inputready fd not set high following NCKEY_EOF #2525

dankamongmen · 2022-01-04T22:43:35Z

I can constantly read NCKEY_EOF, but the inputready fd is never high. If I add a check on input readiness, I never see the NCKEY_EOF. Test with cli1. See also #2521.

The text was updated successfully, but these errors were encountered:

dankamongmen · 2022-01-04T22:59:31Z

so the weird thing is....with the poll(), not only do i block, but i never get the NCKEY_EOF:

GOT: 0x001103f1
GOT: 0x0000006c
GOT: 0x0000006c
GOT: 0x0000006c
GOT: 0x0000006c
GOT: 0x001103f1
GOT: 0x001103f1
GOT: 0x00110424
GOT: 0x00000064
GOT: 0x00110424
GOT: 0x00000064
GOT: 0x00000064
GOT: 0x00000064
GOT: 0x00000071

so we're passing through the poll() any number of times -- NCKEY_EOF simply isn't being generated.

without the poll, we immediately get it, and exit:

GOT: 0x0011056c
2 renders, 440.48µs (147.32µs min, 220.24µs avg, 293.16µs max)
2 rasters, 111.63µs (54.60µs min, 55.81µs avg, 57.03µs max)
2 writes, 70.70µs (26.53µs min, 35.35µs avg, 44.17µs max)
55B (0B min, 27B avg, 55B max) 0 inputs Ghpa: 1
0 failed renders, 0 failed rasters, 0 refreshes, 0 input errors
RGB emits:elides: def 1:24 fg 0:0 bg 0:0
Cell emits:elides: 25:23735 (99.89%) 96.00% 0.00% 0.00%
Bitmap emits:elides: 0:0 (0.00%) 0B (0.00%) SuM: 0 (0.00%)

weird! but surely also the key to the problem.

dankamongmen · 2022-01-04T23:03:22Z

we only generate NCKEY_EOF when we come through internal_get() and hit:

pthread_mutex_lock(&ictx->ilock);                                                                                                 
  while(!ictx->ivalid){                                                                                                             
    if(ictx->stdineof){                                                                                                             
      pthread_mutex_unlock(&ictx->ilock);                                                                                           
      logwarn("read eof on stdin");                                                                                                 
      if(ni){                                                                                                                       
        memset(ni, 0, sizeof(*ni));                                                                                                 
        ni->id = NCKEY_EOF;                                                                                                         
      }                                                                                                                             
      return NCKEY_EOF;                                                                                                             
    }

so if we always hit this and have data available -- which we do, if we're checking poll() beforehand -- we never generate NCKEY_EOF.

the solution here is tripartite:
(1) resolve the main thrust of this bug by ensuring the inputready fd is readable once we hit NCKEY_EOF
(2) once we hit NCKEY_EOF, do not queue further input coming from the terminal
(3) always generate a NCKEY_EOF synthesized event upon hitting eof

dankamongmen · 2022-01-05T02:14:45Z

so yeah it doesn't look like we "generate an NCKEY_EOF" like we do other ncinputs, pushing it into the queue; it's an immediate result returned outside that paradigm. we see it down in read_input_nblock(), called from read_inputs_nblock(), i think process_input() needs to detect the level change, and when it happens, feed a queued event through.

dankamongmen · 2022-01-05T02:29:15Z

we're seeing a livelock in our drone CI when doing notcurses-input < notfcurses-input, and a deadlock when doing notcurses-input < /dev/null. this matches what we're seeing in Debian CI.

i got perf output from a livelock:

  44.34%  notcurses-input  libpthread-2.33.so          [.] __pthread_mutex_unlock_usercnt                                          ▒
  42.98%  notcurses-input  libpthread-2.33.so          [.] __pthread_mutex_lock                                                    ▒
   6.78%  notcurses-input  libnotcurses-core.so.3.0.3  [.] process_ncinput                                                         ◆
   3.28%  notcurses-input  libnotcurses-core.so.3.0.3  [.] process_melange                                                         ▒
   1.21%  notcurses-input  libnotcurses-core.so.3.0.3  [.] pthread_mutex_lock@plt                                                  ▒
   1.05%  notcurses-input  libnotcurses-core.so.3.0.3  [.] pthread_mutex_unlock@plt

dankamongmen · 2022-01-05T03:13:57Z

if we're spinning on pthreads....hrmmm, are we certain that's a livelock, and not just a long-lived process? eh, it's running twenty minutes, so presumably...

