samsungmfp-data: /usr/share/ppd/samsung symlink breaks CUPS

[Martin von Wittich]

Hi,

the samsungmfp-data package installs a symlink /usr/share/ppd/samsung:

Code Select Expand


iserv:~# ll /usr/share/ppd/samsung
lrwxrwxrwx 1 root root 22 17. Sep 16:49 /usr/share/ppd/samsung -> ../cups/model/samsung/


This symlink silently breaks CUPS - when it encounters this symlink, it stops reading in /usr/share/ppd:

Code Select Expand


iserv:~# /usr/lib/cups/daemon/cups-driverd list 1 0 "requested-attributes=all" 2>&1 | grep -e Loading -e Skipping
DEBUG: [cups-driverd] Loading "/usr/share/cups/model"...
DEBUG: [cups-driverd] Loading "/usr/share/cups/model/samsung"...
DEBUG: [cups-driverd] Loading "/usr/share/cups/model/samsung/cms"...
DEBUG: [cups-driverd] Loading "/usr/share/cups/drv"...
DEBUG: [cups-driverd] Loading "/usr/local/share/ppd"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/kyocera"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/lexmark"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/cups-included"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/ghostscript"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/ghostscript/model"...
ERROR: [cups-driverd] Skipping "/usr/share/ppd/samsung": loop detected!


All remaining folders in /usr/share/ppd are now ignored by CUPS. After I've removed the symlink manually, it will again read the remaining folders:

Code Select Expand


iserv:~# rm -f /usr/share/ppd/samsung
iserv:~# /usr/lib/cups/daemon/cups-driverd list 1 0 "requested-attributes=all" 2>&1 | grep -e Loading -e Skipping
DEBUG: [cups-driverd] Loading "/usr/share/cups/model"...
DEBUG: [cups-driverd] Loading "/usr/share/cups/model/samsung"...
DEBUG: [cups-driverd] Loading "/usr/share/cups/model/samsung/cms"...
DEBUG: [cups-driverd] Loading "/usr/share/cups/drv"...
DEBUG: [cups-driverd] Loading "/usr/local/share/ppd"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/kyocera"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/lexmark"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/cups-included"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/ghostscript"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/ghostscript/model"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/brother"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Oki"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Sharp"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Epson"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Oce"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Oce/Oce-VarioPrint_2055PS"...
[...]
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Oce/Oce-PPC3113PS/1"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Toshiba"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Brother"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Lexmark"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Kyocera"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/Kyocera/en"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/HP"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/postscript/KONICA_MINOLTA"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/custom"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/foo2zjs"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/splix"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/splix/xerox"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/splix/dell"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/splix/samsung"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/hplip"...
DEBUG: [cups-driverd] Loading "/usr/share/ppd/hplip/HP"...


[bchemnet]

Thanks.  The symlink is inherited from the original Samsung package.  I'll remove it in the next release (hopefully in one or two weeks).  I don't think it will impact anything else about the driver.

I think this is probably really a bug in CUPS, because I can't see any reason that detection of a loop (which it isn't actually) should impact reading later directories.  But it's easy for me to fix.

In the meantime, anyone who has trouble finding the driver for a second printer while the Samsung driver is installed should remove that link.

[bchemnet]

I have fixed this in the next round of packages I am preparing; look for 4.00.36 and 4.00.39 drivers within a week.