Menu

#319 gscan2pdf crashes with error message: perl: unable to extent pixel cache `No such file or directory' @ fatal/cache.c/CacheSignalHandler/3626

v1.0_(example)
open
nobody
None
5
2019-04-10
2019-04-07
No

Hi Jeffrey,

Another day, another bug.

I'm seeing crashes with both the current version in Debian
experimental (2.5.1-1), and the current version in Debain unstable
(2.3.0-1). The report before is for 2.5.1-1. The error messages in
both cases look identical.

The bit immediately before the crash looks like:

INFO - Scanning 0 pages from 1 with step 1
DEBUG - signal 'started-process' emitted with message: Scanning page 2
DEBUG - signal 'finished-process' emitted with data: scan_pages
perl: unable to extent pixel cache `No such file or directory' @ fatal/cache.c/CacheSignalHandler/3626.
perl: unable to extent pixel cache `Success' @ fatal/cache.c/CacheSignalHandler/3626.

I did the usual

gscan2pdf --log=/tmp/gscan2pdf.log

The log ends with

DEBUG - signal 'finished-process' emitted with data: scan_pages

So the last two lines starting with perl: do not appear in the log.

I was trying to scan at 600 dpi on the gray setting with my Brother
DCP-7065DN flatbed. I'm pretty sure I've never tried to do that
before, but of course I can't be sure this setting is related to the
crash. But I don't recall seeing any crashes before, and I was seeing
crashing more of less immediately after I tried using that setting.

Log attached.

1 Attachments

Related

Bugs: #319

Discussion

  • Faheem Mitha

    Faheem Mitha - 2019-04-07

    Also note that searching for

    fatal/cache.c/CacheSignalHandler/3626
    

    gives hits for Imagemagick, which does not seem like a coincidence., though I was not aware
    that gscan2pdf depended on Imagemagick.

     
  • Jeffrey Ratcliffe

    gscan2pdf uses imagemagick for all the image operations - conversions and the like.

    You are right, this looks like an imagemagick memory problem.

    Therefore, I assume you can workaround the problem by scanning a smaller area, or using a lower resolution.

    As well as helping you get imagemagick working, I'd like to have gscan2pdf not crash, and produce a reasonable error message.

    What is in your /etc/ImageMagick-6/policy.xml ?

    By default, the first thing that happens after scanning an image is that it is converted to PNG to reduce the space used and add some metadata. What happens if you switch off the "convert scanned images to PNG before further processing" in Edit/Preferences/General options ?

     
    • Faheem Mitha

      Faheem Mitha - 2019-04-08

      Hi Jeffrey,

      Here's hoping that replying via email works.

      On Mon, 8 Apr 2019, Jeffrey Ratcliffe wrote:

      gscan2pdf uses imagemagick for all the image operations - conversions and the like.

      I eee. I didn't know that.

      You are right, this looks like an imagemagick memory problem.

      Ok.

      Therefore, I assume you can workaround the problem by scanning a smaller
      area, or using a lower resolution.

      As well as helping you get imagemagick working, I'd like to have
      gscan2pdf not crash, and produce a reasonable error message.

      What is in your /etc/ImageMagick-6/policy.xml ?

      It's the default for Debian stable. I haven't changed anything. Attached -
      here's hoping email attachments convert to ticket attachments.

      By default, the first thing that happens after scanning an image is that
      it is converted to PNG to reduce the space used and add some metadata.
      What happens if you switch off the "convert scanned images to PNG before
      further processing" in Edit/Preferences/General options ?

      It works ok with that option off. If I switch it on again, it promptly
      crashes. Is that helpful? I'm attaching the log file first with that
      option off (no crash), and then with that option on (crash).

      If these attachments don't find their way into the bug report, I'll add
      them manually.

      BTW, on a tangentially related note, someone recently told me that
      SourceForge was "an unusable pile of garbage". Since you're still using
      it, just wonder what you think.


