Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 13 May 2015 07:46:16 +0000
From:      bugzilla-noreply@freebsd.org
To:        freebsd-threads@FreeBSD.org
Subject:   [Bug 200138] [PATCH] Fixed per-thread 'specific' array allocation conflict incurred in libthr by some foreign malloc libraries
Message-ID:  <bug-200138-16-y46bOLKmsb@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-200138-16@https.bugs.freebsd.org/bugzilla/>
References:  <bug-200138-16@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=200138

--- Comment #17 from Konstantin Belousov <kib@FreeBSD.org> ---
(In reply to yuri from comment #16)
I am not sure what 'mapped block of size 128K you mean.  A new thread gets
allocated the stack, which has 1M or 2M size, and struct pthread of size 640B
on 32bit arches.  I already proposed to put the keys table into the struct
pthread, which you did not liked.

I still think that making struct pthread and keys live together is the best
route if malloc() has to be avoided from pthread_setspecific().  I am also fine
with the mmap() patch, despite the fact that some portion of the whole page is
wasted.  I do not want to write a special allocator for this single purpose.

-- 
You are receiving this mail because:
You are the assignee for the bug.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-200138-16-y46bOLKmsb>