• debian-hppa install notes with some CD trouble

    From Meelis Roos@21:1/5 to All on Mon Feb 18 16:00:01 2019
    I tested debian-ports install on HP 9000 A180c (the last 32-bit server with parisc).

    First I needed to find external SCSI CD-ROM with 512-byte block size support - I managed but netboot images would have been much simpler (same goes for sparc64).

    Next, the CD was recognized as bootable by SEARCH IPL, good. It booted too and palo
    chose the correct 32-bit kernel and automatically used my serial console too.

    Initial locale and language dialogs worked fine, until it came to mounting the CD-ROM. That failed and asked if I want to load a driver. sr_mod was there so it was not a module problem. dmesg showed errors about reading past end of device.
    So I wrote another CD (checked SHA256sum of the image to make sure) with wodim -pad
    to pad the secotrs up. Still same problem.

    The logs showed problematic command was
    mount -t iso9660 /dev/sr0 /cdrom
    So I tried it by hand and it failed with "No such file or directory".

    After a manual retry it mounted. I unmounted it an let the installer try again and this time it worked. Why it worked, remains a mystery for me.


    Next stop was at configuring the package manager, after installing base system. I was greeted with the following dialog where I retried multiple times.

    lqqqqqqqqqqqqqqqqu [!!] Configure the package manager tqqqqqqqqqqqqqqqqqk
    x x
    lx /cdrom/:Please insert the disc labeled: 'Debian GNU/Linux 10.0 _Sid_ x
    xx - Unofficial hppa NETINST 20190127-18:59' in the drive '/cdrom/' and x
    xx press enter. x
    xx x
    xx Media change x
    xx <Go Back> <Continue> x
    xx x
    xmqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqj

    This is corresponding log from screen 4 (no I/O errors, just multiple mountings):

    Feb 18 11:45:21 apt-setup: Found 1 package indexes, 0 source indexes, 0 translation indexes and 0 signatures
    Feb 18 11:45:21 apt-setup: This disc is called:
    Feb 18 11:45:21 apt-setup: 'Debian GNU/Linux 10.0 _Sid_ - Unofficial hppa NETINST 20190127-18:59'
    Feb 18 11:45:21 apt-setup: Copying package lists...
    Feb 18 11:45:21 apt-setup: ^MReading Package Indexes... 0%^M
    Feb 18 11:45:21 apt-setup: ^MReading Package Indexes... 0%^M
    Feb 18 11:45:23 apt-setup: ^MReading Package Indexes... 241%^M
    Feb 18 11:45:23 apt-setup: ^MReading Package Indexes... Done^M
    Feb 18 11:45:23 apt-setup:
    Feb 18 11:45:23 apt-setup: Writing new source list
    Feb 18 11:45:23 apt-setup: Source list entries for this disc are:
    Feb 18 11:45:23 apt-setup: deb cdrom:[Debian GNU/Linux 10.0 _Sid_ - Unofficial hppa NETINST 20190127-18:59]/ sid main
    Feb 18 11:45:23 apt-setup: Repeat this process for the rest of the CDs in your set.
    Feb 18 11:45:26 kernel: [ 7011.291838] ISO 9660 Extensions: RRIP_1991A
    Feb 18 11:45:27 kernel: [ 7012.153976] ISO 9660 Extensions: RRIP_1991A
    Feb 18 11:47:00 kernel: [ 7105.302931] ISO 9660 Extensions: RRIP_1991A
    Feb 18 11:48:19 kernel: [ 7184.826600] ISO 9660 Extensions: RRIP_1991A
    Feb 18 12:15:36 kernel: [ 8821.824640] ISO 9660 Extensions: RRIP_1991A

    Retrying package manager configuration from the main menu asked me for network mirror
    so I provided ftp.ports.debian.org and /debian-ports/ and install could continue.

