Re: chg-scsi coredumps when enabling barcode reader.
A simple and dirtly walkaround we use now is to set root suid bit on chg-scsi, this
solves all problems for now.
Regards
Kirill
On Mon, 18 Nov 2002 17:53:11 +0100
Kirill Alder-Ponazdyr <[EMAIL PROTECTED]> wrote:
> On Mon, 18 Nov 2002 11:42:11 -0500
> Galen Johnson <[EMAIL PROTECTED]> wrote:
>
> > Perhaps the barcode device has different ownership/permissions/group
> > than the rest of the devices? I don't have a barcode reader device so
> > I'm just throwing out possiblities...
> >
> > =G=
>
> Not really, the barcodes are being read trough the same device which is used to
>controll the library, and that one is readable / writable by sys group.
>
> Also, the chg-scsi is actually able to read the barcodes (visible in debug) but
>trashes somewhere past that point. I did 2 x trusses, one for root and one for
>amanda, this is what I see when looking at the area near crash:
>
> Amanda User:
>
>~~~
> cut
> write(10, " 1 0 3 7 6 3 5 5 9 7 : #".., 50) = 50
> time() = 1037635597
> write(10, " 1 0 3 7 6 3 5 5 9 7 : #".., 37) = 37
> time() = 1037635597
> write(10, " 1 0 3 7 6 3 5 5 9 7 : d".., 51) = 51
> time() = 1037635597
> write(10, " 1 0 3 7 6 3 5 5 9 7 : #".., 48) = 48
> open64("/etc/amanda/daily.scsi/tape0-slot", O_RDONLY) = 4
> fstat64(4, 0xFFBFF728) = 0
> fstat64(4, 0xFFBFF5D0) = 0
> ioctl(4, TCGETA, 0xFFBFF6B4)Err#25 ENOTTY
> read(4, " 0\n", 8192) = 2
> llseek(4, 0x, SEEK_CUR) = 1
> close(4)= 0
> time() = 1037635597
> write(10, " 1 0 3 7 6 3 5 5 9 7 : #".., 31) = 31
> time() = 1037635597
> write(10, " 1 0 3 7 6 3 5 5 9 7 : u".., 57) = 57
> time() = 1037635597
> write(10, " 1 0 3 7 6 3 5 5 9 7 : u".., 64) = 64
> time() = 1037635597
> time() = 1037635597
> time() = 1037635597
> time() = 1037635597
> time() = 1037635597
> time() = 1037635597
> Incurred fault #6, FLTBOUNDS %pc = 0x000182E4
> siginfo: SIGSEGV SEGV_MAPERR addr=0x0020
> Received signal #11, SIGSEGV [default]
> siginfo: SIGSEGV SEGV_MAPERR addr=0x0020
> end
>
> Root User:
>
>~~~
> cut
> write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 50) = 50
> time() = 1037635622
> write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 37) = 37
> time() = 1037635622
> write(10, " 1 0 3 7 6 3 5 6 2 2 : d".., 51) = 51
> time() = 1037635622
> write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 48) = 48
> open64("/etc/amanda/daily.scsi/tape0-slot", O_RDONLY) = 4
> fstat64(4, 0xFFBFF718) = 0
> fstat64(4, 0xFFBFF5C0) = 0
> ioctl(4, TCGETA, 0xFFBFF6A4)Err#25 ENOTTY
> read(4, " 0\n", 8192) = 2
> llseek(4, 0x, SEEK_CUR) = 1
> close(4)= 0
> time() = 1037635622
> write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 31) = 31
> time() = 1037635622
> write(10, " 1 0 3 7 6 3 5 6 2 2 : u".., 57) = 57
> time() = 1037635622
> write(10, " 1 0 3 7 6 3 5 6 2 2 : u".., 64) = 64
> time() = 1037635622
> time() = 1037635622
> time() = 1037635622
> time() = 1037635622
> time() = 1037635622
> time() = 1037635622
> time()
Re: chg-scsi coredumps when enabling barcode reader.