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

dankamongmen · 2022-01-05T20:21:03Z

ok, so there's one more issue, and this is presumably why our CI is failing. we can get the EOF before we get the flurry of terminal responses, as we probably do when redirecting from /dev/null, see here:

link_numeric:380:adding numeric with follow ; following ]4;                                                                         
link_numeric:393:phi node: 427->428                                                                                                 
link_numeric:394:eta node: 429 philink[;]: 429                                                                                      
insert_path:481:added fixed r 114 as 430                                                                                            
insert_path:481:added fixed g 103 as 431                                                                                            
insert_path:481:added fixed b 98 as 432                                                                                             
insert_path:481:added fixed : 58 as 433                                                                                             
esctrie_make_string:181:made string: 435                          <------ done building automaton                                                                  
block_on_input:2176:blocking on input availability                                                                                  
block_on_input:2233:waiting on 2 fds (ibuf: 0/8192)                                                                                 
block_on_input:2260:poll returned 2                                                                                                 
block_on_input:2274:got events: TI                                                                                                  
read_input_nblock:1871:read 4095B from 8 (4097B left)    <------------ read a bunch of terminal reply                                                                           
read_input_nblock:1862:got EOF on 0                                <------------ read eof on stdin                                                                 
process_escape:1908:initialized automaton to 1                 <------------ process terminal reply                                                                     
process_escape:1920:walk result on 91 ([): 0 4                                                                                      
process_escape:1920:walk result on 55 (7): 0 269                                                                                    
process_escape:1920:walk result on 52 (4): 0 269                                                                                    
process_escape:1920:walk result on 59 (;): 0 270                                                                                    
process_escape:1920:walk result on 49 (1): 0 271                                                                                    
process_escape:1920:walk result on 82 (R): 2 315

and here we never really process the EOF.

dankamongmen · 2022-01-05T20:35:56Z

hrmm, no, i just hadn't brought my local checkout up to speed. once i did, this problem goes away. we're still breaking in CI, though =.

dankamongmen · 2022-01-05T22:22:04Z

so in the drone context where we're seeing this lockup, here's our open fds:

root@2d7b55c0ae13:/proc/3070/cwd# ls -l /proc/1/fd
total 0
lrwx------ 1 root root 64 Jan  5 21:42 0 -> /dev/null
l-wx------ 1 root root 64 Jan  5 21:42 1 -> 'pipe:[5645757]'
l-wx------ 1 root root 64 Jan  5 21:42 2 -> 'pipe:[5645758]'
root@2d7b55c0ae13:/proc/3070/cwd#

dankamongmen · 2022-01-05T22:22:29Z

so different pipes for stdout and stderr, /dev/null for stdin

dankamongmen · 2022-01-05T22:23:17Z

environment:

root@2d7b55c0ae13:/proc/3070/cwd# cat /proc/1/environ 
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/binHOSTNAME=2d7b55c0ae13DRONE_SYSTEM_VERSION=2.7.0DRONE_BUILD_TRIGGER=@hookDRONE_SCRIPT=

if [ ! -z "${DRONE_NETRC_FILE}" ]; then
	echo $DRONE_NETRC_FILE > $HOME/.netrc
	chmod 600 $HOME/.netrc
fi

unset DRONE_SCRIPT
unset DRONE_NETRC_MACHINE
unset DRONE_NETRC_USERNAME
unset DRONE_NETRC_PASSWORD
unset DRONE_NETRC_FILE

set -e

remote_debug() {
	if [ "$?" -ne "0" ]; then
		/usr/drone/bin/tmate -F
	fi
}

if [ "${DRONE_BUILD_DEBUG}" = "true" ]; then
	if [ ! -z "${DRONE_TMATE_HOST}" ]; then
		echo "set -g tmate-server-host $DRONE_TMATE_HOST" >> $HOME/.tmate.conf
		echo "set -g tmate-server-port $DRONE_TMATE_PORT" >> $HOME/.tmate.conf
		echo "set -g tmate-server-rsa-fingerprint $DRONE_TMATE_FINGERPRINT_RSA" >> $HOME/.tmate.conf
		echo "set -g tmate-server-ed25519-fingerprint $DRONE_TMATE_FINGERPRINT_ED25519" >> $HOME/.tmate.conf