    After installing all the packages (pkgsel works fine), the installer again asks for CD labeled
    'Debian GNU/Linux 10.0 _Sid_ - Unofficial hppa NETINST 20190127-18:59'. After choosing
    Back, it told something abou finishing installation and then goes back to main menu with
    "Finish the installation" selected. Selecting something from there (I think I chose PALO installation)
    ejects the CD and tells the installation is finished.

    After reboot, SCSI disk becomes valid for IPL and the systems boots up from the hard disk.

    The kernel waits a lot after
    [ 58.632191] random: crng init done
    but that was the same during installer bootup.

    Finally it boots up fine. zalon7xx gives warnings but that is for another time.


    --
    Meelis Roos <mroos@linux.ee>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Helge Deller@21:1/5 to Meelis Roos on Mon Feb 18 18:10:02 2019
    On 18.02.19 15:24, Meelis Roos wrote:
    I tested debian-ports install on HP 9000 A180c (the last 32-bit server with parisc).

    Thanks for testing!

    First I needed to find external SCSI CD-ROM with 512-byte block size support -
    I managed but netboot images would have been much simpler (same goes for sparc64).


    Adrian mentioned to me, that netboot images are always in the d-i-tarball, e.g.:
    http://ftp.ports.debian.org/debian-ports/pool-hppa/main/d/debian-installer/debian-installer-images_20190118_hppa.tar.gz
    I haven't tried myself.


    Next, the CD was recognized as bootable by SEARCH IPL, good. It booted too and palo
    chose the correct 32-bit kernel and automatically used my serial console too.

    Initial locale and language dialogs worked fine, until it came to mounting the
    CD-ROM. That failed and asked if I want to load a driver. sr_mod was there so it was not a module problem. dmesg showed errors about reading past end of device.
    So I wrote another CD (checked SHA256sum of the image to make sure) with wodim -pad
    to pad the secotrs up. Still same problem.

    The logs showed problematic command was
    mount -t iso9660 /dev/sr0 /cdrom
    So I tried it by hand and it failed with "No such file or directory".

    After a manual retry it mounted. I unmounted it an let the installer try again
    and this time it worked. Why it worked, remains a mystery for me.

    Strange!
    Maybe the "old" CD-ROM drive has hickups ?


    Next stop was at configuring the package manager, after installing base system.
    I was greeted with the following dialog where I retried multiple times.

      lqqqqqqqqqqqqqqqqu [!!] Configure the package manager tqqqqqqqqqqqqqqqqqk    x                                                                       x
      lx /cdrom/:Please insert the disc labeled: 'Debian GNU/Linux 10.0 _Sid_  x
      xx - Unofficial hppa NETINST 20190127-18:59' in the drive '/cdrom/' and  x
      xx press enter.                                                          x
      xx                                                                       x
      xx Media change                                                          x
      xx     <Go Back>                                          <Continue>     x
      xx                                                                       x
      xmqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqj This is corresponding log from screen 4 (no I/O errors, just multiple mountings):

    Feb 18 11:45:21 apt-setup: Found 1 package indexes, 0 source indexes, 0 translation indexes and 0 signatures
    Feb 18 11:45:21 apt-setup: This disc is called:
    Feb 18 11:45:21 apt-setup: 'Debian GNU/Linux 10.0 _Sid_ - Unofficial hppa NETINST 20190127-18:59'
    Feb 18 11:45:21 apt-setup: Copying package lists...
    Feb 18 11:45:21 apt-setup: ^MReading Package Indexes... 0%^M
    Feb 18 11:45:21 apt-setup: ^MReading Package Indexes... 0%^M
    Feb 18 11:45:23 apt-setup: ^MReading Package Indexes... 241%^M
    Feb 18 11:45:23 apt-setup: ^MReading Package Indexes... Done^M
    Feb 18 11:45:23 apt-setup:
    Feb 18 11:45:23 apt-setup: Writing new source list
    Feb 18 11:45:23 apt-setup: Source list entries for this disc are:
    Feb 18 11:45:23 apt-setup: deb cdrom:[Debian GNU/Linux 10.0 _Sid_ - Unofficial hppa NETINST 20190127-18:59]/ sid main
    Feb 18 11:45:23 apt-setup: Repeat this process for the rest of the CDs in your set.

