From: Thomas Wolff @ 2020-08-04 10:02 UTC (permalink / raw)
To: cygwin
Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
> Hi,
>
> I generally kick off multiple (10) mintty sessions, and place them around the screen.
>
> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
>
> Within the last 6 months or so, about 2-3 of them would hang and either mintty/tcsh would not start. I put a "sleep 1" in between each invocation and that seemed to take care of it.
>
> With the latest cygwin update, about 8 of them just hang even with the sleep 1. I put in a "sleep 2", and now everything is coming up again.
>
> Not sure if this is a mintty or tcsh issue, but just wondering if others are seeing this before I start trying to debug this.
I can reproduce such behaviour with /bin/bash (easy cross-check), and in
fact the shell is running in that case (easy test via `echo > .log`), so
I have a vague and unpleasant suspicion it might in fact be related to
mintty although I have no idea how that would happen. To be analysed.
Thomas
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-04 10:02 ` Thomas Wolff
@ 2020-08-05 22:56 ` Thomas Wolff
2020-08-05 23:23 ` Kevin Schnitzius
0 siblings, 1 reply; 20+ messages in thread
From: Thomas Wolff @ 2020-08-05 22:56 UTC (permalink / raw)
To: cygwin
Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
>> Hi,
>>
>> I generally kick off multiple (10) mintty sessions, and place them
>> around the screen.
>>
>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
>>
>> Within the last 6 months or so, about 2-3 of them would hang and
>> either mintty/tcsh would not start. I put a "sleep 1" in between each
>> invocation and that seemed to take care of it.
>>
>> With the latest cygwin update, about 8 of them just hang even with
>> the sleep 1. I put in a "sleep 2", and now everything is coming up
>> again.
>>
>> Not sure if this is a mintty or tcsh issue, but just wondering if
>> others are seeing this before I start trying to debug this.
> I can reproduce such behaviour with /bin/bash (easy cross-check), and
> in fact the shell is running in that case (easy test via `echo >
> .log`), so I have a vague and unpleasant suspicion it might in fact be
> related to mintty although I have no idea how that would happen. To be
> analysed.
When in this state, on the pty from which mintty receives child process
output, select() does not report a ready for reading condition; could it
be related to the recent poll/select patch? I could not reproduce it in
cygwin 3.0.7.
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-05 22:56 ` Thomas Wolff
@ 2020-08-05 23:23 ` Kevin Schnitzius
2020-08-06 11:46 ` Thomas Wolff
0 siblings, 1 reply; 20+ messages in thread
From: Kevin Schnitzius @ 2020-08-05 23:23 UTC (permalink / raw)
To: cygwin
On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff <towo@towo.net> wrote:
>Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
>>> Hi,
>>>
>>> I generally kick off multiple (10) mintty sessions, and place them
>>> around the screen.
>>>
>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
>>>
>>> Within the last 6 months or so, about 2-3 of them would hang and
>>> either mintty/tcsh would not start. I put a "sleep 1" in between each
>>> invocation and that seemed to take care of it.
>>>
>>> With the latest cygwin update, about 8 of them just hang even with
>>> the sleep 1. I put in a "sleep 2", and now everything is coming up
>>> again.
>>>
>>> Not sure if this is a mintty or tcsh issue, but just wondering if
>>> others are seeing this before I start trying to debug this.
>> I can reproduce such behaviour with /bin/bash (easy cross-check), and
>> in fact the shell is running in that case (easy test via `echo >
>> .log`), so I have a vague and unpleasant suspicion it might in fact be
>> related to mintty although I have no idea how that would happen. To be
>> analysed.
> When in this state, on the pty from which mintty receives child process
> output, select() does not report a ready for reading condition; could it
> be related to the recent poll/select patch? I could not reproduce it in
> cygwin 3.0.7.
for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ; do mintty -i /Cygwin-Terminal.ico --position @3 --position 120,$i --size 80x24 /bin/bash & done
This does not fail for me. Nor does it fail from cmd or powershell using a script.
I tried 3.1.4 and 3.1.6 on Windows 10. I tried up to 100 instances of of mintty...
Kevin
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-05 23:23 ` Kevin Schnitzius
@ 2020-08-06 11:46 ` Thomas Wolff
2020-08-06 13:31 ` Thomas Wolff
0 siblings, 1 reply; 20+ messages in thread
From: Thomas Wolff @ 2020-08-06 11:46 UTC (permalink / raw)
To: cygwin
Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff <towo@towo.net> wrote:
>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
>>>> Hi,
>>>>
>>>> I generally kick off multiple (10) mintty sessions, and place them
>>>> around the screen.
>>>>
>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>>> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>>> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
>>>>
>>>> Within the last 6 months or so, about 2-3 of them would hang and
>>>> either mintty/tcsh would not start. I put a "sleep 1" in between each
>>>> invocation and that seemed to take care of it.
>>>>
>>>> With the latest cygwin update, about 8 of them just hang even with
>>>> the sleep 1. I put in a "sleep 2", and now everything is coming up
>>>> again.
>>>>
>>>> Not sure if this is a mintty or tcsh issue, but just wondering if
>>>> others are seeing this before I start trying to debug this.
>>> I can reproduce such behaviour with /bin/bash (easy cross-check), and
>>> in fact the shell is running in that case (easy test via `echo >
>>> .log`), so I have a vague and unpleasant suspicion it might in fact be
>>> related to mintty although I have no idea how that would happen. To be
>>> analysed.
>> When in this state, on the pty from which mintty receives child process
>> output, select() does not report a ready for reading condition; could it
>> be related to the recent poll/select patch? I could not reproduce it in
>> cygwin 3.0.7.