		if [ ! -z "${DRONE_TMATE_AUTHORIZED_KEYS}" ]; then
			echo "$DRONE_TMATE_AUTHORIZED_KEYS" > $HOME/.tmate.authorized_keys
			echo "set -g tmate-authorized-keys \"$HOME/.tmate.authorized_keys\"" >> $HOME/.tmate.conf
		fi
	fi
	trap remote_debug EXIT
fi


echo + "export LANG=en_US.UTF-8"
export LANG=en_US.UTF-8

echo + "export TERM=xterm"
export TERM=xterm

echo + "mkdir build"
mkdir build

echo + "cd build"
cd build

echo + "cmake .. -DCMAKE_BUILD_TYPE=Release -DUSE_STATIC=off -DUSE_GPM=on -DUSE_QRCODEGEN=on -DDFSG_BUILD=on"
cmake .. -DCMAKE_BUILD_TYPE=Release -DUSE_STATIC=off -DUSE_GPM=on -DUSE_QRCODEGEN=on -DDFSG_BUILD=on

echo + "make -j2"
make -j2

echo + "./notcurses-input < /dev/null"
./notcurses-input < /dev/null

echo + "env NOTCURSES_LOGLEVEL=7 ./notcurses-input < notcurses-input"
env NOTCURSES_LOGLEVEL=7 ./notcurses-input < notcurses-input

echo + "./notcurses-input < notcurses-input"
./notcurses-input < notcurses-input
DRONE_SOURCE_BRANCH=dankamongmen/fucking-inputDRONE_COMMIT_AUTHOR_AVATAR=https://avatars.githubusercontent.com/u/143473?v=4DRONE_STAGE_OS=linuxCI_COMMIT_REF=refs/heads/dankamongmen/fucking-inputCI_PARENT_BUILD_NUMBER=0DRONE_REMOTE_URL=https://github.com/dankamongmen/notcurses.gitDRONE_BUILD_NUMBER=10248DRONE_REPO_NAMESPACE=dankamongmenCI_REPO_LINK=https://github.com/dankamongmen/notcursesDRONE_REPO_PRIVATE=falseCI_COMMIT_SHA=53d95ac10aa983bb379c7f55e493bd18ce8ab5fcCI_COMMIT_AUTHOR_NAME=nick blackDRONE_GIT_HTTP_URL=https://github.com/dankamongmen/notcurses.gitCI_REPO_NAME=dankamongmen/[email protected]_COMMIT_MESSAGE=[drone] try and get some useful logsDRONE_BRANCH=dankamongmen/fucking-inputCI_COMMIT_AUTHOR_AVATAR=https://avatars.githubusercontent.com/u/143473?v=4DRONE_REPO_LINK=https://github.com/dankamongmen/notcursesDRONE_BUILD_STATUS=successDRONE_STAGE_TYPE=dockerGIT_AUTHOR_NAME=nick blackDRONE_REPO_OWNER=dankamongmenCI=trueDRONE_BUILD_CREATED=1641418920DRONE_COMMIT_MESSAGE=[drone] try and get some useful logsDRONE_COMMIT=53d95ac10aa983bb379c7f55e493bd18ce8ab5fcDRONE_WORKSPACE_BASE=/drone/srcDRONE_COMMIT_AUTHOR_NAME=nick