musl 1.1.10 uselocale regression, patch

musl 1.1.10 uselocale regression, patch

[Rich Felker]

[-- Attachment #1: Type: text/plain, Size: 105 bytes --]

A new regression in uselocale made it into musl 1.1.10. All users
should apply the attached patch.

Rich

[-- Attachment #2: musl-1.1.10-uselocale-0-fix.diff --]
[-- Type: text/plain, Size: 827 bytes --]

From 63f4b9f18f3674124d8bcb119739fec85e6da005 Mon Sep 17 00:00:00 2001
From: Timo Teräs <timo.teras@iki.fi>
Date: Fri, 05 Jun 2015 07:39:42 +0000
Subject: fix uselocale((locale_t)0) not to modify locale

commit 68630b55c0c7219fe9df70dc28ffbf9efc8021d8 made the new locale to
be assigned unconditonally resulting in crashes later on.
---
diff --git a/src/locale/uselocale.c b/src/locale/uselocale.c
index b70a0c1..0fc5ecb 100644
--- a/src/locale/uselocale.c
+++ b/src/locale/uselocale.c
@@ -8,9 +8,7 @@ locale_t __uselocale(locale_t new)
 	locale_t old = self->locale;
 	locale_t global = &libc.global_locale;
 
-	if (new == LC_GLOBAL_LOCALE) new = global;
-
-	self->locale = new;
+	if (new) self->locale = new == LC_GLOBAL_LOCALE ? global : new;
 
 	return old == global ? LC_GLOBAL_LOCALE : old;
 }
--
cgit v0.9.0.3-65-g4555

Re: musl 1.1.10 uselocale regression, patch

[Anthony G. Basile]

On 6/5/15 12:26 PM, Rich Felker wrote:
> A new regression in uselocale made it into musl 1.1.10. All users
> should apply the attached patch.
>
> Rich
>

I'm skipping 1.1.9 for all the gentoo stages (mostly because the mips 
bug was a show stopper).  I can backport this patch, but is there a new 
release on the horizon given that this is a significant bug?

-- 
Anthony G. Basile, Ph. D.
Chair of Information Technology
D'Youville College
Buffalo, NY 14201
(716) 829-8197

Re: musl 1.1.10 uselocale regression, patch

[Rich Felker]

On Sat, Jun 06, 2015 at 11:44:55AM -0400, Anthony G. Basile wrote:
> On 6/5/15 12:26 PM, Rich Felker wrote:
> >A new regression in uselocale made it into musl 1.1.10. All users
> >should apply the attached patch.
> 
> I'm skipping 1.1.9 for all the gentoo stages (mostly because the
> mips bug was a show stopper).  I can backport this patch, but is
> there a new release on the horizon given that this is a significant
> bug?

I'm not planning to insert another release cycle to deal with it. If
users are unhappy with having to download and apply the patch
separately, I might make (and sign) a 1.1.10a tarball for convenience.
Let me know if that's something you want to see happen.

Rich

Re: musl 1.1.10 uselocale regression, patch

[Anthony G. Basile]

On 6/6/15 12:31 PM, Rich Felker wrote:
> On Sat, Jun 06, 2015 at 11:44:55AM -0400, Anthony G. Basile wrote:
>> On 6/5/15 12:26 PM, Rich Felker wrote:
>>> A new regression in uselocale made it into musl 1.1.10. All users
>>> should apply the attached patch.
>>
>> I'm skipping 1.1.9 for all the gentoo stages (mostly because the
>> mips bug was a show stopper).  I can backport this patch, but is
>> there a new release on the horizon given that this is a significant
>> bug?
>
> I'm not planning to insert another release cycle to deal with it. If
> users are unhappy with having to download and apply the patch
> separately, I might make (and sign) a 1.1.10a tarball for convenience.
> Let me know if that's something you want to see happen.
>
> Rich
>

Nope.  I don't need that.  The point is if a new release was immenant I 
would just wait because there's some work invovled in building and 
pushing these stages out, and I like doing one for each release of musl.

So we're good :)

-- 
Anthony G. Basile, Ph. D.
Chair of Information Technology
D'Youville College
Buffalo, NY 14201
(716) 829-8197