> for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ; do mintty -i /Cygwin-Terminal.ico --position @3 --position 120,$i --size 80x24 /bin/bash & done
>
> This does not fail for me. Nor does it fail from cmd or powershell using a script.
>
> I tried 3.1.4 and 3.1.6 on Windows 10. I tried up to 100 instances of of mintty...
Thanks for testing. It is in fact hard to reproduce, maybe also
depending on system load (speculating).
I could reproduce one case of one of three terminals being unresponsive
also with xterm.
Does anybody familiar with pty/select or recent changes have any idea?
Thomas
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-06 11:46 ` Thomas Wolff
@ 2020-08-06 13:31 ` Thomas Wolff
2020-08-06 14:26 ` Corinna Vinschen
2020-08-11 4:59 ` Takashi Yano
0 siblings, 2 replies; 20+ messages in thread
From: Thomas Wolff @ 2020-08-06 13:31 UTC (permalink / raw)
To: cygwin
Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
> Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
>> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
>> <towo@towo.net> wrote:
>>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
>>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
>>>>> Hi,
>>>>>
>>>>> I generally kick off multiple (10) mintty sessions, and place them
>>>>> around the screen.
>>>>>
>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>>> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>>>> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>>>> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
>>>>>
>>>>> Within the last 6 months or so, about 2-3 of them would hang and
>>>>> either mintty/tcsh would not start. I put a "sleep 1" in between each
>>>>> invocation and that seemed to take care of it.
>>>>>
>>>>> With the latest cygwin update, about 8 of them just hang even with
>>>>> the sleep 1. I put in a "sleep 2", and now everything is coming up
>>>>> again.
>>>>>
>>>>> Not sure if this is a mintty or tcsh issue, but just wondering if
>>>>> others are seeing this before I start trying to debug this.
>>>> I can reproduce such behaviour with /bin/bash (easy cross-check), and
>>>> in fact the shell is running in that case (easy test via `echo >
>>>> .log`), so I have a vague and unpleasant suspicion it might in fact be
>>>> related to mintty although I have no idea how that would happen. To be
>>>> analysed.
>>> When in this state, on the pty from which mintty receives child process
>>> output, select() does not report a ready for reading condition;
>>> could it
>>> be related to the recent poll/select patch? I could not reproduce it in
>>> cygwin 3.0.7.
>> for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ; do
>> mintty -i /Cygwin-Terminal.ico --position @3 --position 120,$i
>> --size 80x24 /bin/bash & done
>>
>> This does not fail for me. Nor does it fail from cmd or powershell
>> using a script.
>>
>> I tried 3.1.4 and 3.1.6 on Windows 10. I tried up to 100 instances
>> of of mintty...
> Thanks for testing. It is in fact hard to reproduce, maybe also
> depending on system load (speculating).
> I could reproduce one case of one of three terminals being
> unresponsive also with xterm.
And it also happens if I drop select() from mintty (and use just
non-blocking read()).
> Does anybody familiar with pty/select or recent changes have any idea?
?
> Thomas
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-06 13:31 ` Thomas Wolff
@ 2020-08-06 14:26 ` Corinna Vinschen
2020-08-06 14:33 ` Thomas Wolff
2020-08-11 4:59 ` Takashi Yano
1 sibling, 1 reply; 20+ messages in thread
From: Corinna Vinschen @ 2020-08-06 14:26 UTC (permalink / raw)
To: cygwin
On Aug 6 15:31, Thomas Wolff wrote:
> Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
> > Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
> > > On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
> > > <towo@towo.net> wrote:
> > > > Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
> > > > > Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
> > > > > > Hi,
> > > > > >
> > > > > > I generally kick off multiple (10) mintty sessions, and place them
> > > > > > around the screen.
> > > > > >
> > > > > > C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> > > > > --position @3 --position 120,0 --size 80x71 /bin/tcsh &
> > > > > > C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> > > > > > --position @3 --position 715,0 --size 80x45 /bin/tcsh &
> > > > > > C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> > > > > > --position @3 --position 715,660 --size 80x24 /bin/tcsh &
> > > > > >
> > > > > > Within the last 6 months or so, about 2-3 of them would hang and
> > > > > > either mintty/tcsh would not start. I put a "sleep 1" in between each
> > > > > > invocation and that seemed to take care of it.
> > > > > >
> > > > > > With the latest cygwin update, about 8 of them just hang even with
> > > > > > the sleep 1. I put in a "sleep 2", and now everything is coming up
> > > > > > again.
> > > > > >
> > > > > > Not sure if this is a mintty or tcsh issue, but just wondering if
> > > > > > others are seeing this before I start trying to debug this.
> > > > > I can reproduce such behaviour with /bin/bash (easy cross-check), and
> > > > > in fact the shell is running in that case (easy test via `echo >
> > > > > .log`), so I have a vague and unpleasant suspicion it might in fact be
> > > > > related to mintty although I have no idea how that would happen. To be
> > > > > analysed.
> > > > When in this state, on the pty from which mintty receives child process
> > > > output, select() does not report a ready for reading condition;
> > > > could it
> > > > be related to the recent poll/select patch? I could not reproduce it in
> > > > cygwin 3.0.7.
> > > for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ; do
> > > mintty -i /Cygwin-Terminal.ico --position @3 --position 120,$i
> > > --size 80x24 /bin/bash & done
> > >
> > > This does not fail for me. Nor does it fail from cmd or powershell
> > > using a script.
> > >
> > > I tried 3.1.4 and 3.1.6 on Windows 10. I tried up to 100 instances
> > > of of mintty...
> > Thanks for testing. It is in fact hard to reproduce, maybe also
> > depending on system load (speculating).
> > I could reproduce one case of one of three terminals being unresponsive
> > also with xterm.
> And it also happens if I drop select() from mintty (and use just
> non-blocking read()).
