NetBSD Problem Report #57798
From ryo@tetera.org Wed Dec 27 12:44:36 2023
Return-Path: <ryo@tetera.org>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
(using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits))
(Client CN "mail.NetBSD.org", Issuer "mail.NetBSD.org CA" (not verified))
by mollari.NetBSD.org (Postfix) with ESMTPS id 7332A1A9238
for <gnats-bugs@gnats.NetBSD.org>; Wed, 27 Dec 2023 12:44:36 +0000 (UTC)
Message-Id: <4e184a7f8c1ccb4adb263909604ea124.ryo@tetera.org>
Date: Wed, 27 Dec 2023 21:44:29 +0900
From: ryo@tetera.org
Reply-To: ryo@tetera.org
To: gnats-bugs@NetBSD.org
Subject: With src/share/locale/ctype/en_US.UTF-8.src, wcwidth() returns 3 when ja_JP.UTF-8 locale is used
X-Send-Pr-Version: 3.95
>Number: 57798
>Category: lib
>Synopsis: With src/share/locale/ctype/en_US.UTF-8.src, wcwidth() returns 3 when ja_JP.UTF-8 locale is used
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: lib-bug-people
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Dec 27 12:45:00 +0000 2023
>Closed-Date: Sun Jan 14 16:48:06 +0000 2024
>Last-Modified: Sun Jan 14 16:48:06 +0000 2024
>Originator: Ryo ONODERA
>Release: NetBSD 10.99.10
>Organization:
Ryo ONODERA // ryo@tetera.org
PGP fingerprint = 82A2 DC91 76E0 A10A 8ABB FD1B F404 27FA C7D1 15F3
>Environment:
System: NetBSD castella 10.99.10 NetBSD 10.99.10 (DTRACE9) #0: Mon Dec 25 05:18:50 JST 2023 ryoon@castella:/usr/world/10.99/amd64/obj/sys/arch/amd64/compile/DTRACE9 amd64
Architecture: x86_64
Machine: amd64
>Description:
The following test program prints 3 for 兆 character.
=== === === === === === === === === === === === === === === ===
#include <stdio.h>
#include <wchar.h>
#include <stdlib.h>
#include <locale.h>
int
main(void)
{
const char *str = "兆";// -> 3, should be 2
setlocale(LC_ALL, "ja_JP.UTF-8");
//setlocale(LC_ALL, "C");
wchar_t cp;
int len = mbtowc(&cp, str, 5);
if (len < 1)
return -1;
int w = wcwidth(cp);
printf("w = %d\n", w);
return 0;
}
=== === === === === === === === === === === === === === === ===
TODIGIT < 0x5146 1000000000000 >
in
src/share/locale/ctype/en_US.UTF-8.src
was introduced in r1.8 and it seems that it makes wcwidth of 兆 character as 3.
>How-To-Repeat:
>Fix:
I have no idea.
>Release-Note:
>Audit-Trail:
From: Rin Okuyama <rokuyama.rk@gmail.com>
To: gnats-bugs@netbsd.org, lib-bug-people@netbsd.org, gnats-admin@netbsd.org,
netbsd-bugs@netbsd.org
Cc:
Subject: Re: lib/57798: With src/share/locale/ctype/en_US.UTF-8.src, wcwidth()
returns 3 when ja_JP.UTF-8 locale is used
Date: Wed, 27 Dec 2023 22:52:04 +0900
兆 (U+5146) is a Chinese symbol for 1,000,000,000,000.
I've confirmed that wcwidth(3) wrongly returns 3 for U+5146.
while it correctly returns 2 for 億 (U+5140: 100,000,000).
wcwidth(3) also returns 3 for U+16B60 (a Pahawn Hmong
character for 10,000,000,000).
These failures should be due to broken TODIGIT supports in
mklocale(1). Its man page says:
> TODIGIT Defines a map from runes to their digit value.
> (snip) Only values up to 255 are allowed.
OpenBSD has already dropped support to TODIGIT from mklocale(1):
https://github.com/OpenBSD/src/commit/4efe9bdeb34
If this commit is mechanically applied to netbsd-10,
wcwidth(3) correctly reports 2 for U+5146.
I will commit it and send a pullup request to netbsd-10,
if there's no objections.