On Mon, 18 Nov 2002 11:42:11 -0500
Galen Johnson <[EMAIL PROTECTED]> wrote:
> Perhaps the barcode device has different ownership/permissions/group
> than the rest of the devices? I don't have a barcode reader device so
> I'm just throwing out possiblities...
>
> =G=
Not really, the barcodes are being read trough the same device which is used to
controll the library, and that one is readable / writable by sys group.
Also, the chg-scsi is actually able to read the barcodes (visible in debug) but
trashes somewhere past that point. I did 2 x trusses, one for root and one for amanda,
this is what I see when looking at the area near crash:
Amanda User:
~~~
cut
write(10, " 1 0 3 7 6 3 5 5 9 7 : #".., 50) = 50
time() = 1037635597
write(10, " 1 0 3 7 6 3 5 5 9 7 : #".., 37) = 37
time() = 1037635597
write(10, " 1 0 3 7 6 3 5 5 9 7 : d".., 51) = 51
time() = 1037635597
write(10, " 1 0 3 7 6 3 5 5 9 7 : #".., 48) = 48
open64("/etc/amanda/daily.scsi/tape0-slot", O_RDONLY) = 4
fstat64(4, 0xFFBFF728) = 0
fstat64(4, 0xFFBFF5D0) = 0
ioctl(4, TCGETA, 0xFFBFF6B4)Err#25 ENOTTY
read(4, " 0\n", 8192) = 2
llseek(4, 0x, SEEK_CUR) = 1
close(4)= 0
time() = 1037635597
write(10, " 1 0 3 7 6 3 5 5 9 7 : #".., 31) = 31
time() = 1037635597
write(10, " 1 0 3 7 6 3 5 5 9 7 : u".., 57) = 57
time() = 1037635597
write(10, " 1 0 3 7 6 3 5 5 9 7 : u".., 64) = 64
time() = 1037635597
time() = 1037635597
time() = 1037635597
time() = 1037635597
time() = 1037635597
time() = 1037635597
Incurred fault #6, FLTBOUNDS %pc = 0x000182E4
siginfo: SIGSEGV SEGV_MAPERR addr=0x0020
Received signal #11, SIGSEGV [default]
siginfo: SIGSEGV SEGV_MAPERR addr=0x0020
end
Root User:
~~~
cut
write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 50) = 50
time() = 1037635622
write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 37) = 37
time() = 1037635622
write(10, " 1 0 3 7 6 3 5 6 2 2 : d".., 51) = 51
time() = 1037635622
write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 48) = 48
open64("/etc/amanda/daily.scsi/tape0-slot", O_RDONLY) = 4
fstat64(4, 0xFFBFF718) = 0
fstat64(4, 0xFFBFF5C0) = 0
ioctl(4, TCGETA, 0xFFBFF6A4)Err#25 ENOTTY
read(4, " 0\n", 8192) = 2
llseek(4, 0x, SEEK_CUR) = 1
close(4)= 0
time() = 1037635622
write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 31) = 31
time() = 1037635622
write(10, " 1 0 3 7 6 3 5 6 2 2 : u".., 57) = 57
time() = 1037635622
write(10, " 1 0 3 7 6 3 5 6 2 2 : u".., 64) = 64
time() = 1037635622
time() = 1037635622
time() = 1037635622
time() = 1037635622
time() = 1037635622
time() = 1037635622
time() = 1037635622
time() = 1037635622
write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 42) = 42
time() = 1037635622
write(10, " 1 0 3 7 6 3 5 6 2 2 : #".., 34) = 34
time() = 1037635622
write(10, " S C S I _ E x e c u t e".., 38) = 38
time() = 1037635622
write(10, " 0 0", 3) = 3
time() = 1037635622
write(10, " 0 0", 3) = 3
time() = 1037635622
write(10, " 0 0", 3) = 3
time() = 1037635622
write(10, " 0 0", 3) = 3
time()
Re: chg-scsi coredumps when enabling barcode reader.