> > Does anybody familiar with pty/select or recent changes have any idea?
> ?
Something with the pseudo console? Can you test w/o pseudo console code,
just to check?
Other than that we probably need some bisecting to find the cause.
Corinna
--
Corinna Vinschen
Cygwin Maintainer
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-06 14:26 ` Corinna Vinschen
@ 2020-08-06 14:33 ` Thomas Wolff
2020-08-07 15:34 ` Takashi Yano
0 siblings, 1 reply; 20+ messages in thread
From: Thomas Wolff @ 2020-08-06 14:33 UTC (permalink / raw)
To: cygwin
Am 06.08.2020 um 16:26 schrieb Corinna Vinschen:
> On Aug 6 15:31, Thomas Wolff wrote:
>> Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
>>> Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
>>>> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
>>>> <towo@towo.net> wrote:
>>>>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
>>>>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
>>>>>>> Hi,
>>>>>>>
>>>>>>> I generally kick off multiple (10) mintty sessions, and place them
>>>>>>> around the screen.
>>>>>>>
>>>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>>>>> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
>>>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>>>>>> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
>>>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
>>>>>>> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
>>>>>>>
>>>>>>> Within the last 6 months or so, about 2-3 of them would hang and
>>>>>>> either mintty/tcsh would not start. I put a "sleep 1" in between each
>>>>>>> invocation and that seemed to take care of it.
>>>>>>>
>>>>>>> With the latest cygwin update, about 8 of them just hang even with
>>>>>>> the sleep 1. I put in a "sleep 2", and now everything is coming up
>>>>>>> again.
>>>>>>>
>>>>>>> Not sure if this is a mintty or tcsh issue, but just wondering if
>>>>>>> others are seeing this before I start trying to debug this.
>>>>>> I can reproduce such behaviour with /bin/bash (easy cross-check), and
>>>>>> in fact the shell is running in that case (easy test via `echo >
>>>>>> .log`), so I have a vague and unpleasant suspicion it might in fact be
>>>>>> related to mintty although I have no idea how that would happen. To be
>>>>>> analysed.
>>>>> When in this state, on the pty from which mintty receives child process
>>>>> output, select() does not report a ready for reading condition;
>>>>> could it
>>>>> be related to the recent poll/select patch? I could not reproduce it in
>>>>> cygwin 3.0.7.
>>>> for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ; do
>>>> mintty -i /Cygwin-Terminal.ico --position @3 --position 120,$i
>>>> --size 80x24 /bin/bash & done
>>>>
>>>> This does not fail for me. Nor does it fail from cmd or powershell
>>>> using a script.
>>>>
>>>> I tried 3.1.4 and 3.1.6 on Windows 10. I tried up to 100 instances
>>>> of of mintty...
>>> Thanks for testing. It is in fact hard to reproduce, maybe also
>>> depending on system load (speculating).
>>> I could reproduce one case of one of three terminals being unresponsive
>>> also with xterm.
>> And it also happens if I drop select() from mintty (and use just
>> non-blocking read()).
>>> Does anybody familiar with pty/select or recent changes have any idea?
>> ?
> Something with the pseudo console? Can you test w/o pseudo console code, just to check?
Hmm, CYGWIN=disable_pcon does not seem to disable ConPTY support for me
any more...
>
> Other than that we probably need some bisecting to find the cause.
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-03 22:13 Race condition hangs on multiple mintty/tcsh? Brad Wetmore Brad Wetmore
2020-08-04 10:02 ` Thomas Wolff
@ 2020-08-06 18:07 ` Brad Wetmore
2020-08-06 18:48 ` Thomas Wolff
2020-08-06 19:25 ` Brad Wetmore
1 sibling, 2 replies; 20+ messages in thread
From: Brad Wetmore @ 2020-08-06 18:07 UTC (permalink / raw)
To: cygwin
[I apologize if this doesn't thread correctly: I've been getting the digest, and can't respond directly. And I have no idea how my name ended up in the Subject line: Rookie Move!]
More debug information, and a workaround/solution:
I am running:
Windows 10 Version 1909, fully patched.
Cygwin v3.1.6-1
Openssh v8.3p1-1
mintty v3.2.0-1
tcsh v6.22.02-1
The issue seems to be a race condition setting up ssh-agent, and then immediately starting up the the mintty/tcsh terminals. My guess is that something in ssh-agent is not fully setup yet, and is causing read hiccups in mintty.
My startup script looks like:
#!/bin/sh
if [ -z "$SSH_AGENT_PID" ] ; then
echo "Starting ssh-agent"
eval $(ssh-agent)
ssh-add
#sleep 3
fi
C:/cygwin64/bin/mintty.exe -i /usr/share/mintty/icon/wsl.ico \
--position 0,0 --size 80x24 --fg black --bg white /bin/tcsh &
#sleep 1
...9 more invocations deleted...
Without the sleep 3 after starting ssh-agent, I see one of three conditions:
1. The mintty window is created, but immediately crashes printing a dozen or so "Use "exit" to leave tcsh" before it does. At least that I think it says, it happens fast! ;)
2. The mintty window is created, and is ready for input. On the first keypress, it prints a dozen or so "Use "exit" to leave tcsh" and crashes.
3. The mintty window is created, and hangs. No prompt.
I get conditions #1/2 when I don't pause between mintty invocations.
I get condition #2 when I pause for 1 second between mintty invocations.
If I pause for 3 seconds after starting ssh-agent, then everything works as expected.
Note this issue occurs only during the first setup of 10 terminals. Running it again does not tickle the issue, probably because ssh-agent is fully up.
Hope this helps.