Thanks,
rin
From: Valery Ushakov <uwe@stderr.spb.ru>
To: Rin Okuyama <rokuyama.rk@gmail.com>
Cc: gnats-bugs@netbsd.org
Subject: Re: lib/57798: With src/share/locale/ctype/en_US.UTF-8.src,
wcwidth() returns 3 when ja_JP.UTF-8 locale is used
Date: Wed, 27 Dec 2023 16:52:28 +0300
On Wed, Dec 27, 2023 at 22:52:04 +0900, Rin Okuyama wrote:
> https://github.com/OpenBSD/src/commit/4efe9bdeb34
>
> If this commit is mechanically applied to netbsd-10,
> wcwidth(3) correctly reports 2 for U+5146.
Please, can you leave a comment somewhere around the new "DIGITMAP
mapignore" rule that mentions that we now ignore this information, so
that it's obvious without consulting the version history?
And of course s/Ox/Nx/ in the man page diff.
Thanks!
-uwe
From: "Rin Okuyama" <rin@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/57798 CVS commit: src/usr.bin/mklocale
Date: Thu, 28 Dec 2023 03:49:35 +0000
Module Name: src
Committed By: rin
Date: Thu Dec 28 03:49:35 UTC 2023
Modified Files:
src/usr.bin/mklocale: mklocale.1 yacc.y
Log Message:
mklocale: XXX: Neglect TODIGIT at the moment
PR lib/57798
It was implemented with an assumption that all digit characters
can be mapped to numerical values <= 255.
This is no longer true for Unicode, and results in, e.g., wrong
return values of wcwidth(3) for U+5146 or U+16B60.
As a workaround, neglect TODIGIT for now, as done for OpenBSD:
https://github.com/OpenBSD/src/commit/4efe9bdeb34
XXX
At least netbsd-10 should be fixed, but it requires some tests.
To generate a diff of this commit:
cvs rdiff -u -r1.17 -r1.18 src/usr.bin/mklocale/mklocale.1
cvs rdiff -u -r1.34 -r1.35 src/usr.bin/mklocale/yacc.y
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: Rin Okuyama <rokuyama.rk@gmail.com>
To: Valery Ushakov <uwe@stderr.spb.ru>, ryo@tetera.org
Cc: gnats-bugs@netbsd.org, netbsd-bugs@NetBSD.org
Subject: Re: lib/57798: With src/share/locale/ctype/en_US.UTF-8.src, wcwidth()
returns 3 when ja_JP.UTF-8 locale is used
Date: Thu, 28 Dec 2023 12:57:08 +0900
On 2023/12/27 22:52, Valery Ushakov wrote:
> On Wed, Dec 27, 2023 at 22:52:04 +0900, Rin Okuyama wrote:
>
>> https://github.com/OpenBSD/src/commit/4efe9bdeb34
>>
>> If this commit is mechanically applied to netbsd-10,
>> wcwidth(3) correctly reports 2 for U+5146.
>
> Please, can you leave a comment somewhere around the new "DIGITMAP
> mapignore" rule that mentions that we now ignore this information, so
> that it's obvious without consulting the version history?
>
> And of course s/Ox/Nx/ in the man page diff.
Thank you uwe@ for careful review!
I've committed it to -current. Let us see what happens for a while.
It would be nice if we can pull this up (in some form) to netbsd-10,
but I'm afraid there's no enough time left before 10.0 release...
ryoon@, do you have some ideas to test this change? For what
application have you encountered this problem? Is it fixed now
without regression?
Thanks,
rin
From: Ryo ONODERA <ryo@tetera.org>
To: Rin Okuyama <rokuyama.rk@gmail.com>, Valery Ushakov <uwe@stderr.spb.ru>
Cc: gnats-bugs@netbsd.org
Subject: Re: lib/57798: With src/share/locale/ctype/en_US.UTF-8.src,
wcwidth() returns 3 when ja_JP.UTF-8 locale is used
Date: Thu, 28 Dec 2023 13:27:53 +0900
Hi,
Rin Okuyama <rokuyama.rk@gmail.com> writes:
> On 2023/12/27 22:52, Valery Ushakov wrote:
>> On Wed, Dec 27, 2023 at 22:52:04 +0900, Rin Okuyama wrote:
>>
>>> https://github.com/OpenBSD/src/commit/4efe9bdeb34
>>>
>>> If this commit is mechanically applied to netbsd-10,
>>> wcwidth(3) correctly reports 2 for U+5146.
