SCM

[#315650] Samsung C1860FW---scanning hangs when scanning large files over tcp with xerox_mfp

View Trackers | Bugs | Export CSV

Date:
2017-03-11 00:23
Priority:
3
State:
Open
Submitted by:
Matthew Trescott (mtres19-guest)
Assigned to:
Alex Belkin (abc-guest)
Category:
backends (drivers)
Group:
None
Resolution:
None
Summary:
Samsung C1860FW---scanning hangs when scanning large files over tcp with xerox_mfp

Detailed description
When I try to scan color images from the flatbed at high resolutions (≥ 300 dpi), scanimage hangs and takes 100% CPU (displayed as 25% due to having a quad-core system). The same happens with simple-scan and skanlite, so not a front-end problem.

System: Arch Linux 64-bit, fully upgraded. Scanner: Samsung C1860FW multifunction. Setup: [code]tcp 192.168.1.x[/code] in xerox_mfp.conf (where x is the last part of the LAN IP address). Note that this was also a problem in Kubuntu 16.04 LTS as of just a few days ago. I don't think this was previously an issue, but I might have simply been using lower resolutions and not noticed.

It seemed that changing the --buffer-size option on scanimage to something less than the default helped it get a little farther before freezing up, but I'm not sure; with 32 KB scanimage got to ~16% before freezing, with 4 KB it got to ~53% once but then ~25% the next time. So not sure if that has anything to do with it.

If needed I could run it under a debugger like valgrind or GDB but I don't know how to use those tools so I'd have to have the exact commands to run.

Followups: Sort comments antichronologically

Message
Date: 2017-03-11 00:29
Sender: Matthew Trescott

Note that I do not receive any error message from scanimage, even with the -v flag.
Date: 2017-03-20 11:37
Sender: Olaf Meeuwissen

Hi Alex,

Time permitting, could you take a look at this?
Date: 2017-03-20 11:49
Sender: Alex Belkin

Matthew,

What version of sane? You may run under gdb, first install sane debuginfo package if present in your distro (don't know how it should eb achieved for Arch linux), then:

# gdb
(gdb) file scanimage
(gdb) run <type scanimage options here w/o angly brackets>

when is start to load cpu press ^C and you will return to gdb prompt:

(gdb) bt

copy paste backtrace output here.
Date: 2017-03-21 00:12
Sender: Matthew Trescott

Hi,

Here's the backtrace from version 1.0.25:
#0 0x00007ffff687f3c0 in __write_nocancel () from /usr/lib/libc.so.6
#1 0x00007ffff6814f97 in _IO_new_file_write () from /usr/lib/libc.so.6
#2 0x00007ffff6814232 in new_do_write () from /usr/lib/libc.so.6
#3 0x00007ffff68156ee in __GI__IO_file_xsputn () from /usr/lib/libc.so.6
#4 0x00007ffff67ec778 in buffered_vfprintf () from /usr/lib/libc.so.6
#5 0x00007ffff67e9a26 in vfprintf () from /usr/lib/libc.so.6
#6 0x00007ffff67f1dd4 in fprintf () from /usr/lib/libc.so.6
#7 0x00000000004051cf in scan_it (ofp=0x7ffff6b405e0 <_IO_2_1_stdout_>) at scanimage.c:1451
#8 0x0000000000407953 in main (argc=3, argv=0x7fffffffe6f8) at scanimage.c:2558

However, I also tried the git version, and there's no hangs, so I guess this issue is already resolved!

Thanks,

Matthew

Attached Files:

Changes:

Field Old Value Date By
assigned_tonone2017-03-20 11:37olaf-guest
Powered By FusionForge