Created attachment 11137 [details]
Patch that went into master
Move level 0 catch-all debug statement from behind a switch clause inside the switch clause as default case.
This fixes an issue that resulted in the log being flooded with those messages in case someone puts a file with an illegal UTF8 encoding (eg '\xA0test') in a toplevel directory:
[2015/05/12 18:17:26.865144, 0] ../lib/util/charset/convert_string.c:438(convert_string_talloc_handle)
Conversion error: Incomplete multibyte sequence(<A0>test)
There's already an explicit EILSEQ case with a log message at loglevel 3, so afaict the level 0 log statement after the switch clause doesn't make sense.
Otoh, I'd think it would be worthwhile logging a useful diagnostic with the full file patch which is what patch 2/2 does.
As a result, with both patches in place we'd write this to the logs:
[2015/06/09 13:18:51.429654, 1, pid=3461] ../source3/smbd/trans2.c:2379(smbd_dirptr_lanman2_entry)
Conversion error: illegal character: dir/test?
Created attachment 11140 [details]
Patch for 4.2 cherry-picked from master
No patch for 4.1 as that branch misses prerequisite fixes introduced in bug 10775. :)
Re-assigning to Karolin for inclusion in 4.2.next.
Pushed to autobuild-v4-2-test.
(In reply to Karolin Seeger from comment #3)
Pushed to v4-2-test.
Closing out bug report.