Brad
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-06 18:07 ` Brad Wetmore
@ 2020-08-06 18:48 ` Thomas Wolff
2020-08-06 19:25 ` Brad Wetmore
1 sibling, 0 replies; 20+ messages in thread
From: Thomas Wolff @ 2020-08-06 18:48 UTC (permalink / raw)
To: cygwin
Am 06.08.2020 um 20:07 schrieb Brad Wetmore via Cygwin:
> [I apologize if this doesn't thread correctly: I've been getting the digest, and can't respond directly. And I have no idea how my name ended up in the Subject line: Rookie Move!]
>
> More debug information, and a workaround/solution:
>
> I am running:
>
> Windows 10 Version 1909, fully patched.
> Cygwin v3.1.6-1
> Openssh v8.3p1-1
> mintty v3.2.0-1
> tcsh v6.22.02-1
>
> The issue seems to be a race condition setting up ssh-agent, and then immediately starting up the the mintty/tcsh terminals. My guess is that something in ssh-agent is not fully setup yet, and is causing read hiccups in mintty.
>
> My startup script looks like:
>
> #!/bin/sh
>
> if [ -z "$SSH_AGENT_PID" ] ; then
> echo "Starting ssh-agent"
> eval $(ssh-agent)
> ssh-add
> #sleep 3
> fi
>
> C:/cygwin64/bin/mintty.exe -i /usr/share/mintty/icon/wsl.ico \
> --position 0,0 --size 80x24 --fg black --bg white /bin/tcsh &
> #sleep 1
> ...9 more invocations deleted...
>
> Without the sleep 3 after starting ssh-agent, I see one of three conditions:
>
> 1. The mintty window is created, but immediately crashes printing a dozen or so "Use "exit" to leave tcsh" before it does. At least that I think it says, it happens fast! ;)
>
> 2. The mintty window is created, and is ready for input. On the first keypress, it prints a dozen or so "Use "exit" to leave tcsh" and crashes.
>
> 3. The mintty window is created, and hangs. No prompt.
>
> I get conditions #1/2 when I don't pause between mintty invocations.
> I get condition #2 when I pause for 1 second between mintty invocations.
>
> If I pause for 3 seconds after starting ssh-agent, then everything works as expected.
>
> Note this issue occurs only during the first setup of 10 terminals. Running it again does not tickle the issue, probably because ssh-agent is fully up.
>
> Hope this helps.
You had previously reported the issue without ssh involved. Does this
make it reproducible more often? Otherwise let's not make things
unnecessarily complicated.
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-06 18:07 ` Brad Wetmore
2020-08-06 18:48 ` Thomas Wolff
@ 2020-08-06 19:25 ` Brad Wetmore
2020-08-13 3:08 ` Brad Wetmore
1 sibling, 1 reply; 20+ messages in thread
From: Brad Wetmore @ 2020-08-06 19:25 UTC (permalink / raw)
To: cygwin
Thomas Wolff wrote:
> You had previously reported the issue without ssh involved. Does this
> make it reproducible more often? Otherwise let's not make things
> unnecessarily complicated.
I didn't mention it originally because I didn't expect that would have an impact, but then I successfully tried 4 times with the ssh-agent/sleep 3. That's when I sent the email with what seemed to be a good workaround.
But after trying it the 5th time following your email, 1 of the 10 didn't pass. 🙁
I took out the ssh, and am seeing it there too, so that was not the issue. Drat.
So the delay helped, but didn't completely solve the problem. I have been stalling before starting the cygwin termincal/script for some minutes after a reboot, but that doesn't seem to make a difference either.
I can send you the script if you want to play with it.
Brad
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-06 14:33 ` Thomas Wolff
@ 2020-08-07 15:34 ` Takashi Yano
0 siblings, 0 replies; 20+ messages in thread
From: Takashi Yano @ 2020-08-07 15:34 UTC (permalink / raw)
To: cygwin
Hi Thomas,
On Thu, 6 Aug 2020 16:33:36 +0200
Thomas Wolff wrote:
> Am 06.08.2020 um 16:26 schrieb Corinna Vinschen:
> > On Aug 6 15:31, Thomas Wolff wrote:
> >> Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
> >>> Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
> >>>> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
> >>>> <towo@towo.net> wrote:
> >>>>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
> >>>>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
> >>>>>>> Hi,
> >>>>>>>
> >>>>>>> I generally kick off multiple (10) mintty sessions, and place them
> >>>>>>> around the screen.
> >>>>>>>
> >>>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>>>> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
> >>>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>>>>> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
> >>>>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>>>>> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
> >>>>>>>
> >>>>>>> Within the last 6 months or so, about 2-3 of them would hang and
> >>>>>>> either mintty/tcsh would not start. I put a "sleep 1" in between each
> >>>>>>> invocation and that seemed to take care of it.
> >>>>>>>
> >>>>>>> With the latest cygwin update, about 8 of them just hang even with
> >>>>>>> the sleep 1. I put in a "sleep 2", and now everything is coming up
> >>>>>>> again.
> >>>>>>>
> >>>>>>> Not sure if this is a mintty or tcsh issue, but just wondering if
> >>>>>>> others are seeing this before I start trying to debug this.
> >>>>>> I can reproduce such behaviour with /bin/bash (easy cross-check), and
> >>>>>> in fact the shell is running in that case (easy test via `echo >
> >>>>>> .log`), so I have a vague and unpleasant suspicion it might in fact be
> >>>>>> related to mintty although I have no idea how that would happen. To be
> >>>>>> analysed.
> >>>>> When in this state, on the pty from which mintty receives child process
> >>>>> output, select() does not report a ready for reading condition;
> >>>>> could it
> >>>>> be related to the recent poll/select patch? I could not reproduce it in
> >>>>> cygwin 3.0.7.