An update to this problem: When I run amanda as root user this does not happen. I wonder why, amanda user has all rights on tape devices and loader device (it is in sys group). It is also the owner of configuration, debug, log and so on directories, the "amanda" group is allowed to execute every single binary from amanda directory. Again, it does work when no barcode reading is involved, what is so special in chg-scsi to require root rights when it has to access barcode reader ? Regards ----- Kirill Alder-Ponazdyr SGI / SUN UNIX Consultant Codeangels Solutions Phone : +41 43 844 90 10 Fax : +41 43 844 90 12 Mobile: +41 79 370 89 30 On Sun, 17 Nov 2002 16:08:39 +0100 Kirill Alder-Ponazdyr <[EMAIL PROTECTED]> wrote: > Greetings, > > We have a following setup: > > Solaris 9 based Netra 1205 machine > Amanda 2.4.3 > Overland Data LibraryPro with 2 x AIT2 drives (4.20 firmware). > > The problem is, that chg-scsi crashes with segmentation fault when we enable barcode >reader (havebarcode 1) for that library, it works well without barcode enabling. > > Does anyone has an idea why this crash happens ? > > Thanks, > > Kirill > > Truss ending: > ~~~ > Incurred fault #6, FLTBOUNDS %pc = 0x000182E4 > siginfo: SIGSEGV SEGV_MAPERR addr=0x0020 > Received signal #11, SIGSEGV [default] > siginfo: SIGSEGV SEGV_MAPERR addr=0x0020 > ~~~ > > changer.conf: > ~~~ > number_configs 1 > eject 0 # Tapedrives need an eject command > sleep 10 # Seconds to wait until the tape gets ready > cleanmax10 # How many times could a cleaning tape get used > debuglevel9 > havebarcode 1 > changerdev /dev/scsi/changer/c1t1d0 > labelfile /opt/amanda/etc/amanda/daily/labelfile > > # > # Next comes the data for drive 0 > # > config 0 > drivenum0 > dev /dev/rmt/0cbn > startuse0 > enduse 11 > statfile/opt/amanda/etc/amanda/daily/tape0-slot > cleanfile /opt/amanda/etc/amanda/daily/tape0-clean > usagecount /opt/amanda/etc/amanda/daily/totaltime > tapestatus /opt/amanda/etc/amanda/daily/tape0-status > ~~~ > > > debug file: > ~~~ > chg-scsi: debug 1 pid 2012 ruid 1020 euid 1020: start at Sun Nov 17 15:59:16 2002 > chg-scsi: $Id: chg-scsi.c,v 1.6.2.22.2.7.2.6 2002/08/26 13:57:39 martinea Exp $ > 1037545156:scsi-changer-driver: $Id: scsi-changer-driver.c,v 1.1.2.27.2.7.2.9 >2002/08/26 13:57:40 martinea Exp $ > ARG [0] : /opt/amanda/libexec/chg-scsi > ARG [1] : -slot > ARG [2] : 0 > Number of configurations: 1 > Tapes need eject: No > raw: 0 > Inv. auto update: No > raw: 0 > barcode reader : Yes > raw: 1 > Emulate Barcode : No > raw: 0 > debug level : 9 > Tapes need sleep: 10 seconds > Cleancycles : 10 > Changerdevice : /dev/scsi/changer/c1t1d0 > Labelfile : /opt/amanda/etc/amanda/daily/labelfile > Tapeconfig Nr: 0 > Drivenumber : 0 > Startslot : 0 > Endslot : 11 > Cleanslot : -1 > Devicename: /dev/rmt/0cbn > changerident : none > SCSITapedev : none > tapeident : none > statfile : /opt/amanda/etc/amanda/daily/tape0-status > Slotfile : /opt/amanda/etc/amanda/daily/tape0-slot > Cleanfile : /opt/amanda/etc/amanda/daily/tape0-clean > Usagecount: /opt/amanda/etc/amanda/daily/totaltime > 1037545156:# START OpenDevice > 1037545156:OpenDevice : /dev/scsi/changer/c1t1d0 > 1037545156:# START DecodeSCSI > SCSI_ExecuteCommand : INQUIRY 12 00 00 00 38 00 > 1037545156:# STOP DecodeSCSI > 1037545156:ioctl ret (0) > 1037545156:using ident = generic_changer, type = Generic driver changer >[generic_changer] > 1037545156:# STOP OpenDevice > 1037545156:# START OpenDevice > 1037545156:OpenDevice : /dev/rmt/0cbn > ioctl on 4 failed, errno 1, ret -1 > # START DecodeSense > SCSI_ExecuteCommand:Sense Keys > ErrorCode 00 > Valid 0 > ASC 00 > ASCQ 00 > Sense key 00 > No Sense > 1037545156:ioctl ret (-1) > 1037545156:# STOP OpenDevice failed > 1037545156:# ST
Re: chg-scsi coredumps when enabling barcode reader.
