DoS mail?

DoS mail?

[Colin Marquardt]

[-- Attachment #1: Type: text/plain, Size: 3210 bytes --]

Hi,

I just seem to have gotten one of those Gnus DoS mails.  Gnus sits
there with 100% CPU after "Generating Summary...done".  It's no
general hanging, just for that one group. It's my second-youngest
mail there, so "2 g" (and "3 g" etc.) hangs it while "1 g" works
fine. C-g works fine too, no mails though. This has never happened
to me before.

This was with Oort 0.01, and I remember ShengHuo fixing something
for mail like that (but that was for mail I think). I thus upgraded
to current CVS, but it's still the same.

I listed my messages, and the 2nd-youngest one was some spam from
Korea. I moved it away from ~/Mail, and copied another message over
it so as to create no gap. I moved the Incoming* file away which contained
that message. Still... no cookie.

This is with Emacs 21.1.1 on Debian unstable.

Here is the last part of the backtrace from when I interrupt it:

Debugger entered--Lisp error: (quit)
  backquote-listify(((0 quote (quote dummy)) (0 quote gnus-summary-make-false-root) (0 quote eq)) (0))
  backquote-process((eq gnus-summary-make-false-root (quote dummy)))
  backquote-process((and (eq gnus-summary-make-false-root (quote dummy)) (get-text-property (point) (quote gnus-intangible))))
  backquote-process((if (and (eq gnus-summary-make-false-root ...) (get-text-property ... ...)) 0 (gnus-data-level (gnus-data-find ...))))
  #[(arg) "Á\b!A‡" [arg backquote-process] 2 663713]((if (and (eq gnus-summary-make-false-root ...) (get-text-property ... ...)) 0 (gnus-data-level (gnus-data-find ...))))
  (\` (if (and ... ...) 0 (gnus-data-level ...)))
  (lambda (&optional number) "Return the level of thread that starts with article NUMBER." (\` (if ... 0 ...)))()
  (gnus-summary-thread-level)
  (< level (gnus-summary-thread-level))
  (and (not (eobp)) (< level (gnus-summary-thread-level)))
  (while (and (not ...) (< level ...)) (forward-line way))
  (let ((level ...) (way ...) (beg ...)) (forward-line way) (while (and ... ...) (forward-line way)) (if (eobp) (progn ... nil) (setq beg ...) (prog1 ... ...)))
  gnus-summary-go-to-next-thread(nil)
  (and (> n 0) (gnus-summary-go-to-next-thread backward))
  (while (and (> n 0) (gnus-summary-go-to-next-thread backward)) (decf n))
  (let ((backward ...) (n ...)) (while (and ... ...) (decf n)) (unless silent (gnus-summary-position-point)) (when (and ... ...) (gnus-message 7 "No more threads")) n)
  gnus-summary-next-thread(1 t)
  (zerop (gnus-summary-next-thread 1 t))
  (not (zerop (gnus-summary-next-thread 1 t)))
  (setq end (not (zerop ...)))
  (while (not end) (when (or ... ...) (gnus-summary-hide-thread)) (setq end (not ...)))
  (let ((end nil)) (while (not end) (when ... ...) (setq end ...)))
  (save-excursion (goto-char (point-min)) (let (...) (while ... ... ...)))
  gnus-summary-hide-all-threads(nil)
  (progn (gnus-summary-hide-all-threads (if ... ... nil)))
  (if (and gnus-show-threads gnus-thread-hide-subtree) (progn (gnus-summary-hide-all-threads ...)))
  (when (and gnus-show-threads gnus-thread-hide-subtree) (gnus-summary-hide-all-threads (if ... ... nil)))
  gnus-summary-maybe-hide-threads()


I'm attaching the mail gzipped.

TIA,
  Colin


[-- Attachment #2: GnusDoS.txt.gz --]
[-- Type: application/octet-stream, Size: 2898 bytes --]

Re: DoS mail?

[Colin Marquardt]

Colin Marquardt <colin@marquardt-home.de> writes:

> This was with Oort 0.01, and I remember ShengHuo fixing something
> for mail like that (but that was for mail I think). I thus upgraded

Agh, this should read "but that was for displaying the mail body".

> I listed my messages, and the 2nd-youngest one was some spam from
> Korea. I moved it away from ~/Mail, and copied another message over
> it so as to create no gap. I moved the Incoming* file away which contained
> that message. Still... no cookie.

Seems that it just took very long. I finally saw the old subject
with the Korean letters in the Summary, and the body was the mail
that I copied over that spam. I'm sure that I had waited longer when
I tried to enter the Group initially (before I had overwritten that
message). Very strange. What could cause such a delay?

Cheers,
  Colin

-- 
YYURYYUBICURYY4ME.

Re: DoS mail?

[Lars Magne Ingebrigtsen]

Colin Marquardt <colin@marquardt-home.de> writes:

> Here is the last part of the backtrace from when I interrupt it:
>
> Debugger entered--Lisp error: (quit)
>   backquote-listify(((0 quote (quote dummy)) (0 quote gnus-summary-make-false-root) (0 quote eq)) (0))

This means that you're running Gnus uncompiled, which will make it
very slow indeed.

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: DoS mail?

[Colin Marquardt]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Colin Marquardt <colin@marquardt-home.de> writes:
>
>> Here is the last part of the backtrace from when I interrupt it:
>>
>> Debugger entered--Lisp error: (quit)
>>   backquote-listify(((0 quote (quote dummy)) (0 quote gnus-summary-make-false-root) (0 quote eq)) (0))
>
> This means that you're running Gnus uncompiled, which will make it
> very slow indeed.

Maybe, but I never saw something like that (maybe taking 10 mins to
display a summary), and I wasn't running Gnus byte compiled before
either. The old Gnus was a gnus-CURRENT-20010227.

Cheers,
  Colin

-- 
YYURYYUBICURYY4ME.

Re: DoS mail?

[Lars Magne Ingebrigtsen]

Colin Marquardt <colin@marquardt-home.de> writes:

> Maybe, but I never saw something like that (maybe taking 10 mins to
> display a summary), and I wasn't running Gnus byte compiled before
> either. The old Gnus was a gnus-CURRENT-20010227.

Does it take 10 minutes to display the summary if you run Gnus
byte-compiled? 

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen