bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[Eli Zaretskii]

> Date: Mon, 19 Apr 2021 13:35:25 -0500
> From:  David Greene via "Bug reports for GNU Emacs,
>  the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>
> 
> I've attached TRAMP verbose 6 output.

Thanks, but please in the future compress such huge files before you
send them.

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[Michael Albinus]

David Greene <dag@obbligato.org> writes:

Hi David,

> So I tried emacs 28. No lockups at all. Yay! But every single compile
> failes with exit code 9. This isn't an error from the compilation
> itself. I can run the compilation under emacs 27 and it completes (that
> is, when it doesn't cause a lockup) and I can run it in a terminal and
> it completes. I suspect that whatever was causing lockups in emacs 27 is
> now causing signal 9 in 28. Perhaps this has something to do with the
> new process handling in 28?
>
> I have tried disabling as much of my config as I can but nothing seems
> to help.
>
> I've attached TRAMP verbose 6 output.

Your prompt on the remote host looks strange:

--8<---------------cut here---------------start------------->8---
> ]51;Adgreene@dgreene-dev:/cb/home/dgreene\
--8<---------------cut here---------------end--------------->8---

Please consult the Tramp manual how to fix this.

> David

Best regards, Michael.

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[dag via Bug reports for GNU Emacs, the Swiss army knife of text editors]

[-- Attachment #1: Type: text/plain, Size: 604 bytes --]

On 2021-04-20 13:16, Michael Albinus <michael.albinus@gmx.de> wrote: > > Your prompt on the remote host looks strange: > > --8<---------------cut here---------------start------------->8--- > > ]51;Adgreene@dgreene-dev:/cb/home/dgreene\ > --8<---------------cut here---------------end--------------->8--- > > Please consult the Tramp manual how to fix this. Thanks Michael. I'll take a look at it. Just to clarify, what is "strange" about it? The control sequences? I've set it up to work with emacs-libvterm. Also, I don't see TRAMP ever hanging on "Waiting for prompt..." it is always after that. David

[-- Attachment #2: Type: text/html, Size: 939 bytes --]

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[Michael Albinus]

dag via "Bug reports for GNU Emacs, the Swiss army knife of text
editors" <bug-gnu-emacs@gnu.org> writes:

> On 2021-04-20 13:16, Michael Albinus <michael.albinus@gmx.de> wrote:
>
>     Your prompt on the remote host looks strange:
>
> --8<---------------cut here---------------start------------->8---
>> ]51;Adgreene@dgreene-dev:/cb/home/dgreene\
> --8<---------------cut here---------------end--------------->8---
>
> Please consult the Tramp manual how to fix this.
>
> Thanks Michael.  I'll take a look at it.  Just to clarify, what is
> "strange" about it?  The control sequences?  I've set it up to work
> with emacs-libvterm.

Yes, the escape sequences. You can check the environment variable
INSIDE_EMACS on the remote host. If Tramp is involved, it tells you there.

> Also, I don't see TRAMP ever hanging on "Waiting for prompt..." it is
> always after that.

Same here, I haven't seen this in the Tramp. But the command ping/pong
between Tramp and the remote host shows some strange entries ...

> David

Best regards, Michael.

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[dag via Bug reports for GNU Emacs, the Swiss army knife of text editors]

[-- Attachment #1: Type: text/plain, Size: 755 bytes --]

On 2021-04-21 09:22, Michael Albinus <michael.albinus@gmx.de> wrote: > Yes, the escape sequences. You can check the environment variable > INSIDE_EMACS on the remote host. If Tramp is involved, it tells you there. Hmm. ssh can't always pass environment variables because the remote admin may have disabled it, so INSIDE_EMACS isn't always reliable. Is testing for TERM == "dumb" sufficient? I will have to see whether my current provided allows me to pass INSIDE_EMACS. > > Also, I don't see TRAMP ever hanging on "Waiting for prompt..." it is > > always after that. > Same here, I haven't seen this in the Tramp. But the command ping/pong > between Tramp and the remote host shows some strange entries ... Ok, definitely worth at least ruling out. David