>>
>> Please, can you leave a comment somewhere around the new "DIGITMAP
>> mapignore" rule that mentions that we now ignore this information, so
>> that it's obvious without consulting the version history?
>>
>> And of course s/Ox/Nx/ in the man page diff.
>
> Thank you uwe@ for careful review!
>
> I've committed it to -current. Let us see what happens for a while.
>
> It would be nice if we can pull this up (in some form) to netbsd-10,
> but I'm afraid there's no enough time left before 10.0 release...
>
> ryoon@, do you have some ideas to test this change? For what
> application have you encountered this problem? Is it fixed now
> without regression?
TODIGIT for U+5146 and similar Kanji characters are always useless
for me.
I have forgotten to mention in the original PR, tmux in NetBSD base
(/usr/bin/tmux) is affected by this problem.
If you set LANG=ja_JP.UTF-8 and start tmux and display $BC{(B character
in the tmux window, tmux exits unexpectedly with '[server exited unexpectedly]'
error message. And all tmux sessions will be lost.
Recent tmux assumes wcwidth always returns <= 2.
See: 'ud->width >2' in utf8_from_data() in src/external/bsd/tmux/dist/utf8.c
/* Get UTF-8 character from data. */
enum utf8_state
utf8_from_data(const struct utf8_data *ud, utf8_char *uc)
{
u_int index;
if (ud->width > 2)
fatalx("invalid UTF-8 width: %u", ud->width);
Thank you.
> Thanks,
> rin
--
Ryo ONODERA // ryo@tetera.org
PGP fingerprint = 82A2 DC91 76E0 A10A 8ABB FD1B F404 27FA C7D1 15F3
From: Brett Lymn <blymn@internode.on.net>
To: gnats-bugs@netbsd.org
Cc: lib-bug-people@netbsd.org, gnats-admin@netbsd.org, netbsd-bugs@netbsd.org,
ryo@tetera.org
Subject: Re: lib/57798: With src/share/locale/ctype/en_US.UTF-8.src,
wcwidth() returns 3 when ja_JP.UTF-8 locale is used
Date: Fri, 29 Dec 2023 07:20:32 +1030
On Thu, Dec 28, 2023 at 04:00:03AM +0000, Rin Okuyama wrote:
>
> ryoon@, do you have some ideas to test this change? For what
> application have you encountered this problem? Is it fixed now
> without regression?
>
Our wide curses relies on wcwidth to determine cursor positioning and call widths, so any
curses based application attempting to display these characters will have a corrupted
display.>
--
Brett Lymn
--
Sent from my NetBSD device.
"We are were wolves",
"You mean werewolves?",
"No we were wolves, now we are something else entirely",
"Oh"
From: Valery Ushakov <uwe@stderr.spb.ru>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: PR/57798 CVS commit: src/usr.bin/mklocale
Date: Fri, 29 Dec 2023 00:16:12 +0300
On Thu, Dec 28, 2023 at 03:50:01 +0000, Rin Okuyama wrote:
> It was implemented with an assumption that all digit characters
> can be mapped to numerical values <= 255.
Unicode has three different "numeric" values for a character
Unicode Character Database
https://unicode.org/reports/tr44/
Numeric_Value is extracted based on the actual numeric value of the
data in field 8 of UnicodeData.txt or the values of the
kPrimaryNumeric, kAccountingNumeric, or kOtherNumeric tags, for
characters listed in the Unihan data files.
Numeric_Type is extracted as follows. If fields 6, 7, and 8 in
UnicodeData.txt are all non-empty, then Numeric_Type=Decimal.
Otherwise, if fields 7 and 8 are both non-empty, then
Numeric_Type=Digit. Otherwise, if field 8 is non-empty, then
Numeric_Type=Numeric. For characters listed in the Unihan data
files, Numeric_Type=Numeric for characters that have
kPrimaryNumeric, kAccountingNumeric, or kOtherNumeric tags. The
default value is Numeric_Type=None.
The intention of TODIGIT is likely to be able to eventually provide
support for something like LC_TIME's alt_digits or glibc printf(3)
extension that provides 'I' modifier for %d and friends - that use
locale-specific digits, say u+0f20..u+0f29 for Tibetan/Dzongkha
locales.