An update to this problem: When I run amanda as root user this does not happen. I wonder why, amanda user has all rights on tape devices and loader device (it is in sys group). It is also the owner of configuration, debug, log and so on directories, the "amanda" group is allowed to execute every single binary from amanda directory. Again, it does work when no barcode reading is involved, what is so special in chg-scsi to require root rights when it has to access barcode reader ? Regards ----- Kirill Alder-Ponazdyr SGI / SUN UNIX Consultant Codeangels Solutions Phone : +41 43 844 90 10 Fax : +41 43 844 90 12 Mobile: +41 79 370 89 30 On Sun, 17 Nov 2002 16:08:39 +0100 Kirill Alder-Ponazdyr <[EMAIL PROTECTED]> wrote: > Greetings, > > We have a following setup: > > Solaris 9 based Netra 1205 machine > Amanda 2.4.3 > Overland Data LibraryPro with 2 x AIT2 drives (4.20 firmware). > > The problem is, that chg-scsi crashes with segmentation fault when we enable barcode >reader (havebarcode 1) for that library, it works well without barcode enabling. > > Does anyone has an idea why this crash happens ? > > Thanks, > > Kirill > > Truss ending: > ~~~ > Incurred fault #6, FLTBOUNDS %pc = 0x000182E4 > siginfo: SIGSEGV SEGV_MAPERR addr=0x0020 > Received signal #11, SIGSEGV [default] > siginfo: SIGSEGV SEGV_MAPERR addr=0x0020 > ~~~ > > changer.conf: > ~~~ > number_configs 1 > eject 0 # Tapedrives need an eject command > sleep 10 # Seconds to wait until the tape gets ready > cleanmax10 # How many times could a cleaning tape get used > debuglevel9 > havebarcode 1 > changerdev /dev/scsi/changer/c1t1d0 > labelfile /opt/amanda/etc/amanda/daily/labelfile > > # > # Next comes the data for drive 0 > # > config 0 > drivenum0 > dev /dev/rmt/0cbn > startuse0 > enduse 11 > statfile/opt/amanda/etc/amanda/daily/tape0-slot > cleanfile /opt/amanda/etc/amanda/daily/tape0-clean > usagecount /opt/amanda/etc/amanda/daily/totaltime > tapestatus /opt/amanda/etc/amanda/daily/tape0-status > ~~~ > > > debug file: > ~~~ > chg-scsi: debug 1 pid 2012 ruid 1020 euid 1020: start at Sun Nov 17 15:59:16 2002 > chg-scsi: $Id: chg-scsi.c,v 1.6.2.22.2.7.2.6 2002/08/26 13:57:39 martinea Exp $ > 1037545156:scsi-changer-driver: $Id: scsi-changer-driver.c,v 1.1.2.27.2.7.2.9 >2002/08/26 13:57:40 martinea Exp $ > ARG [0] : /opt/amanda/libexec/chg-scsi > ARG [1] : -slot > ARG [2] : 0 > Number of configurations: 1 > Tapes need eject: No > raw: 0 > Inv. auto update: No > raw: 0 > barcode reader : Yes > raw: 1 > Emulate