[-- Attachment #2: Type: text/html, Size: 1051 bytes --]

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[Michael Albinus]

dag via "Bug reports for GNU Emacs, the Swiss army knife of text
editors" <bug-gnu-emacs@gnu.org> writes:

>> Yes, the escape sequences. You can check the environment variable
>> INSIDE_EMACS on the remote host. If Tramp is involved, it tells you
>> there.
>
> Hmm.  ssh can't always pass environment variables because the remote
> admin may have disabled it, so INSIDE_EMACS isn't always reliable.  Is
> testing for TERM == "dumb" sufficient?

Yes. But I don't know, whether emacs-libvterm does the same setting.

> David

Best regards, Michael.

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[dag via Bug reports for GNU Emacs, the Swiss army knife of text editors]

[-- Attachment #1: Type: text/plain, Size: 3094 bytes --]

On 2021-04-22 02:32, Michael Albinus <michael.albinus@gmx.de> wrote: > > Hmm. ssh can't always pass environment variables because the remote > > admin may have disabled it, so INSIDE_EMACS isn't always reliable. Is > > testing for TERM == "dumb" sufficient? > Yes. But I don't know, whether emacs-libvterm does the same setting. It seems to. I set the prompt to something really simple: "$ " (debug log snippet below). However, that didn't seem to help with the lockups. What *did* seem to help quite a bit was doing this in my init: ; Have compile mode scroll until the first error. ;;(setq compilation-scroll-output 'first-error) ;;(setq compilation-auto-jump-to-first-error 1) I've been meaning to disable those anyway, but it's a shame these don't seem to work well with TRAMP. Now lockups that need "kill -SIGUSR2" are pretty rare (though they do happen occasionally). I do still need the to <quit> every now and then, which is annoying but not nearly as annoying as before. I do still get hangs on "decoding remote file..." and if I <quit>, visit the resulting buffer (which may or may not be empty) and <find-alternate-file> it will reload without issue. Again, annoying but not a showstopper. I haven't yet tried emacs 28 to see if the above fixes the signal-9-terminated compiles. David 11:01:33.141006 tramp-process-actions (3) # Waiting for prompts from remote shell...done 11:01:33.141091 tramp-maybe-open-connection (3) # Found remote shell prompt on ‘dgreene-dev’ 11:01:33.141157 tramp-open-shell (5) # Opening remote shell ‘/bin/sh’... 11:01:33.141223 tramp-send-command (6) # exec env TERM='dumb' INSIDE_EMACS='27.2,tramp:2.4.5.27.2' ENV='' HISTFILE=~/.tramp_history PROMPT_COMMAND='' PS1=\#\$\ PS2='' PS3='' /bin/sh 11:01:33.202768 tramp-wait-for-regexp (6) # #$ 11:01:33.202879 tramp-send-command (6) # (cd ~/) 2>/dev/null; echo tramp_exit_status $? 11:01:33.258755 tramp-wait-for-regexp (6) # tramp_exit_status 0 #$ 11:01:33.258876 tramp-open-shell (5) # Opening remote shell ‘/bin/sh’...done 11:01:33.258970 tramp-open-connection-setup-interactive-shell (5) # Setting up remote shell environment 11:01:33.259039 tramp-send-command (6) # set +o vi +o emacs 11:01:33.313578 tramp-wait-for-regexp (6) # #$ 11:01:33.313763 tramp-send-command (6) # stty -inlcr -onlcr -echo kill '^U' erase '^H' 11:01:33.368599 tramp-wait-for-regexp (6) # #$ 11:01:33.368721 tramp-send-command (6) # echo foo 11:01:33.423097 tramp-wait-for-regexp (6) # foo #$ 11:01:33.423252 tramp-open-connection-setup-interactive-shell (5) # Setting shell prompt 11:01:33.423416 tramp-send-command (6) # PS1=///e9364282a97cbba0d7fbd76f9c2169f1\#\$ PS2='' PS3='' PROMPT_COMMAND='' 11:01:33.475982 tramp-wait-for-regexp (6) # ///e9364282a97cbba0d7fbd76f9c2169f1#$ 11:01:33.476293 tramp-open-connection-setup-interactive-shell (5) # Checking system information 11:01:33.476541 tramp-send-command (6) # echo \"`uname -sr`\" 2>/dev/null; echo tramp_exit_status $? 11:01:33.530843 tramp-wait-for-regexp (6) # "Linux 3.10.0-1127.19.1.el7.x86_64" tramp_exit_status 0 ///e9364282a97cbba0d7fbd76f9c2169f1#$