> >>>> for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ; do
> >>>> mintty -i /Cygwin-Terminal.ico --position @3 --position 120,$i
> >>>> --size 80x24 /bin/bash & done
> >>>>
> >>>> This does not fail for me. Nor does it fail from cmd or powershell
> >>>> using a script.
> >>>>
> >>>> I tried 3.1.4 and 3.1.6 on Windows 10. I tried up to 100 instances
> >>>> of of mintty...
> >>> Thanks for testing. It is in fact hard to reproduce, maybe also
> >>> depending on system load (speculating).
> >>> I could reproduce one case of one of three terminals being unresponsive
> >>> also with xterm.
> >> And it also happens if I drop select() from mintty (and use just
> >> non-blocking read()).
> >>> Does anybody familiar with pty/select or recent changes have any idea?
> >> ?
> > Something with the pseudo console? Can you test w/o pseudo console code, just to check?
> Hmm, CYGWIN=disable_pcon does not seem to disable ConPTY support for me
> any more...
CYGWIN=disable_pcon needs to be set before opening pty, i.e.
before opening mintty. It works for me with cygwin 3.1.6.
--
Takashi Yano <takashi.yano@nifty.ne.jp>
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-06 13:31 ` Thomas Wolff
2020-08-06 14:26 ` Corinna Vinschen
@ 2020-08-11 4:59 ` Takashi Yano
2020-08-11 9:37 ` Corinna Vinschen
1 sibling, 1 reply; 20+ messages in thread
From: Takashi Yano @ 2020-08-11 4:59 UTC (permalink / raw)
To: cygwin
[-- Attachment #1: Type: text/plain, Size: 3034 bytes --]
Hi Thomas,
On Thu, 6 Aug 2020 15:31:24 +0200
Thomas Wolff wrote:
> Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
> > Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
> >> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
> >> <towo@towo.net> wrote:
> >>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
> >>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
> >>>>> Hi,
> >>>>>
> >>>>> I generally kick off multiple (10) mintty sessions, and place them
> >>>>> around the screen.
> >>>>>
> >>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>> --position @3 --position 120,0 --size 80x71 /bin/tcsh &
> >>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>>> --position @3 --position 715,0 --size 80x45 /bin/tcsh &
> >>>>> C:/cygwin64/bin/mintty.exe -i /Cygwin-Terminal.ico \
> >>>>> --position @3 --position 715,660 --size 80x24 /bin/tcsh &
> >>>>>
> >>>>> Within the last 6 months or so, about 2-3 of them would hang and
> >>>>> either mintty/tcsh would not start. I put a "sleep 1" in between each
> >>>>> invocation and that seemed to take care of it.
> >>>>>
> >>>>> With the latest cygwin update, about 8 of them just hang even with
> >>>>> the sleep 1. I put in a "sleep 2", and now everything is coming up
> >>>>> again.
> >>>>>
> >>>>> Not sure if this is a mintty or tcsh issue, but just wondering if
> >>>>> others are seeing this before I start trying to debug this.
> >>>> I can reproduce such behaviour with /bin/bash (easy cross-check), and
> >>>> in fact the shell is running in that case (easy test via `echo >
> >>>> .log`), so I have a vague and unpleasant suspicion it might in fact be
> >>>> related to mintty although I have no idea how that would happen. To be
> >>>> analysed.
> >>> When in this state, on the pty from which mintty receives child process
> >>> output, select() does not report a ready for reading condition;
> >>> could it
> >>> be related to the recent poll/select patch? I could not reproduce it in
> >>> cygwin 3.0.7.
> >> for i in 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ; do
> >> mintty -i /Cygwin-Terminal.ico --position @3 --position 120,$i
> >> --size 80x24 /bin/bash & done
> >>
> >> This does not fail for me. Nor does it fail from cmd or powershell
> >> using a script.
> >>
> >> I tried 3.1.4 and 3.1.6 on Windows 10. I tried up to 100 instances
> >> of of mintty...
> > Thanks for testing. It is in fact hard to reproduce, maybe also
> > depending on system load (speculating).
> > I could reproduce one case of one of three terminals being
> > unresponsive also with xterm.
> And it also happens if I drop select() from mintty (and use just
> non-blocking read()).
> > Does anybody familiar with pty/select or recent changes have any idea?
I looked into this problem. After much struggle, I think
I have found a workaround for this issue.
I am not sure why this solves the issue at all, however,
this works for me.
Could you please test a patch attached?
--
Takashi Yano <takashi.yano@nifty.ne.jp>
[-- Attachment #2: 0001-Cygwin-pty-Add-a-workaround-for-issue-of-starting-a-.patch --]
[-- Type: application/octet-stream, Size: 2139 bytes --]
From cc451648bbc7f395fa00cb2b97351d1513827f82 Mon Sep 17 00:00:00 2001
From: Takashi Yano <takashi.yano@nifty.ne.jp>
Date: Tue, 11 Aug 2020 13:16:52 +0900
Subject: [PATCH] Cygwin: pty: Add a workaround for issue of starting a lot of
mintty.
- If a lot of mintty are started in a short time from a mintty, some
of them hang with empty screen, crash immediately or hang on exiting
mintty. The following report seems to be related to this issue.
https://cygwin.com/pipermail/cygwin/2020-August/245751.html
The cause is not clear at all, but this patch seems to solve the
issue.
