Re: Race condition hangs on multiple mintty/tcsh? Brad Wetmore

[Thomas Wolff <towo@towo.net>]

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.