[-- Attachment #2: Type: text/html, Size: 4146 bytes --]

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[dag via Bug reports for GNU Emacs, the Swiss army knife of text editors]

[-- Attachment #1: Type: text/plain, Size: 1696 bytes --]

On 2021-04-22 11:10, dag <dag@obbligato.org> wrote: > > > > > I've been meaning to disable those anyway, but it's a shame these don't seem to work well > > with TRAMP. Now lockups that need "kill -SIGUSR2" are pretty rare (though they do > > happen occasionally). > > And...I just had a -SIGUSR2 lockup. :). Here's a backtrace: > > > > > > Debugger entered--entering a function: > > > > * tramp-signal-hook-function(quit nil) > > > > accept-process-output(#<process compilation> 0 nil t) > > > > tramp-accept-process-output(#<process compilation> 0) > > > > tramp-interrupt-process(#<process compilation> nil) > > > > interrupt-process(#<process compilation>) > > > > compilation-start("git be bic all-devrel-local" compilation-mode nil nil) > > > > apply(compilation-start ("git be bic all-devrel-lo
 cal" compilation-mode nil nil)) > > > > recompile(nil) > > > > funcall-interactively(recompile nil) > > > > call-interactively(recompile nil nil) > > > > command-execute(recompile) > > > > The debug

 log ends with this: 11:11:36.384026 tramp-send-command (6) # \readlink --canonicalize-missing /net/dgreene-dev/<redacted> 2>/dev/null; echo tramp_exit_status $? 11:15:42.087353 tramp-accept-process-output (1) # Quit: "Quit", "" 11:15:42.087652 tramp-accept-process-output (1) # Quit: "Quit", "" 11:15:42.087766 tramp-accept-process-output (1) # Quit: "Quit", "" 11:15:42.087879 tramp-accept-process-output (1) # Quit: "Quit", "" 11:15:42.088020 tramp-accept-process-output (1) # Quit: "Quit", "" 11:15:42.088130 tramp-accept-process-output (1) # Quit: "Quit", "" 11:15:42.088243 tramp-accept-process-output (1) # Quit: "Quit", "" Followed by hundreds of thousands of the "Quit" message. David

[-- Attachment #2: Type: text/html, Size: 2483 bytes --]

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[Michael Albinus]

dag via "Bug reports for GNU Emacs, the Swiss army knife of text
editors" <bug-gnu-emacs@gnu.org> writes:

Hi David,

> I haven't yet tried emacs 28 to see if the above fixes the
> signal-9-terminated compiles.

What you could try is Tramp 2.5.0.3 from GNU ELPA. There are some fixes
in the context of remote compilation; maybe it helps you.

> David

Best regards, Michael.

