Bug 167977 - [smbfs] mount_smbfs results are differ when utf-8 or UTF-8 local encoding's name is used
Summary: [smbfs] mount_smbfs results are differ when utf-8 or UTF-8 local encoding's n...
Status: Closed FIXED
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: 9.0-RELEASE
Hardware: Any Any
: Normal Affects Only Me
Assignee: John Baldwin
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2012-05-16 20:40 UTC by Alexey Kovalenko
Modified: 2014-12-24 17:02 UTC (History)
4 users (show)

See Also:

Attachments
kiconv_case.patch (1.10 KB, patch)
2014-06-05 13:52 UTC, John Baldwin 		no flags Details | Diff
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Alexey Kovalenko 2012-05-16 20:40:01 UTC 
When .nsmbrc contains

[default]
charsets=UTF-8:cp866

remote (windows 2000) filenames with "№" character are displayed ok
( "123№123" ), but when 

[default]
charsets=utf-8:cp866

(note lo-case) filenames are garbaged ( same name as above wiil be "123?"). 

Thereis no SMBFS nor NETSMB options in kernel config
options LIBICONV is.

"№" is 3-byte in UTF-8. XLAT16_ACCEPT_3BYTE_CHR isn't used at build time.

ldserver# smbutil -v lc
SMB connections:
VC: \\SSS\USER
(root:wheel) 711
    state:    Session active
    flags:    0x0428 <private|encpwd>
    usecount: 1
    dialect:  6 (NT LM 0.12, Windows for Workgroups 3.1a, NT LANMAN 1.0)
    smode:    3
    caps:     0xf3fd <SMB_CAP_RAW_MODE|SMB_CAP_UNICODE|SMB_CAP_LARGE_FILES|SMB_CAP_NT_SMBS|SMB_CAP_NT_FIND>
    maxmux:   50
    maxvcs:   1
    Share:    D$(root:wheel) 711
        flags:    0x0004 <>
        usecount: 1

How-To-Repeat: mount remote smb share, look at files with "&#8470;" in name
Comment 1 Mark Linimon freebsd_committer freebsd_triage 2012-05-16 21:22:26 UTC 
Responsible Changed
From-To: freebsd-bugs->freebsd-fs

Over to maintainer(s).
Comment 2 buganini 2012-05-16 22:08:14 UTC 
How about change line 105 and 107 in /usr/src/sys/libkern/iconv_ucs.c
from strcmp to strcasecmp ?

Regards,
Buganini
Comment 3 Semen Soldatov 2014-06-05 12:15:25 UTC 
(In reply to buganini from comment #2)
> How about change line 105 and 107 in /usr/src/sys/libkern/iconv_ucs.c
> from strcmp to strcasecmp ?
> 
> Regards,
> Buganini

with strcasecmp all works ok.
Comment 4 John Baldwin freebsd_committer freebsd_triage 2014-06-05 13:52:46 UTC 
Created attachment 143407 [details]
kiconv_case.patch

I think this is correct as iconv(1) accepts case-insensitive encoding names.  I think there is one other place that should also use strcasecmp() (when looking to see if we already have an existing encoding mapping that can be reused).
Comment 5 Semen Soldatov 2014-06-06 04:38:31 UTC 
(In reply to John Baldwin from comment #4)
> Created attachment 143407 [details]
> kiconv_case.patch
> 
> I think this is correct as iconv(1) accepts case-insensitive encoding names.
> I think there is one other place that should also use strcasecmp() (when
> looking to see if we already have an existing encoding mapping that can be
> reused).

Thx, i set strcasecmp in iconv_ucs.c and utf8 encoding map works ok. if i set koi8-r without patch strcasecmp in iconv.c then filename broken. your patch work fine.
Comment 6 commit-hook freebsd_committer freebsd_triage 2014-06-09 19:28:00 UTC 
A commit references this bug:

Author: jhb
Date: Mon Jun  9 19:27:48 UTC 2014
New revision: 267291
URL: http://svnweb.freebsd.org/changeset/base/267291

Log:
  Use strcasecmp() instead of strcmp() when checking user-supplied encoding
  names so that encoding names are treated as case-insensitive.  This allows
  the use of 'utf-8' instead of 'UTF-8' for example and matches the behavior
  of iconv(1).

  PR:		167977
  Submitted by:	buganini@gmail.com
  MFC after:	1 week

Changes:
  head/sys/libkern/iconv.c
  head/sys/libkern/iconv_ucs.c
Comment 7 John Baldwin freebsd_committer freebsd_triage 2014-06-09 19:29:22 UTC 
Committed to HEAD, thanks!
Comment 8 commit-hook freebsd_committer freebsd_triage 2014-06-27 20:40:05 UTC 
A commit references this bug:

Author: jhb
Date: Fri Jun 27 20:39:46 UTC 2014
New revision: 267980
URL: http://svnweb.freebsd.org/changeset/base/267980

Log:
  MFC 267291:
  Use strcasecmp() instead of strcmp() when checking user-supplied encoding
  names so that encoding names are treated as case-insensitive.  This allows
  the use of 'utf-8' instead of 'UTF-8' for example and matches the behavior
  of iconv(1).

  PR:		167977

Changes:
_U  stable/10/
  stable/10/sys/libkern/iconv.c
  stable/10/sys/libkern/iconv_ucs.c
_U  stable/9/sys/
  stable/9/sys/libkern/iconv.c
  stable/9/sys/libkern/iconv_ucs.c