blackDRONE_COMMIT_REF=refs/heads/dankamongmen/fucking-inputDRONE_REPO_BRANCH=masterDRONE_DOCKER_VOLUME_ID=drone-umi3O9KRd3GBRptjD3QiDRONE_BUILD_PARENT=0DRONE_STAGE_NAME=debian-unstableDRONE_COMMIT_AFTER=53d95ac10aa983bb379c7f55e493bd18ce8ab5fcDRONE_BUILD_EVENT=pushCI_COMMIT_AUTHOR=dankamongmenDRONE_STAGE_STARTED=1641418921CI_BUILD_CREATED=1641418920DRONE_BUILD_LINK=https://drone.dsscaw.com:4443/dankamongmen/notcurses/10248DRONE_WORKSPACE=/drone/srcCI_BUILD_FINISHED=0CI_REPO_REMOTE=https://github.com/dankamongmen/[email protected]_BUILD_LINK=https://github.com/dankamongmen/notcurses/compare/25c54f875331...53d95ac10aa9DRONE_SYSTEM_PROTO=httpsDRONE_BUILD_FINISHED=1641418940DRONE_BUILD_DEBUG=falseDRONE_COMMIT_BEFORE=25c54f875331ac09a1b3c57239ec810c8bb2019cDRONE_DOCKER_NETWORK_ID=drone-hulmydnAlRkDgQ6DDagKCI_BUILD_NUMBER=10248DRONE_STAGE_FINISHED=1641418940DRONE_GIT_SSH_URL=git@github.com:dankamongmen/notcurses.gitDRONE_STAGE_STATUS=successDRONE_STAGE_ARCH=amd64CI_REPO_PRIVATE=falseDRONE_TARGET_BRANCH=dankamongmen/fucking-inputDRONE_STEP_NAME=debian-buildGIT_COMMITTER_NAME=nick blackCI_BUILD_EVENT=pushDRONE_STAGE_KIND=pipelineDRONE_SYSTEM_HOSTNAME=drone.dsscaw.com:4443DRONE_COMMIT_SHA=53d95ac10aa983bb379c7f55e493bd18ce8ab5fcGIT_COMMITTER_EMAIL=dankamongmen@gmail.comDRONE_STAGE_VERSION=3GIT_TERMINAL_PROMPT=0DRONE_STAGE_MACHINE=vps1DRONE_STAGE_NUMBER=1DRONE_CARD_PATH=/dev/stdoutDRONE=trueDRONE_STEP_NUMBER=2DRONE_REPO_NAME=notcursesDRONE_DEPLOY_ID=0DRONE_BUILD_STARTED=1641418940CI_COMMIT_BRANCH=dankamongmen/fucking-inputDRONE_COMMIT_BRANCH=dankamongmen/fucking-inputDRONE_COMMIT_LINK=https://github.com/dankamongmen/notcurses/compare/25c54f875331...53d95ac10aa9DRONE_REPO_VISIBILITY=publicCI_REMOTE_URL=https://github.com/dankamongmen/notcurses.gitDRONE_COMMIT_AUTHOR=dankamongmenDRONE_SYSTEM_HOST=drone.dsscaw.com:4443CI_BUILD_STATUS=pendingCI_REPO=dankamongmen/notcursesCI_BUILD_STARTED=0DRONE_COMMIT_AUTHOR_EMAIL=dankamongmen@gmail.comDRONE_REPO=dankamongmen/notcursesHOME=/rootroot@2d7b55c0ae13:/proc/3070/cwd#