---
winsup/cygwin/spawn.cc | 15 +++++++++++----
1 file changed, 11 insertions(+), 4 deletions(-)
diff --git a/winsup/cygwin/spawn.cc b/winsup/cygwin/spawn.cc
index 840ec4a86..e70ceb86d 100644
--- a/winsup/cygwin/spawn.cc
+++ b/winsup/cygwin/spawn.cc
@@ -177,7 +177,7 @@ find_exec (const char *name, path_conv& buf, const char *search,
/* Utility for child_info_spawn::worker. */
static HANDLE
-handle (int fd, bool writing)
+handle (int fd, bool writing, bool iscygwin)
{
HANDLE h;
cygheap_fdget cfd (fd);
@@ -188,6 +188,11 @@ handle (int fd, bool writing)
h = INVALID_HANDLE_VALUE;
else if (!writing)
h = cfd->get_handle ();
+ else if (cfd->get_major () == DEV_PTYS_MAJOR && iscygwin)
+ {
+ fhandler_pty_slave *ptys = (fhandler_pty_slave *)(fhandler_base *) cfd;
+ h = ptys->get_output_handle_cyg ();
+ }
else
h = cfd->get_output_handle ();
@@ -625,9 +630,11 @@ child_info_spawn::worker (const char *prog_arg, const char *const *argv,
/* Set up needed handles for stdio */
si.dwFlags = STARTF_USESTDHANDLES;
- si.hStdInput = handle ((in__stdin < 0 ? 0 : in__stdin), false);
- si.hStdOutput = handle ((in__stdout < 0 ? 1 : in__stdout), true);
- si.hStdError = handle (2, true);
+ si.hStdInput = handle ((in__stdin < 0 ? 0 : in__stdin), false,
+ iscygwin ());
+ si.hStdOutput = handle ((in__stdout < 0 ? 1 : in__stdout), true,
+ iscygwin ());
+ si.hStdError = handle (2, true, iscygwin ());
si.cb = sizeof (si);
--
2.27.0
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-11 4:59 ` Takashi Yano
@ 2020-08-11 9:37 ` Corinna Vinschen
2020-08-11 10:02 ` Takashi Yano
0 siblings, 1 reply; 20+ messages in thread
From: Corinna Vinschen @ 2020-08-11 9:37 UTC (permalink / raw)
To: cygwin
On Aug 11 13:59, Takashi Yano via Cygwin wrote:
> Hi Thomas,
>
> On Thu, 6 Aug 2020 15:31:24 +0200
> Thomas Wolff wrote:
> > Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
> > > Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
> > >> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
> > >> <towo@towo.net> wrote:
> > >>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
> > >>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
> > >>>>> Hi,
> > >>>>>
> > >>>>> I generally kick off multiple (10) mintty sessions, and place them
> > >>>>> around the screen.
> > >>>>> [...]
> > > I could reproduce one case of one of three terminals being
> > > unresponsive also with xterm.
> > And it also happens if I drop select() from mintty (and use just
> > non-blocking read()).
> > > Does anybody familiar with pty/select or recent changes have any idea?
>
> I looked into this problem. After much struggle, I think
> I have found a workaround for this issue.
>
> I am not sure why this solves the issue at all, however,
> this works for me.
>
> Could you please test a patch attached?
Shall I commit the patch and create a developer snapshot for testing?
Corinna
--
Corinna Vinschen
Cygwin Maintainer
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-11 9:37 ` Corinna Vinschen
@ 2020-08-11 10:02 ` Takashi Yano
2020-08-11 10:30 ` Corinna Vinschen
0 siblings, 1 reply; 20+ messages in thread
From: Takashi Yano @ 2020-08-11 10:02 UTC (permalink / raw)
To: cygwin
On Tue, 11 Aug 2020 11:37:58 +0200
Corinna Vinschen wrote:
> On Aug 11 13:59, Takashi Yano via Cygwin wrote:
> > Hi Thomas,
> >
> > On Thu, 6 Aug 2020 15:31:24 +0200
> > Thomas Wolff wrote:
> > > Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
> > > > Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
> > > >> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
> > > >> <towo@towo.net> wrote:
> > > >>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
> > > >>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
> > > >>>>> Hi,
> > > >>>>>
> > > >>>>> I generally kick off multiple (10) mintty sessions, and place them
> > > >>>>> around the screen.
> > > >>>>> [...]
> > > > I could reproduce one case of one of three terminals being
> > > > unresponsive also with xterm.
> > > And it also happens if I drop select() from mintty (and use just
> > > non-blocking read()).
> > > > Does anybody familiar with pty/select or recent changes have any idea?
> >
> > I looked into this problem. After much struggle, I think
> > I have found a workaround for this issue.
> >
> > I am not sure why this solves the issue at all, however,
> > this works for me.
> >
> > Could you please test a patch attached?
>
> Shall I commit the patch and create a developer snapshot for testing?
Thanks so much. That would be more convenient for potential testers.
--
Takashi Yano <takashi.yano@nifty.ne.jp>
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-11 10:02 ` Takashi Yano
@ 2020-08-11 10:30 ` Corinna Vinschen
2020-08-11 17:37 ` Thomas Wolff
0 siblings, 1 reply; 20+ messages in thread
From: Corinna Vinschen @ 2020-08-11 10:30 UTC (permalink / raw)
To: cygwin
On Aug 11 19:02, Takashi Yano via Cygwin wrote:
> On Tue, 11 Aug 2020 11:37:58 +0200
> Corinna Vinschen wrote:
> > On Aug 11 13:59, Takashi Yano via Cygwin wrote:
> > > Hi Thomas,
> > >
> > > On Thu, 6 Aug 2020 15:31:24 +0200
> > > Thomas Wolff wrote:
> > > > Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
> > > > > Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
> > > > >> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
> > > > >> <towo@towo.net> wrote:
> > > > >>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
> > > > >>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
> > > > >>>>> Hi,
> > > > >>>>>
> > > > >>>>> I generally kick off multiple (10) mintty sessions, and place them
> > > > >>>>> around the screen.
> > > > >>>>> [...]
> > > > > I could reproduce one case of one of three terminals being
> > > > > unresponsive also with xterm.