    ^^^ That message is strange...

    Feb 18 11:45:26 kernel: [ 7011.291838] ISO 9660 Extensions: RRIP_1991A
    Feb 18 11:45:27 kernel: [ 7012.153976] ISO 9660 Extensions: RRIP_1991A
    Feb 18 11:47:00 kernel: [ 7105.302931] ISO 9660 Extensions: RRIP_1991A
    Feb 18 11:48:19 kernel: [ 7184.826600] ISO 9660 Extensions: RRIP_1991A
    Feb 18 12:15:36 kernel: [ 8821.824640] ISO 9660 Extensions: RRIP_1991A

    Retrying package manager configuration from the main menu asked me for network mirror
    so I provided ftp.ports.debian.org and /debian-ports/ and install could continue.

    We are still waiting for this fix: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=879130


    After installing all the packages (pkgsel works fine), the installer again asks for CD labeled
    'Debian GNU/Linux 10.0 _Sid_ - Unofficial hppa NETINST 20190127-18:59'. After choosing
    Back, it told something abou finishing installation and then goes back to main menu with
    "Finish the installation" selected. Selecting something from there (I think I chose PALO installation)
    ejects the CD and tells the installation is finished.

    After reboot, SCSI disk becomes valid for IPL and the systems boots up from the hard disk.

    The kernel waits a lot after
    [   58.632191] random: crng init done
    but that was the same during installer bootup.

    Yes, that's a common problem, that the random number pool needs to get filled. It's not parisc specific.

    Finally it boots up fine. zalon7xx gives warnings but that is for another time.

    What kind of warnings? Do you have the log?
    Does the A500 uses zalon?

    Helge

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Meelis Roos@21:1/5 to All on Mon Feb 18 18:40:01 2019
    Adrian mentioned to me, that netboot images are always in the d-i-tarball, e.g.:
    http://ftp.ports.debian.org/debian-ports/pool-hppa/main/d/debian-installer/debian-installer-images_20190118_hppa.tar.gz
    I haven't tried myself.

    Will try some time.

    Feb 18 11:45:21 apt-setup: Found 1 package indexes, 0 source indexes, 0 translation indexes and 0 signatures
    Feb 18 11:45:21 apt-setup: This disc is called:
    Feb 18 11:45:21 apt-setup: 'Debian GNU/Linux 10.0 _Sid_ - Unofficial hppa NETINST 20190127-18:59'
    Feb 18 11:45:21 apt-setup: Copying package lists...
    Feb 18 11:45:21 apt-setup: ^MReading Package Indexes... 0%^M
    Feb 18 11:45:21 apt-setup: ^MReading Package Indexes... 0%^M
    Feb 18 11:45:23 apt-setup: ^MReading Package Indexes... 241%^M
    Feb 18 11:45:23 apt-setup: ^MReading Package Indexes... Done^M
    Feb 18 11:45:23 apt-setup:
    Feb 18 11:45:23 apt-setup: Writing new source list
    Feb 18 11:45:23 apt-setup: Source list entries for this disc are:
    Feb 18 11:45:23 apt-setup: deb cdrom:[Debian GNU/Linux 10.0 _Sid_ - Unofficial hppa NETINST 20190127-18:59]/ sid main
    Feb 18 11:45:23 apt-setup: Repeat this process for the rest of the CDs in your set.

    ^^^ That message is strange...

    /dev/sr0: UUID="2019-01-27-19-00-03-00" LABEL="Debian 10.0 hppa n" TYPE="iso9660"

    Perhaps the label is truncated?

    Finally it boots up fine. zalon7xx gives warnings but that is for another time.

    What kind of warnings? Do you have the log?

