* Libc in CVS
@ 1997-11-24 4:41 ralf
1997-11-24 9:46 ` Alan Cox
0 siblings, 1 reply; 11+ messages in thread
From: ralf @ 1997-11-24 4:41 UTC (permalink / raw
To: linux
Hi all,
with the bug fixes I checked in during the last few days libc is now
pretty stable. In particular the problems with wrong relocations in
the X libraries and all libpam clients are working now.
Still to do:
- some of the fixes are not very nice or correct at all
- build binaries for redistribution. Can somebody take care of this?
Given that this opens the way to useable native X libraries, which makes
building a large number of other RPM packages possible, these patches
are quite a breakthough.
Ralf
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 4:41 Libc in CVS ralf
@ 1997-11-24 9:46 ` Alan Cox
1997-11-24 11:05 ` ralf
0 siblings, 1 reply; 11+ messages in thread
From: Alan Cox @ 1997-11-24 9:46 UTC (permalink / raw
To: ralf; +Cc: linux
> Given that this opens the way to useable native X libraries, which makes
> building a large number of other RPM packages possible, these patches
> are quite a breakthough.
With luck it'll also fix the rpm-2.4.8 core dump bug which seems to be
bad linking. I'll turn the RPM factory back on
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 9:46 ` Alan Cox
@ 1997-11-24 11:05 ` ralf
1997-11-24 11:30 ` Alan Cox
0 siblings, 1 reply; 11+ messages in thread
From: ralf @ 1997-11-24 11:05 UTC (permalink / raw
To: Alan Cox; +Cc: linux
On Mon, Nov 24, 1997 at 09:46:35AM +0000, Alan Cox wrote:
> > Given that this opens the way to useable native X libraries, which makes
> > building a large number of other RPM packages possible, these patches
> > are quite a breakthough.
>
> With luck it'll also fix the rpm-2.4.8 core dump bug which seems to be
> bad linking.
My rpm has never dumped core but it was complaining about nonexisting
users while creating the binary rpm file even though the user was
existing thus failing to build. That rpm problem is fixed.
> I'll turn the RPM factory back on
All the source RPM packages on linus are known to build and work for
little endian targets. Some of them, for example ncompress, have special
modifications to support big endian targets.
Ralf
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 11:05 ` ralf
@ 1997-11-24 11:30 ` Alan Cox
1997-11-24 13:18 ` ralf
0 siblings, 1 reply; 11+ messages in thread
From: Alan Cox @ 1997-11-24 11:30 UTC (permalink / raw
To: ralf; +Cc: alan, linux
> > With luck it'll also fix the rpm-2.4.8 core dump bug which seems to be
> > bad linking.
>
> My rpm has never dumped core but it was complaining about nonexisting
> users while creating the binary rpm file even though the user was
> existing thus failing to build. That rpm problem is fixed.
Yep thats the bug - I also got some crashes. It was caused by getpw* failing
in peculiar ways. My RPM binary has its own
> All the source RPM packages on linus are known to build and work for
> little endian targets. Some of them, for example ncompress, have special
> modifications to support big endian targets.
Im actually trying to build a full RedHat 5.0 by early December so Im picking
up your changes to stuff like bash indirectly via redhat on the whole. I can
then merge anything that escaped and together we can make sure they are in
RH5.1.
Obviously as of about Dec1 the amount of answers I can get from RedHat will
noise dive for a couple of months
Alan
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 11:30 ` Alan Cox
@ 1997-11-24 13:18 ` ralf
1997-11-24 14:09 ` Alan Cox
1997-11-24 15:55 ` Erik Troan
0 siblings, 2 replies; 11+ messages in thread
From: ralf @ 1997-11-24 13:18 UTC (permalink / raw
To: Alan Cox; +Cc: linux, linux-mips
On Mon, Nov 24, 1997 at 11:30:41AM +0000, Alan Cox wrote:
> > > With luck it'll also fix the rpm-2.4.8 core dump bug which seems to be
> > > bad linking.
> >
> > My rpm has never dumped core but it was complaining about nonexisting
> > users while creating the binary rpm file even though the user was
> > existing thus failing to build. That rpm problem is fixed.
>
> Yep thats the bug - I also got some crashes. It was caused by getpw* failing
> in peculiar ways. My RPM binary has its own
For rpm the trick is easy, just don't use a static linked binary.