      [bugs:#319] gscan2pdf crashes with error message: perl: unable to extent pixel cache `No such file or directory' @ fatal/cache.c/CacheSignalHandler/3626

      Status: open
      Group: v1.0_(example)
      Created: Sun Apr 07, 2019 07:47 PM UTC by Faheem Mitha
      Last Updated: Sun Apr 07, 2019 07:53 PM UTC
      Owner: nobody
      Attachments:

      • gscan2pdf.log (182.9 kB; text/x-log)

      Hi Jeffrey,

      Another day, another bug.

      I'm seeing crashes with both the current version in Debian
      experimental (2.5.1-1), and the current version in Debain unstable
      (2.3.0-1). The report before is for 2.5.1-1. The error messages in
      both cases look identical.

      The bit immediately before the crash looks like:

      INFO - Scanning 0 pages from 1 with step 1
      DEBUG - signal 'started-process' emitted with message: Scanning page 2
      DEBUG - signal 'finished-process' emitted with data: scan_pages
      perl: unable to extent pixel cache No such file or directory' @ fatal/cache.c/CacheSignalHandler/3626. perl: unable to extent pixel cacheSuccess' @ fatal/cache.c/CacheSignalHandler/3626.

      I did the usual

      gscan2pdf --log=/tmp/gscan2pdf.log

      The log ends with

      DEBUG - signal 'finished-process' emitted with data: scan_pages

      So the last two lines starting with perl: do not appear in the log.

      I was trying to scan at 600 dpi on the gray setting with my Brother
      DCP-7065DN flatbed. I'm pretty sure I've never tried to do that
      before, but of course I can't be sure this setting is related to the
      crash. But I don't recall seeing any crashes before, and I was seeing
      crashing more of less immediately after I tried using that setting.

      Log attached.


      Sent from sourceforge.net because you indicated interest in https://sourceforge.net/p/gscan2pdf/bugs/319/

      To unsubscribe from further messages, please visit https://sourceforge.net/auth/subscriptions/

       

      Related

      Bugs: #319

  • Jeffrey Ratcliffe

    Here, it looks as though imagemagick is running out of tmp:

    https://www.imagemagick.org/discourse-server/viewtopic.php?t=25064

    Having scanned the image without the to-png option, please copy it out of the /tmp/gscan2pdf-xxxx directory to somewhere it won't get deleted, and see if you can reproduce the problem from the command file:

    convert test.pnm test.png
    

    If that produces the same error,

    df
    

    Might tell where you are short of space and

    convert -debug cache test.pnm test.png
    

    might give more info.

    "Unusable" is a little harsh and, at least for me, inaccurate. I don't see the advantage (to me) of moving the project to (say) github.

     
    • Faheem Mitha

      Faheem Mitha - 2019-04-10

      Hi Jeffrey,

      On Wed, 10 Apr 2019, Jeffrey Ratcliffe wrote:

      Here, it looks as though imagemagick is running out of tmp:

      https://www.imagemagick.org/discourse-server/viewtopic.php?t=25064

      Having scanned the image without the to-png option, please copy it out
      of the /tmp/gscan2pdf-xxxx directory to somewhere it won't get deleted,
      and see if you can reproduce the problem from the command file:

      convert test.pnm test.png

      If that produces the same error,

      df

      Might tell where you are short of space and

      convert -debug cache test.pnm test.png

      might give more info.

      Yes, it's probably running out of space in /tmp.

      As it happens, I was trying to backport the Gimp 2.10.8 Debian source from
      unstable to stable yesterday. The tests run during build were producing
      mysterious errors, but that's another story. Anyway, the build crapped out
      towards the end, because it ran out of space on /, which also contains
      /tmp. So I removed a bunch of old and obsolete packages (Debian lets old
      and obsolete packages in many cases hang around indefinitely, gcc,
      clang/LLVM, for example). So now I have around 3.1 GB free, and it
      probably won't show the same problem. But I'll try your instructions in a
      bit. Of course, I could temporarily reduce the amount of space available
      on / and see if I can reproduce.

      In any case, the message produced is really unclear.

      Have you tested with an inadequate amount of /tmp space to see what will
      happen?

      "Unusable" is a little harsh and, at least for me, inaccurate. I don't
      see the advantage (to me) of moving the project to (say) github.

      Yes, GH is the new hotness. Everyone and their cousin is on it.

       

Log in to post a comment.

MongoDB Logo MongoDB