[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: glibc packages up for testing



Hello!

On Thu, Jul 05, 2007 at 06:34:03PM +0200, Michael Banck wrote:
> I've built glibc-2.5 with gcc-4.1 now, using Thomas Schwinge's patches.
> The result is a
> deb http://people.debian.org/~mbanck/hurd-glibc/ ./
> 
> I've tested them briefly, if nobody reports issues with them

I could only do very light testing, but no problems so far.

> I will upload them to gnuab.

Go ahead, I'd say.

I think I updated that glibc patch for making it work with GCC 4.1 some
more to avoid a compiler warning.  I'll see to publish that patch
somewhen soon.  But don't let that stop you.

> whether the Hurd built against those boots up fine.

Done that, also only very light testing, but as well no problems so far.


Also, in the mean time you published glibc packages that have been built
without forcefully disabling ``-z relro''.  Again, only (even more) light
testing, but as well no problems found.

Also built a Hurd package against that one and it also works.  (I.e. it
booted...)

However: this does not necessarily mean that the ``relro'' stuff
magically fixed itself, because I seem to remember that, at the times
when we noticed that breakage for the first time -- which may have been
one and a half years ago or so, at the time when Debian GNU/hurd was
upgraded from using GNU binutils version (I think) 2.15 to 2.16 -- this
breakage also wasn't visible with the standard Debian glibc packages, but
only when using the unstripped library files for linking!  (And only with
`ext2fs.static' or (perhaps) some other statically linked executables
that are using cthreads.)  So I can't advice whether Debian (Hurd) should
assume that the ``-z relro'' enabled glibc packages will always work or
not.


Regards,
 Thomas

Attachment: signature.asc
Description: Digital signature


Reply to: