Override NGROUPS_MAX

[Sébastien Morand]

Hi,

I'm still giving a try recompiling with NGROUPS_MAX et KI_NGROUPS
modified on a test computer ;-). Then I'll install samba from ports.

Test are concluant until there. I change NGROUPS_MAX and KI_NGROUPS to
1024 and compile all user land and packages required. Can connect to
samba with 18 groups (not yet tried more).

Let you know further tests.
Regards,
Sebastien

[Theo de Raadt]

Post by Sébastien Morand

I'm still giving a try recompiling with NGROUPS_MAX et KI_NGROUPS
modified on a test computer ;-). Then I'll install samba from ports.

Test are concluant until there. I change NGROUPS_MAX and KI_NGROUPS to
1024 and compile all user land and packages required. Can connect to
samba with 18 groups (not yet tried more).

Congratulations. You are no longer running OpenBSD. Your system
has a significant incompatibility, and now we cannot accept any
bug reports from you anymore. Any bug you hit might be due to that
change you made. You own the change.

[Artturi Alm]

Post by Theo de Raadt
Congratulations. You are no longer running OpenBSD. Your system
has a significant incompatibility, and now we cannot accept any
bug reports from you anymore. Any bug you hit might be due to that
change you made. You own the change.

How about config(8)?
"Use of an alternative kernel configuration is not recommended."
Rather mildly written, if it is to be understood like that.

-Artturi

(i'm not good w/man-pages, so..:)

diff --git a/usr.sbin/config/main.c b/usr.sbin/config/main.c
index 33d82e1..71c9fc9 100644
--- a/usr.sbin/config/main.c
+++ b/usr.sbin/config/main.c
@@ -110,6 +110,9 @@ main(int argc, char *argv[])
if (pledge("stdio rpath wpath cpath flock", NULL) == -1)
err(1, "pledge");

+ (void)fprintf(stderr, "Any bug you hit might be due to that change"
+ " you made.\n\tYou own the change.\n");
+
pflag = eflag = uflag = fflag = 0;
while ((ch = getopt(argc, argv, "egpfb:s:o:u")) != -1) {
switch (ch) {

[Artturi Alm]

The person didn't make a simple config change they made
a change to the actual kernal code.
Huge difference.

thank you for your reply, but that's no answer to my question, and
you have no sense for my lack of humour it seems.
the Huge difference is between kernel and kernal code, now have you used
config to make changes to usar code? you should test the diff i sent, if so.

-Artturi

Post by Artturi Alm

Post by Theo de Raadt
Congratulations. You are no longer running OpenBSD. Your system
has a significant incompatibility, and now we cannot accept any
bug reports from you anymore. Any bug you hit might be due to that
change you made. You own the change.

How about config(8)?
"Use of an alternative kernel configuration is not recommended."
Rather mildly written, if it is to be understood like that.
-Artturi
(i'm not good w/man-pages, so..:)
diff --git a/usr.sbin/config/main.c b/usr.sbin/config/main.c
index 33d82e1..71c9fc9 100644
--- a/usr.sbin/config/main.c
+++ b/usr.sbin/config/main.c
@@ -110,6 +110,9 @@ main(int argc, char *argv[])
if (pledge("stdio rpath wpath cpath flock", NULL) == -1)
err(1, "pledge");
+ (void)fprintf(stderr, "Any bug you hit might be due to that
change"
+ " you made.\n\tYou own the change.\n");
+
pflag = eflag = uflag = fflag = 0;
while ((ch = getopt(argc, argv, "egpfb:s:o:u")) != -1) {
switch (ch) {

[Philip Guenther]

Post by Artturi Alm

The person didn't make a simple config change they made
a change to the actual kernal code.
Huge difference.

thank you for your reply, but that's no answer to my question, and
you have no sense for my lack of humour it seems.
the Huge difference is between kernel and kernal code, now have you used
config to make changes to usar code? you should test the diff i sent, if

so.

There are a spectrum of possible changes that someone may make from
inconsequential to forked project. Your diff made running config(8)
emit a warning even if the user had made *no* changes to a provided
GENERIC config, effectively claiming that we deny support if they
don't ship a kernel theo builds, but that is *not* the case. Indeed,
we provide errata including kernel patches for the last couple
releases. We are not that dogmatic.

What's we're saying is that while many changes have no effect on
support or will be gladly merged into the base in some form, other
changes go beyond what the project can or will support. There is no
easily stated rule for this and the effective rule changes in various
ways as our understanding changes and as the world changes. For
example, a year ago changes that would fail on static-lib-only archs
would not be accepted; now they are. Theo is saying that the world
and the project will need to change in many, many ways before
Sébastien's diff would be supported and we don't see that happening in
the foreseeable future.


Philip Guenther

[Theo de Raadt]

Post by Philip Guenther

Post by Artturi Alm

The person didn't make a simple config change they made
a change to the actual kernal code.
Huge difference.

thank you for your reply, but that's no answer to my question, and
you have no sense for my lack of humour it seems.
the Huge difference is between kernel and kernal code, now have you used
config to make changes to usar code? you should test the diff i sent, if

so.
There are a spectrum of possible changes that someone may make from
inconsequential to forked project. Your diff made running config(8)
emit a warning even if the user had made *no* changes to a provided
GENERIC config, effectively claiming that we deny support if they
don't ship a kernel theo builds, but that is *not* the case. Indeed,
we provide errata including kernel patches for the last couple
releases. We are not that dogmatic.
What's we're saying is that while many changes have no effect on
support or will be gladly merged into the base in some form, other
changes go beyond what the project can or will support. There is no
easily stated rule for this and the effective rule changes in various
ways as our understanding changes and as the world changes. For
example, a year ago changes that would fail on static-lib-only archs
would not be accepted; now they are. Theo is saying that the world
and the project will need to change in many, many ways before
Sébastien's diff would be supported and we don't see that happening in
the foreseeable future.

I'll add something more.

We already suffer from low quality in many bug reports. People very
often forget to mention they have tweaks of their own. Resizing such
an array could lead to many unknown consequences, which we don't want
to think though. Our function in this ecosystem definately does not
include dealing with other people's bullshit decisions...

[Sébastien Morand]

Hi,

Post by Theo de Raadt
Congratulations. You are no longer running OpenBSD. Your system
has a significant incompatibility, and now we cannot accept any
bug reports from you anymore. Any bug you hit might be due to that
change you made. You own the change.

This is true, thanks for the reminder although I was aware of this matter.

I'm not deploying this change widely in my company, all my others
OpenBSD servers still use unmodified 5.9 kernel and packages.

I won't report any bug from this machine nor from related computer
(particularly using NFS protocol even with unmodified kernel and
packages).

Regards,
Sébastien