Barcode : No > raw: 0 > debug level : 9 > Tapes need sleep: 10 seconds > Cleancycles : 10 > Changerdevice : /dev/scsi/changer/c1t1d0 > Labelfile : /opt/amanda/etc/amanda/daily/labelfile > Tapeconfig Nr: 0 > Drivenumber : 0 > Startslot : 0 > Endslot : 11 > Cleanslot : -1 > Devicename: /dev/rmt/0cbn > changerident : none > SCSITapedev : none > tapeident : none > statfile : /opt/amanda/etc/amanda/daily/tape0-status > Slotfile : /opt/amanda/etc/amanda/daily/tape0-slot > Cleanfile : /opt/amanda/etc/amanda/daily/tape0-clean > Usagecount: /opt/amanda/etc/amanda/daily/totaltime > 1037545156:# START OpenDevice > 1037545156:OpenDevice : /dev/scsi/changer/c1t1d0 > 1037545156:# START DecodeSCSI > SCSI_ExecuteCommand : INQUIRY 12 00 00 00 38 00 > 1037545156:# STOP DecodeSCSI > 1037545156:ioctl ret (0) > 1037545156:using ident = generic_changer, type = Generic driver changer >[generic_changer] > 1037545156:# STOP OpenDevice > 1037545156:# START OpenDevice > 1037545156:OpenDevice : /dev/rmt/0cbn > ioctl on 4 failed, errno 1, ret -1 > # START DecodeSense > SCSI_ExecuteCommand:Sense Keys > ErrorCode 00 > Valid 0 > ASC 00 > ASCQ 00 > Sense key 00 > No Sense > 1037545156:ioctl ret (-1) > 1037545156:# STOP OpenDevice failed > 1037545156:# ST
Re: chg-scsi coredumps when enabling barcode reader.
Here is gcc -v: Reading specs from /opt/sfw/lib/gcc-lib/sparc-sun-solaris2.9/2.95.3/specs gcc version 2.95.3 20010315 (release) And that was not working before ether. For now we have rolled back to chg-zd-mtx, but would like to use chg-scsi if it would be working. Regards - Kirill Alder-Ponazdyr SGI / SUN UNIX Consultant Codeangels Solutions Phone : +41 43 844 90 10 Fax : +41 43 844 90 12 Mobile: +41 79 370 89 30 On Sun, 17 Nov 2002 11:08:54 -0500 Galen Johnson <[EMAIL PROTECTED]> wrote: > Kirill Alder-Ponazdyr wrote: > > >Greetings, > > > >We have a following setup: > > > >Solaris 9 based Netra 1205 machine > >Amanda 2.4.3 > >Overland Data LibraryPro with 2 x AIT2 drives (4.20 firmware). > > > >The problem is, that chg-scsi crashes with segmentation fault when we enable >barcode reader (havebarcode 1) for that library, it works well without barcode >enabling. > > > >Does anyone has an idea why this crash happens ? > > > >Thanks, > > > >Kirill > > > > > > > What version of GCC did you use to compile this? If you used gcc3.2, > I'm not sure that the code is ready for that...but I'm not sure. Have > you had it working before? > > =G= > > >
chg-scsi coredumps when enabling barcode reader.
