Re: "Calculated and written ECMA-119 tree end differ" under very specific circumstances

[Thomas Schmitt]

Hi,

Valtteri Vuorikoski wrote:
> Hi, looks like "-for_backup -xattr on" + zisofs completes successfully.

A good hint for the bug hunt.

But the current 1.5.5 seems to be dangerous.
I advise you to try xorriso-1.5.4 with "-for_backup -xattr on".

I am using -for_backup 5 times a day. Two of them are with zisofs.
The buggy state served me fine for about 6 years.
Now the semi-corrected state of xorriso-1.5.5 fails after only three tries.

---------------------------------------------------------------------------

My 11 o'clock backup (without zisofs) just spoiled its new track on BD-R:

  libisofs: WARNING : Calculated and written ECMA-119 tree end differ: 11150384 
<> 11150383

(If i only could test for this before writing happens. Grrr.)

The program ended early and warned
  libburn : WARNING : Last session on media is still open.

But inspection of the medium (without eject and reload) yielded no new
session at all.

Nevertheless, the next attempt to do the backup by xorriso-1.5.4 failed
with
  libburn : NOTE : Closing open session before writing new one
  ...
  libburn : WARNING : Upcoming track: No Next-Writable-Address
  ...
  libburn : FATAL : SCSI error on write(0,32): [5 21 00] Illegal request. Lba 
out of range.
  libburn : FATAL : CDB= WRITE(10) : 2a 00 00 00 00 00 00 00 20 00  : 
dxfer_len= 65536

Afterwards there was a new session #161 on the BD-R, obviously bearing the
start of the ISO filesystem, but being too small for the filesystem size.

This state was suitable for a xorriso-1.5.4 run with "-load session 160"
to get a neat base for the -update_r commands. The result passed a test by

  xorriso -for_backup -indev /dev/sr4 -check_md5_r sorry / --

But now there is an ugly stain on the Table-of-content:
  ...
  ISO session  : 159 ,  11042112 ,     66263s , HOME_2021_01_31_120204
  ISO session  : 160 ,  11108544 ,     31474s , HOME_2021_02_01_110638
  xorriso : WARNING : Session 161 bears ISO image size 38089s larger than track 
size 5504s
  ISO session  : 161 ,  11140192 ,      5504s , HOME_2021_02_02_111521
  ISO session  : 162 ,  11145696 ,     38089s , HOME_2021_02_02_112447
  Media summary: 162 sessions, 11183936 data blocks, 21.3g data, 2022m free
  Media nwa    : 11183936s

With a BD-RE i could polish it away by

  xorriso -outdev /dev/sr4 -truncate_overwritable session 160 new

and then doing a fresh update session afterwards.
On the BD-R it will stay forever and remind me of my clumsiness combined
with my courage to do my backups with what i upload to the web.
(I have 5 media in use for daily backups. So i hope to learn about bugs
before all of them are unrepairable.)


Have a nice day :)

Thomas