But I don't really know much about those areas of locales...
-uwe
From: "Rin Okuyama" <rin@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/57798 CVS commit: src/usr.bin/mklocale
Date: Fri, 5 Jan 2024 02:38:07 +0000
Module Name: src
Committed By: rin
Date: Fri Jan 5 02:38:06 UTC 2024
Modified Files:
src/usr.bin/mklocale: mklocale.1 yacc.y
Log Message:
mklocale(1): Add range check for TODIGIT, rather than disabling it
PR lib/57798
Digit value specified by TODIGIT is storaged as lowest 8 bits of
_RuneType, see lib/libc/locale/runetype_file.h:
https://nxr.netbsd.org/xref/src/lib/libc/locale/runetype_file.h#56
The symptom reported in the PR is due to missing range check for
this value; values of 256 and above were mistakenly treated as
other flag bits in _RuneType.
For example, U+5146 has numerical value 1000,000,000,000 ==
0xe8d4a51000 where __BITS(30, 31) == _RUNETYPE_SW3 are turned on.
This is why wcwidth(3) returned 3 for this character.
This apparently affected not only character width, but also other
attributes storaged in _RuneType.
IIUC, digit value attributes in _RuneType have never been utilized
until now, but preserve these if digit fits within (0, 256). This
should be safer for pulling this up into netbsd-10. Also, these
attributes may be useful to implement some I18N features as
suggested by uwe@ in the PR.
netbsd-[98] is not affected as these use old UTF-8 ctype definitions.
To generate a diff of this commit:
cvs rdiff -u -r1.18 -r1.19 src/usr.bin/mklocale/mklocale.1
cvs rdiff -u -r1.35 -r1.36 src/usr.bin/mklocale/yacc.y
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: Rin Okuyama <rokuyama.rk@gmail.com>
To: gnats-bugs@netbsd.org, lib-bug-people@netbsd.org, gnats-admin@netbsd.org,
netbsd-bugs@netbsd.org, ryo@tetera.org
Cc: ryo@tetera.org, Brett Lymn <blymn@internode.on.net>,
Valery Ushakov <uwe@stderr.spb.ru>, Martin Husemann <martin@duskware.de>
Subject: Re: PR/57798 CVS commit: src/usr.bin/mklocale
Date: Fri, 5 Jan 2024 12:05:53 +0900
I've committed a revised fix to -current.
As wrote in the commit log, the original problem affected not
only wcwidth(3), but also other attributes storaged in _RuneType.
I'll send a pullup request to netbsd-10 tomorrow, if there's no
objections.
On 2023/12/29 6:40, Valery Ushakov wrote:
> Unicode has three different "numeric" values for a character
>
> Unicode Character Database
> https://unicode.org/reports/tr44/
>
> Numeric_Value is extracted based on the actual numeric value of the
> data in field 8 of UnicodeData.txt or the values of the
> kPrimaryNumeric, kAccountingNumeric, or kOtherNumeric tags, for
> characters listed in the Unihan data files.
>
> Numeric_Type is extracted as follows. If fields 6, 7, and 8 in
> UnicodeData.txt are all non-empty, then Numeric_Type=Decimal.
> Otherwise, if fields 7 and 8 are both non-empty, then
> Numeric_Type=Digit. Otherwise, if field 8 is non-empty, then
> Numeric_Type=Numeric. For characters listed in the Unihan data
> files, Numeric_Type=Numeric for characters that have
> kPrimaryNumeric, kAccountingNumeric, or kOtherNumeric tags. The
> default value is Numeric_Type=None.
>
> The intention of TODIGIT is likely to be able to eventually provide
> support for something like LC_TIME's alt_digits or glibc printf(3)
> extension that provides 'I' modifier for %d and friends - that use
> locale-specific digits, say u+0f20..u+0f29 for Tibetan/Dzongkha
> locales.
>
> But I don't really know much about those areas of locales...
Thank you for info. As far as I can see, most of characters in
problem are categorized to Numeric_Type=Numeric, and it seems
difficult to distinguish these with, e.g., [0-9a-f].
