This is the mail archive of the libc-hacker@sources.redhat.com mailing list for the glibc project.

Note that libc-hacker is a closed list. You may look at the archives of this list, but subscription and posting are not open.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]

Re: [ia64-tools] Function descriptor and dlopen/dlclose under ia64.

To: ia64-tools at napali dot hpl dot hp dot com
Subject: Re: [ia64-tools] Function descriptor and dlopen/dlclose under ia64.
From: "H . J . Lu" <hjl at valinux dot com>
Date: Mon, 11 Sep 2000 23:46:02 -0700
Cc: GNU C Library <libc-hacker at sourceware dot cygnus dot com>
References: <20000911193218.A29487@valinux.com>

On Mon, Sep 11, 2000 at 07:32:18PM -0700, H . J . Lu wrote:
> I may know why "make check" failed in iconvdata. The failed testcase
> is tst-loading. It does 5000 iterations of random dlopen/dlclose.
> The first few iterations are ok. It fails under ia64 after a few
> hundreds of iterations.
> 
> Remember the official function descriptor, which is created at the
> fly by the dynamic linker. All DSOes use the same official function
> descriptor. If an official function descriptor is created by dloepn a
> DSO, we cannot free it via dlclose if it is used by another DSO later
> on. Basically, we need a reference count on the official function
> descriptor. Otherwise, we will be in deep trouble with dlopen/dlclose.
> 

Apparently, for some reason, the ia64 port in glibc 2.2 left out some
stuff in my working ia64 glibc 2.1 port. I will take a closer look at
the ia64 port in glibc 2.2 and provide a patch when I find time.


-- 
H.J. Lu (hjl@gnu.org)

Follow-Ups:
- Re: [ia64-tools] Function descriptor and dlopen/dlclose under ia64.
  - From: Jes Sorensen

References:
- Function descriptor and dlopen/dlclose under ia64.
  - From: H . J . Lu

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]