> > > > And it also happens if I drop select() from mintty (and use just
> > > > non-blocking read()).
> > > > > Does anybody familiar with pty/select or recent changes have any idea?
> > >
> > > I looked into this problem. After much struggle, I think
> > > I have found a workaround for this issue.
> > >
> > > I am not sure why this solves the issue at all, however,
> > > this works for me.
> > >
> > > Could you please test a patch attached?
> >
> > Shall I commit the patch and create a developer snapshot for testing?
>
> Thanks so much. That would be more convenient for potential testers.
Done. Please try the latest snapshot from https://cygwin.com/snapshots/
Thanks,
Corinna
--
Corinna Vinschen
Cygwin Maintainer
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-11 10:30 ` Corinna Vinschen
@ 2020-08-11 17:37 ` Thomas Wolff
2020-08-12 20:27 ` Thomas Wolff
0 siblings, 1 reply; 20+ messages in thread
From: Thomas Wolff @ 2020-08-11 17:37 UTC (permalink / raw)
To: cygwin
On 11.08.2020 12:30, Corinna Vinschen wrote:
> On Aug 11 19:02, Takashi Yano via Cygwin wrote:
>> On Tue, 11 Aug 2020 11:37:58 +0200
>> Corinna Vinschen wrote:
>>> On Aug 11 13:59, Takashi Yano via Cygwin wrote:
>>>> Hi Thomas,
>>>>
>>>> On Thu, 6 Aug 2020 15:31:24 +0200
>>>> Thomas Wolff wrote:
>>>>> Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
>>>>>> Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
>>>>>>> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
>>>>>>> <towo@towo.net> wrote:
>>>>>>>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
>>>>>>>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
>>>>>>>>>> Hi,
>>>>>>>>>>
>>>>>>>>>> I generally kick off multiple (10) mintty sessions, and place them
>>>>>>>>>> around the screen.
>>>>>>>>>> [...]
>>>>>> I could reproduce one case of one of three terminals being
>>>>>> unresponsive also with xterm.
>>>>> And it also happens if I drop select() from mintty (and use just
>>>>> non-blocking read()).
>>>>>> Does anybody familiar with pty/select or recent changes have any idea?
>>>> I looked into this problem. After much struggle, I think
>>>> I have found a workaround for this issue.
>>>>
>>>> I am not sure why this solves the issue at all, however,
>>>> this works for me.
>>>>
>>>> Could you please test a patch attached?
>>> Shall I commit the patch and create a developer snapshot for testing?
>> Thanks so much. That would be more convenient for potential testers.
> Done. Please try the latest snapshot from https://cygwin.com/snapshots/
Thanks. I could not reproduce the issue on one system in ~27 cases. I'd
like to retest on another system where I could previously produce the
issue more frequently, but that's currently broken...
Thomas
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-11 17:37 ` Thomas Wolff
@ 2020-08-12 20:27 ` Thomas Wolff
0 siblings, 0 replies; 20+ messages in thread
From: Thomas Wolff @ 2020-08-12 20:27 UTC (permalink / raw)
To: cygwin
Am 11.08.2020 um 19:37 schrieb Thomas Wolff:
>
> On 11.08.2020 12:30, Corinna Vinschen wrote:
>> On Aug 11 19:02, Takashi Yano via Cygwin wrote:
>>> On Tue, 11 Aug 2020 11:37:58 +0200
>>> Corinna Vinschen wrote:
>>>> On Aug 11 13:59, Takashi Yano via Cygwin wrote:
>>>>> Hi Thomas,
>>>>>
>>>>> On Thu, 6 Aug 2020 15:31:24 +0200
>>>>> Thomas Wolff wrote:
>>>>>> Am 06.08.2020 um 13:46 schrieb Thomas Wolff:
>>>>>>> Am 06.08.2020 um 01:23 schrieb Kevin Schnitzius via Cygwin:
>>>>>>>> On Wednesday, August 5, 2020, 06:56:48 PM EDT, Thomas Wolff
>>>>>>>> <towo@towo.net> wrote:
>>>>>>>>> Am 04.08.2020 um 12:02 schrieb Thomas Wolff:
>>>>>>>>>> Am 04.08.2020 um 00:13 schrieb Brad Wetmore via Cygwin:
>>>>>>>>>>> Hi,
>>>>>>>>>>>
>>>>>>>>>>> I generally kick off multiple (10) mintty sessions, and
>>>>>>>>>>> place them
>>>>>>>>>>> around the screen.
>>>>>>>>>>> [...]
>>>>>>> I could reproduce one case of one of three terminals being
>>>>>>> unresponsive also with xterm.
>>>>>> And it also happens if I drop select() from mintty (and use just
>>>>>> non-blocking read()).
>>>>>>> Does anybody familiar with pty/select or recent changes have any
>>>>>>> idea?
>>>>> I looked into this problem. After much struggle, I think
>>>>> I have found a workaround for this issue.
>>>>>
>>>>> I am not sure why this solves the issue at all, however,
>>>>> this works for me.
>>>>>
>>>>> Could you please test a patch attached?
>>>> Shall I commit the patch and create a developer snapshot for testing?
>>> Thanks so much. That would be more convenient for potential testers.
>> Done. Please try the latest snapshot from https://cygwin.com/snapshots/
> Thanks. I could not reproduce the issue on one system in ~27 cases.
> I'd like to retest on another system where I could previously produce
> the issue more frequently, but that's currently broken...
I gave it another 22 test instances on my other machine (the more
sensitive one), no issue reproduced with the patch.