dankamongmen · 2022-01-05T22:25:53Z

environment of actual notcurses-input:

./notcurses-inputroot@2d7b55c0ae13:/proc/3070/cwd# cat /proc/3070/environ 
DRONE_COMMIT_AUTHOR_AVATAR=https://avatars.githubusercontent.com/u/143473?v=4DRONE_BRANCH=dankamongmen/fucking-inputGIT_COMMITTER_NAME=nick blackDRONE_SYSTEM_HOST=drone.dsscaw.com:[email protected]:dankamongmen/[email protected]_BUILD_DEBUG=falseDRONE_STAGE_VERSION=3CI=trueHOSTNAME=2d7b55c0ae13DRONE_STAGE_OS=linuxDRONE_REPO_NAMESPACE=dankamongmenDRONE_REPO_LINK=https://github.com/dankamongmen/notcursesCI_BUILD_NUMBER=10248DRONE_TARGET_BRANCH=dankamongmen/fucking-inputDRONE_COMMIT_AUTHOR=dankamongmenCI_BUILD_STARTED=0DRONE_GIT_HTTP_URL=https://github.com/dankamongmen/notcurses.gitCI_BUILD_LINK=https://github.com/dankamongmen/notcurses/compare/25c54f875331...53d95ac10aa9DRONE_COMMIT_BRANCH=dankamongmen/fucking-inputHOME=/rootOLDPWD=/drone/srcDRONE_REPO_PRIVATE=falseDRONE_STAGE_TYPE=dockerDRONE_SYSTEM_PROTO=httpsGIT_AUTHOR_NAME=nick blackDRONE_STEP_NUMBER=2DRONE_BUILD_STATUS=successDRONE_REPO_VISIBILITY=publicCI_PARENT_BUILD_NUMBER=0DRONE_WORKSPACE_BASE=/drone/srcDRONE_COMMIT_BEFORE=25c54f875331ac09a1b3c57239ec810c8bb2019cDRONE_STAGE_ARCH=amd64CI_BUILD_EVENT=pushCI_COMMIT_SHA=53d95ac10aa983bb379c7f55e493bd18ce8ab5fcCI_REPO_NAME=dankamongmen/notcursesCI_BUILD_CREATED=1641418920CI_COMMIT_AUTHOR_EMAIL=dankamongmen@gmail.comCI_COMMIT_REF=refs/heads/dankamongmen/fucking-inputDRONE_SOURCE_BRANCH=dankamongmen/fucking-inputDRONE_STAGE_NAME=debian-unstableDRONE_STAGE_STATUS=successDRONE_CARD_PATH=/dev/stdoutDRONE_COMMIT_MESSAGE=[drone] try and get some useful logsDRONE_REPO_BRANCH=masterCI_BUILD_FINISHED=0DRONE_SYSTEM_HOSTNAME=drone.dsscaw.com:4443DRONE_REMOTE_URL=https://github.com/dankamongmen/notcurses.gitCI_REPO=dankamongmen/notcursesTERM=xtermDRONE_BUILD_TRIGGER=@hookCI_COMMIT_AUTHOR_NAME=nick blackDRONE_REPO_OWNER=dankamongmenDRONE_STAGE_KIND=pipelinePATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/binCI_COMMIT_AUTHOR_AVATAR=https://avatars.githubusercontent.com/u/143473?v=4DRONE=trueDRONE_BUILD_NUMBER=10248DRONE_BUILD_STARTED=1641418940DRONE_BUILD_PARENT=0DRONE_BUILD_LINK=https://drone.dsscaw.com:4443/dankamongmen/notcurses/10248DRONE_WORKSPACE=/drone/srcDRONE_COMMIT_AFTER=53d95ac10aa983bb379c7f55e493bd18ce8ab5fcLANG=en_US.UTF-8CI_REPO_LINK=https://github.com/dankamongmen/notcursesCI_COMMIT_AUTHOR=dankamongmenDRONE_STAGE_STARTED=1641418921DRONE_STAGE_NUMBER=1DRONE_COMMIT=53d95ac10aa983bb379c7f55e493bd18ce8ab5fcCI_REPO_REMOTE=https://github.com/dankamongmen/notcurses.gitGIT_TERMINAL_PROMPT=0CI_COMMIT_BRANCH=dankamongmen/fucking-inputCI_REPO_PRIVATE=falseDRONE_BUILD_EVENT=pushDRONE_BUILD_CREATED=1641418920DRONE_COMMIT_SHA=53d95ac10aa983bb379c7f55e493bd18ce8ab5fcDRONE_REPO_NAME=notcursesDRONE_COMMIT_AUTHOR_EMAIL=dankamongmen@gmail.comDRONE_COMMIT_REF=refs/heads/dankamongmen/fucking-inputDRONE_STAGE_MACHINE=vps1CI_BUILD_STATUS=pendingGIT_COMMITTER_EMAIL=dankamongmen@gmail.comDRONE_DOCKER_VOLUME_ID=drone-umi3O9KRd3GBRptjD3QiDRONE_DEPLOY_ID=0PWD=/drone/src/buildDRONE_BUILD_FINISHED=1641418940DRONE_COMMIT_LINK=https://github.com/dankamongmen/notcurses/compare/25c54f875331...53d95ac10aa9CI_COMMIT_MESSAGE=[drone] try and get some useful logsDRONE_STEP_NAME=debian-buildDRONE_REPO=dankamongmen/notcursesDRONE_SYSTEM_VERSION=2.7.0DRONE_COMMIT_AUTHOR_NAME=nick blackDRONE_DOCKER_NETWORK_ID=drone-hulmydnAlRkDgQ6DDagKDRONE_STAGE_FINISHED=1641418940CI_REMOTE_URL=https://github.com/dankamongmen/notcurses.gitroot@2d7b55c0ae13:/proc/3070/cwd#

and now we just have more pipes:

root@2d7b55c0ae13:/proc/3070/cwd# ls -l /proc/3070/fd
total 0
lr-x------ 1 root root 64 Jan  5 22:25 0 -> /dev/null
l-wx------ 1 root root 64 Jan  5 22:25 1 -> 'pipe:[5645757]'
l-wx------ 1 root root 64 Jan  5 22:25 2 -> 'pipe:[5645758]'
lr-x------ 1 root root 64 Jan  5 22:25 3 -> 'pipe:[5655568]'
l-wx------ 1 root root 64 Jan  5 22:25 4 -> 'pipe:[5655568]'
lr-x------ 1 root root 64 Jan  5 22:25 5 -> 'pipe:[5655569]'
l-wx------ 1 root root 64 Jan  5 22:25 6 -> 'pipe:[5655569]'
root@2d7b55c0ae13:/proc/3070/cwd#

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

dankamongmen · 2022-01-06T01:21:24Z

+ ./notcurses-input < /dev/null
530	(Bnotcurses 3.0.3 on xterm (Linux 5.11.13-x86_64-linode143)
531	20 rows 76 cols 8 colors
532	gcc-11.2.0 (LE)
533	terminfo 6.2.20210905 libdeflate 1.8 GPM 2.1.0
534	avformat 58.76.100 avutil 56.70.100 swscale 5.9.100 avcodec 58.134.100
535	mouse_setup:21:no tty, not emitting mouse control
536	
537	IVALID: 0 EOF: 0 IREAD: 0
538	WHILE IVALID: 0 EOF: 0 IREAD: 0
539	WAITING

so we're sitting on the icond condition variable. hrmm.

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

dankamongmen added bug Something isn't working input readin' dem bytes labels Jan 4, 2022

dankamongmen added this to the 3.1.0 milestone Jan 4, 2022

dankamongmen self-assigned this Jan 4, 2022

dankamongmen added a commit that referenced this issue Jan 4, 2022

[input] don't queue further terminal input following stdineof #2525

15122f3

dankamongmen added a commit that referenced this issue Jan 5, 2022

Post to the readyness fd for EOF

b459af7

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

dankamongmen mentioned this issue Jan 5, 2022

Handle EOF on stdin better #2528

Merged

dankamongmen added a commit that referenced this issue Jan 5, 2022

[input] don't queue further terminal input following stdineof #2525

a64250c

dankamongmen added a commit that referenced this issue Jan 5, 2022

Post to the readyness fd for EOF

46a062f

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

dankamongmen added a commit that referenced this issue Jan 5, 2022

[input] don't queue further terminal input following stdineof #2525

ef4eba0

dankamongmen added a commit that referenced this issue Jan 5, 2022

Post to the readyness fd for EOF

80e4acc

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

dankamongmen added a commit that referenced this issue Jan 6, 2022

[input] don't queue further terminal input following stdineof #2525

a6d98ba

dankamongmen added a commit that referenced this issue Jan 6, 2022

Post to the readyness fd for EOF

bcc2352

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

dankamongmen added a commit that referenced this issue Jan 8, 2022

[input] don't queue further terminal input following stdineof #2525

7c9e08d

dankamongmen added a commit that referenced this issue Jan 8, 2022

Post to the readyness fd for EOF

9618f8c

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

dankamongmen closed this as completed in #2528 Jan 8, 2022

dankamongmen added a commit that referenced this issue Jan 8, 2022

[input] don't queue further terminal input following stdineof #2525

28cd88f

dankamongmen added a commit that referenced this issue Jan 8, 2022

Post to the readyness fd for EOF

4bccb59

When we get to the EOF on stdin, we need account for that with the readiness fd. Post to it upon the EOF transition. Closes #2525.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

inputready fd not set high following NCKEY_EOF #2525

inputready fd not set high following NCKEY_EOF #2525

dankamongmen commented Jan 4, 2022

dankamongmen commented Jan 4, 2022

dankamongmen commented Jan 4, 2022 •

edited

Loading

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 6, 2022

inputready fd not set high following NCKEY_EOF #2525

inputready fd not set high following NCKEY_EOF #2525

Comments

dankamongmen commented Jan 4, 2022

dankamongmen commented Jan 4, 2022

dankamongmen commented Jan 4, 2022 • edited Loading

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 5, 2022

dankamongmen commented Jan 6, 2022

dankamongmen commented Jan 4, 2022 •

edited

Loading