Unfortunately the Redhat guys seem to think static binaries are a good
idea and install a static rpm by default. Which it is not, not even
without a buggy dynamic linker.
> > All the source RPM packages on linus are known to build and work for
> > little endian targets. Some of them, for example ncompress, have special
> > modifications to support big endian targets.
>
> Im actually trying to build a full RedHat 5.0 by early December so Im picking
> up your changes to stuff like bash indirectly via redhat on the whole. I can
> then merge anything that escaped and together we can make sure they are in
> RH5.1.
Ok. I stopped communicating with the Redhat guys shortly before I left the
US. Essentially they received a part of the patches that I don't classify
as ugly hacks or work in progress.
Some more packages like the binutils, gcc or libc are based on different
versions and we should try to base our work on the same versions as Redhat
does.
> Obviously as of about Dec1 the amount of answers I can get from RedHat will
> noise dive for a couple of months
:-)
Ralf
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 13:18 ` ralf
@ 1997-11-24 14:09 ` Alan Cox
1997-11-24 15:55 ` Erik Troan
1 sibling, 0 replies; 11+ messages in thread
From: Alan Cox @ 1997-11-24 14:09 UTC (permalink / raw
To: ralf; +Cc: alan, linux, linux-mips
> For rpm the trick is easy, just don't use a static linked binary.
> Unfortunately the Redhat guys seem to think static binaries are a good
> idea and install a static rpm by default. Which it is not, not even
> without a buggy dynamic linker.
Static rpm 2.4.8 +fixes works
Dynamic 2.4.8 crashes on startup every time
> Ok. I stopped communicating with the Redhat guys shortly before I left the
> US. Essentially they received a part of the patches that I don't classify
> as ugly hacks or work in progress.
No problem
> Some more packages like the binutils, gcc or libc are based on different
> versions and we should try to base our work on the same versions as Redhat
> does.
Right now Ive got chunks of bastardised 4.8 (eg with -O not -O2 to avoid
gcc blowing up on zsh) but hopefully I can knock those out over time
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 13:18 ` ralf
1997-11-24 14:09 ` Alan Cox
@ 1997-11-24 15:55 ` Erik Troan
1997-11-24 17:16 ` ralf
1997-11-24 21:35 ` David S. Miller
1 sibling, 2 replies; 11+ messages in thread
From: Erik Troan @ 1997-11-24 15:55 UTC (permalink / raw
To: linux-mips; +Cc: Alan Cox, linux
On Mon, 24 Nov 1997 ralf@lappi.waldorf-gmbh.de wrote:
> For rpm the trick is easy, just don't use a static linked binary.
> Unfortunately the Redhat guys seem to think static binaries are a good
> idea and install a static rpm by default. Which it is not, not even
> without a buggy dynamic linker.
A static RPM has saved my ass *many* times, and it would irresponsible
for us not to ship it static. Glibc ought to be able to generate
static binaries. If it can't, it's broken.
Erik
-------------------------------------------------------------------------------
| "For the next two hours, VH1 will be filled with foul-mouthed, |
| crossdressing Australians. Viewer discretion is advised." |
| |
| Erik Troan = ewt@redhat.com = ewt@sunsite.unc.edu |
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 15:55 ` Erik Troan
@ 1997-11-24 17:16 ` ralf
1997-11-24 21:35 ` David S. Miller
1 sibling, 0 replies; 11+ messages in thread
From: ralf @ 1997-11-24 17:16 UTC (permalink / raw
To: linux-mips; +Cc: ewt, Alan Cox, linux
On Mon, Nov 24, 1997 at 10:55:25AM -0500, Erik Troan wrote:
> On Mon, 24 Nov 1997 ralf@lappi.waldorf-gmbh.de wrote:
> > For rpm the trick is easy, just don't use a static linked binary.
> > Unfortunately the Redhat guys seem to think static binaries are a good
> > idea and install a static rpm by default. Which it is not, not even
> > without a buggy dynamic linker.
>
> A static RPM has saved my ass *many* times, and it would irresponsible
> for us not to ship it static.
I'm not arguing against static binaries because we've got a bug in the
libs.
- static binaries contain syscalls and therefore make it very difficult,
if not impossible to modify the kernel interfaces.
The SVID btw., explicitly forbids embedding syscalls into ABI compliant
binaries. While SVID compliance is not directly mandatory for Linux,
binary compatibility issues with IRIX (or Solaris or ...) etc. might
somewhen enforce modifications to the syscall interface resulting in
broken static binaries.