    Yes, I have a fully working system now and the errors are repeatable. Will try latest git first :)

    Does the A500 uses zalon?

    No, Symbios Logic 53C875.
    --
    Meelis Roos

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Meelis Roos@21:1/5 to All on Tue Feb 19 10:20:01 2019
    Initial locale and language dialogs worked fine, until it came to mounting the
    CD-ROM. That failed and asked if I want to load a driver. sr_mod was there so
    it was not a module problem. dmesg showed errors about reading past end of device.
    So I wrote another CD (checked SHA256sum of the image to make sure) with wodim -pad
    to pad the secotrs up. Still same problem.

    The logs showed problematic command was
    mount -t iso9660 /dev/sr0 /cdrom
    So I tried it by hand and it failed with "No such file or directory".

    After a manual retry it mounted. I unmounted it an let the installer try again
    and this time it worked. Why it worked, remains a mystery for me.

    Strange!
    Maybe the "old" CD-ROM drive has hickups ?
    I have since tested it in as many ways I could - mounted and unmounted it repeatedly,
    tar'ed all the files from it etc. It is slow but always works.

    The only CD-related errors in dmesg are in the form of not being able to read past end of device:

    [ 35.835421] sr 0:0:6:0: [sr0] tag#46 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=0x08
    [ 35.918389] sr 0:0:6:0: [sr0] tag#46 Sense Key : 0x3 [current]
    [ 35.973370] sr 0:0:6:0: [sr0] tag#46 ASC=0x6 ASCQ=0x0
    [ 36.019042] sr 0:0:6:0: [sr0] tag#46 CDB: opcode=0x28 28 00 00 01 9c 2c 00 00 02 00
    [ 36.094680] print_req_error: I/O error, dev sr0, sector 422064 flags 80700
    [ 36.164130] attempt to access beyond end of device
    [ 36.205653] sr0: rw=0, want=422072, limit=422064
    [ 36.245094] Buffer I/O error on dev sr0, logical block 52758, async page read

    # isoinfo -d dev=/dev/sr0
    CD-ROM is in ISO 9660 format
    System id: LINUX
    Volume id: Debian 10.0 hppa n
    Volume set id:
    Publisher id:
    Data preparer id:
    Application id: GENISOIMAGE ISO 9660/HFS FILESYSTEM CREATOR (C) 1993 E.YOUNGDALE (C) 1997-2006 J.PEARSON/J.SCHILLING (C) 2006-2007 CDRKIT TEAM
    Copyright File id:
    Abstract File id:
    Bibliographic File id:
    Volume set size is: 1
    Volume set sequence number is: 1
    Logical block size is: 2048
    Volume size is: 105501
    NO Joliet present
    Rock Ridge signatures version 1 found

    This shows the file system is 105501*4=422004 512-byte sectors in size and should not cause the accesses beyond end of device.

    udev's /lib/udev/cdrom_id does not cause these errors either, at least when I run it.

    # wodim -toc
    Device was not specified. Trying to find an appropriate drive...
    Detected CD-R drive: /dev/cdrw
    Using /dev/cdrom of unknown capabilities
    Device type : Removable CD-ROM
    Version : 2
    Response Format: 2
    Capabilities : SYNC LINKED
    Vendor_info : 'PLEXTOR '
    Identification : 'CD-R PX-W1210S'
    Revision : '1.00'
    Device seems to be: Generic mmc CD-RW.
    Using generic SCSI-3/mmc CD-R/CD-RW driver (mmc_cdr).
    Driver flags : MMC SWABAUDIO BURNFREE
    Supported modes: TAO PACKET SAO SAO/R96P SAO/R96R RAW/R16 RAW/R96P RAW/R96R first: 1 last 1
    track: 1 lba: 0 ( 0) 00:02:00 adr: 1 control: 4 mode: 1 track:lout lba: 105518 ( 422072) 23:28:68 adr: 1 control: 4 mode: -1


    So it seems something else in the boot process tries to read the very end of the CD and fails. But
    nothing depends on it in practice.