On 2023/12/29 5:50, Brett Lymn wrote:
> Our wide curses relies on wcwidth to determine cursor positioning and
call widths, so any
> curses based application attempting to display these characters will
have a corrupted
> display.>
Yeah, /usr/bin/vi gets confused when edit message that contains
U+5146 actually ;)
I've roughly checked output from -d option of mklocale(1).
Width (and other attribute fields) seems fixed now, as far as
I can see.
On 2023/12/28 13:30, Ryo ONODERA wrote:
> See: 'ud->width >2' in utf8_from_data() in
src/external/bsd/tmux/dist/utf8.c
>
> /* Get UTF-8 character from data. */
> enum utf8_state
> utf8_from_data(const struct utf8_data *ud, utf8_char *uc)
> {
> u_int index;
>
> if (ud->width > 2)
> fatalx("invalid UTF-8 width: %u", ud->width);
Oops. I'm not pretty sure whether this is a good programming
practice, but this was actually useful to find out the problem ;)
Thanks,
rin
State-Changed-From-To: open->pending-pullups
State-Changed-By: rin@NetBSD.org
State-Changed-When: Sat, 06 Jan 2024 15:46:22 +0000
State-Changed-Why:
[pullup-10 #538] [SERIOUS] Fix mklocale(1) for lib/57798
https://releng.netbsd.org/cgi-bin/req-10.cgi?show=538
netbsd-[98] are not affected.
From: "Martin Husemann" <martin@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/57798 CVS commit: [netbsd-10] src/usr.bin/mklocale
Date: Sun, 14 Jan 2024 15:15:00 +0000
Module Name: src
Committed By: martin
Date: Sun Jan 14 15:15:00 UTC 2024
Modified Files:
src/usr.bin/mklocale [netbsd-10]: mklocale.1 yacc.y
Log Message:
Pull up following revision(s) (requested by rin in ticket #538):
usr.bin/mklocale/yacc.y: revision 1.35
usr.bin/mklocale/yacc.y: revision 1.36
usr.bin/mklocale/mklocale.1: revision 1.18
usr.bin/mklocale/mklocale.1: revision 1.19
mklocale: XXX: Neglect TODIGIT at the moment
PR lib/57798
It was implemented with an assumption that all digit characters
can be mapped to numerical values <= 255.
This is no longer true for Unicode, and results in, e.g., wrong
return values of wcwidth(3) for U+5146 or U+16B60.
As a workaround, neglect TODIGIT for now, as done for OpenBSD:
https://github.com/OpenBSD/src/commit/4efe9bdeb34
XXX
At least netbsd-10 should be fixed, but it requires some tests.
mklocale(1): Add range check for TODIGIT, rather than disabling it
PR lib/57798
Digit value specified by TODIGIT is storaged as lowest 8 bits of
_RuneType, see lib/libc/locale/runetype_file.h:
https://nxr.netbsd.org/xref/src/lib/libc/locale/runetype_file.h#56
The symptom reported in the PR is due to missing range check for
this value; values of 256 and above were mistakenly treated as
other flag bits in _RuneType.
For example, U+5146 has numerical value 1000,000,000,000 ==
0xe8d4a51000 where __BITS(30, 31) == _RUNETYPE_SW3 are turned on.
This is why wcwidth(3) returned 3 for this character.
This apparently affected not only character width, but also other
attributes storaged in _RuneType.
IIUC, digit value attributes in _RuneType have never been utilized
until now, but preserve these if digit fits within (0, 256). This
should be safer for pulling this up into netbsd-10. Also, these
attributes may be useful to implement some I18N features as
suggested by uwe@ in the PR.
netbsd-[98] is not affected as these use old UTF-8 ctype definitions.
To generate a diff of this commit:
cvs rdiff -u -r1.17 -r1.17.16.1 src/usr.bin/mklocale/mklocale.1
cvs rdiff -u -r1.34 -r1.34.8.1 src/usr.bin/mklocale/yacc.y
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
State-Changed-From-To: pending-pullups->closed
State-Changed-By: riastradh@NetBSD.org
State-Changed-When: Sun, 14 Jan 2024 16:48:06 +0000
State-Changed-Why:
fixed and pulled up to 10, not needed for 9 or 8
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.47 2022/09/11 19:34:41 kim Exp $
$NetBSD: gnats_config.sh,v 1.9 2014/08/02 14:16:04 spz Exp $
Copyright © 1994-2024
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.