Greetings, We have a following setup: Solaris 9 based Netra 1205 machine Amanda 2.4.3 Overland Data LibraryPro with 2 x AIT2 drives (4.20 firmware). The problem is, that chg-scsi crashes with segmentation fault when we enable barcode reader (havebarcode 1) for that library, it works well without barcode enabling. Does anyone has an idea why this crash happens ? Thanks, Kirill Truss ending: ~~~ Incurred fault #6, FLTBOUNDS %pc = 0x000182E4 siginfo: SIGSEGV SEGV_MAPERR addr=0x0020 Received signal #11, SIGSEGV [default] siginfo: SIGSEGV SEGV_MAPERR addr=0x0020 ~~~ changer.conf: ~~~ number_configs 1 eject 0 # Tapedrives need an eject command sleep 10 # Seconds to wait until the tape gets ready cleanmax10 # How many times could a cleaning tape get used debuglevel 9 havebarcode 1 changerdev /dev/scsi/changer/c1t1d0 labelfile /opt/amanda/etc/amanda/daily/labelfile # # Next comes the data for drive 0 # config 0 drivenum0 dev /dev/rmt/0cbn startuse0 enduse 11 statfile/opt/amanda/etc/amanda/daily/tape0-slot cleanfile /opt/amanda/etc/amanda/daily/tape0-clean usagecount /opt/amanda/etc/amanda/daily/totaltime tapestatus /opt/amanda/etc/amanda/daily/tape0-status ~~~ debug file: ~~~ chg-scsi: debug 1 pid 2012 ruid 1020 euid 1020: start at Sun Nov 17 15:59:16 2002 chg-scsi: $Id: chg-scsi.c,v 1.6.2.22.2.7.2.6 2002/08/26 13:57:39 martinea Exp $ 1037545156:scsi-changer-driver: $Id: scsi-changer-driver.c,v 1.1.2.27.2.7.2.9 2002/08/26 13:57:40 martinea Exp $ ARG [0] : /opt/amanda/libexec/chg-scsi ARG [1] : -slot ARG [2] : 0 Number of configurations: 1 Tapes need eject: No raw: 0 Inv. auto update: No raw: 0 barcode reader : Yes raw: 1 Emulate Barcode : No raw: 0 debug level : 9 Tapes need sleep: 10 seconds Cleancycles : 10 Changerdevice : /dev/scsi/changer/c1t1d0 Labelfile : /opt/amanda/etc/amanda/daily/labelfile Tapeconfig Nr: 0 Drivenumber : 0 Startslot : 0 Endslot : 11 Cleanslot : -1 Devicename: /dev/rmt/0cbn changerident : none SCSITapedev : none tapeident : none statfile : /opt/amanda/etc/amanda/daily/tape0-status Slotfile : /opt/amanda/etc/amanda/daily/tape0-slot Cleanfile : /opt/amanda/etc/amanda/daily/tape0-clean Usagecount: /opt/amanda/etc/amanda/daily/totaltime 1037545156:# START OpenDevice 1037545156:OpenDevice : /dev/scsi/changer/c1t1d0 1037545156:# START DecodeSCSI SCSI_ExecuteCommand : INQUIRY 12 00 00 00 38 00 1037545156:# STOP DecodeSCSI 1037545156:ioctl ret (0) 1037545156:using ident = generic_changer, type = Generic driver changer [generic_changer] 1037545156:# STOP OpenDevice 1037545156:# START OpenDevice 1037545156:OpenDevice : /dev/rmt/0cbn ioctl on 4 failed, errno 1, ret -1 # START DecodeSense SCSI_ExecuteCommand:Sense Keys ErrorCode 00 Valid 0 ASC 00 ASCQ 00 Sense key 00 No Sense 1037545156:ioctl ret (-1) 1037545156:# STOP OpenDevice failed 1037545156:# STOP OpenDevice (nothing found) !! warning open of /dev/rmt/0cbn: failed DecodeModeSense : Element Address Assignment Page Medium Transport Element Address 0 Number of Medium Transport Elements 1 First Storage Element Address 1 Number of Storage Elements 19 First Import/Export Element Address 224 Number of ImportExport Elements1 First Data Transfer Element Address 240 Number of Data Transfer Elements 2 DecodeModeSense : MT can store data cartridges 0 DecodeModeSense : ST can store data cartridges 1 DecodeModeSense : IE can store data cartridges 1 DecodeModeSense : DT can store data cartridges 1 DecodeModeSense : MT to MT 0 DecodeModeSense : MT to ST 0 DecodeModeSense : MT to IE 0 DecodeModeSense : MT to DT 0 DecodeModeSense : ST to MT 1 DecodeModeSense : ST to MT 1 DecodeModeSense : ST to DT 1 DecodeModeSense : IE to MT 0 DecodeModeSense : IE to ST 1 DecodeModeSense : IE to ST 1 DecodeModeSense : IE to ST 1 DecodeModeSense : DT to MT 0 DecodeModeSense : DT to ST 1 DecodeModeSense : DT to IE 1 DecodeModeSense : DT to DT 1 1037545157:# START DecodeSCSI SCSI_ExecuteCommand : TEST UNIT READY 00 00 00 00 00 00 1037545157:# STOP DecodeSCSI 1037545157:ioctl