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-06 19:25 ` Brad Wetmore
@ 2020-08-13 3:08 ` Brad Wetmore
2020-08-25 18:45 ` Brad Wetmore
0 siblings, 1 reply; 20+ messages in thread
From: Brad Wetmore @ 2020-08-13 3:08 UTC (permalink / raw)
To: cygwin
Hi Carinna/Thomas/Takashi,
> > > > > > Does anybody familiar with pty/select or recent changes have any idea?
> > > >
> > > > I looked into this problem. After much struggle, I think
> > > > I have found a workaround for this issue.
> > > >
> > > > I am not sure why this solves the issue at all, however,
> > > > this works for me.
> > > >
> > > Could you please test a patch attached?
> > >
> > > Shall I commit the patch and create a developer snapshot for testing?
> >
> > Thanks so much. That would be more convenient for potential testers.
>
> Done. Please try the latest snapshot from https://cygwin.com/snapshots/
Thanks all for looking into this. I dropped the .dll into my c:\cygwin64\bin directory and wasn't able to reproduce over 5 reboots. The minute I put back the original .dll, it failed again.
Let me know if there's anything else I can do to help.
Best to you all,
Brad
________________________________
From: Brad Wetmore <bradfordwetmore@hotmail.com>
Sent: Thursday, August 6, 2020 12:25 PM
To: cygwin@cygwin.com <cygwin@cygwin.com>
Subject: Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
Thomas Wolff wrote:
> You had previously reported the issue without ssh involved. Does this
> make it reproducible more often? Otherwise let's not make things
> unnecessarily complicated.
I didn't mention it originally because I didn't expect that would have an impact, but then I successfully tried 4 times with the ssh-agent/sleep 3. That's when I sent the email with what seemed to be a good workaround.
But after trying it the 5th time following your email, 1 of the 10 didn't pass. 🙁
I took out the ssh, and am seeing it there too, so that was not the issue. Drat.
So the delay helped, but didn't completely solve the problem. I have been stalling before starting the cygwin termincal/script for some minutes after a reboot, but that doesn't seem to make a difference either.
I can send you the script if you want to play with it.
Brad
^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
2020-08-13 3:08 ` Brad Wetmore
@ 2020-08-25 18:45 ` Brad Wetmore
0 siblings, 0 replies; 20+ messages in thread
From: Brad Wetmore @ 2020-08-25 18:45 UTC (permalink / raw)
To: cygwin
Just installed the latest cygwin (3.1.7-1), and my issue seems to be solved. Thanks to Thomas, Kevin, Corinna, and Takashi for looking in it.
Brad
________________________________
From: Brad Wetmore <bradfordwetmore@hotmail.com>
Sent: Wednesday, August 12, 2020 8:08 PM
To: cygwin@cygwin.com <cygwin@cygwin.com>
Subject: Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
Hi Carinna/Thomas/Takashi,
> > > > > > Does anybody familiar with pty/select or recent changes have any idea?
> > > >
> > > > I looked into this problem. After much struggle, I think
> > > > I have found a workaround for this issue.
> > > >
> > > > I am not sure why this solves the issue at all, however,
> > > > this works for me.
> > > >
> > > Could you please test a patch attached?
> > >
> > > Shall I commit the patch and create a developer snapshot for testing?
> >
> > Thanks so much. That would be more convenient for potential testers.
>
> Done. Please try the latest snapshot from https://cygwin.com/snapshots/
Thanks all for looking into this. I dropped the .dll into my c:\cygwin64\bin directory and wasn't able to reproduce over 5 reboots. The minute I put back the original .dll, it failed again.
Let me know if there's anything else I can do to help.
Best to you all,
Brad
________________________________
From: Brad Wetmore <bradfordwetmore@hotmail.com>
Sent: Thursday, August 6, 2020 12:25 PM
To: cygwin@cygwin.com <cygwin@cygwin.com>
Subject: Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore
Thomas Wolff wrote:
> You had previously reported the issue without ssh involved. Does this
> make it reproducible more often? Otherwise let's not make things
> unnecessarily complicated.
I didn't mention it originally because I didn't expect that would have an impact, but then I successfully tried 4 times with the ssh-agent/sleep 3. That's when I sent the email with what seemed to be a good workaround.
But after trying it the 5th time following your email, 1 of the 10 didn't pass. 🙁
I took out the ssh, and am seeing it there too, so that was not the issue. Drat.
So the delay helped, but didn't completely solve the problem. I have been stalling before starting the cygwin termincal/script for some minutes after a reboot, but that doesn't seem to make a difference either.
I can send you the script if you want to play with it.
Brad
^ permalink raw reply [flat|nested] 20+ messages in thread
end of thread, other threads:[~2020-08-25 18:45 UTC | newest]
Thread overview: 20+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-08-03 22:13 Race condition hangs on multiple mintty/tcsh? Brad Wetmore Brad Wetmore
2020-08-04 10:02 ` Thomas Wolff
2020-08-05 22:56 ` Thomas Wolff
2020-08-05 23:23 ` Kevin Schnitzius
2020-08-06 11:46 ` Thomas Wolff
2020-08-06 13:31 ` Thomas Wolff
2020-08-06 14:26 ` Corinna Vinschen
2020-08-06 14:33 ` Thomas Wolff
2020-08-07 15:34 ` Takashi Yano
2020-08-11 4:59 ` Takashi Yano
2020-08-11 9:37 ` Corinna Vinschen
2020-08-11 10:02 ` Takashi Yano
2020-08-11 10:30 ` Corinna Vinschen
2020-08-11 17:37 ` Thomas Wolff
2020-08-12 20:27 ` Thomas Wolff
2020-08-06 18:07 ` Brad Wetmore
2020-08-06 18:48 ` Thomas Wolff
2020-08-06 19:25 ` Brad Wetmore
2020-08-13 3:08 ` Brad Wetmore
2020-08-25 18:45 ` Brad Wetmore
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).