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[dag via Bug reports for GNU Emacs, the Swiss army knife of text editors]

[-- Attachment #1: Type: text/plain, Size: 1497 bytes --]

On 2021-04-22 14:18, Michael Albinus <michael.albinus@gmx.de> wrote: > > What you could try is Tramp 2.5.0.3 from GNU ELPA. There are some fixes > in the context of remote compilation; maybe it helps you. What is the correct way to do that with use-package/straight.el to override the built-in TRAMP? I have this: (use-package tramp :straight t :config ;;(setq tramp-verbose 6) (setq tramp-default-method "ssh") ;;(setq tramp-shell-prompt-pattern "\\(?:^\\|\r\\)[^]#$%>\n]*#?[]#$%>].* *\\(^[\\[[0-9;]*[a-zA-Z] *\\)*") (setq vc-ignore-dir-regexp (format "\\(%s\\)\\|\\(%s\\)" vc-ignore-dir-regexp tramp-file-name-regexp)) ;; Honor remote PATH. (add-to-list 'tramp-remote-path 'tramp-own-remote-path) ;; Allow ssh connections to persist. ;; ;; This seems to maybe cause tramp to hang a lot. (customize-
 set-variable 'tramp-use-ssh-controlmaster-options nil) ;; (customize-set-variable 'tramp-use-ssh-controlmaster-options t) (if (eq window-system 'w32) ;(setq tramp-default-method "ssh") (setq tramp-d

efault-method "scp") ;(setq tramp-default-method "scpx") (setq tramp-default-method "ssh")) ;;(setq tramp-default-method "ssh") ;;(tramp-change-syntax 'simplified) ;;(setq tramp-verbose 10) (autoload #'tramp-register-crypt-file-name-handler "tramp-crypt") ) The autoload I got from here to solve the tramp-crypt error: https://www.reddit.com/r/emacs/comments/l2lw7o/tramp_blocking_emacs/ Even so, tramp-version reports 2.4.5.27.2. I verified the package straight.el downloaded is 2.5.0.3. David

[-- Attachment #2: Type: text/html, Size: 2839 bytes --]

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[Michael Albinus]

dag via "Bug reports for GNU Emacs, the Swiss army knife of text
editors" <bug-gnu-emacs@gnu.org> writes:

Hi David,

> What is the correct way to do that with use-package/straight.el to
> override the
> built-in TRAMP?  I have this:

Sorry, I don't use use-package and straight. I would simply open the
package manager by "M-x list-packages", mark Tramp 2.5.0.3 (hit "i"),
and execute it (hit "x").

> The autoload I got from here to solve the tramp-crypt error:
> https://www.reddit.com/r/emacs/comments/l2lw7o/tramp_blocking_emacs/
>
> Even so, tramp-version reports 2.4.5.27.2.  I verified the package
> straight.el
> downloaded is 2.5.0.3.

Hmm, here I would need more details. Could you show what happens when
the Tramp package is compiled during installation?

> David

Best regards, Michael.

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[Michael Albinus]

Michael Albinus <michael.albinus@gmx.de> writes:

Hi David,

>> What is the correct way to do that with use-package/straight.el to
>> override the
>> built-in TRAMP?  I have this:
>
> Sorry, I don't use use-package and straight. I would simply open the
> package manager by "M-x list-packages", mark Tramp 2.5.0.3 (hit "i"),
> and execute it (hit "x").
>
>> The autoload I got from here to solve the tramp-crypt error:
>> https://www.reddit.com/r/emacs/comments/l2lw7o/tramp_blocking_emacs/
>>
>> Even so, tramp-version reports 2.4.5.27.2.  I verified the package
>> straight.el
>> downloaded is 2.5.0.3.
>
> Hmm, here I would need more details. Could you show what happens when
> the Tramp package is compiled during installation?

Anything else I could do for you? Otherwise, I'd like to close the bug.

Btw, There's meanwhile Tramp 2.5.0.5 on GNU ELPA.

>> David

Best regards, Michael.

bug#47896: 27.2; TRAMP hang/lockup/fault on compile/file access

[Michael Albinus]

Michael Albinus <michael.albinus@gmx.de> writes:

Hi David,

> Anything else I could do for you? Otherwise, I'd like to close the bug.

No further information, so I'm closing the bug.

Best regards, Michael.