    Sector numbers are strange though - 422072 sectors in tack but 422064 is causing problems - maybe because of Linux doing bigger requests.

    dd with bs=2048 reads
    105516+0 records in
    105516+0 records out
    and 105516 < 105518. Why 105516?


    --
    Meelis Roos <mroos@linux.ee>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Thomas Schmitt@21:1/5 to Meelis Roos on Tue Feb 19 11:10:01 2019
    Hi,

    Meelis Roos wrote:
    The only CD-related errors in dmesg are in the form of not being able to
    read past end of device:
    [...]
    [ 35.918389] sr 0:0:6:0: [sr0] tag#46 Sense Key : 0x3 [current]
    [ 35.973370] sr 0:0:6:0: [sr0] tag#46 ASC=0x6 ASCQ=0x0
    [ 36.019042] sr 0:0:6:0: [sr0] tag#46 CDB: opcode=0x28 28 00 00 01 9c 2c 00 00 02 00
    [...]
    track:lout lba: 105518 ( 422072) 23:28:68 adr: 1 control: 4 mode: -1

    This is probably the CD-TAO-Read-Ahead bug.
    It can be avoided by burning the CD with write type SAO (wodim option -sao).


    Most CD drives report the media size including the two TAO Run-Out blocks, which are not readable by SCSI command 28h READ(10). This is due to an ambiguity of SCSI/MMC specs for command 25h READ CAPACITY which only some
    few drives interpret like Linux expects.
    See my futile attempt to bring it to the attention of linux-scsi
    http://marc.info/?l=linux-scsi&m=145666692729714&w=2

    The remedy would be reading the last two blocks always separately from
    previous blocks and to interpret their failure as End-Of-Medium if they are
    not ASC 0x11 (read error) or ASC 0x15 (positioning error). No reliable prediction of the readability of these sectors seems possible.


    (I could need contact to a kernel developer for this and some other
    problems like:
    - Concurrent burn slowdown
    http://marc.info/?l=linux-scsi&m=135705061804384&w=2
    - Failure to wait for loaded tray to become ready
    - ioctl(BLKRRPART) refusing to re-assess CD/DVD/BD size after burning
    - undocumented old SCSI command feature used to determine start of last
    session.
    )


    Have a nice day :)

    Thomas

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Thomas Schmitt@21:1/5 to Meelis Roos on Tue Feb 19 12:40:01 2019
    Hi,

    i had a second look at the details and would now bet on blkid cautiously
    trying to read the end of the medium for things like GPT backup block.
    If this is true, then the kernel log message is harmless.


    Meelis Roos wrote:
    So it seems something else in the boot process tries to read the very end of the CD and fails. But nothing depends on it in practice.

    About 3 years ago, Archlinux shipped in its ISOs a version of blkid which wanted to access the last two blocks of the medium. I discussed this with
    Karel Zak in spring 2016. He changed blkid to do something like

    $ LIBBLKID_DEBUG=all ./blkid -p /dev/sr2
    ...
    17182: libblkid: LOWPROBE: CDROM: read sector 1436248 failed Input/output error
    17182: libblkid: LOWPROBE: CDROM: reduce size from 735363072 to 735358976.
    17182: libblkid: LOWPROBE: ready for low-probing, offset=0, size=735358976
    ...


    Sector numbers are strange though - 422072 sectors in tack but 422064 is causing problems - maybe because of Linux doing bigger requests.

    The reported READ(10) command requested only two blocks:
    [ 36.019042] sr 0:0:6:0: [sr0] tag#46 CDB: opcode=0x28 28 00 00 01 9c 2c 00 00 02 00
    Starting Logical Block address is 0x019c2c = 105516. Transfer Length is 0x002.

    So it is not the classical "Read-Ahead" situation but would well match
    the behavior of blkid.
    Assumed that your blkid is already fixed, this error report would be the harmless trace of blkid's cautious read attempt.


    dd with bs=2048 reads
    105516+0 records in
    105516+0 records out
    and 105516 < 105518. Why 105516?

