From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/3070#note_1871493816
Please bring the pieces of the last commit ("rh_flags: Fix of build errors and
warnings") to the individual commits. While some projects do such cumulative
fixes at the end of a MR, we
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2403#note_1438205693
The only non-rt core networking related change seems to be that
CONFIG_INET_TABLE_PERTURB_ORDER is being set to 16. I don't know what caused
this to be added but this is the default value
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2490#note_1432295679
This slipped my radar. Thanks for the ping, Patrick!
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an email to
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2490#note_1409821891
That looks much better. And yes, I think it's something for upstream.
IMHO Documentation/kbuild/kbuild.rst documents HOSTCFLAGS as something that
can be overriden from the make command
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2490#note_1409378737
I wonder whether it's not a problem in the upstream commits playing weird
games with the makefile variables. For example, this sentence in upstream
commit 13e07691a16f
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2286#note_1273813728
Scratch that, this is not about the bpf object files but about the (host
native) binary running them. Sorry for my misunderstanding.
___
kernel
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2286#note_1273804191
This is a good workaround. I wonder though whether it should not be more
generic. We already have custom code to install the bpf selftest files (since
the upstream `make install` does not
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2041#note_1124642864
I find this useful when working on a bug report. By getting /proc/config.gz,
I'm 100% sure I've got the correct config. No space for mishandling a script
fetching from /boot, no space for
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2041#note_1124638254
I'd argue that if you're running 1000 VMs, 50 MB hardly matters. The kernels
are what, 30 GB at that point? 0.1% is still 0.1%.
___
kernel
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2021#note_1114008796
Thanks!
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an email to kernel-le...@lists.fedoraproject.org
Fedora
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/2021#note_1110837378
Good idea.
The name is a bit misleading, since the option applies to all selftests we
build, not just the bpf ones. I mentioned bpf in the description, as the
problem with failing build
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1858#note_980591912
Please put this to common/generic and delete ark/generic/CONFIG_DEBUG_NET.
___
kernel mailing list -- kernel@lists.fedoraproject.org
To
From: Jiri Benc
Enable net reference count trackers in all debug kernels
A leaked netdev or netns entry is a common source of bugs, which are
hard to debug. The refcount tracker for them is already enabled for
Fedora debug kernels. Enable it also for ARK debug kernels.
Signed-off-by: Jiri Benc
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1619#note_846679718
This leaves the semicolon here in the GENKSYMS pass. I don't think we want
that; the kABI checksum would be generated from:
```
union {
unsigned long xyz[_size];
;
}
```
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1619#note_846001310
Sabrina, I suggest enabling CONFIG_RH_KABI_SIZE_ALIGN_CHECKS only in
redhat/configs/ark and disabling them in redhat/configs/fedora. That config
option is not useful for Fedora.
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1422#note_698589843
Acked-by: Jiri Benc
(via approve button)
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an email to
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1422#note_694688949
Acked-by: Jiri Benc
(via approve button)
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an email to
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1422#note_694678839
The other file is just a list of modules that should not be put into the main
rpm and it's the correct thing to do.
___
kernel mailing list --
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1226#note_692091567
Acked-by: Jiri Benc
(via approve button)
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an email to
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1325#note_659345171
Acked-by: Jiri Benc
(via approve button)
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an email to
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1325#note_659253997
@bnemeth, maybe leave this enabled in Fedora and disable for ARK only?
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1311#note_659027149
Acked-by: Jiri Benc
(via approve button)
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an email to
From: Jiri Benc
kernel.spec: fix build of samples/bpf
Due to some change of the kernel makefiles, "samples/bpf" does not work
anymore as a make target. Use M= instead.
Signed-off-by: Jiri Benc
diff --git a/redhat/kernel.spec.template b/redhat/kernel.spec.template
index blahblah..blahblah
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/994#note_561200965
Thanks for putting the explanation in the patch description. I'm okay
with this approach.
___
kernel mailing list --
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/994#note_554516072
Thanks for the link to the upstream discussion. I read the whole thread
and it doesn't seem to me the idea was rejected. Just a different
implementation was requested. We should pursue the
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1029#note_552917820
From a quick look, this seems reasonable even for RHEL. However, it
needs a verification. Which is easy, you take the rpms built before and
after the patch set and compare their content
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/987#note_546473354
Looks good. Thanks!
___
kernel mailing list -- kernel@lists.fedoraproject.org
To unsubscribe send an email to
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/994#note_546471110
That's a good point. I agree that this should be submitted upstream
first.
___
kernel mailing list -- kernel@lists.fedoraproject.org
To
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/987#note_541631004
Don, you're disabling CONFIG_NET_ACT_CTINFO and CONFIG_NET_ACT_MPLS
options everywhere but they should stay enabled in Fedora. Keep them as
'm' in common/generic and disable them in
From: Jiri Benc on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/987#note_540836903
Thanks for letting me know! After some digging, I found the reason. In
RHEL 8, CONFIG_NET_ACT_CTINFO and CONFIG_NET_ACT_MPLS are disabled and
those modules do not appear in
30 matches
Mail list logo
