On Sat, 27 Jul 2019 04:59:43 +0000 (UTC) DJ Lucas via blfs-support 
<[email protected]> wrote:

> On 7/26/2019 9:47 AM, Stephen Berman via blfs-support wrote:
>> In reviewing what your plugin instructions install, I noticed several
>> libraries were missing that had been installed when I had previously
>> installed the plugin using the hp-plugin script, and which are also
>> installed in my openSUSE Tumbleweed, namely
>> hpgt2500_ntdcmsdll-x86_64.so, libsane-hp2000S1-x86_64.so.1.0.25,
>> libsane-hpgt2500-x86_64.so.1.0.27 and libjpeg-x86_64.so.9.2.0.
>
> Running hp-setup should download any needed binary blobs.

That's also what it says here:
https://developers.hp.com/hp-linux-imaging-and-printing/howtos/install#howtocheck11
but it didn't work that way for me: hp-setup simply found the device and
set up the print and fax queues and then exited.  To install the plugin
I've had to directly run hp-plugin or another hp-* script that calls
this; see below.

>> So I
>> added the first three to /usr/lib/sane and the libjpeg to /usr/lib, made
>> symlinks leaving out the architecture and version bits.  Then I ran
>> `scanimage >image.pnm' again and it immediately segfaulted.  By process
>> of elimination I determined that having
>> libsane-hpgt2500-x86_64.so.1.0.27 in /usr/lib/sane is what causes the
>> segfault, while with either or both of the other two libraries in
>> /usr/lib/sane, there is no segfault but hplip thinks no plugin is
>> installed.  (The presence or absence of libjpeg.so.9 didn't make a
>> difference.)
> What does ldd reveal?

Nothing that strikes me as problematic:

libsane-hp2000S1-x86_64.so.1.0.25
        linux-vdso.so.1 (0x00007ffd7dfe0000)
        libusb-1.0.so.0 => /usr/lib/libusb-1.0.so.0 (0x00007f818612c000)
        libjpeg.so.9 => /usr/lib/libjpeg.so.9 (0x00007f8185ef0000)
        libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x00007f8185d65000)
        libm.so.6 => /lib/libm.so.6 (0x00007f8185c21000)
        libgcc_s.so.1 => /usr/lib/libgcc_s.so.1 (0x00007f8185c07000)
        libc.so.6 => /lib/libc.so.6 (0x00007f8185a46000)
        libudev.so.1 => /lib/libudev.so.1 (0x00007f8185a1e000)
        libpthread.so.0 => /lib/libpthread.so.0 (0x00007f81859fd000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f818643d000)
hpgt2500_ntdcmsdll-x86_64.so
        linux-vdso.so.1 (0x00007ffcd7f3f000)
        libstdc++.so.6 => /usr/lib/libstdc++.so.6 (0x00007f3b362ab000)
        libc.so.6 => /lib/libc.so.6 (0x00007f3b360ea000)
        libm.so.6 => /lib/libm.so.6 (0x00007f3b35fa6000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f3b36684000)
        libgcc_s.so.1 => /usr/lib/libgcc_s.so.1 (0x00007f3b35f8c000)
libsane-hpgt2500-x86_64.so.1.0.27
        linux-vdso.so.1 (0x00007ffdbe7f0000)
        libpthread.so.0 => /lib/libpthread.so.0 (0x00007f0c1481d000)
        libdl.so.2 => /lib/libdl.so.2 (0x00007f0c14818000)
        libusb-1.0.so.0 => /usr/lib/libusb-1.0.so.0 (0x00007f0c147fd000)
        libc.so.6 => /lib/libc.so.6 (0x00007f0c1463c000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f0c14b74000)
        libudev.so.1 => /lib/libudev.so.1 (0x00007f0c14616000)

> That said, did you try the Arch PKGBUILD? They have a patch to remove some of
> the closed source image processing stuff. It's been a bit, but last time I
> tried it, seemed to work.
>
> https://git.archlinux.org/svntogit/packages.git/tree/trunk?h=packages/hplip
>
> Arch is usually my first stop if I can't figure something out now days.

Wayne Blaszczyk also mentioned the Arch hplip package to me offlist, and
I may try it at some point, but it appears that I've managed to get the
scanner working now, as follows (details for the record):

First, I (once again) deleted the print and fax queues in CUPS, then I
ran hp-setup, which reinstalled those queues but, as noted above, did
not install or even mention the plugin.  Without installing the plugin I
then tried, following a suggestion of Wayne's, hp-scan instead of
scanimage, but the result was this:

Using device hpaio:/usb/HP_ColorLaserJet_MFP_M278-M281?serial=VNBNM1K0H8
Opening connection to device...
error: SANE: Error during device I/O (code=9)

Then, instead of running hp-plugin I tried a couple of other hp-*
commands and one, hp-diagnose_plugin, failed with:

error:  No system tray detected on this system.  Unable to start,
exiting.

but also popped up a window asking whether to install the plugin, so I
now went ahead with that and the plugin was once again installed.  Then
I reran hp-scan and it immediately segfaulted, just as scanimage had
done previously.  Then I deleted libsane-hpgt2500-x86_64.so.1.0.27 and
the symlinks to it from /usr/lib/sane, ran hp-scan again -- and now it
worked!  Then I ran xsane and it worked too.

So now I know the scanner works with my BLFS (but I don't know why it
didn't work after I had previously uninstalled the same library, so I
wonder how reliable it is).  It seems likely, since that library is
installed in openSUSE Tumbleweed, where the scanner also works, that the
patches in openSUSE disable that library; maybe it's the same with the
Arch hplip package.  This conclusion is also suggested by the following
observation: I also have openSUSE Leap 15.1 on the same machine and the
scanner works there too, but the hplip version is 3.18.6 and after
installing the plugin the three libraries noted above are not included.
So they apparently only come with later versions of the plugin, and one
of them breaks SANE -- maybe.  Anyway, I'm glad I finally have the
scanner working in BLFS (at least for the moment...), and really
appreciate the feedback I've gotten here.

Steve Berman
-- 
http://lists.linuxfromscratch.org/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to