    With older kernels you would get an I/O error. With newer kernels there
    is an attempt to catch the Read-Ahead bug which sometimes works and sometimes does not. In
    https://marc.info/?l=linux-scsi&m=145666692729714&w=2
    i wrote
    "I do not find the promise for the "error_sector" information in SPC-3
    or MMC-5. [...]
    The correction in sr.c seems (empirically) to work with TAO CD tracks of
    which the number of blocks is divisible by 2. All valid blocks can be
    read by read(2) and mmap(2).
    Nevertheless ioctl(BLKGETSIZE) still returns the size as deduced from
    READ CAPACITY.
    So young blkid explicitly demands the two unreadble blocks and thus fails."


    Have a nice day :)

    Thomas

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Meelis Roos@21:1/5 to All on Thu Mar 21 16:50:01 2019
    I tested debian-ports install on HP 9000 A180c (the last 32-bit server with parisc).

    Thanks for testing!

    First I needed to find external SCSI CD-ROM with 512-byte block size support -
    I managed but netboot images would have been much simpler (same goes for sparc64).


    Adrian mentioned to me, that netboot images are always in the d-i-tarball, e.g.:
    http://ftp.ports.debian.org/debian-ports/pool-hppa/main/d/debian-installer/debian-installer-images_20190118_hppa.tar.gz
    I haven't tried myself.

    I tried that now that I found another (faster) disk to try install on.

    The image boots fine, loads its components etc until it gets to choosing mirror.
    I entered ftp.ports.debian.org and /devian-ports/, it did not find buster but I selected unstable. It still fails, with the following in the log:

    Mar 21 15:13:16 anna[1165]: WARNING **: bad d-i Packages file
    Mar 21 15:13:16 anna[1165]: 2019-03-21 15:13:16 URL:http://ftp.ports.debian.org/debian-ports//dists/unstable/Release [
    Mar 21 15:13:17 anna[1165]: 2019-03-21 15:13:17 URL:http://ftp.ports.debian.org/debian-ports//dists/unstable/Release.g
    Mar 21 15:13:17 net-retriever: gpgv: Signature made Thu Mar 21 13:58:29 2019 UTC
    Mar 21 15:13:17 net-retriever:
    Mar 21 15:13:17 net-retriever: gpgv: using RSA key 66571731B5A71F91C501F3FDDA1B2CEA81DCBC61
    Mar 21 15:13:17 net-retriever: gpgv: Can't check signature: No public key
    Mar 21 15:13:17 net-retriever:
    Mar 21 15:13:17 net-retriever: error: Bad signature on /tmp/net-retriever-1200-Release.

    Any idea how to add another key there? apt-key and gpg seem to be missing from the netinst miniroot.

    # cd /usr/share/keyrings/
    /usr/share/keyrings # ls -l
    lrwxrwxrwx 1 root root 26 Mar 21 15:12 archive.gpg -> debian-archive-keyring.gpg
    -rw-r--r-- 1 root root 5106 Oct 28 17:22 debian-archive-jessie-automatic.gpg
    -rw-r--r-- 1 root root 5115 Oct 28 17:22 debian-archive-jessie-security-automatic.gpg
    -rw-r--r-- 1 root root 2763 Oct 28 17:22 debian-archive-jessie-stable.gpg
    -rw-r--r-- 1 root root 30142 Oct 28 17:22 debian-archive-keyring.gpg
    -rw-r--r-- 1 root root 7443 Oct 28 17:22 debian-archive-stretch-automatic.gpg
    -rw-r--r-- 1 root root 7452 Oct 28 17:22 debian-archive-stretch-security-automatic.gpg
    -rw-r--r-- 1 root root 2263 Oct 28 17:22 debian-archive-stretch-stable.gpg

    --
    Meelis Roos <mroos@linux.ee>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)