- many static binaries will use dlopen() to load _shared_ libraries under
certain circumstances. Rpm is just one of them. For example try with
your static rpm
rm -f /lib/*.so*
rpm --install ftp://ftp.whitehat.org/libc-2.0.5-.rpm
Won't work, because you don't have your shared libc and libnss anymore ...
I admit that this type of desaster recovery was nicer with Linux libc
which doesn't do the dlopen() thing.
> Glibc ought to be able to generate static binaries. If it can't, it's broken.
Yes, and the bugs have been fixed.
Ralf
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 15:55 ` Erik Troan
1997-11-24 17:16 ` ralf
@ 1997-11-24 21:35 ` David S. Miller
1997-11-24 21:58 ` Erik Troan
1 sibling, 1 reply; 11+ messages in thread
From: David S. Miller @ 1997-11-24 21:35 UTC (permalink / raw
To: ewt; +Cc: linux-mips, alan, linux
Date: Mon, 24 Nov 1997 10:55:25 -0500 (EST)
From: Erik Troan <ewt@redhat.com>
A static RPM has saved my ass *many* times, and it would
irresponsible for us not to ship it static. Glibc ought to be able
to generate static binaries. If it can't, it's broken.
That was a nice thing back in the pre-glibc days. But since so many
static binaries (including RPM) pull in the NIS stuff dynamically via
dlopen() due to how GLIBC works, your ass will no longer get saved the
way it used to.
Later,
David S. Miller
davem@dm.cobaltmicro.com
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 21:35 ` David S. Miller
@ 1997-11-24 21:58 ` Erik Troan
1997-11-24 23:22 ` ralf
0 siblings, 1 reply; 11+ messages in thread
From: Erik Troan @ 1997-11-24 21:58 UTC (permalink / raw
To: David S. Miller; +Cc: linux-mips, alan, linux
On Mon, 24 Nov 1997, David S. Miller wrote:
> That was a nice thing back in the pre-glibc days. But since so many
> static binaries (including RPM) pull in the NIS stuff dynamically via
> dlopen() due to how GLIBC works, your ass will no longer get saved the
> way it used to.
At least you can install the packages though. At the very worst,
getpwnam() will fail and RPM will use root.root for all of the files,
but at least they will be there. That's surely better then not being
able to install packages at all.
Erik
-------------------------------------------------------------------------------
| "For the next two hours, VH1 will be filled with foul-mouthed, |
| crossdressing Australians. Viewer discretion is advised." |
| |
| Erik Troan = ewt@redhat.com = ewt@sunsite.unc.edu |
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: Libc in CVS
1997-11-24 21:58 ` Erik Troan
@ 1997-11-24 23:22 ` ralf
0 siblings, 0 replies; 11+ messages in thread
From: ralf @ 1997-11-24 23:22 UTC (permalink / raw
To: linux-mips; +Cc: David S. Miller, alan, linux
On Mon, Nov 24, 1997 at 04:58:22PM -0500, Erik Troan wrote:
> > That was a nice thing back in the pre-glibc days. But since so many
> > static binaries (including RPM) pull in the NIS stuff dynamically via
> > dlopen() due to how GLIBC works, your ass will no longer get saved the
> > way it used to.
>
> At least you can install the packages though. At the very worst,
> getpwnam() will fail and RPM will use root.root for all of the files,
> but at least they will be there. That's surely better then not being
> able to install packages at all.
I just checked mount(8), it's a shared binary. I guess only the fewest
Redhat installations it is possible to access the packages without using
nss services or mount.
Aside, if everything gets installed as root.root in absence of the nss
functionality this open a theoretical security problem for some packages.
Ralf
^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~1997-11-24 23:30 UTC | newest]
Thread overview: 11+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
1997-11-24 4:41 Libc in CVS ralf
1997-11-24 9:46 ` Alan Cox
1997-11-24 11:05 ` ralf
1997-11-24 11:30 ` Alan Cox
1997-11-24 13:18 ` ralf
1997-11-24 14:09 ` Alan Cox
1997-11-24 15:55 ` Erik Troan
1997-11-24 17:16 ` ralf
1997-11-24 21:35 ` David S. Miller
1997-11-24 21:58 ` Erik Troan
1997-11-24 23:22 ` ralf
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.